Magic Squares in The Tenth Century: Jacques Sesiano

Sources and Studies in the History of Mathematics
and Physical Sciences
Jacques Sesiano
Magic Squares
in the Tenth
Century
Two Arabic Treatises by Ant. ākī
and Būzjānī
Sources and Studies in the History of Mathematics
and Physical Sciences
Series editor
Jed Z. Buchwald
Associate editor
J.L. Berggren
J. Lützen
J. Renn
Advisory Board
C. Fraser
T. Sauer
A. Shapiro
Sources and Studies in the History of Mathematics and Physical Sciences
was inaugurated as two series in 1975 with the publication in Studies of Otto
Neugebauer’s seminal three-volume History of Ancient Mathematical Astronomy,
which remains the central history of the subject. This publication was followed the
next year in Sources by Gerald Toomer’s transcription, translation (from the Arabic),
and commentary of Diocles on Burning Mirrors. The two series were eventually
amalgamated under a single editorial board led originally by Martin Klein (d. 2009)
and Gerald Toomer, respectively two of the foremost historians of modern and
ancient physical science. The goal of the joint series, as of its two predecessors, is
to publish probing histories and thorough editions of technical developments in
mathematics and physics, broadly construed. Its scope covers all relevant work
from pre-classical antiquity through the last century, ranging from Babylonian
mathematics to the scientific correspondence of H. A. Lorentz. Books in this series
will interest scholars in the history of mathematics and physics, mathematicians,
physicists, engineers, and anyone who seeks to understand the historical
underpinnings of the modern physical sciences.
More information about this series at http://www.springer.com/series/4142

Jacques Sesiano
Magic Squares in the Tenth

Century
Two Arabic Treatises by Anṭākī and Būzjānī
Jacques Sesiano
Département de mathématiques
Station 8
Ecole polytechnique fédérale
CH-1015 Lausanne
seziano@bk.ru
ISSN 2196-8810 ISSN 2196-8829 (electronic)

Sources and Studies in the History of Mathematics and Physical Sciences
ISBN 978-3-319-52113-8 ISBN 978-3-319-52114-5 (eBook)
DOI 10.1007/978-3-319-52114-5
Library of Congress Control Number: 2017935465
© Springer International Publishing AG 2017

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the
material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the
editors give a warranty, express or implied, with respect to the material contained herein or for any errors or
omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
Printed on acid-free paper
This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Table of contents
Part I: Tenth-century construction methods
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
§1. General notions on magic squares . . . . . . . . . . . . . . . . . . . . . . . . . 3
§2. The two texts from the tenth century . . . . . . . . . . . . . . . . . . . . . 9
A. Text A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1. Author and manuscript . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2. The Introduction to Arithmetic by Nicomachos . . . . . . . . 11
3. Contents of text A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
B. Text B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Chapter I. Ordinary magic squares . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

§1. Construction of odd-order squares . . . . . . . . . . . . . . . . . . . . . . . 20
A. Particular case of order 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1. Uniqueness of the square of order 3 . . . . . . . . . . . . . . . . . . 20
2. Construction of the square of order 3 ‘by displacement’ . 20
3. A construction ‘without displacement’ . . . . . . . . . . . . . . . 21
B. Square of order 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
C. Later developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
§2. Construction of even-order squares . . . . . . . . . . . . . . . . . . . . . . . 26
A. Square of order 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1. Constructions by displacement . . . . . . . . . . . . . . . . . . . . . . 26
2. A construction without displacement . . . . . . . . . . . . . . . . 27
B. Square of order 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
C. Square of order 8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
D. Allusion to a generalization . . . . . . . . . . . . . . . . . . . . . . . . . . 31
1. Evenly even orders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2. Evenly odd orders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
E. Later developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
1. Equalization rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2. Case of evenly even orders . . . . . . . . . . . . . . . . . . . . . . . . . 37
3. Case of evenly odd orders . . . . . . . . . . . . . . . . . . . . . . . . . . 40
v
vi Table of contents
Chapter II. Bordered magic squares . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

§1. Construction of odd-order bordered squares . . . . . . . . . . . . . . . 43
A. Empirical discovery of a method . . . . . . . . . . . . . . . . . . . . . . 43
1. Square of order 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2. Higher odd-order squares . . . . . . . . . . . . . . . . . . . . . . . . . . 46
B. Description of the general method . . . . . . . . . . . . . . . . . . . . 48
C. Mathematical basis for this method . . . . . . . . . . . . . . . . . . . 49
§2. Construction of even-order bordered squares . . . . . . . . . . . . . . . 51
A. Construction of bordered squares of evenly even orders . . . 51
1. Description of the method in A and B . . . . . . . . . . . . . . . 51
2. Description of the general method . . . . . . . . . . . . . . . . . . . 53
3. Mathematical basis for this method . . . . . . . . . . . . . . . . . 54
(α) General theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
(β) The tenth-century method . . . . . . . . . . . . . . . . . . . . . . 58
B. Construction of bordered squares of evenly odd orders . . . 59
1. Particular case of order 6 . . . . . . . . . . . . . . . . . . . . . . . . . . 59
2. Description of a general method in A . . . . . . . . . . . . . . . . 61
3. Other description of the general method . . . . . . . . . . . . . 61
4. Mathematical basis for this method . . . . . . . . . . . . . . . . . 62
Chapter III. Separation by parity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

§1. Structure of the rhomb according to B . . . . . . . . . . . . . . . . . . . . 67
A. The rows of the rhomb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
B. Largest square within the rhomb . . . . . . . . . . . . . . . . . . . . . 68
§2. The partially empirical construction of B . . . . . . . . . . . . . . . . . 69
§3. Filling the rhomb according to A . . . . . . . . . . . . . . . . . . . . . . . . 72
§4. Placing the even numbers around the rhomb . . . . . . . . . . . . . . . 74
A. Situation after filling the rhomb . . . . . . . . . . . . . . . . . . . . . . 74
1. Determining the number of remaining empty cells . . . . . 74
2. Determining the sum required . . . . . . . . . . . . . . . . . . . . . . 75
B. Rules for placing the even numbers . . . . . . . . . . . . . . . . . . . . 78
C. Case of the order n = 4t + 1 (with t ≥ 2) . . . . . . . . . . . . . . . 85
1. First border . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
2. Other borders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
3. Recapitulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4. Completing the construction . . . . . . . . . . . . . . . . . . . . . . . 92
D. Case of the order n = 4t + 3 (with t ≥ 1) . . . . . . . . . . . . . . 95
1. First border . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Table of contents vii
2. Other borders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
3. Recapitulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
4. Completing the construction . . . . . . . . . . . . . . . . . . . . . . 101
E. Particular case of order 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
Chapter IV. Composite magic squares . . . . . . . . . . . . . . . . . . . . . . . . . 104

§1. Subsquares having same size and unequal sums . . . . . . . . . . . 104
§2. Subsquares having same size and equal sums . . . . . . . . . . . . . 107
§3. Parts having different sizes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
A. Method of the cross . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
B. Method of the central square . . . . . . . . . . . . . . . . . . . . . . . . 113
Part II: Translation

Editorial procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Text A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Chapter I of Book III . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
Chapter II. Science of the magic square . . . . . . . . . . . . . . . . . . . . . 143
§1. Science of odd-order (squares) . . . . . . . . . . . . . . . . . . . . . . . . 143
§2. Science of even-order (squares) . . . . . . . . . . . . . . . . . . . . . . . 164
Composite magic squares . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
Chapter III. Determining the hidden number . . . . . . . . . . . . . . . . . 180
Text B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
(Introduction) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
(Construction of ordinary magic squares) . . . . . . . . . . . . . . . . . . . . 210
(Construction of bordered magic squares) . . . . . . . . . . . . . . . . . . . 222
(Particular cases of odd-order squares) . . . . . . . . . . . . . . . . . . . . . . 240
(Particular cases of even-order squares) . . . . . . . . . . . . . . . . . . . . . 250
Opuscule on the magic square by al-Kharaqı̄ . . . . . . . . . . . . . . . . . . . . 253
Part III: Arabic texts

Editorial procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
Text A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
Chapter I of Book III . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
Chapter II. Science of the magic square . . . . . . . . . . . . . . . . . . . . . 282
§1. Science of odd-order (squares) . . . . . . . . . . . . . . . . . . . . . . . . 282
§2. Science of even-order (squares) . . . . . . . . . . . . . . . . . . . . . . . 299
Composite magic squares . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
viii Table of contents
Chapter III. Determining the hidden number . . . . . . . . . . . . . . . . . 316

Text B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335
(Introduction) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335
(Construction of ordinary magic squares) . . . . . . . . . . . . . . . . . . . . 337
(Construction of bordered magic squares) . . . . . . . . . . . . . . . . . . . 349
(Particular cases of odd-order squares) . . . . . . . . . . . . . . . . . . . . . . 368
(Particular cases of even-order squares) . . . . . . . . . . . . . . . . . . . . . 378
Opuscule on the magic square by al-Kharaqı̄ . . . . . . . . . . . . . . . . . . . . 383
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389
Part I
Tenth-century construction methods

Introduction
§1. General notions on magic squares
A magic square is a square divided into a square number of cells in
which natural numbers, all different, are arranged in such a way that the
same sum is found in each horizontal row, each vertical row, and each of
the two main diagonals.
A square with n cells on each side, thus n2 cells altogether, is said to
have the order n. The constant sum to be found in each row is called the
magic sum of this square.
Usually what is written in a square of order n, thus with n2 cells, are
the first n2 natural numbers. Since the sum of all these numbers equals
n2 (n2 + 1)
,
2
the sum in each row, thus the magic sum for such a square, will be
n(n2 + 1)
Mn = .
2
I. A square displaying this magic sum in the 2n + 2 aforesaid rows is an

ordinary magic square (Fig. 1). It meets the minimal number of required
conditions, and such a square can be constructed for any given order
n ≥ 3 (a magic square of order 2 is not possible with different numbers).
1 31 22 15 30 12
2 32 14 23 29 11
3 33 24 13 28 10
34 4 16 21 9 27
35 5 17 20 8 26
36 6 18 19 7 25
Fig. 1
But there are magic squares which display further properties.

II. A bordered magic square is one where removal of its successive borders
leaves each time a magic square (Fig. 2). With an odd-order square,
after removing each (odd-order) border in turn, we shall finally reach the
smallest possible square, that of order 3. With an even-order square, that
will be one of order 4 (the border cannot be removed since there is no
3
4 General notions on magic squares
magic square of order 2). For order n ≥ 5, bordered squares are always
possible.
92 17 4 95 8 91 12 87 16 83
99 76 31 22 77 26 73 30 69 2
1 20 64 41 36 63 40 59 81 100
3 19 67 58 47 51 46 34 82 98
96 80 33 52 45 57 48 68 21 5
7 78 35 49 56 44 53 66 23 94
90 27 62 43 54 50 55 39 74 11
13 72 42 60 65 38 61 37 29 88
86 32 70 79 24 75 28 71 25 15
18 84 97 6 93 10 89 14 85 9
Fig. 2
As seen above, the magic sum for a square of order n filled with the
2
n first natural numbers is
n(n2 + 1)
Mn = .
2
2
Clearly, the average sum in each case is n 2+1 . Accordingly, for m cells,
this average sum should be m times that quantity; this will be called the
sum due for m cells. Thus, the mth border (m ≥ 5) of a bordered square
of order n will contain in each row its sum due, namely
m(n2 + 1)
Mn(m) =
2
if the main square is filled with the n2 first natural numbers. The sum
in a row in one border will therefore differ from the next one by n2 + 1.
With this in mind, we clearly see the structure of bordered squares: the
elements of the same border which are opposite (horizontally, vertically,
or diagonally for the corner cells) are what we shall call complements,
that is they add up to n2 + 1 if the main square has the order n. For
example, in each border of the above figure, the sum of opposite elements
is 101.
III. In both the above cases the main diagonals must always contain the
magic sum. But there are magic squares in which broken diagonals also
make the magic sum.1 Such squares are called pandiagonal (Fig. 3). They
1
A broken diagonal is a pair of diagonal rows, on either side of and parallel to a
main diagonal, comprising n cells altogether.
Introduction 5
are possible for any odd order from 5 on, and for any even order divisible
by 4 , from 4 on. Thus, in this figure, each of the two broken diagonals
starting with 22 —namely 22, 2, 43, . . . , 33, 55 and 22, 38, 41, . . . , 19—
makes the sum 260.
11 22 47 50 9 24 45 52
38 59 2 31 40 57 4 29
18 15 54 43 20 13 56 41
63 34 27 6 61 36 25 8
3 30 39 58 1 32 37 60
46 51 10 23 48 49 12 21
26 7 62 35 28 5 64 33
55 42 19 14 53 44 17 16
Fig. 3
In such squares any lateral row can be moved to the other side: the
main diagonals of the new square will be magic since they were already
magic as broken diagonals. By repeating such vertical and/or horizontal
moves we are able to place any element in any given cell of the square.
IV. Even more particular are the composite squares: the main square
comprises subsquares which, taken individually, are also magic (Fig. 4).
The possibility of such an arrangement depends on the divisibility of the
order of the main square.
Fig. 4
By general construction methods we mean ways of directly construct-
ing a square of given order and given type, that is, belonging to one of the
aforementioned types: once the empty square is drawn, a few, easily re-
membered instructions will enable us to place the sequence of consecutive
numbers without any computation or recourse to trial and error.
Now, considering one of the above types of magic square, there ex-
ists no general method applicable to any order. General methods are
applicable at most to one of three categories of order, which are:
6 General notions on magic squares
— The squares of odd orders, also called odd squares, thus with n = 2k+1,
of which the smallest is the square of order 3.
— The squares of evenly even orders, also called evenly even squares,
thus with n = 4k, of which the smallest is the square of order 4.
— The squares of evenly odd orders, also called evenly odd squares, thus
with n = 4k + 2, of which the smallest is the square of order 6.
These methods are, by definition, applicable whatever the size of the
order; but they may require some adapting for squares of lower orders,
like 3 or 4, sometimes also 6 and 8.
Remark. Speaking of different methods suggests that a square of given
order may take different aspects. As a matter of fact, the number of
possible configurations (excluding mere inversions and rotations of the
square) rapidly increases with the size of the order. Whereas there is
just one form of the magic square of order 3, there are already 880 for
the order 4, as discovered in 1693.2
The beginning of the science of magic squares is unknown. No extant
ancient Greek text deals with magic squares, no allusion whatsoever is
made to them in antiquity, and the only Greek text preserved is a Byzan-
tine writing by Manuel Moschopoulos (ca. 1300), who appears to draw
his knowledge from some Arabic or Persian text.3 Furthermore, direct
allusions to Greek studies by Arabic authors seem fanciful. The earliest
magic square appears in China at the beginning of our era, but it is a
square of order 3; higher order squares do not occur there before the 12th
century, and are clearly of Arabic or Persian origin. The same holds for
Indian magic squares.4
In the 7th century the game of chess arrived in Persia from India,
where it had been invented sometime before. It appears to be rapidly con-
nected with the construction of magic squares: the early Arabic treatises
often describe the successive placing of numbers in small-order squares by
means of chess moves.5 General construction methods existed already by
the middle of the 10th century, as attested by our two texts. But they are
limited to bordered and composite squares: ordinary magic squares are
constructed only for small orders (3, 4, 5, 6), and by methods applicable
to one order only. (This may be the reason for the early appearance of
2
By Frénicle de Bessy in his Table generale des quarrez de quatre.
3
Moschopoulos’ text has been edited and translated by Tannery; on its origin, see
our study of it.
4
General survey in our Les carrés magiques, pp. 8–10 (Russian edition, pp. 15–18).
5
For our texts, see A.II.37 and B.4, B.5, B6, B.8vi, B.9iii, B.15i, B.26iii.
Introduction 7
composite squares, as an attempt to reduce the construction of larger,

non-bordered squares to that of smaller-order ones.) But the most elabo-
rate result obtained by that time is the construction of bordered squares
where the odd numbers appear separated from the even ones, the former
being grouped together in a rhomb within the main square. No later
treatise till modern times will ever match such a stage of development.
Separating in a similar way the numbers according to parity will occur in
the 11th century for ordinary magic squares; but there the construction
is much simpler. This at least shows that separation by parity has been
of permanent interest.6
There will be further development in the 11th and early 12th century,
with the discovery of general methods for ordinary magic squares, by far
the simplest way to construct magic squares. Various constructions of
pandiagonal squares also emerge, although what is then seen as charac-
teristic is not the magic property of all diagonals but the possibility of
choosing the place of a given number within the square, a mere conse-
quence of the pandiagonality, as already mentioned (p. 5). Squares with
non-consecutive numbers also begin to appear. The origin of this has
to do with the association of Arabic letters with numerical values —an
adaptation of the Greek numerical system (Fig. 5), also used in early Is-
lamic times, before the adoption of the Indian numerals. Since to each
letter of a word, or to each word of a sentence, was attributed a numerical
quantity, thus to the word or the sentence a certain sum, this word or
sentence could be placed as such in the cells of a certain row of a square.
The task was then to complete the square numerically so that it would
display in each row the sum represented by the word or sentence —a
mathematically interesting problem since this is not always possible.7
Later centuries in Islamic countries saw mainly improvements and re-
finements of existing methods. Meanwhile, however, popular use of magic
squares as talismans grew apace. Authors of many shorter texts are just
not interested in the mathematical properties or, even, the construction
of magic squares, their purpose being purely practical if not commercial.
They present a few magic squares, most commonly squares of the orders
3 to 9 associated with the seven then known planets (including Moon
and Sun) of which they embodied the respective, good or evil, qualities
—fully described in these texts. The reader is taught to draw one of these
squares on some appropriate material (the nature of which can increase
6
On these latter constructions, see Les carrés magiques, pp. 33–35 (Russian edition,
pp. 42–44).
7
Les carrés magiques, pp. 182–220 (Russian edition, pp. 195–233).
8 The two texts from the tenth century
its efficacy), if possible at some astrologically predetermined time (which

also increases the effect); he will then put it in or near something be-
longing to the chosen person, beneficiary or victim.8 It was the arrival of
such texts in late mediaeval Europe which first aroused interest in, and
study of, such squares there and also gave rise to use of the term ‘magic’
—formerly also ‘planetary’, which we find still employed by Fermat.9
α β γ δ ε ϛ ζ η θ

1 2 3 4 5 6 7 8 9
ι κ λ μ ν ξ ο π ϟ

10 20 30 40 50 60 70 80 90
ρ σ τ υ φ χ ψ ω ϡ

100 200 300 400 500 600 700 800 900
α

1000
Fig. 5
We thus see that ‘magic’ applications do not seem to have played a
significant rôle at the outset. The magic square (wafq al-a‘dād: ‘har-
monious disposition of numbers’, as it was then called) was originally
studied as a branch of number theory, just as were the perfect and ami-
cable numbers so highly esteemed by Pythagoreans and their followers in
late antiquity. In short, such research was at the time considered as being
perfectly serious and had not yet been tainted with the sad reputation
later associated with their practical use. We are nevertheless left with
the question of when actual methods of construction first appeared: the
two 10th-century texts to be studied here, paradoxically, far from provid-
ing an answer, merely complicate matters: at the same time, we witness,
reproduced by a very competent author, the first steps in the study of magic
squares (this is our text B) and, reported by a mathematician of unequal
value, some general methods but also that highly elaborate method never to
be surpassed later (this is our text A). In short, the tenth century, for the
history of magic squares, gives the impression of being both a beginning and
an end.
8
Typical examples in our Magic squares for daily life.
9
Varia opera mathematica, p. 176; Œuvres complètes, II, p. 194.
Introduction 9
§2. The two texts from the tenth century

These two texts are, as just hinted at, very different. One, B, is
very didactic —indeed, its author explains very well how to construct
the magic squares he presents, and mostly justifies his manner of pro-
ceeding. It is perfectly adequate for a beginner. Its only weakness is
that the knowledge it presents is embryonic: it teaches just one general
construction method, namely for odd-order bordered squares. The other
text, A, is of little use to the reader. For the bordered magic squares
of the first orders, it merely indicates the place of each number, without
justification and rarely hinting at general procedures; still, the attentive
reader will deduce them and thus be able to construct bordered squares
of any possible order. On the other hand, A is unique in that it attests
to the surprising results already obtained in the science of magic squares
by the tenth century, thereby suggesting a much earlier, possibly Greek
time for the first discoveries. These two treatises thus complete each other;
they are certainly the most important testimonies on the early history of that
science, when it was still in the hands of mathematicians and not yet reduced
to popular interest or superstition.
A. Text A
1. Author and manuscript
The University of Ankara, or, more precisely its Faculty of Language,
History and Geography (Dil ve Tarih-Coǧrafya Fakültesi), was donated
by a well-read book collector, Ismail Saib Sencer (d. 1940), more than
ten thousand manuscripts divided into two collections. Among the first,
one finds the manuscript I, 5311, of 84 leaves, written on paper in the
13th century (7th century of the hegira). It contains ten treatises, all
on mathematical subjects (geometry, algebra, magic squares), and copied
by the same hand. The first of these (fol. 1r –36r ), by ‘Alı̄ ibn Ah.mad
al-Ant.ākı̄, is our text A. There are between 18 and 24 lines on a page,
mostly 20 or 21. The copyist sometimes used red ink, namely for some
headings, numerical symbols and framing the figures; he did it later on,
and that is why one correction is in red (l. 863 in our edition of the text)
and some spaces left for numbers were insufficient (ll. 1165, 1221, 1317).
The writing (naskhı̄) is quite legible, except in some worm-eaten places.
The copy is on the whole in good condition, and the copyist is responsible
for only a few textual omissions which he partly or completely remedied
upon rereading. We may, though, remark that there is some disorder in
the Arabic text on the leaves 18 and 19, and that only half of fol. 18r
contains text. It appears that there has been an error in an earlier copy,
just reproduced by the present copyist, together with, in red ink, the
explanation (fol. 18v , ll. 830–831 in our Arabic text): This part is left
blank for the figure, where K is (written) in red on the left-hand page;
now the place referred to is not found in the present copy.
In this copy at least, the text did not arouse much interest: the mar-
gins contain no readers’ observations, merely, in the copyist’s own hand,
words or sentences originally omitted. The only extraneous annotations
are in a few figures of squares, and are of limited interest, such as cal-
culating the number of cells in a 12 × 12 square. The ancestors of our
manuscript had more success: notes, once written in the margin by read-
ers, have then been incorporated into the text, though mostly in an in-
appropriate place. But such interventions are mostly devoted to the less
difficult parts; the others prompted no comments.
Such is the form in which text A is preserved. As for its author,
al-Ant.ākı̄, we know that he lived in the tenth century, for Ibn al-Nadı̄m
(the writer of a bibliographic work compiled 987/8, known by the name
of Fihrist, and our main source on the authors of the first three centuries
of the hegira) gives the time of his death as 376 of the hegira, thus 987; he
also reports the titles of some of his works, including three on reckoning,
one on the problem of weights,10 one on ‘the cubes’, a commentary on
Euclid,11 and his ‘commentary on the Arithmetic’ (tafsı̄r al-arithmāt.ı̄qı̄),
that is, on the Introduction to Arithmetic by Nicomachos.12 Only this
last work, our A, is known to survive (though incompletely) today.
Opinions about al-Ant.ākı̄’s ability as a mathematician are, let us put
it this way, somewhat divergent. While the biographer Ibn al-Qift.ı̄ (d.
1248/9) praises it highly, the mathematician al-Nasawı̄, in the late 10th
century, finds one of his writings unclear and verbose.13 Our text will
attest to the plausibility of such a discrepancy: it gives the impression
that we have here, on the one hand, a rather weak mathematician (parts I
and III) and, on the other hand (in some of part II), a superior mind who
very concisely, but correctly, explains the difficult construction of special
magic squares (see in particular A.II.12–A.II.35, A.II.44–A.II.54). In
such cases, it may safely be assumed that the text reflects the source
10
Possibly determination of the least number of weights needed to weigh a given
integral weight; see our Récréations mathématiques au Moyen Âge, pp. 39–47.
11
His knowledge of Euclid’s Elements is evident from the first part of the present
text; but his comprehension of Euclid seems rather weak.
12
Fihrist, vol. I, p. 284; Suter, Mathematiker, p. 64; Sezgin, Geschichte, V, p. 310
(and VII, Nachträge, p. 407); Woepcke, Propagation, p. 493 (or Etudes, II, p. 425).
13
A treatise of his on reckoning is confus et d’une longueur excessive (Woepcke,
ibid.).
Introduction 11
from which the author takes his information. This is, for the history
of magic squares, excellent news: the most difficult subjects must be
transmitted by A unchanged, as he found them, and thus reproduce what
was already known much earlier since he is unlikely to have plagiarized
his contemporaries.
The text presented and analyzed here is, as said above, not the whole
of al-Ant.ākı̄’s commentary on Nicomachos’ Introduction to Arithmetic.
For, as indicated by the title, that is only its third book, thus the third
part of it, and also the last since the colophon clearly indicates that
the work ends there. Since A comments on Nicomachos’ Introduction to
Arithmetic, and Thābit ibn Qurra’s Arabic translation is quite faithful to
the Greek original, we shall give a brief outline of it. That will enable us
to note points of convergence or divergence, if any, between the original
text and its commentary.14 The reader will keep in mind that, according
to its Greek meaning, arithmetic has to do with what we could call today
number theory, and not arithmetical calculations.
2. The Introduction to arithmetic by Nicomachos
Nicomachos of Gerasa is a relatively late Greek mathematician (end of
the 1st century). His Introduction to Arithmetic (᾿Αριθμητικὴ Εἰσαγωγή)
was widely known during late antiquity, and some Greek commentaries on
it have survived: those by Iamblichos, Asclepios of Tralles, J. Philoponos,
and Proclos. The Introduction had a notable influence later as well. It
was first translated into Syriac, then, in the 9th century, into Arabic by
Thābit ibn Qurra. Its rôle in mediaeval Europe is important as well.
There it was, until the 12th century, the only surviving work of ancient
Greek mathematics, through an adaptation by Boëtius (who, in the early
6th century partly summarized the original text and added remarks).15
In short, by means of Nicomachos’ book, ancient Greek, Byzantine, Ara-
bic and mediaeval European readers could learn some fundamentals of
number theory.
This notoriety does not go, though, hand in hand with the mathe-
matical level. It is in no way comparable to that of classical authors like
Euclid (around −300) and Archimedes (around −250) —or also Diophan-
tos (who, unlike the two others, lived after Nicomachos). With Nicoma-
chos, mathematics seems to take a step backwards: an opinion becomes
14
We adopt the division of the text used by the modern editor of the Greek text,
Hoche; the editor of the Arabic translation, Kutsch, refers page by page to Hoche’s
edition.
15
A reported earlier Latin translation by Apuleius is not preserved.
a theorem and a test serves as a demonstration. A typical example is the

subject of perfect numbers, which are numbers equal to the sum of their
divisors (excluding the number itself). Euclid had demonstrated that, if
we take consecutive powers of 2 from 1 on, thus 1, 2, 22 , . . . , 2n−1 , and
their sum, namely 2n −1, happens to be a prime number, then its product
with the last term added, thus 2n−1 (2n − 1), will be a perfect number.
We are informed through Nicomachos that the Greeks knew (at least) the
first four ones: 6 (= 1 + 2 + 3), 28 (= 1 + 2 + 4 + 7 + 14), 496, 8128. Now,
observing that each of them occurs between two consecutive powers of
10, and that they end alternately in 6 and 8, Nicomachos asserted that
all perfect numbers must belong to one, and only one, of the successive
decimal orders and must obey the same alternance for the final digit. Of
all this, it is only true that they will end (but not alternately) by 6 and 8;
as to decimal scale —which has no other origin than our ten fingers— it
cannot have any relation to a property inherent in number such as divis-
ibility. In short, Nicomachos’ Introduction though undoubtedly of use to
anyone wishing to acquire elementary notions of number theory, will also
delude others who expect thereby to be trained in mathematical think-
ing. It embodies perfectly what would be called today ‘democratization
of knowledge’.
Book I begins with some philosophical thoughts. It proceeds with the
classification of natural numbers according to kinds of parity. Among
odd numbers, there are the prime and incomposite (our ‘primes’) and
the secondary and composite (our ‘composite’); the numbers secondary
and composite in themselves but prime and incomposite with others are
our ‘composite and relatively prime’. The even numbers may be evenly
even, evenly odd, oddly even, according to whether they have the form
2n , 2(2k +1), 2m (2k +1) (with m ≥ 2). More practical, perhaps also more
useful since becoming thus widespread, are the description of the sieve of
Eratosthenes (around −250) for finding the sequence of consecutive prime
numbers —of which Euclid had already demonstrated the infinite quan-
tity in his Elements (proposition IX, 20)— and how to find the greatest
common divisor of two numbers (Elements, VII, 1), if any —it will be 1
if the two numbers are relatively prime. Then it defines abundant, defi-
cient and perfect numbers, the sums of the divisors of which (excluding
the number itself) are, respectively, greater than, less than, equal to, the
number considered. Book I ends with a long section, somewhat tedious
and irrelevant to the present study, on the different kinds of numerical
ratios.
Book II discusses in particular geometric progressions. One then finds
Introduction 13
the theory of (so-called) figured numbers, first plane then solid (a natural
number is represented, according to the quantity of its units, by dots
arranged so as to represent a regular geometric figure), and the properties
inferred thereby. It is also observed that, considering the sequence of
consecutive odd numbers starting with 1, the first, the sum of the next
two, the sum of the next three, and so on, give the sequence of consecutive
integral cubes. Book II ends with an enumeration of the different kinds
of mean proportionals —another somewhat tedious conclusion.
3. Contents of text A
Book III of al-Ant.ākı̄’s Commentary consists of three distinct parts
without any connection between them.
Part A.I. It consists of an enumeration of definitions, statements of propo-
sitions and identities. Definitions and propositions are chiefly taken from
Euclid’s Elements, mostly from the so-called arithmetical Books since
they have more to do with number theory than geometry:
— from Book VII are taken definitions, namely 8 (≈ A.I.6), 9 (≈ A.I.5 ),
11 (≈ A.I.7 ), 12 (A.I.1), 15 (A.I.3), 16 (A.I.31), 18 (A.I.90); and
propositions, namely 2 corollary (A.I.19), 4 (A.I.21), 5 (A.I.22), 6
(A.I.23), 7 (≈ A.I.26), 8 (≈ A.I.27), 9 (A.I.24), 10 (A.I.25), 11 (A.I.28),
14 (A.I.29), 17-18 (A.I.32), 23 (A.I.33), 24 (A.I.34), 25 (A.I.35), 27
(≈ A.I.36), 29 (A.I.38), 30 (A.I.39), 32 (A.I.37), 34 (≈ A.I.40 & 48),
35 (A.I.50), 36 (A.I.49), 37 (A.I.51);
— from Book VIII are taken propositions 1 (A.I.52), 2 cor. (A.I.53), 3
(A.I.54), 5 (A.I.55), 7 (A.I.56), 8 (A.I.57), 9 (A.I.58), 11 (A.I.59),
12 (A.I.61), 13 (A.I.63), 14 (A.I.64), 15 (A.I.66), 16 (A.I.65), 17
(A.I.67), 18 (A.I.68), 19 (A.I.69), 20 (A.I.70), 21 (A.I.71), 22 (A.I.72),
23 (A.I.73), 24 (A.I.75), 25 (A.I.76), 26 (A.I.77), 27 (A.I.78);
— finally, from Book IX, propositions 1 (A.I.81), 2 (A.I.82), 4–5 (A.I.86),
8 (A.I.91), 9 (A.I.92), 10 (A.I.93), 11 (A.I.41), 12 (A.I.42), 13 (A.I.43),
15 (A.I.45), 16 (A.I.46), 17 (A.I.47), 20 (A.I.44), 24 (A.I.8), 25 (A.I.9),
26 (A.I.10), 27 (A.I.11), 28 (A.I.5), 29 (A.I.7), 30 (A.I.13), 33 (A.I.14),
35 (A.I.95), 36 (A.I.94).
We also find some propositions which are not in the original version
of the Elements but occur in Greek commentaries (available in Arabic)
on it: A.I.12, A.I.74, A.I.80, A.I.83–A.I.85.
Also found, as A.I.124–A.I.133, are the first ten propositions of Book

II of the Elements, which, though demonstrated geometrically there, in
fact correspond to algebraic identities.16

Apart from these assertions, there is a set of arithmetical identities
(A.I.96–A.I.123), some of which are taken from the algebraic books by
al-Khwārizmı̄, c. 820 (A.I.96, A.I.105) or Abū Kāmil, c. 880 (A.I.105–
A.I.108, A.I.111, A.I.112, A.I.115–A.I.119, A.I.123). When expressed in
symbols, they are often banal —though less so for the mediaeval reader
faced with them in their verbal form.
This first part sometimes casts doubts on al-Ant.ākı̄’s abilities. As
pointed out in some of our notes, certain formulations are strange or non-
sensical (A.I.18, A.I.20, A.I.26, A.I.115, A.I.119), incomplete (A.I.43,
A.I.92, A.I.94) or imprecise (A.I.41, A.I.48, A.I.60, A.I.109, A.I.128,
A.I.132); numerical examples are not well chosen (A.I.29, A.I.42, A.I.95);
definitions of terms do not appear together with the first use of these terms
(A.I.30, A.I.55) or are omitted (A.I.33, A.I.55, A.I.68, A.I.69); a demon-
stration is wrong (A.I.44), as well as two assertions (A.I.87, A.I.88). Fi-
nally, one identity (A.I.117) has been taken without much thought since
one condition in Abū Kāmil’s problem has been repeated despite being
wholly irrelevant to the identity. (We find a similar case in A.I.134.)
Since the items in A.I are, with rare exceptions, merely stated, that
did not leave much room for the imagination of early readers; thus, the
manuscript reproduces only a few interpolations (A.I.17, A.I.20, A.I.35–
A.I.36, A.I.111, A.I.121, A.I.132).
Part A.II. By far more interesting for us is the second part, on the con-
struction of magic squares. Its reading is difficult for two reasons: one of
its subjects (the construction of bordered squares with separation by par-
ity) is too advanced for the reader, and not only the average one; the other
constructions of magic squares may be understood, but are rendered un-
necessarily tedious because of their presentation. A 12th-century reader
who had read several treatises on magic squares expresses his opinion
as follows: As for other (authors), among whom al-Ant.ākı̄, I found that
they (merely) said to put such and such a number in such and such a cell
without explaining the reason for that, although this subject is difficult
to understand and requires to be memorized.17 Reading al-Ant.ākı̄’s con-
struction of usual bordered squares (without separation) fully confirms
this reproach. It goes so far that not only does he list all places of lesser
numbers but also those of their complements, which is superfluous since
16
Their algebraic versions are commonly found in mediaeval textbooks; see note 193,
p. 140.
17
See our Une compilation arabe, p. 163, lines 239–241 of the Arabic text.
Introduction 15
they are anyway placed opposite (see above, p. 4).

Part A.III. The last subject is that of hidden numbers, that is, of guessing
a number thought of by a person. The asker has him perform various
computations on that number; from the result, communicated by the
person, he will be able to find out what number the person had thought
of. Of course, there must be several kinds of operation to prevent the
person from discovering the link between the instructions and the final
result. This part was by far the most successful, as may be inferred
from early readers’ traces, which often remedy the imprecision of the
account. It is indeed not as tedious as the first and more accessible than
the second; and its recreational and social purpose would naturally make
it more appealing.
Comparing now the content of Nicomachos’ work with that of al-
Ant.ākı̄’s Commentary will lead to a surprising discovery, namely that
there is no connection between the two works —at least not in Book
III of al-Ant.ākı̄’s Commentary since that is all we know. A.I, with its
theorems and identities, recalls some other Arabic texts, which begin
with such lists;18 but then Euclid and not Nicomachos would rightly be
indicated as the source for the propositions. As for A.III, we know that
such problems existed in antiquity —indeed, they left traces in an author
of the early Middle Ages, namely Bede (c. 700)19 — and sections of our
text are sometimes found word for word in a writing a century older,
the Treatise on the determination of hidden numbers by the philosopher
and mathematician al-Kindı̄ (d. around 870).20 But such problems have
nothing to do with Nicomachos’ Introduction. A.II, on magic squares,
is also completely foreign to the work by Nicomachos. Finally, we guess
from an allusion to the previous part of the work (the only one, in A.I.94)
that the subject of perfect numbers was discussed there, which is indeed a
subject found in Nicomachos —but also, and more scientifically, in Euclid
(see above, p. 12).
From this we learn two things. The first is that al-Ant.ākı̄ associates with
Greek antiquity —through Nicomachos— three subjects of which two have
attested links with Greek mathematics and the third none, as far as we know.
The second is that al-Ant.ākı̄ takes the essence of his work from other sources,
as, by the way, his often weak mathematical knowledge obliges him to.
18
Characteristically al-Karajı̄’s Badı̄‘, Book I.
19
Tropfke, Geschichte, p. 643; see below, notes to A.III.14 and A.III.15. A Byzantine
example is also appended to Hoche’s edition of Nicomachos (pp. 152–153).
20
Risāla fı̄ istikhrāj al-a‘dād al-mud.marra. Almost identical or very similar are al-
Ant.ākı̄’s sections A.III.2, 3, 4, 6, (7), 15, 16, (17), 23, 25, 28, 33, (34), (35), 36.
Remark. It has recently been suggested21 that, first, there could be some
allusions by Nicomachos and Iamblichos to magical arrangement in
the square of order 3 and, second, that the science of magic squares
may have been studied in Jewish circles. As to the first, although
these Greek allusions are in themselves not convincing, they can no
longer be dismissed a priori. As for the Jewish influence, it is hard
to believe that, considering the strength of tradition and transmis-
sion in the Jewish community, and their fondness for numerology, no
trace would have survived. Indeed, wherever magic squares are found
in later (mediaeval) Jewish writings, they are obviously copied from
earlier Arabic writings.
B. Text B
Abū’l-Wafā’ al-Būzjānı̄ (940–997/8) is one of the most famous math-
ematicians of Islamic civilization. From his work done whilst in Baghdad
(his name indicates his Persian origin) part has survived, sufficiently to
testify to his broad field of mathematical activity: there are writings on
arithmetic, both theoretical (number theory) and practical (reckoning),
geometrical constructions, plane and spherical trigonometry as well as
their applications to astronomy.22 Apart from that there is his treatise
on magic squares, completely devoted to the subject —unlike A’s single
chapter.23 The manuscript preserving it, of the 13th century as it seems,
is in good condition; but the copyist has omitted all figures of magic
squares. Fortunately, since the constructions are accurately described in
the text, most could be reconstituted. This is, without doubt, a charac-
teristic of the text: everything is thoroughly discussed and in a highly
pedagogical way. Indeed, it represents an excellent introduction to the
study of magic squares, and if the present reader wishes to look at the
translations before reading the commentary, he should begin with text B.
Even if general methods are few (presumably because not yet known to
the author), he will be quite satisfied with the examples given. In this
respect, B is superior to text A. But, as already said (p. 9), the two
treatises complete each other: one because it presents fully developed
methods (though either without any justification or only in a extremely
concise manner), the other because it presents methods which are not
general but shown to have a clear and solid basis.
21
N. Vinel, (Iamblichi) In Nicom. Arithmeticam, pp. 23–35.
22
Listed e.g. by Suter, Mathematiker, pp. 71–72; Sezgin, Geschichte, V, pp. 321–325.
23
Manuscript Istanbul, Aya Sofya 4843, fol. 23v –31v , 38r –56v . B’s text was edited
by us in its entirety some twenty years ago. The most relevant parts are reproduced
here.
Introduction 17
Text B is divided into four parts. In the Introduction we find various

preliminaries on the construction of magic squares and the treatment
of the square of order 3. Chapter I is devoted to odd-order squares,
and ends, after the treatment of a few examples, with a general method
inferred from the examples. Chapter II treats the even-order squares; it
takes them order by order, without distinguishing between the two types
of even order, and without proposing any applicable general method.
Finally, Chapter III presents ‘curiosities’, which are in fact extensions
and particular constructions; we are told how to construct composite
squares, and also odd-order bordered squares displaying separation by
parity, but this —unlike in A— leaves too much to trial and error.
Remark. The fragments as we present them do not always follow their
order in the original text. We have arranged them to suit the order
in which subjects are treated in our present study: ordinary magic
squares (B.3–B.11), bordered magic squares (B.12–B.19), bordered
odd-order squares with separation by parity (B.20–B.23), composite
squares (B.24–B.26).
The way B treats his subject enables us to reconstitute the beginning

of the science of magic squares (which in fact concurs with what may
safely be conjectured). The construction of bordered squares begins with
the smallest possible square (of order 3 or 4), and a border is constructed
around this square. Thus, from the square of order 3, we pass to order 5,
then to order 7 and so on, from an odd order to the next odd order. From
this appears quite naturally a way of placing the numbers for any odd
order, border by border. The same is done for the even orders, starting
with the smallest possible square, that of order 4. But a general method
does not appear as before, the successive (even) orders considered being
too few to suggest a treatment for either one of the two types of even
order, although B gives some indications. However, by the 10th century,
general methods for bordered squares had already been discovered, as is
seen from A.
So much for bordered squares. As for ordinary magic squares, their
study is dismissed by A at the outset as being too complicated (A.II.2).
But the form early attempts took appears quite clearly from B. He does
not know any general method, and so constructs the squares individually.
He does it as was customary for ordinary magic squares before the dis-
covery of general methods; that is, starting with the natural square of the
corresponding order —thus the square containing the consecutive num-
bers in natural sequence— and then proceeding with exchanges between
opposite rows in order to obtain in each the required sum. Moreover,

since B does not find a method applicable to different orders, he consid-
ers in more detail each order and tries to find other configurations —all
possible ones, in fact. This will disappear altogether with the discovery,
later on, of general methods: authors will no longer concentrate solely
on one specific order. From both texts it thus appears that there was
no knowledge of general methods for ordinary magic squares in the 10th
century. By the way, had the two authors realized how simple their con-
structions were, they would not have rejected their study as being too
complicated for the beginner (A) or bothered themselves with various
constructions valid for just one order (B).
Text B is not seen to have left much trace later. Indeed, since general
methods for constructing ordinary magic squares were discovered in the
11th century, and very easy to apply, B became rapidly outdated. The
same happened to his bordered squares constructed order by order —
and all the more rapidly in that general methods already existed in his
time (A). Of the particular constructions, at most that of the smallest
bordered square with separation by parity might have inspired a later
reader (al-Kharaqı̄, our text C, appended to the extracts of B). As for
the construction of composite squares, it would be difficult to attribute
their later use to B, for at that time they were apparently already in
widespread use.
Chapter I. Ordinary magic squares
The two tenth-century texts construct magic squares with the num-
bers taken in their natural order beginning with 1. But this is merely a
particular case since the same construction could be applied to any arith-
metical progression with any chosen initial number. This was indeed
known then: when B defines the magic square, he explicitly mentions the
general case and gives examples of such squares (B.1).
The construction of ordinary magic squares, which was to attain a

remarkable development from the 11th century on, is almost insignificant
in the 10th century; as already said, it is on the one hand seen as difficult
for the beginner (A.II.2) and, on the other hand, less esteemed than the
construction of bordered squares: whereas the arrangement in bordered
squares is called ‘regular’, the ordinary magic arrangement is ‘not regular’
according to B (B.2i). To construct such a square, as mentioned by A
when stating its difficulty, we are first to construct the natural square
of the order considered, that is, the square filled with the numbers to
be placed in the magic square, but simply taken in their natural order
(Fig. 6); since it appears that the sum in the rows, either horizontal
or vertical, is in one half less than the magic sum and correspondingly
more in the other, exchanges will be made between opposite rows so as
to eliminate these differences. This is what, according to A, is difficult
for the beginner. The examples given by B, for the orders 3, 4, 5, 6,
8 indeed confirm that: each order has to be treated separately, and the
rules applied to one order will not be valid for another. Once again, we
are far from the easy methods to be developed in the 11th century, with
the natural sequence of numbers being placed directly in the square to be
constructed, thus without resorting to the natural square.
1 2 3 4 5 6
7 8 9 10 11 12
13 14 15 16 17 18
19 20 21 22 23 24
25 26 27 28 29 30
31 32 33 34 35 36
Fig. 6
© Springer International Publishing AG 2017 19
J. Sesiano, Magic Squares in the Tenth Century, Sources and Studies in the History
of Mathematics and Physical Sciences, DOI 10.1007/978-3-319-52114-5_1
20 Construction of odd-order squares
§1. Construction of odd-order squares

A. Particular case of order 3
1. Uniqueness of the square of order 3
Supposing that the square is to be filled with the first nine natural
numbers, we immediately know that the magic sum will be M3 = 15
(above, p. 3; or B.2ii). Considering that r will occupy the central cell,
the sum in each of the four pairs of opposite cells will be M3 − r (Fig.
7). Thus, the sum on the whole square will be on the one hand 3M3 , on
the other 4(M3 − r) + r. Equating these two expressions and putting, as
required in our case, M3 = 15, we find r = 5. This is indeed the result
arrived at in B (B.3i), but rather by postulating that 5 must be in the
middle since such is its place in the natural square.24 Then we are left
with pairs of complements adding up to 10 which are to be placed around
the centre in opposite cells.
∗ † ◦ 2 9 4
r 7 5 3
◦ † ∗ 6 1 8
Fig. 7 Fig. 8
Filling the remainder of the square is not difficult, for it suffices to
consider the place 1 must occupy. It can be only in the middle cell of a
lateral row or in a corner, and (to make the sum 15) 9 will face it. If,
first, it is in a corner cell, we are to find two pairs of numbers making the
complement, thus 14; if, second, it is in a middle cell, only one. Now of
the still available numbers only one pair, namely 6 and 8, makes 14. Thus
1 must be in a middle cell with 6 and 8 as its neighbours, say 6 on the left
(the other choice just inverts the figure). That enables us to determine
the occupants of the other cells (Fig. 8). This is exactly the reasoning
found in B.3ii–iii. Incidentally, this deduction of the construction shows
that the magic square of order 3 with the first nine numbers admits of
only one form, others being just rotations or inversions of it.
2. Construction of the square of order 3 ‘by displacement’
This logical deduction in B is followed by two ways of obtaining the
same magic square but this time by displacement, thus by moving the
numbers in the natural square of order 3 (B.4).
24
With odd orders n, the median number is always in the central cell for bordered
squares, often for ordinary squares.
Ordinary magic squares 21
(a) Consider the natural square of order 3 (Fig. 9). Leaving the occupant
of the central cell where it is, we move those of the border. First, we
move the numbers in the corner cells to their respective knight’s cell by
turning in the same direction, namely two cells towards a corner and one
cell sideways (Fig. 10). Then we take the elements formerly in those cells
and transfer them to the next corner, moving this time in the opposite
direction (Fig. 11). The result is the square already seen, but inverted.
1 2 3 9 4 2
4 5 6 3 5 7
7 8 9 1 8 6
Fig. 9 Fig. 10 Fig. 11

(b) Considering again the natural square, we move each number in the
border to the next cell by turning in the same direction (Fig. 12). Since
then the sum of the occupants of two corners in any row and the occupant
of the middle cell in the opposite row makes 15, we shall just exchange
the occupants of opposite middle cells (Fig. 13).
4 1 2 4 9 2
7 5 3 3 5 7
8 9 6 8 1 6
Fig. 12 Fig. 13
3. A construction ‘without displacement’
B adds to these two ways a method which avoids using the natural
square (B.5): we are told to place 5 in the central cell, 1 in the middle
of a lateral row, 2 in the knight’s cell of 1, and 3 in the queen’s cell of
1 (the next diagonally), namely, B adds, in the knight’s cell of 2 (this
determines the cell). Having placed these four numbers, completing the
square is straightforward.
As to A, he also tells us where to place the numbers; but, unlike B, he
does it for all occupants, without mentioning that the places of some may
determine those of others (A.II.3). Such tedious instructions are common
with him, and he was blamed for that by a later reader (above, p. 14).
B. Square of order 5
The two subsequent examples are historically quite interesting since
(apart from the particular case of order 3) they are the only ones where a
displacement method valid for just one odd order is fully described. They
have a common feature, made evident by the author (B.6i), which will
recur in all early displacement methods for n > 3, odd or even: since the
diagonals of the natural square already make the magic sum, they will
remain unchanged, thus leaving us with equalizing only the vertical and
horizontal rows. (This apparent simplification turned out in fact, as we
shall see below, to be an obstacle to further developments, at least for
odd-order squares.)
(a) Taking the natural square for order 5 (Fig. 14), we first exchange
the elements of the inner square (excepting those of the diagonals) with
those of their bishop’s cells (that is, next but one diagonally) by turning
in the same direction (Fig. 15). Then, considering the pairs of remaining
elements (corners excepted) in opposite lateral rows, we exchange their
places (Fig. 16). See B.6ii.
1 2 3 4 5 1 2 3 12 5 1 23 24 12 5
6 7 8 9 10 18 7 20 9 10 18 7 20 9 11
11 12 13 14 15 11 4 13 22 15 10 4 13 22 16
16 17 18 19 20 16 17 6 19 8 15 17 6 19 8
21 22 23 24 25 21 14 23 24 25 21 14 2 3 25
Fig. 14 Fig. 15 Fig. 16

(b) In the natural square, we have, as just seen, dealt with the pairs 12 &
4, 18 & 6, 14 & 22, 8 & 20, with each element in the bishop’s cell of the
other, and exchanged their places. Here we take 14 & 2, 8 & 20, 12 & 24,
18 & 6 and exchange their places —in other words, we invert the places of
each pair adjoining, on either side, a diagonal, here (say) the descending
one (Fig. 17). We then proceed as before with the lateral opposite rows
(Fig. 18). See B.6iii.
1 14 3 4 5 1 14 22 23 5
18 7 20 9 10 18 7 20 9 11
11 24 13 2 15 10 24 13 2 16
16 17 6 19 8 15 17 6 19 8
21 22 23 12 25 21 3 4 12 25
Fig. 17 Fig. 18
These displacements correspond exactly to what A found fault with in
the methods starting from the natural square (A.II.2): performing rather
complicated exchanges between vertical and horizontal opposite rows (see
note 526).
C. Later developments
B has thus carried out specific moves in the natural square for each of
the first two odd orders. But neither his moves for the square of order 3
nor those for the (less particular) order 5 led to an extension to other odd
orders. The constant features of his two treatments for the order 5 are,
first, to leave the main diagonals as they are and, second, to exchange
a pair of numbers between symmetrical, vertical and horizontal, rows.
For order 3 there is also one exchange between opposite rows; but in this
case the natural diagonals could not be maintained, for —owing to the
single possible configuration— we were obliged to replace them by the
(natural) middle rows. This latter step, in fact, prefigured the discovery
of a general method for odd orders, seen a few decennaries later.
Indeed, the two following properties are verified in each natural square
of odd order.
I. The main diagonals, but also the broken diagonals, make the magic
sum.
Thus, in our Fig. 19, the sum of 2, 8, 14, 20, 21, or of 11, 7, 3, 24, 20,
equals M5 = 65.
This is easily verified generally. Any diagonal of the natural square,
either main or broken, contains each unit of the order, from 1 to n, and
each multiple of the order, from 0 · n to (n − 1) · n (Fig. 20); since the
sum of all these elements is
n(n + 1) (n − 1)n
1 + 2 + · · · + n + n · 0 + 1 + · · · + (n − 1) = +n ,
2 2
n

we have indeed 2 n + 1 + n2 − n = Mn .
II. The two middle rows make the magic sum.
Indeed, the median column contains, as units of the order, n+12 uni-
formly, as well as each multiple of the order; its sum will therefore be
the same as above. The median line contains each unit and n times the
quantity n(n−1)
2 ; its sum will thus again be the same.
1 2 3 4 5 0,1 0,2 0,3 0,4 0,5
6 7 8 9 10 1,1 1,2 1,3 1,4 1,5
11 12 13 14 15 2,1 2,2 2,3 2,4 2,5
16 17 18 19 20 3,1 3,2 3,3 3,4 3,5
21 22 23 24 25 4,1 4,2 4,3 4,4 4,5
Fig. 19 Fig. 20
Now it would have sufficed —as had been done by B for the square
of order 3— to exchange the rôles of the median rows and the main
diagonals and to have in mind the property of the broken diagonals to
reach the easiest general construction, which no longer even requires use
of the natural square as a starting-point. It was Ibn al-Haytham (around
965-1041) who deduced it, or at least explained it.25
Let us consider, for instance once again for the order 5, the natural
square (Fig. 21), and let us thus put its middle rows as diagonals of
the square to be constructed. The main diagonals then meet the magic
condition. We next wish to determine the other occupants of the square,
for instance, first, that in the top of the last-but-one (right-hand) column.
For this, we shall use the property of broken diagonals. Since (Fig. 22)
the first line contains 11 and 3, we know that if it also contains 7, 20, 24,
which are the other three terms of their broken diagonal in the natural
square, it will make the magic sum. On the other hand 8 and 14, in
the column considered, belong to the broken diagonal 2, 8, 14, 20, 21.
Their common element is 20, which therefore will have its place at the
intersection of the line and the column considered. We shall proceed in
the same way for the other elements, at first disregarding the median
rows where there is only one known element. These median rows will be
considered last: in them, we shall simply write the remaining elements,
like 7 for the first line after 24 has been placed. (It will be seen that the
median rows contain the elements of the diagonals in the natural square.)
In this way we shall have constructed the magic square of Fig. 23.
1 2 3 4 5 11 20 3 11 24 7 20 3
6 7 8 9 10 12 8 4 12 25 8 16
11 12 13 14 15 13 17 5 13 21 9
16 17 18 19 20 18 14 10 18 1 14 22
21 22 23 24 25 23 15 23 6 19 2 15
Fig. 21 Fig. 22 Fig. 23

In short, the magic property of the diagonals in the constructed square
merely reproduces the magic property of the median rows in the natural
square, while the magic property of the other rows originates in their
reconstituting (succession of the elements excepted), the diagonals, main
or broken, of the natural square. Previous attempts involving exchanges
25
See our Les carrés magiques, pp. 25–27 (Russian edition, pp. 33–35); or (Arabic
text and translation) our Une compilation arabe, pp. 163–164, 181–182 (Arabic text
also in the Russian edition, pp. 279–280).
between opposite rows in the natural square of odd order may thus be
abandoned.
In this very same way we may construct any odd-order square, such
as that of Fig. 24, with n = 9. Furthermore, and as already hinted at, we
can do so without resorting to the natural square, and proceed as follows
in an empty square of the order considered:
— put 1 underneath the central cell;
— put the subsequent numbers descending diagonally;
— when the side of the square is reached (as with 4 and then 5 in Fig.
24), move to the opposite side and place the next number in the cell which
would have been reached by the diagonal move had the same square been
juxtaposed with that of the figure, then resume the diagonal placing;
— when, after placing a sequence of n elements (thus nine in our example)
a cell already occupied is reached, go down (whatever the order) two cells
in the same column, and then continue diagonally.
37 78 29 70 21 62 13 54 5
6 38 79 30 71 22 63 14 46
47 7 39 80 31 72 23 55 15
16 48 8 40 81 32 64 24 56
57 17 49 9 41 73 33 65 25
26 58 18 50 1 42 74 34 66
67 27 59 10 51 2 43 75 35
36 68 19 60 11 52 3 44 76
77 28 69 20 61 12 53 4 45
Fig. 24
The placement will end with n2 opposite 1, thus on the other side of
the central cell. As a matter of fact, it is a general feature of a square
constructed in this way to be symmetrical, that is, with two elements
adding up to n2 + 1 being placed in opposite cells relative to the centre;
2
accordingly, n 2+1 must be in the central cell.
Remark. In this construction the only moment of hesitation might occur
when the last cell of the descending diagonal is reached: the next cell
would be the top cell of the descending diagonal if it were not already
occupied; we shall therefore remain in the same column but have to
move on to its second upper cell.
26 Construction of even-order squares
It is clear that neither A nor B knew of such a method. The first

would not have spoken of a method presenting difficulty for the beginner
(A.II.2) and the second would not have presented two rather inconvenient
methods, only valid for one order, particularly since he elsewhere appears
to be anxious to explain a method to improve the student’s skill and (also)
intended for those who prefer to save themselves the trouble of working
out which numbers to arrange in the square (B.17, for bordered squares).
Remark. Ibn al-Haytham could well mark the turning point between the
method presenting difficulty for the beginner and the straightforward
placing we have just seen. The anonymous writer of the 12th century
who relates Ibn al-Haytham’s method says that his displacement of
elements between broken diagonals (and no longer between vertical and
horizontal rows) is carried out under long conditions which it would
take time to mention and the realization of which presents difficulties
for the beginner. This indeed singularly recalls A’s criticism of the
methods for ordinary magic squares. The difference is that Ibn al-
Haytham paved the way for a general and always applicable method
even if, for its explanation, he had to justify it with long conditions.
§2. Construction of even-order squares

The impossibility of a magic square of order 2 is simply stated by A
(A.II.36) but justified by the other author (B.7): the magic sum, 5, is
found in the diagonals of the natural square of order 2, and it should be
found, if possible, in the other rows; now putting a number in one corner
will impose the same quantity, namely its complement to 5, to occupy all
three other cells. Thus, in fact, the same sum in all rows will be possible
only if all four numbers are equal, which is contrary to the requirement
of different numbers (B.1). The first possible even magic square will thus
be that of order 4, which is constructed by our authors in various ways.
Indeed, as said before (pp. 6, 20), only the square of order 3 has a single
possible configuration.
A. Square of order 4
1. Constructions by displacement
(a) In B.8 a magic arrangement is obtained in the 4 × 4 square ‘by dis-
placement’, using two properties of the natural square (Fig. 25): first, as
always, the content of each of the two main diagonals makes the required
sum, 34; second, the content of two end cells of one row and that of the
two middle cells of the opposite row also make this sum (thus 2, 3 with
13, 16, or 6, 10 with 3, 15). Therefore, we shall keep the two diagonals as
they are, according to the first property, and use the second property to
equalize first the lateral rows of the border (Fig. 26), then the inner rows
(Fig. 27).
1 2 3 4 1 14 15 4 1 15 14 4
5 6 7 8 8 6 7 5 12 6 7 9
9 10 11 12 12 10 11 9 8 10 11 5
13 14 15 16 13 2 3 16 13 3 2 16
Fig. 25 Fig. 26 Fig. 27

(b) B also constructs this square more rapidly by combining, in a single
move, those two properties: he exchanges diagonally the elements in the
natural diagonals, thereby inverting them (Fig. 28). This, by a rotation
of 180◦ , gives the same square as before.
16 2 3 13
5 11 10 8
9 7 6 12
4 14 15 1
Fig. 28
Finally, B observes that the magic sum is also found elsewhere in the
magic square obtained (B.8v): (α) in the pairs of middle lateral cells
and those opposite (like 12, 8, 9, 5 in Fig. 27); (β) in the pairs of cells
adjoining two opposite corners (like 15, 12, 2, 5); (γ) in each 2 × 2 square
in a corner (like 1, 15, 12, 6). Here we may remark that these properties
are not all of the same kind: the first two occur in any magic square of
order 4 (and also in the natural square) whereas the last is not verified
in all constructions.
2. A construction without displacement
4 14 7 9 1 14 11 8
15 1 12 6 15 4 5 10
10 8 13 3 6 9 16 3
5 11 2 16 12 7 2 13
Fig. 29 Fig. 30
A, according to his initial rejection of displacements in the natural
square, does not propose a method as above. The only construction he
describes is that of Fig. 29 (here inverted). As usual, he merely indicates
each place individually (A.II.37), even though to construct this square

there is no need to know more than where to put the first four numbers
(see below). B constructs Fig. 30 (here inverted), which is older than his
treatise since he tells us that he has found it in his predecessors (B.9i).
These two figures are not fundamentally different since they are just
variants of a single pandiagonal structure: they can be deduced from
Fig. 31 by shifting the 4 × 4 frame in one direction or the other. Any
square thus found will be magic (and pandiagonal), the only change being
that a formerly broken diagonal becomes a main diagonal, and a given
number occupies another cell in the resulting square. As already noted
(p. 7), this latter property was highly considered at that time, whereas
pandiagonality apparently aroused no interest.
Remark. The manuscript of A has two other figures. One (a 22 in A.II.37)
corresponds to the same pandiagonal structure. The other (a 23) is of
another type (it displays a different pandiagonal structure, see Fig. 39
below).
13 2 7 12 13 2 7 12 13 2 7 12
3 16 9 6 3 16 9 6 3 16 9 6
10 5 4 15 10 5 4 15 10 5 4 15
8 11 14 1 8 11 14 1 8 11 14 1
13 2 7 12 13 2 7 12 13 2 7 12
3 16 9 6 3 16 9 6 3 16 9 6
10 5 4 15 10 5 4 15 10 5 4 15
8 11 14 1 8 11 14 1 8 11 14 1
13 2 7 12 13 2 7 12 13 2 7 12
3 16 9 6 3 16 9 6 3 16 9 6
10 5 4 15 10 5 4 15 10 5 4 15
8 11 14 1 8 11 14 1 8 11 14 1
Fig. 31
This possibility of choosing the initial cell leads B to four types of
arrangement (B.9ii, note 545): with the first number, here 1, occupying
any corner cell; with it in anyone of the central cells; with it in any middle
one of the lateral columns; and with it in any middle one of the lateral
lines. In our figures 32, 35, 38, the ∗ show where to place the first four
numbers and the † the four subsequent ones, the sequence of placing being
simply reversed. The other cases are obtained by rotation.
∗ † 1 14 11 8 4 15 10 5
† ∗ 12 7 2 13 9 6 3 16
† ∗ 6 9 16 3 7 12 13 2
∗ † 15 4 5 10 14 1 8 11
Fig. 32 Fig. 33 Fig. 34
∗ † 4 15 10 5 1 14 11 8
∗ † 14 1 8 11 15 4 5 10
† ∗ 7 12 13 2 6 9 16 3
† ∗ 9 6 3 16 12 7 2 13
Fig. 35 Fig. 36 Fig. 37
∗ † 1 12 13 8 4 9 16 5
† ∗ 15 6 3 10 14 7 2 11
∗ † 4 9 16 5 1 12 13 8
† ∗ 14 7 2 11 15 6 3 10
Fig. 38 Fig. 39 Fig. 40
This arrangement of the first eight numbers gives the explanation for
the magic property: in the resulting squares each line contains the same
sum, 9, and each one of alternate columns contains the same sum as its
conjugate; thus, by writing the complements in the bishop’s cells of one
of two conjugate rows, we shall complete the required sum in the other
conjugate. As to the magic sum of the main diagonals, it is evident since
they comprise pairs of complements.
B observes that any square of four cells in such a figure also contains
the magic sum M4 = 34, and this is what is called full and complete
magic arrangement (B.9i). Indeed, any four-cell square in Fig. 31 makes
the sum M4 . Fig. 41 and Fig. 42 make clear the origin of that for Fig.
36 and Fig. 39, as also the magic property of the broken diagonals (M
stands for M4 ).
M M
4 M
−2 M
−7 5 1 2
−5 2
−4 8
2 2
M M
M
−3 1 8 M
−6 2
−2 6 3 2
−7
2 2
M M
7 M
−5 M
−4 2 4 2
−8 2
−1 5
2 2
M M
M
−8 6 3 M
−1 2
−3 7 2 2
−6
2 2
Fig. 41 Fig. 42
Remarks.
(1) Fig. 33 is the one we shall adopt for our commentary, being easier to
memorize: starting from a corner with 1, the moves successively used
are the knight’s, the queen’s, the knight’s (and the same in reverse for
the sequence 5, . . . , 8).
(2) B twice mentions that a square of order 4 can be made the foundation
(as.l) for magic arrangement in other even squares (B.8i & B.9i). Once
this is, banally, when he is dealing with the construction of bordered
even squares, where the square of order 4, whatever its form, plays the
same rôle as the square of order 3 for odd orders: that of a core to be
successively surrounded by borders (see below, pp. 43–47). Another
time, and less banally, it is for the construction of composite squares,
since any evenly even square may be constructed by means of such
subsquares (see below, pp. 107–108).
B. Square of order 6
In the construction of the square of order 6 will be seen the limits of
B’s method by displacement. He operates on the natural square in two
steps, moving first the numbers of the inner square of order 4 then those
of the border. The way he does that is quite accurately described in B.10.
1 5 33 34 32 6
30 25 +18
8 9 10 11 8 28 9 11 −18 24 13 0
14 15 16 17 20 15 16 23 0 18 19 0
20 21 22 23 17 21 22 14 0 7 12 −18
26 27 28 29 26 10 27 29 +18 31 35 4 3 2 36
−3 0 0 +3 +3 0 0 −3
Fig. 43 Fig. 44 Fig. 45

Fig. 43 shows the original occupants in the central part of the natural
square and Fig. 44 the result of the displacements (B.10iii). Since the
magic sum in the square of order 6 is 3 · 37 = 111, the rows of the

inner square would accordingly make, if surrounded by a border, 2 · 37 =
74. But we are dealing here with ordinary magic squares, where that
should not be the case. The differences are indicated in the figure (the
two diagonals, being kept from the natural square, make the required
sum). The subsequent displacements (B.10iv) will fill the outer border
and eliminate the differences (Fig. 45).
Remark. These numerical differences are of course intentional since B
wants to construct an ordinary magic square. Indeed, for the inner
square of the natural square of order 6 he could have applied the same
moves he had used to construct the square of order 4 in methods (a)
and (b); but then, having still to add uniformly 37, he would have
obtained a bordered square of order 6.
C. Square of order 8
To obtain the square of order 8 ‘by displacements’ (B.11), the previous
construction is repeated for the inner part, the numbers moved being this
time those of the natural square of order 8 (B.11i–ii). As a result, the
square of order 6 displays exactly the sum due, and completing the square
will just mean adding a border which uniformly increases each row by
65 (Fig. 46). In this sense, the construction is a failure, unlike that in
the previous 6 × 6 square. Even if B has recognized some equalization
principles, his method could not, as we shall see, be extended to an even
order larger than 6.
1 58 59 5 4 62 63 8
16 10 14 52 53 51 15 49
24 47 19 45 20 22 42 41
33 39 35 28 29 38 26 32
25 31 30 36 37 27 34 40
48 18 43 21 44 46 23 17
56 50 54 13 12 11 55 9
57 7 6 60 61 3 2 64
Fig. 46
D. Allusion to a generalization
B seems to have had two general rules; at least, that is what can be
inferred from his two examples for small orders (6 and 8) and his assertion
that one should proceed like that for higher even orders (B.11i & iv).
1. Evenly even orders

Considering the natural square of order 4k (Fig. 47, n = 12, thus
k = 3), without touching the corner-cell elements (which belong to the
diagonals), we exchange diagonally 2k of the 4k elements, k in each half
(Fig. 48); the four lateral rows will thus make the required sum. Then,
we invert the succession of all elements within one horizontal and one
vertical row (Fig. 49); with that, the pairs of opposite elements in lines
and columns are also equalized. But since pairs of complements now
occupy opposite cells, we have de facto constructed a border.
1 2 3 4 5 6 7 8 9 10 11 12
13 24
25 36
37 48
49 60
61 72
73 84
85 96
97 108
109 120
121 132
133 134 135 136 137 138 139 140 141 142 143 144 Fig. 47
1 143 142 141 5 6 7 8 136 135 134 12
132 121 +108
120 109 +84
108 97 +60
49 60 −36
61 72 −12
73 84 +12
85 96 +36
48 37 −60
36 25 −84
24 13 −108
133 11 10 9 137 138 139 140 4 3 2 144

Fig. 48
+9 +7 +5 −3 −1 +1 +3 −5 −7 −9
1 134 135 136 8 7 6 5 141 142 143 12
24 121 0
36 109 0
48 97 0
85 60 0
73 72 0
61 84 0
49 96 0
108 37 0
120 25 0
132 13 0
133 11 10 9 137 138 139 140 4 3 2 144

Fig. 49
0 0 0 0 0 0 0 0 0 0
Applied to order 8 (Fig. 50–52), these transformations give successively
1 2 3 4 5 6 7 8 1 63 62 4 5 59 58 8
9 16 56 49
17 24 48 41
25 32 25 32
33 40 33 40
41 48 24 17
49 56 16 9
57 58 59 60 61 62 63 64 57 7 6 60 61 3 2 64
Fig. 50 Fig. 51
1 58 59 5 4 62 63 8
16 49
24 41
33 32
25 40
48 17
56 9
57 7 6 60 61 3 2 64
Fig. 52
which is indeed the border previously obtained by B (Fig. 46).
2. Evenly odd orders

If the natural square is of order n = 4k + 2 (Fig. 53, n = 14, thus
k = 3), we shall first exchange diagonally k +1 elements between opposite
rows (Fig. 54). We then exchange the places of one such element and its
opposite (Fig. 55); the lateral rows then make the required sum. We may
further invert the succession of the elements within one row, excepting
the pair last moved and the symmetrical pair (Fig. 56); this will give the
sum n2 + 1 in any two opposite cells, except in the latter two pairs.
1 2 3 4 5 6 7 8 9 10 11 12 13 14
15 28
29 42
43 56
57 70
71 84
85 98
99 112
113 126
127 140
141 154
155 168
169 182
183 184 185 186 187 188 189 190 191 192 193 194 195 196 Fig. 53
1 2 3 193 192 191 190 189 188 187 186 12 13 14
15 28
29 42
154 141
140 127
126 113
112 99
98 85
84 71
70 57
56 43
155 168
169 182
183 184 185 11 10 9 8 7 6 5 4 194 195 196

Fig. 54
1 2 3 11 192 191 190 189 188 187 186 12 13 14
15 28
29 42
154 141
140 127
126 113
112 99
98 85
84 71
70 57
43 56
155 168
169 182
183 184 185 193 10 9 8 7 6 5 4 194 195 196 Fig. 55
1 13 12 11 187 188 189 190 191 192 186 3 2 14
15 182
29 168
154 141 +98
140 57
126 71
112 85
98 99
84 113
70 127
43 56 −98
155 42
169 28
183 184 185 193 10 9 8 7 6 5 4 194 195 196

Fig. 56
+7 −7
Applied to order 6 (Fig. 57–59), this method gives, successively,

1 2 3 4 5 6 1 35 34 33 32 6 1 5 33 34 32 6
7 12 30 25 30 25
13 18 24 19 24 13
19 24 18 13 18 19
25 30 12 7 7 12
31 32 33 34 35 36 31 5 4 3 2 36 31 35 4 3 2 36
Fig. 57 Fig. 58 Fig. 59

the last of which is the border seen in B (Fig. 45).

As already observed, this method will not work with n > 6. The
border of order 6 will only make up for the differences in the inner square
and so the border of order 8 will necessarily have to be equalized by itself,
as for a bordered square; the same will hold for all subsequent borders.
E. Later developments
1. Equalization rules
As a matter of fact, B was not so very far from discovering a general
construction method for even orders, or at least evenly even ones. He
had noticed two properties of the natural square of order 4: that the
main diagonals make the magic sum and that two corner cells and the
two middle ones of the opposite row do too. This is in fact the particular
case of two general properties of the natural square of even order, which
are the following.
I. The sum of the elements in the diagonals, main or broken, equals the
magic sum for the order considered.
We have already verified this property for odd orders (p. 23). It was
then applied to easily obtain the magic sum in all horizontal and vertical
rows by reconstituting the broken diagonals.
II. The sum of half the elements in a row and the other half of the
elements in the symmetrically placed row equals the magic sum.
Indeed, the elements of two lines placed symmetrically being (for i ≤ n2 )
(i − 1)n + 1, (i − 1)n + 2, (i − 1)n + 3, . . . , (i − 1)n + n
(n − i)n + 1, (n − i)n + 2, (n − i)n + 3, . . . , (n − i)n + n

the sum of the first will be
n(n + 1) n
(i − 1)n2 + = [2ni − 2n + n + 1]
2 2
n 2 n
= [n + 1 − n2 + 2ni − n] = Mn − n(n − 2i + 1) ,
2 2
and that of the second (replacing i by n − i + 1)
n(n + 1) n
(n − i)n2 + = Mn + n(n − 2i + 1) .
2 2
Now the difference displayed by two of their elements vertically aligned is
uniformly n(n − 2i + 1), which, when multiplied by n2 , equals the deficit
in the upper line and the excess opposite. Therefore exchanging between
two lines placed symmetrically half of their elements vertically aligned
will produce the magic sum in them.
The same reasoning is applicable to the columns. The elements of the
jth left-hand column and of its opposite, the right-hand (n − j + 1)th
column (j ≤ n2 ), are
j, j + n, j + 2n, . . . , j + (n − 1)n
n − j + 1, 2n − j + 1, 3n − j + 1, . . . , n2 − j + 1
the sums of which are, respectively,
(n − 1)n n
nj + n = [2j + n2 − n]
2 2
n 2 n
= [n + 1 + 2j − n − 1] = Mn − n − 2j + 1 ,
2 2
(n − 1)n n
n(n − j + 1) + n = Mn + n − 2j + 1 .
2 2
Since, here again, the difference between two elements horizontally aligned
is n − 2j + 1, exchanging between two columns placed symmetrically half
of their elements horizontally aligned will eliminate their difference.
We have seen that, for equalizing the rows in the case of odd orders,
the property principally used was that of the (main or broken) diagonals
of the natural square, that of the middle rows being used only to meet the
condition in the diagonals. In the case of even orders, it is the comple-
mentarity of opposite rows which will play the main rôle, with half their
elements being exchanged. There is here, though, one difficulty: if one
starts with the exchanges between lines, opposite columns will no longer
display everywhere the same difference between opposite cells, which is a
necessary condition for the exchanges. We may remedy that situation as
follows.
2. Case of evenly even orders
Let us consider (Fig. 60) four elements placed symmetrically relative
to the median axes, thus horizontally at the same distance from the verti-
cal axis and vertically at the same distance from the horizontal axis. We
first carry out the vertical exchange of these pairs (Fig. 61); thus doing,
we have reduced the difference between the two opposite lines by its n4 th
part —since each exchange of one element with its opposite contributes a
reduction in the difference by its n2 th part— while the difference between
the two columns remains unchanged. Then we carry out the horizon-
tal exchange of the elements displaced (Fig. 62); that will not alter the
sum in the lines but lead to a reduction by its n4 th part in the difference
between the two columns containing the elements considered. Now it
then appears that these four exchanges are equivalent to two exchanges,
namely of each of the initial elements with the diagonally opposite one.
Thus, effecting, between two opposite lines of a square of evenly even or-
der n, n4 pairs of such diagonal exchanges will completely eliminate their
difference; and that will, at the same time, eliminate the n4 th part for
each pair of columns involved. Extending that procedure to the whole
square so as to effect exactly n2 exchanges between opposite lines and
opposite columns, namely n4 in each line of a quadrant, we shall have a
magic square. An embryonic stage of this is seen in B; since he restricts
himself to one border of the order considered, that is all he equalizes.
a b c d d c
◦ ◦ ◦ ◦ ◦ ◦
◦ ◦ ◦ ◦ ◦ ◦
c d a b b a
Fig. 60 Fig. 61 Fig. 62

Remark. It is not necessary for all exchanges to be diagonal. There can
be some that are diagonal, or also solely horizontal and vertical ones,
provided in all cases that there are in each pair of symmetric rows n2
exchanges. If the exchanges involve elements of the diagonals, they
must involve, as above, two pairs placed symmetrically in order to
maintain in each diagonal the sum n2 + 1 displayed in it by pairs of
opposite elements.
The magic square of order 4 obtained ‘by displacement’ in B is the

simplest example of such exchanges (as becomes apparent with the in-
version of the natural diagonals —method b in the constructions by dis-
placement). With the reasoning to be described now, this square can be
constructed directly, without resorting to the natural square. Let us mark
with dots in an empty square of order 4 its diagonals (Fig. 63). Then,
starting in a corner, say the upper left-hand one, let us count its cells
and write in the cells which are marked each number reached; this being
done, let us count the cells again, but starting from the opposite corner
and writing each time the number reached in the blank cell. We shall
obtain the square of Fig. 64, identical to the square found by B (Fig. 27).
If we count the cells in reverse order —that is, start by filling the cells
with dots from the lower right-hand corner— we shall obtain the square
found next by B (Fig. 28), thus the same one rotated by 180◦ .
• • 1 15 14 4
• • 12 6 7 9
• • 8 10 11 5
• • 13 3 2 16
Fig. 63 Fig. 64
This method can easily be extended to any square of order 4k, as
was already done in the 11th century.26 We need merely to mark the
diagonals in all subsquares of order 4 and then count the cells as above
from two opposite corners (Fig. 65 & Fig. 66). Here too, the dots signify
the places unchanged and the blank cells the diagonal exchanges; but,
again, reversing their rôles would just produce the same square, rotated
by 180◦ .
• • • • 1 63 62 4 5 59 58 8
• • • • 56 10 11 53 52 14 15 49
• • • • 48 18 19 45 44 22 23 41
• • • • 25 39 38 28 29 35 34 32
• • • • 33 31 30 36 37 27 26 40
• • • • 24 42 43 21 20 46 47 17
• • • • 16 50 51 13 12 54 55 9
• • • • 57 7 6 60 61 3 2 64
Fig. 65 Fig. 66
Another less particular way, which enables us to obtain without much

thought various configurations of blanks (or dots) indicating diagonal
exchanges, is the following. Consider, in the square of order 4k, its first
quadrant, of order 2k. In it, we shall put dots in such a way that there
will be exactly k in each line and in each column. The simplest way to
do it is to put k dots in the first line, and then carry on with writing
dots along the corresponding broken diagonals of the quadrant; this will
26
ensure that there are exactly k dots in each line and in each column of the
quadrant. Then we reproduce, in each of the remaining quadrants, the
same arrangement mirror-image-wise; we thus again obtain, in the square
as a whole, a configuration displaying a central symmetry. Counting
the cells as before from two opposite corners will carry out the required
exchanges. Here are examples of configurations obtained in this way for
order 8 (Fig. 67–70).
• • • • • • • •
• • • • • • • •
• • • • • • • •
• • • • • • • •
• • • • • • • •
• • • • • • • •
• • • • • • • •
• • • • • • • •
Fig. 67 Fig. 68
• • • • • • • •
• • • • • • • •
• • • • • • • •
• • • • • • • •
• • • • • • • •
• • • • • • • •
• • • • • • • •
• • • • • • • •
Fig. 69 Fig. 70
3. Case of evenly odd orders
Such an easily obtained configuration is now no longer possible: since
the square is of order n = 4k + 2, its quadrant is of order 2k + 1 and
the upper row cannot be half filled with dots. One can, though, resort
to a subterfuge: let us put in the first line of the first quadrant k + 1
dots, or rather, this time, × to indicate that they are diagonal exchanges.
Next, just as we did earlier, we reproduce these signs first diagonally in
the whole quadrant, then in the whole square by reproducing the arrange-
ment (Fig. 71). Now the number of diagonal exchanges, 2k + 2 by pairs of
symmetric rows, would be appropriate for a square of order 4k + 4; since
the order here is 4k + 2, with just its half, thus 2k + 1, required horizontal
and vertical exchanges, we shall keep 2k diagonal exchanges, which will
simultaneously effect 2k horizontal and 2k vertical exchanges, but reduce
the two supplementary diagonal exchanges, one to a purely vertical ex-
change and the other to a purely horizontal exchange, by returning their
displaced elements to the neighbouring half; this is the meaning of the
signs — and | in Fig. 72 (n = 10, thus k = 2), while the remaining ×

further play their rôles of diagonal exchanges. One may verify that there
are indeed in each line of the whole square 5 (= 2k + 1) vertical exchanges
(signs × and |) and, in each column, 5 horizontal exchanges (× and —).
The diagonals have been left out of these exchanges.
× × × × × × × — | × — ×
× × × × × × | × — — × ×
× × × × × × — | × × × —
× × × × × × × — | × — ×
× × × × × × × — | × — ×
× × × × × × × × | × × ×
× × × × × × × × | × × ×
× × × × × × × | × × × ×
× × × × × × | × × × × ×
× × × × × × × × | × × ×
Fig. 71 Fig. 72
Applying this to the square of order 10 gives, respectively, Fig. 73 and
Fig. 74.
1 2 98 97 96 95 94 93 9 10 1 2 98 94 6 95 97 93 9 10
90 12 13 87 86 85 84 18 19 81 20 12 13 87 85 86 84 18 19 81
80 79 23 24 76 75 27 28 72 71 71 29 23 24 76 75 27 28 72 80
70 69 68 34 35 36 37 63 62 61 70 62 38 34 35 36 37 63 69 61
41 59 58 57 45 46 54 53 52 50 41 59 53 47 45 46 54 58 52 50
51 49 48 47 55 56 44 43 42 60 51 49 48 57 55 56 44 43 42 60
40 39 38 64 65 66 67 33 32 31 40 39 68 64 65 66 67 33 32 31
30 29 73 74 26 25 77 78 22 21 30 79 73 74 26 25 77 78 22 21
20 82 83 17 16 15 14 88 89 11 90 82 83 17 16 15 14 88 89 11
91 92 8 7 6 5 4 3 99 100 91 92 8 7 96 5 4 3 99 100
Fig. 73 Fig. 74
An embryonic idea of such exchanges is found in B: he has effected

k + 1 diagonal exchanges in the lateral rows of each quadrant (Fig. 54),
then inverted vertically and horizontally two elements already moved (Fig.
55). But here again —and the final reversal (Fig. 56) shows it— this was
carried out for equalizing not only the rows of the outer border but also,
as far as possible, opposite cells; and the idea of extending such exchanges
to the whole natural square is absent. This will come, at the latest, by
the end of the 11th century; the last difficulty posed by the construction
of ordinary magic squares will thus be removed.27
27
Chapter II. Bordered magic squares
It will appear from A that general methods for bordered squares of
any order were definitely known in the 10th century, whereas B knows
a general method only for odd orders. For odd as well as even orders,
B considers each square in turn, starting with the smallest, 3 or 4, and
shows how to add successive borders around it. For odd orders, this will
lead him to his general way of arranging the numbers. Here we have
a clear testimony on the earlier history of magic squares, namely how a
general method was empirically arrived at. Moreover, searching for various
ways of filling the small orders will lead B to attempting to enumerate the
different possible configurations. Making such investigations was natural
when each order was examined separately; it was, however, no longer
of major interest once general methods had been discovered. All this
we have already noted, but it is important enough to deserve a second
mention.
§1. Construction of odd-order bordered squares

A, in his description of the construction (A.II.3–A.II.6), tells us where
to put the successive numbers, cell by cell, without explanation, as usual
(see above, p. 14). On the contrary, B explains how to construct a square
border by border (B.12–B.16). First of all, we are to determine the num-
ber of cells the border considered contains and what their occupants will
be (B.13). Since a border of order n contains 2n + 2(n − 2) = 4n − 4
cells, we then know that we are to put in it 2n − 2 pairs of comple-
2
ments, that is, 2n − 2 ‘small’ numbers (less than n 2+1 , the median, to be
put in the central cell), and 2n − 2 ‘large’ numbers, their complements.
Therefore, if we begin with the outer border, its occupants will be the
first 2n − 2 natural numbers and their complements, and the next bor-
der, with 2(n − 2) + 2(n − 4) = 4n − 12 cells altogether, will receive the
next two sequences of numbers, and so on with similar decreases for the
subsequent borders (B.13iii–iv).
A. Empirical discovery of a method
1. Square of order 5
Suppose then we wish to fill a bordered square of order n = 5 (B.15),
putting the first small numbers in the outer border while the median
number is in the central cell and its neighbours in the border of order 3.
Filling the central square of order m = 3 is easy since there is just one
44 Construction of odd-order bordered squares
2
possible arrangement. So we shall put n 2+1 = 13 in the central cell, then,
in the border, its 4m − 4 = 8 neighbours, 2m − 2 = 4 on either side of
the median, that is,
9 10 11 12
17 16 15 14,
each one with its place as seen, say with the smallest in the middle of the
bottom row (Fig. 75).
4 6 4 24 23 8 6
10 17 12 10 17 12 19 10 17 12 7
15 13 11 15 13 11 21 15 13 11 5
14 9 16 14 9 16 1 14 9 16 25
20 22 20 2 3 18 22
Fig. 75 Fig. 76 Fig. 77

The sum of a pair of opposite numbers indeed makes n2 + 1
= 26, and
such is the sum pairs of complements are to make in the border of order
5. These pairs are
1 2 3 4 5 6 7 8
25 24 23 22 21 20 19 18.
We know that the sum in each border row must be M5 = 65, for the author
has taught how to calculate it whatever the size of the border (B.2). The
first placing to consider is for the common corner cells; here B remarks
that we may not place just any pair of small numbers in consecutive
corners (B.15vi): with the pairs 1 & 4, 1 & 2, 4 & 5, 6 & 7 we would
not be able to complete the border with the still available numbers. On
the other hand, the pairs 1 & 3, 3 & 7, 2 & 8, 4 & 6, 5 & 7, 6 & 8
are admissible. Of those, B chooses 4 and 6 (Fig. 76) —and for a good
reason: they are easy to remember, with the value of the order between
them, just like the occupants of the upper corner cells in the square of 3
(Fig. 8). B must then choose, from among the remaining numbers, three
adding up to 55 for the top row and three adding up to 41 for the left-
hand row. With the complements in place, the whole square is completed
(Fig. 77).
Remarks.
(1) The six choices for consecutive corners proposed by B are indeed the
only possible. They lead to ten borders (disregarding banal changes
of place within the rows). The nine others are represented in our
Bordered magic squares 45
figures 78–86. In the manuscript of B, which has no figure drawn,

there is a blank space with enough room either for six figures (if the
author found it unnecessary to give more than one possibility for each
admissible corner-cell pair) or, more likely, for a table enumerating
all possibilities. In the latter case, B would have anticipated Bernard
Violle who, in the early 19th century, is seen to have listed them.28
(2) With 4 and 6 occupying the corners, there are two possibilities (Fig.
77 and Fig. 85); with B’s explicit mention of the numbers used, we
know which his was, but at this point he does not actually allude to a
second configuration.
(3) Having given all, or some, of the possible configurations, he then

(B.15viii) considers how many different aspects the border would take
by keeping the same numbers within the rows but modifying their
places. Unfortunately, the result he obtained does not appear in the
manuscript. We know for sure that he considered there to be six
possibilities for each row, thus six both for the top row and a lateral
one (the place of the complements is fixed by their opposite). He might
thus have found 36 arrangements for each border, to be multiplied by
his number of possibilities for two consecutive corners.
1 22 20 19 3 1 22 21 18 3 3 25 24 6 7
2 24 2 24 4 22
21 5 20 6 21 5
18 8 19 7 18 8
23 4 6 7 25 23 4 5 8 25 19 1 2 20 23
Fig. 78 Fig. 79 Fig. 80
3 25 22 8 7 5 25 24 4 7 5 24 23 6 7
2 24 3 23 1 25
21 5 20 6 22 4
20 6 18 8 18 8
19 1 4 18 23 19 1 2 22 21 19 2 3 20 21
Fig. 81 Fig. 82 Fig. 83
28
Traité complet, I, p. 63.
2 25 23 7 8 4 25 23 7 6 6 24 23 4 8
4 22 2 24 1 25
21 5 21 5 21 5
20 6 18 8 19 7
18 1 3 19 24 20 1 3 19 22 18 2 3 22 20
Fig. 84 Fig. 85 Fig. 86

2. Higher odd-order squares
In order for this construction to become easier for the student, B
treats the example of a square of order 9, or, rather, of the borders of
order 7 and 9 since the inner arrangement is already known (B.16).
The central square of order 5, of 25 cells, thus contains twelve numbers
on either side of 41, altogether from 29 to 53; and even though we already
know how to place them, the author tells us again (B.16ii–iii).
8 80 78 76 75 12 14 16 10
67 22 64 62 61 26 28 24 15
69 55 32 52 51 36 34 27 13
71 57 47 38 45 40 35 25 11
73 59 49 43 41 39 33 23 9
5 19 29 42 37 44 53 63 77
3 17 48 30 31 46 50 65 79
1 58 18 20 21 56 54 60 81
72 2 4 6 7 70 68 66 74
Fig. 87
The next numbers to be placed, in the border of order 7, will be the
subsequent twenty-four numbers, namely (B.16iv)
17 18 19 20 21 22 23 24 25 26 27 28
65 64 63 62 61 60 59 58 57 56 55 54
while the thirty-two for the next border of order 9 will be (B.16v)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
81 80 79 78 77 76 75 74 73 72 71 70 69 68 67 66.
In accordance with his previous choices, B takes for the corner cells
in the border of order 7 the sixth and the eighth of the smaller numbers
listed, thus 22 and 24, and for those of order 9 the eighth and the tenth of
the second list, thus 8 and 10. Then he is left with completing the required
sums in the rows. We know his choices since he explicitly mentions them.
He thus obtains the square of Fig. 87.
B then goes on to describe in general terms a method for those who
prefer to save themselves the trouble of working out (by calculation) which
numbers to arrange in the square (B.17). The successive constructions
have indeed made apparent a certain regularity in the occupation of the
rows, which may be easily remembered. This is what he explains by
means of a geometrical figure (a mere visual aid).
Consider, according to the figure of the text, a square ABCD of odd
order, with its medial rows EF and GH (Fig. 88). We write the con-
secutive numbers alternately in the rows comprising a corner (say A),
excepting the medial and corner cells, until half of these cells are filled.
The next four numbers are written, respectively, in E, B, H, D, then the
subsequent ones alternately in the rows meeting at D, keeping the same
parity as in the opposite rows. The cells opposite are each time left empty
for the complements. One then begins to fill in the same way the next
inner border, starting with the smaller number attained, and so on until
the central cell is reached, with the median number in it. After placing
the complements, we shall have filled the whole square. Note that with
this geometric figure we see the movement to be followed for filling the
borders.
B F D
G H
A E C
Fig. 88
Here is the only empirical deduction we know of a general method:
from the successive fillings of several borders, involving each time choices and
calculations, we have made the transition to a few instructions enabling us to
place directly all numbers in succession. This case is particularly favorable,
with odd border succeeding to odd border and, since the square is a
bordered one, each border being filled separately and independently.
B. Description of the general method

As appears from his text, B leaves the place of numbers within the
rows optional. Starting at the corners, we have put them in the successive
cells, which is the arrangement chosen by A in his figures a 2–a 5. (The
only difference in the description is that, since he fills the squares from
the inside out, his smaller numbers are placed in decreasing order from
the median number.)
10 120 118 116 114 113 14 16 18 20 12
103 28 100 98 96 95 32 34 36 30 19
105 87 42 84 82 81 46 48 44 35 17
107 89 75 52 72 71 56 54 47 33 15
109 91 77 67 58 65 60 55 45 31 13
111 93 79 69 63 61 59 53 43 29 11
7 25 39 49 62 57 64 73 83 97 115
5 23 37 68 50 51 66 70 85 99 117
3 21 78 38 40 41 76 74 80 101 119
1 92 22 24 26 27 90 88 86 94 121
110 2 4 6 8 9 108 106 104 102 112
Fig. 89
Although the procedure should by now be clear enough, it is worth
repeating if it is, as it then was, to be memorized. Starting from a corner
(the lower left-hand corner in Fig. 89, n = 11), we put the numbers
alternately in the two rows until we arrive at the middle cells. The next
number is written in the middle cell just reached, then one fills the cell
of the opposite corner (belonging to the initial row) then the middle cell
in the opposite side, then the still empty corner cell of the upper row.
Beginning then after the middle cell already filled and moving towards
that corner from the middle of the two contiguous rows, we resume the
alternate placement. Half of the border cells are then occupied; the other
half will be filled by writing in each the complement to n2 + 1 of the
number opposite, horizontally or vertically (diagonally for the angles).
The next border, once again odd, will be filled in the same way, and so
on to the square of order 3, filled as usual.29 In this way, each removal
29
It will be seen that the above placing can also be applied to the border of order 3
considering that it comprises only corner and middle cells.
of a border will reduce the magic sum by n2 + 1, and the square which
remains will also be magic.
C. Mathematical basis for this method
Texts of that time do not present any mathematical justification for
the configurations attained: an empirical verification is sufficient, and we
have seen how the result was obtained.
Its general validity can be formally established as follows. Let us take
a square of order n, with n = 2k + 1 since the square is of odd order. It
will be filled with n2 = 4k2 +4k +1 numbers, namely, as we already know,
2
the smaller numbers, less than the median term n 2+1 = 2k2 + 2k + 1, and
the larger numbers, higher than the median, which are the complements
of the former, and placed according to what has been determined for the
smaller numbers. The outer border of this square of order n comprises
2n + 2(n − 2) = 4n − 4 cells, thus 8k cells, to be occupied by the first 4k
smaller numbers and their 4k complements.
In order to have a more or less balanced distribution of the smaller
numbers and their complements in the outer border, we shall first con-
sider attributing to each of its rows a sequence of k smaller numbers.
Further, to facilitate memorization, we shall choose a sequence display-
ing regularity; here, we shall take ascending sequences of numbers having
the same parity.30 The four sequences to k terms will then be
1, 3, . . . , 2k − 1 with sum k2
2, 4, . . . , 2k with sum k2 + k
2k + 1, 2k + 3, . . . , 4k − 1 with sum 3k2
2k + 2, 2k + 4, . . . , 4k with sum 3k2 + k.
For our equalization of the rows, we shall take only the k − 1 first terms
of the two smaller sequences and the k − 1 last terms of the two larger
ones, which we shall write, one by one, in each row in turn, so that
two opposite rows will contain terms having the same parity. The four
remaining terms are kept for two (consecutive) angles and for two other
cells, say (consecutive) middle ones: their location will serve to equalize
the rows. Only these four numbers might break the otherwise uniform
parity found in the rows’ occupants.
The four previous sequences, now reduced to k − 1 terms each, will
then be
30
Some later Arabic texts choose sequences following the natural order; see Les
carrés magiques, pp. 124–125, 129–130 (Russian edition, pp. 134–136, 140–141).
{ag } : 1, 3, . . . , 2k − 3 with sum k2 − 2k + 1

{bh } : 2, 4, . . . , 2k − 2 with sum k2 − k
{ci } : 2k + 3, 2k + 5, . . . , 4k − 1 with sum 3k2 − 2k − 1
{dj } : 2k + 4, 2k + 6, . . . , 4k with sum 3k2 − k − 2.
A {dj } D
{ci }
{ag }
{bh } B
Fig. 90
Putting these four sequences successively in each row (Fig. 90), and saving
the opposite cells for their complements, we shall thus be left with four
unplaced smaller terms (2k − 1, 2k, 2k + 1, 2k + 2) and their four cells
(A, B, C, D).
Consider now the sums thus attained. The sum in the left-hand col-
umn is (taking into account the complements to be found there, that is,
besides the k − 1 complements of the {ci }, those of the numbers in C and
D)

ag + A + (n2 + 1) − C + (n2 + 1) − D + (k − 1)(n2 + 1) − ci
= k2 − 2k + 1 + A − C − D + (k + 1)(n2 + 1) − 3k2 + 2k + 1
= A − C − D − 2k2 + 2 + (k + 1)(n2 + 1).
Since this must equal the magic sum of the square considered, Mn =
(k + 12 )(n2 + 1), we shall have
1
C + D − A + 2k2 − 2 = 2 (n2 + 1) = 2k2 + 2k + 1
and therefore
C + D − A = 2k + 3. (∗)
As for the sum in the upper row, it will be

dj + A + D + (k − 1)(n2 + 1) − bh + (n2 + 1) − B
= 3k2 − k − 2 + A + D − B − k2 + k + k(n2 + 1)
= A + D − B + 2k2 − 2 + k(n2 + 1).
Putting this, in order to have the magic sum, equal to Mn = (k + 12 )(n2 +

1), we find
A + D − B + 2k2 − 2 = 12 (n2 + 1) = 2k2 + 2k + 1,
and therefore
A + D − B = 2k + 3. (∗∗)
We now have two relations for the determination of the numbers due
to occupy the cells A, B, C, D, and we know the set of their admissible
values, namely 2k − 1, 2k, 2k + 1, 2k + 2. It appears that only two
combinations enable us to obtain the sum 2k + 3 in (∗) and (∗∗), namely
(2k + 2) + (2k + 1) − 2k = 2k + 3,
(2k + 2) + 2k − (2k − 1) = 2k + 3.
Comparing this with the relations (∗) and (∗∗) leads us to put
A = 2k, B = 2k − 1, C = 2k + 1, D = 2k + 2.
Except for the corners, the place of the smaller numbers in the rows
is optional —as long as the opposite cells are left empty. They may be
placed according to their order of succession, which will help to memorize
the arrangement.
§2. Construction of even-order bordered squares

A. Construction of bordered squares of evenly even orders
1. Description of the method in A and B
As B observes, the magic arrangement of the square of 4 differs from
that of the other evenly even squares (B.13vi). This is true for bordered
squares (not for the ordinary magic arrangement, as we have seen be-
fore): since there is no magic square of order 2, we cannot place pairs of
complements in opposite rows. The square of order 4 must therefore be
filled as a whole, also as seen before.
For a general evenly even square, A takes the smaller numbers to
be placed in the outer border in decreasing order, for he supposes the
inner part to have been filled first. He describes that in A.II.42–A.II.43
(Fig. 91, with n = 12, and 22 the first number to be placed since the
border comprises 44 cells). He takes groups of four consecutive numbers
of which he places the extremes (say) on the top and the two middle
on the bottom, until he is left with six empty cells within the row. Of
the next six numbers, the first two are put above and the others below
—leaving, as usual, the opposite cells empty. A then fills, successively,
the two corners, the cell underneath the corner just filled, then one cell
on the other side. Returning to the opposite column, he completes the
52 Construction of even-order bordered squares
two columns by means of groups of four consecutive numbers, the two

extremes on one side and the two middle on the other. These smaller
numbers and their complements fill the outer border completely.
11 22 124 125 19 18 17 129 130 131 132 12
10 135
136 9
8 137
138 7
139 6
5 140
4 141
142 3
143 2
1 144
133 123 21 20 126 127 128 16 15 14 13 134
Fig. 91
B constructs only the square of order 8 (B.19). He puts in the upper
corner cells the number of the order and its predecessor, then considers
the missing sum and searches among the numbers still available for those
which make the required sum. The border of Fig. 92 is his result (apart,
perhaps, from arrangement within the rows).
8 1 2 59 60 61 62 7 7 14 13 53 54 55 56 8
9 56 6 59
55 10 60 5
54 11 4 61
12 53 62 3
13 52 63 2
51 14 1 64
58 64 63 6 5 4 3 57 57 51 52 12 11 10 9 58
Fig. 92 Fig. 93
In A (Fig. 93), the arrangement is essentially the same, with the cor-
ners occupied by the order number and its predecessor; but with the
square being filled from within, the ascending sequence found in B be-
comes a descending one, from 14 on.

Remark. B fills the border of order 8 just after the border of order 6. He
thus does not distinguish between the two types of even order. Nor
does he explain an arrangement whereby the evenly even border could
be filled without computations. If text A had not been preserved, we
would not now know about the existence in the 10th century of a general
method for bordered squares of evenly even orders.
2. Description of the general method
15 1 255 254 4 5 251 250 8 9 10 246 245 244 243 16
240 17
18 239
19 238
237 20
236 21
22 235
23 234
233 24
232 25
26 231
27 230
229 28
228 29
30 227
241 256 2 3 253 252 6 7 249 248 247 11 12 13 14 242
Fig. 94
A thus attests that the following general construction method was
known in the 10th century. Consider an empty border of order n = 4k,
with k ≥ 3 (which excludes the orders 8 and, of course, 4 which cannot
be bordered). Starting (say) at the upper left corner cell, we put (Fig.
94) the successive numbers by groups of four, the first one next to the
corner, the second and the third on the bottom and the fourth at the top,
and so on until there remains in the top row (excluding the corners) six
empty cells; we then write the next two numbers above and the next four
below. We then fill the upper corners, first left then right. We place the
next number below this last one, the next number on the other side, and
resume placing groups of four consecutive numbers as before.
This general method also applies to order 8; we shall just begin di-
rectly with the six cells (Fig. 95).
7 1 2 62 61 60 59 8
56 9
10 55
11 54
53 12
52 13
14 51
57 64 63 3 4 5 6 58
Fig. 95
3. Mathematical basis for this method
(α) General theory
Basically, what should be obtained when constructing borders of even
orders is, after a preliminary placement, repeated use of what we shall
call neutral placements. Consider a sequence of four consecutive numbers
of which we place the extremes in one row and the middle ones in the
opposite row (Fig. 96); after writing in their complements, not only will
the sum due be obtained in cells facing one another, but also in the
four cells of each row. Therefore, if, after a preliminary placement, we
reach a situation where completing rows may be carried out by neutral
placements, filling them will be easy and applicable to any order. This
preliminary placement must, however, obey some conditions.
α n2 + 1 − (α + 1) n2 + 1 − (α + 2) α+3 =⇒ 2(n2 + 1)
n2 + 1 − α α+1 α+2 n2 + 1 − (α + 3) =⇒ 2(n2 + 1)
Fig. 96
(i) The preliminary placement must be such that, after writing in the
2
complements, each row displays its sum due, thus m · n 2+1 if m cells are
filled.
(ii) The preliminary placement must be uniform to be applicable to any

order, excepting at most small orders.
(iii) The number of cells filled with this preliminary placement must be
as small as possible in order for the method to be applicable, as far as
possible, to the smallest orders.
(iv) The preliminary placement must settle the question of the corner
cells, each common to two rows.
(v) Evidently, the number of cells left empty after the preliminary place-
ment must be divisible by 4 for subsequent neutral placements.
(vi) The numbers left must form groups of four consecutive numbers in
order to facilitate neutral placements.
(vii) Finally, for the outer border, the numbers placed preliminarily should,
ideally, be the first natural numbers, for that will make the neutral place-
ments straightforward. But this condition may be disregarded if the pre-
liminary placement made with other numbers is particularly easy to re-
member.
Let us apply these general instructions to the present case, that of
evenly even orders. Consider thus a square of order n = 4k with k ≥ 2.
There are various possibilities for the preliminary placement.
(a) We shall have as quantity of empty cells a multiple of 4 by placing
two numbers in each row, two of which must occupy consecutive corner
cells (Fig. 97). Indeed, after placing the complements, four cells in each
row will be occupied and 4(k − 1) will still be empty.
a c
e
f
b d
Fig. 97
If the numbers thus placed are the first six natural numbers, we shall
have only to memorize their positions and the subsequent filling will just
consist in writing groups of four numbers taken in consecutive order.
Consider thus this placement of the first six numbers. Designate by the
sign prime their complements, that of a being thus a = (n2 + 1) − a.
After writing in the complements, we must have the sum due for four
cells filled, thus, for the first line and the first column,
a + b + d + c = 2(n2 + 1),
a + c + f + e = 2(n2 + 1).
Since b + d = 2(n2 + 1) − b − d and c + f = 2(n2 + 1) − c − f , the two
previous relations become

a+c=b+d
a + e = c + f,
a and c being common to the two equalities since they occupy the corners.
Now solving this pair of equations with the first six natural numbers gives
only two solutions: 2 and 3 in the corners with 1 and 4 in the opposite
row, or 4 and 5 in the corners with 3 and 6 in the opposite row (the
place of the two remaining numbers is immediately determined). As we
see, memorizing this placement is particularly easy since we are merely
to keep in mind the occupants of the two upper corners, the position of
the other numbers being inferred thereby. The disadvantage, if any, is
the limited number of configurations. This arrangement is attested from
the 11th century.31
(b) Writing two more small numbers within the top and bottom rows,
as in Fig. 98, will leave 4(k − 2) empty cells in the horizontal rows and,
as before, 4(k − 1) in the columns. This preliminary placement of ten
numbers —again to be memorized if we are to fill the border without
much thought— is also applicable to the order 8. As for the pair of
equations for their determination, it will be

a+b+c+d= e+f +g+h
a + i = d + j,
a b c d
i
j
e f g h
Fig. 98
31
with this time —if we still restrict ourselves to the first ten natural
numbers— 128 solutions (Fig. 99a–99d). But memorizing the places of
ten numbers is no longer as easy as before.
a b c d i j e f g h a b c d i j e f g h
1 2 10 9 1 5 6 8 7 4 3 44 5 6 3 4 9 10 8 7 2 1
2 2 10 8 1 6 7 9 5 4 3 45 6 10 5 1 3 8 9 7 4 2
3 2 10 6 1 8 9 7 5 4 3 46 6 9 4 1 5 10 8 7 3 2
4 2 8 7 1 9 10 6 5 4 3 47 6 10 2 3 5 8 9 7 4 1
5 3 10 8 2 4 5 9 7 6 1 48 6 9 7 3 1 4 10 8 5 2
6 3 10 7 2 5 6 9 8 4 1 49 6 9 1 3 7 10 8 5 4 2
7 3 10 5 2 7 8 9 6 4 1 50 6 8 7 3 2 5 10 9 4 1
8 3 10 4 2 8 9 7 6 5 1 51 6 8 5 3 4 7 10 9 2 1
9 3 9 8 2 5 6 10 7 4 1 52 6 8 2 3 7 10 9 5 4 1
10 3 9 6 2 7 8 10 5 4 1 53 6 10 4 5 2 3 9 8 7 1
11 3 8 5 2 9 10 7 6 4 1 54 6 8 7 5 1 2 10 9 4 3
12 3 7 6 2 9 10 8 5 4 1 55 6 7 1 5 8 9 10 4 3 2
13 4 10 9 1 2 5 8 7 6 3 56 6 4 3 5 9 10 8 7 2 1
14 4 10 8 1 3 6 9 7 5 2 57 7 10 5 2 1 6 9 8 4 3
15 4 10 6 1 5 8 9 7 3 2 58 7 9 4 2 3 8 10 6 5 1
16 4 10 5 1 6 9 8 7 3 2 59 7 8 3 2 5 10 9 6 4 1
17 4 9 7 1 5 8 10 6 3 2 60 7 10 3 4 2 5 9 8 6 1
18 4 9 5 1 7 10 8 6 3 2 61 7 10 2 4 3 6 9 8 5 1
19 4 8 7 1 6 9 10 5 3 2 62 7 9 1 4 5 8 10 6 3 2
20 4 8 6 1 7 10 9 5 3 2 63 7 8 1 4 6 9 10 5 3 2
21 4 10 9 3 1 2 8 7 6 5 64 7 10 3 6 1 2 9 8 5 4
22 4 10 2 3 8 9 7 6 5 1 65 7 10 1 6 3 4 9 8 5 2
23 4 9 5 3 6 7 10 8 2 1 66 7 9 4 6 1 2 10 8 5 3
24 4 8 7 3 5 6 10 9 2 1 67 7 9 2 6 3 4 10 8 5 1
25 4 7 5 3 8 9 10 6 2 1 68 7 9 1 6 4 5 10 8 3 2
26 4 6 5 3 9 10 8 7 2 1 69 7 8 5 6 1 2 10 9 4 3
27 5 10 8 2 1 4 9 7 6 3 70 7 8 4 6 2 3 10 9 5 1
28 5 10 3 2 6 9 8 7 4 1 71 7 8 2 6 4 5 10 9 3 1
29 5 9 7 2 3 6 10 8 4 1 72 7 5 1 6 8 9 10 4 3 2
30 5 9 6 2 4 7 10 8 3 1 73 7 4 2 6 8 9 10 5 3 1
31 5 9 3 2 7 10 8 6 4 1 74 7 4 1 6 9 10 8 5 3 2
32 5 8 4 2 7 10 9 6 3 1 75 7 3 2 6 9 10 8 5 4 1
33 5 10 7 4 1 2 9 8 6 3 76 8 10 3 1 2 9 7 6 5 4
34 5 10 6 4 2 3 9 8 7 1 77 8 7 6 1 2 9 10 5 4 3
35 5 10 2 4 6 7 9 8 3 1 78 8 7 5 1 3 10 9 6 4 2
36 5 10 1 4 7 8 9 6 3 2 79 8 9 4 3 1 6 10 7 5 2
37 5 9 8 4 1 2 10 7 6 3 80 8 7 2 3 5 10 9 6 4 1
38 5 9 7 4 2 3 10 8 6 1 81 8 10 2 5 1 4 9 7 6 3
39 5 9 3 4 6 7 10 8 2 1 82 8 9 3 5 1 4 10 7 6 2
40 5 9 2 4 7 8 10 6 3 1 83 8 9 1 5 3 6 10 7 4 2
41 5 8 1 4 9 10 7 6 3 2 84 8 6 3 5 4 7 10 9 2 1
42 5 7 3 4 8 9 10 6 2 1 85 8 4 3 5 6 9 10 7 2 1
43 5 7 2 4 9 10 8 6 3 1 86 8 4 2 5 7 10 9 6 3 1
Fig. 99a Fig. 99b

a b c d i j e f g h a b c d i j e f g h
87 8 9 1 7 2 3 10 6 5 4 108 9 5 2 8 3 4 10 7 6 1
88 8 6 5 7 1 2 10 9 4 3 109 9 4 1 8 5 6 10 7 3 2
89 8 6 4 7 2 3 10 9 5 1 110 9 3 2 8 5 6 10 7 4 1
90 8 6 2 7 4 5 10 9 3 1 111 9 3 1 8 6 7 10 5 4 2
91 8 4 3 7 5 6 10 9 2 1 112 10 8 1 3 2 9 7 6 5 4
92 8 2 1 7 9 10 6 5 4 3 113 10 6 4 3 1 8 9 7 5 2
93 9 7 5 2 1 8 10 6 4 3 114 10 5 4 3 2 9 8 7 6 1
94 9 6 4 2 3 10 8 7 5 1 115 10 7 2 5 1 6 9 8 4 3
95 9 8 3 4 1 6 10 7 5 2 116 10 6 1 5 3 8 9 7 4 2
96 9 7 2 4 3 8 10 6 5 1 117 10 6 2 7 1 4 9 8 5 3
97 9 6 1 4 5 10 8 7 3 2 118 10 6 1 7 2 5 9 8 4 3
98 9 8 2 6 1 4 10 7 5 3 119 10 5 3 7 1 4 9 8 6 2
99 9 8 1 6 2 5 10 7 4 3 120 10 5 1 7 3 6 9 8 4 2
100 9 7 3 6 1 4 10 8 5 2 121 10 4 3 7 2 5 9 8 6 1
101 9 5 2 6 4 7 10 8 3 1 122 10 4 2 7 3 6 9 8 5 1
102 9 4 2 6 5 8 10 7 3 1 123 10 3 1 7 5 8 9 6 4 2
103 9 3 1 6 7 10 8 5 4 2 124 10 2 1 7 6 9 8 5 4 3
104 9 7 1 8 2 3 10 6 5 4 125 10 5 1 9 2 3 8 7 6 4
105 9 6 3 8 1 2 10 7 5 4 126 10 4 3 9 1 2 8 7 6 5
106 9 6 1 8 3 4 10 7 5 2 127 10 3 1 9 4 5 8 7 6 2
107 9 5 4 8 1 2 10 7 6 3 128 10 2 1 9 5 6 8 7 4 3
Fig. 99c Fig. 99d
(c) We could also place fourteen numbers, in particular the first four-
teen natural numbers, by putting the four supplementary numbers in the
columns. Again, that would be applicable to the order 8. Generally, we
could continue attributing pairs of numbers to the rows, but that would
merely hinder memorization and increasingly exclude the smallest orders.
(β) The tenth-century method
If this tenth-century method seems fundamentally different, that is in
appearance only: it reduces de facto to the placement of ten numbers,
which are just not the first ten. We indeed see in Fig. 94 that we have,
in the lines and columns respectively of this border of order 16,

15 + 9 + 10 + 16 = 11 + 12 + 13 + 14
15 + 18 = 16 + 17.
Generally, we shall put at the top n − 1, n − 7, n − 6, n, on the bottom
n − 5, n − 4, n − 3, n − 2, and in the two columns n + 2 (besides n − 1) and
n + 1 (besides n). Leaving out these ten consecutive numbers, the 8k − 12
smaller numbers still to be placed in the border always form groups of
four consecutive numbers, namely 1, . . . , n − 8, thus 4k − 8 numbers, and
n + 3, . . . , 2n − 2, thus 4k − 4 numbers, all of which will be arranged with
neutral placements, in the horizontal rows for the first group and in the
columns for the second. As a result, we have a method which is certainly
easier to remember than the placement of ten numbers, and also easier
to apply: bearing in mind the regularity of the placement in the outer
border, we shall be able not only to repeat it for any outer border, but also
for any inner evenly even border.
Remarks.
(1) The arrangement seen in Fig. 95 follows the previous general method;
but it can also be reduced to placing the first ten numbers.32
(2) In later times this general method is taught using another mnemonic
aid. Starting at the upper corner cell, we put 1 next to it (Fig. 94),
then the subsequent numbers by pairs alternately on the bottom and
at the top until there remain in the top row (excluding the corners)
seven empty cells; we then write three numbers above and four below.
We then fill the upper corners, first left then right. We place the next
number below this last one, then the others alternately by pairs in the
columns, the placing ending with the last on the left. The square of
order 8 then becomes a particular case, with two numbers instead of
three being placed above in the first step.33
B. Construction of bordered squares of evenly odd orders
1. Particular case of order 6
The general method to be explained here may also be applied to order
6. Put 1 on the bottom, (say) next to the left-hand corner, 2 at the top,
3 on the bottom, 4 on the right, the subsequent two in the upper corners,
first left. Putting the next two numbers, 7 and 8, on the left, 9 below
and 10 on the right, we shall have filled three cells in each row and the
complements will fill the whole border (Fig. 100).
5 36 2 34 28 6
7 30
8 29
27 10
33 4
31 1 35 3 9 32
Fig. 100
32
This would be number 92 in our list of solutions (Fig. 99), completed by the neutral
placement 11, . . . , 14.
33
The border given by A (a 24) is, except for inversion of the columns,
that of Fig. 101, thus practically the same. Indeed, in Fig. 100 we have
modified the place of certain numbers within the rows in order for the
arrangement to obey the general rule we shall explain; for a bordered
square, as already said, the place of the elements within the rows (that is,
excepting the corners) is arbitrary, provided the complements are moved
accordingly.
5 36 2 34 28 6 5 36 34 28 2 6
27 10 33 4
8 29 27 10
7 30 8 29
33 4 7 30
31 1 35 3 9 32 31 1 3 9 35 32
Fig. 101 Fig. 102

As for B (B.18), after filling the inner square of order 4, he considers
the numbers still to be placed,
1 2 3 4 5 6 7 8 9 10
36 35 34 33 32 31 30 29 28 27,
and begins by choosing those of two consecutive corners. It seems to him
preferable to put in the first corner of this square (. . . ) that which equals
the side of the square in which we wish to arrange the numbers, and in the
second corner the number preceding it (B.18v). His choice, which is thus
also that of A, is just made by analogy with the other even borders.34
But B mentions other possibilities, such as 1 and 2, or 9 and 10, and
excludes some (here the manuscript is lacunary).35 The arrangement is
pursued as usual, by choosing numbers which will make up the required
sum. His choices, besides 5 and 6 in the corners, are not indicated; but, of
the two possibilities (b 37, b 38), he will probably have chosen that of Fig.
102, which is the arrangement he adopts when constructing the larger
square of order 8. Thus the 6 × 6 squares in A and B can be said to be
essentially the same.
34
For the other categories of order, the occupants of the upper corner cells of the
border were also chosen relative to the order number: on either side of it if odd; the
number itself and its predecessor if evenly even > 4.
35
In fact, the only condition is that the numbers have different parity. See the
subsequent table in Fig. 106a–106d.
2. Description of a general method in A

Whereas B does not go farther than the order 6, A considers the
general case and thus attests to the existence of a general method for
constructing evenly odd squares in the tenth century. Indeed, he describes
(A.II.39) how to fill any border n = 4k + 2 starting from the (highest)
smaller number arrived at after filling the inner square, which will always
be even.36 He places it below the (for us) right-hand upper corner cell
(Fig. 103), then the next even numbers alternately in the columns until
he has filled half of the inner cells (leaving, as usual, the opposite cells
empty). He puts the next one (odd) below the cell just filled, the next
two (the order of the border and its predecessor) in the right and left
upper corner cells, then resumes the alternate placement of even numbers
in the columns until he reaches 4. As for the odd numbers, they will
fill, in descending sequence, the two lines alternately, starting on the
bottom —just omitting the two numbers already placed and, for the
immediately subsequent number, remaining in the same line— until 3
is placed. Putting 2 above and 1 below, we shall have placed all small
numbers and just be left with adding the complements.
9 100 2 98 5 94 88 15 84 10
83 18
16 85
87 14
12 89
11 90
93 8
6 95
97 4
91 1 99 3 96 7 13 86 17 92
Fig. 103
3. Other description of the general method

The same method will be found in another form in the following cen-
turies.37
36
Since A places the numbers in descending sequence, he begins on the right. The
descending sequence of smaller numbers begins with 18 in this border of order 10.
37
13 196 2 194 5 190 9 186 180 19 176 23 172 14
6 191
10 187
15 182
16 181
20 177
24 173
171 26
175 22
179 18
185 12
189 8
193 4
183 1 195 3 192 7 188 11 17 178 21 174 25 184
Fig. 104
The arrangement will be only slighly modified within the columns
since the numbers are taken in increasing order and the movement adopted
for placing them is different (Fig. 104, n = 14): put 1 (or the first num-
ber to be placed if it is not the outer border) on the bottom, next to the
left-hand corner cell, 2 above, 3 below, then turn around the border for
placing the next numbers, until n − 2 if the (even) order of the border
is n. Place the next two numbers in the upper corners (the one placed
lastly is thus n, for us on the right). Write the next two numbers in the
column on the other side, then resume the same cyclical movement until
half the cells are filled. This is thus applicable from the order 6.
4. Mathematical basis for this method
In order to leave a number of empty cells divisible by 4, and thus com-

plete the border with neutral placements, we may put ten numbers, three
in each row including the corner cells. After placing of the complements,
there will be six numbers in each row, and this will completely fill the
border of order 6 or leave 4(k − 1) empty cells in each row for the order
n = 4k + 2 (k ≥ 2).
With the ten numbers of Fig. 105, we are to find a solution to the system

a+b+c = d+e+f
a + g + h = c + i + j.
a b c
g
h
i
j
d e f
Fig. 105
Restricting ourselves to the first ten numbers, we shall already find 140
possibilities (Fig. 106a–106d). Some of these are represented in Arabic
texts of the eleventh century and later.38
a b c g h i j d e f a b c g h i j d e f
1 1 9 2 6 10 7 8 3 4 5 23 1 7 10 6 8 2 3 4 5 9
2 1 10 2 7 8 5 9 3 4 6 24 1 7 10 5 9 2 3 4 6 8
3 1 8 4 7 9 3 10 2 5 6 25 2 8 3 6 9 4 10 1 5 7
4 1 9 4 7 8 2 10 3 5 6 26 2 9 3 4 10 6 7 1 5 8
5 1 10 4 5 9 3 8 2 6 7 27 2 9 3 4 10 5 8 1 6 7
6 1 10 4 6 8 2 9 3 5 7 28 2 10 3 6 7 4 8 1 5 9
7 1 5 6 8 10 4 9 2 3 7 29 2 10 3 4 9 5 7 1 6 8
8 1 8 6 5 10 3 7 2 4 9 30 2 4 5 8 10 6 9 1 3 7
9 1 9 6 4 10 2 7 3 5 8 31 2 6 5 7 9 3 10 1 4 8
10 1 10 6 4 9 3 5 2 7 8 32 2 7 5 6 9 4 8 1 3 10
11 1 3 8 9 10 5 7 2 4 6 33 2 9 5 3 10 4 6 1 7 8
12 1 5 8 7 10 4 6 2 3 9 34 2 10 5 4 8 3 6 1 7 9
13 1 6 8 7 9 4 5 2 3 10 35 2 10 5 3 9 1 8 4 6 7
14 1 7 8 6 9 3 5 2 4 10 36 2 3 7 8 10 4 9 1 5 6
15 1 7 8 5 10 2 6 3 4 9 37 2 5 7 6 10 3 8 1 4 9
16 1 9 8 6 7 2 4 3 5 10 38 2 6 7 5 10 1 9 3 4 8
17 1 9 8 3 10 2 4 5 6 7 39 2 8 7 4 9 3 5 1 6 10
18 1 10 8 5 7 2 3 4 6 9 40 2 9 7 4 8 1 6 3 5 10
19 1 4 10 8 9 3 5 2 6 7 41 2 10 7 3 8 1 5 4 6 9
20 1 4 10 8 9 2 6 3 5 7 42 2 3 9 7 10 4 6 1 5 8
21 1 5 10 7 9 3 4 2 6 8 43 2 4 9 6 10 1 8 3 5 7
22 1 6 10 7 8 2 4 3 5 9 44 2 7 9 3 10 1 5 4 6 8
Fig. 106a Fig. 106b
38
Les carrés magiques, pp. 142–144, 147 (Russian edition, pp. 154–155, 159).
a b c g h i j d e f a b c g h i j d e f
45 2 8 9 5 7 1 4 3 6 10 93 4 6 9 3 8 1 5 2 7 10
46 3 5 4 6 10 7 8 1 2 9 94 4 8 9 2 7 1 3 5 6 10
47 3 6 4 7 8 5 9 1 2 10 95 5 2 6 7 8 4 10 1 3 9
48 3 8 4 6 7 2 10 1 5 9 96 5 9 6 1 7 3 4 2 8 10
49 3 9 4 2 10 5 6 1 7 8 97 5 1 8 6 9 2 10 3 4 7
50 3 9 4 5 7 1 10 2 6 8 98 5 3 8 6 7 1 9 2 4 10
51 3 10 4 5 6 2 8 1 7 9 99 5 4 8 3 9 2 7 1 6 10
52 3 10 4 5 6 1 9 2 7 8 100 5 4 8 2 10 3 6 1 7 9
53 3 4 6 7 9 5 8 1 2 10 101 5 6 8 1 9 3 4 2 7 10
54 3 5 6 7 8 2 10 1 4 9 102 5 7 8 3 6 2 4 1 9 10
55 3 7 6 4 9 2 8 1 5 10 103 5 1 10 6 8 2 7 3 4 9
56 3 7 6 5 8 1 9 2 4 10 104 5 2 10 4 9 1 7 3 6 8
57 3 8 6 2 10 4 5 1 7 9 105 5 3 10 4 8 1 6 2 7 9
58 3 10 6 1 9 2 5 4 7 8 106 5 6 10 2 7 1 3 4 8 9
59 3 4 8 5 10 1 9 2 6 7 107 6 1 7 4 10 5 8 2 3 9
60 3 5 8 4 10 2 7 1 6 9 108 6 1 7 5 9 3 10 2 4 8
61 3 6 8 4 9 1 7 2 5 10 109 6 2 7 5 8 3 9 1 4 10
62 3 7 8 2 10 1 6 4 5 9 110 6 2 7 3 10 4 8 1 5 9
63 3 1 10 8 9 4 6 2 5 7 111 6 3 7 4 8 2 9 1 5 10
64 3 2 10 7 9 4 5 1 6 8 112 6 3 7 4 8 1 10 2 5 9
65 3 2 10 7 9 1 8 4 5 6 113 6 4 7 3 8 1 9 2 5 10
66 3 4 10 6 8 2 5 1 7 9 114 6 4 7 1 10 2 8 3 5 9
67 3 4 10 5 9 1 6 2 7 8 115 6 8 7 3 4 1 5 2 9 10
68 3 5 10 6 7 2 4 1 8 9 116 6 9 7 1 5 2 3 4 8 10
69 3 6 10 5 7 1 4 2 8 9 117 6 1 9 5 8 3 7 2 4 10
70 3 8 10 4 6 1 2 5 7 9 118 6 1 9 3 10 2 8 4 5 7
71 4 2 5 8 9 6 10 1 3 7 119 6 2 9 5 7 1 8 3 4 10
72 4 3 5 6 10 7 8 1 2 9 120 6 4 9 3 7 2 5 1 8 10
73 4 7 5 2 10 3 8 1 6 9 121 6 5 9 1 8 2 4 3 7 10
74 4 7 5 3 9 1 10 2 6 8 122 6 7 9 2 5 1 3 4 8 10
75 4 8 5 2 9 3 7 1 6 10 123 7 1 8 3 9 5 6 2 4 10
76 4 8 5 1 10 3 7 2 6 9 124 7 1 8 2 10 5 6 3 4 9
77 4 9 5 2 8 3 6 1 7 10 125 7 2 8 5 6 1 9 3 4 10
78 4 9 5 3 7 1 8 2 6 10 126 7 2 8 1 10 4 6 3 5 9
79 4 10 5 3 6 1 7 2 8 9 127 7 3 8 1 9 4 5 2 6 10
80 4 10 5 1 8 2 6 3 7 9 128 7 5 8 2 6 3 4 1 9 10
81 4 2 7 6 10 5 8 1 3 9 129 7 6 8 3 4 1 5 2 9 10
82 4 3 7 6 9 2 10 1 5 8 130 7 1 10 3 8 2 6 4 5 9
83 4 5 7 3 10 2 8 1 6 9 131 7 1 10 2 9 3 5 4 6 8
84 4 5 7 3 10 1 9 2 6 8 132 7 2 10 1 9 3 4 5 6 8
85 4 6 7 3 9 1 8 2 5 10 133 7 3 10 1 8 2 4 5 6 9
86 4 6 7 2 10 1 8 3 5 9 134 8 1 9 3 7 4 5 2 6 10
87 4 8 7 1 9 2 5 3 6 10 135 8 1 9 4 6 2 7 3 5 10
88 4 9 7 3 6 1 5 2 8 10 136 8 2 9 4 5 1 7 3 6 10
89 4 1 9 6 10 3 8 2 5 7 137 8 2 9 3 6 1 7 4 5 10
90 4 2 9 5 10 3 7 1 6 8 138 8 3 9 1 7 2 5 4 6 10
91 4 3 9 6 8 2 7 1 5 10 139 9 1 10 2 6 3 4 5 7 8
92 4 5 9 2 10 1 6 3 7 8 140 9 2 10 3 4 1 5 6 7 8
Fig. 106c Fig. 106d

Remark. As in the case of evenly even orders, it would seem at first sight
simpler to put in the first ten numbers and then continue with placing
the subsequent numbers by groups of four; as a matter of fact, it is
more convenient to learn a continuous movement which can just be
repeated in all other evenly odd borders.
Let us first examine which are the ten numbers belonging to the pre-
liminary placement. In Fig. 104 (n = 14, thus k = 3), we see that

13 + 2 + 14 = 1 + 11 + 17
13 + 15 + 16 = 14 + 12 + 18 ;
generally, the three occupants of the upper row will be n − 1, 2, n, and
those of the lower row 1, n − 3, n + 3; in the columns, repeating the
numbers in the upper corner cells, we find n − 1, n + 1, n + 2, and n,
n − 2, n + 4. With the complements, each row will contain its sum due,
3(n2 + 1), and we shall be left with placing groups of four numbers in
discontinuous sequences.
Consider, then, these discontinuous sequences. In the horizontal rows,
it appears that, disregarding the preliminary placement, the upper row
contains, distributed on its two sides, the k − 1 pairs 5 & 2n − 5, 9 &
2n − 9, . . . , n − 5 & 2n − (n − 5), adding up altogether to (k − 1)2n
and leaving 2(k − 1) empty cells. Similarly, the bottom row receives the
k − 1 pairs 3 & 2n − 3, 7 & 2n − 7, . . . , n − 7 & 2n − (n − 7), here again
adding up to (k − 1)2n and leaving 2(k − 1) empty cells. Since these two
sums are equal, so will be the sums of the complements, the placing of
which will thus equalize the two horizontal rows. Indeed, the sum of the
k − 1 pairs in one row together with the 2(k − 1) complements of the
other makes 2(k − 1)(n2 + 1), and, with the 3(n2 + 1) of the preliminary
placement, (2k + 1)(n2 + 1) = Mn since n = 4k + 2. We may also consider
these discontinuous sequences as forming neutral placements, each group
of four terms consisting of two pairs in arithmetical progression, namely
3, 5, 2n − 5, 2n − 3, then 7, 9, 2n − 9, 2n − 7, and so on to n − 7, n − 5,
2n − (n − 5), 2n − (n − 7).
In the columns, after the initial placement, the pairs to be placed, all
of them even numbers, are, for one column, 6 & 2n − 4, 10 & 2n − 8,
. . . , n − 4 & 2n − (n − 6), and for the other 4 & 2n − 2, 8 & 2n − 6, . . . ,
n − 6 & 2n − (n − 8); in both cases, the sum resulting from these k − 1
pairs is (k − 1)(2n + 2), and writing the complements of one column in
the 2(k − 1) empty cells of the other will make 2(k − 1)(n2 + 1), thus, with
the 3(n2 + 1) of the initial placement, Mn . Here again, the discontinuous
sequences form neutral placements, namely 4, 6, 2n − 4, 2n − 2, then 8,
10, 2n − 8, 2n − 6, and so on to n − 6, n − 4, 2n − (n − 6), 2n − (n − 8).

Remark. The arrangement in the square of 6 seen in Fig. 100–102 follows
the previous general method, but can also be reduced to placing the
first ten numbers (number 95 in our list of solutions).
Chapter III. Separation by parity
This admirable construction also involves bordered squares, but with
the odd numbers all contained within a rhomb having its corners on the
middle of each side of the main square whereas all even numbers fill the
cells outside the rhomb (Fig. 107). This configuration naturally occurs in
the square of order 3 (Fig. 108), which certainly led to the search for an
analogous configuration in odd squares of higher orders. Such an arrange-
ment must have been well known in the 10th century since both authors
present it (A.II.7–35, B.20–23) and B asserts that it is used by mathe-
maticians (B.20i). But their treatment is very different. Whereas B has
no real method of filling, leaving much room for trial and error, A sets
out precise rules, thereby avoiding any hesitation or repeated attempts.
In other words, we learn from A a true method of placement since, once
the (odd) order is known, the construction can be performed right away.
Indeed, filling the rhomb with the odd numbers follows precise rules re-
lated to those seen for common bordered squares of odd orders, while for
the even numbers a precisely determined initial placement is followed by
groups of neutral placements, just as in the construction of even-order
bordered squares. But in this case the initial placement is, by far, more
complex to determine than for bordered even-order squares.
Remark. Even if A describes how to fill such a square, he surely in no
way contributed to the development of this theory.
2 9 4
ev
en
en
ev
odd 7 5 3
6 1 8
ev
en
en
ev
Fig. 107 Fig. 108

§1. Structure of the rhomb according to B
Take, generally, a square of odd order n = 2k + 1, which therefore
contains n2 = 4k2 + 4k + 1 cells. The rhomb will contain as many cells
as there are odd numbers, thus 2k2 + 2k + 1. The quantity of 2k2 + 2k
even numbers will be equally divided among the corners, thus 12 k(k + 1)
elements in each corner —an integer, since the product of k by k + 1 is
even.
A. The rows of the rhomb
B considers this rhomb as an oblique square, of which he describes
the length of the rows, starting with the least possible order of the main
68 Structure of the rhomb according to B
square (B.21i).39 As B observes, the side of the rhomb within the square
of order 3 is 2, that within the square of order 5 is 3, and so on; thus,
when we wish to know the side of such an oblique square, we add 1 to the
side of the (main) square and take half the result. As to the inner rows
of the rhomb, they are alternately equal to this quantity and less by one.
In our terms, if the order of the main square is n = 2k + 1, the rhomb
will have 2k + 1 rows containing alternately k + 1 and k cells, beginning
with k + 1 in the lateral rows.
B. Largest square within the rhomb
In the square of order 3, the rhomb includes the central cell (Fig. 108).
For higher orders, the rhomb will also include a largest inner square. But,
as B observes (B.21ii), each square (of this kind) appears (inside the
main square) according to the succession of consecutive odd numbers for
a couple of squares. Thus, he goes on, the largest inner square for orders
5 and 7 will have side 3, for orders 9 and 11, 5, and so on. In our terms,
the largest inner square will have the order 2t + 1 both for n = 4t + 1 and
n = 4t + 3, that is, it will remain the same if, when n = 2k + 1, k takes
two consecutive values, first even then odd. The only difference displayed
by the inner square common to this pair of orders is that its corner cells
are cut by the sides of the rhomb in the first case and fully enclosed in the
second (Fig. 109 & Fig. 110, t = 2, thus with the two consecutive orders
9 and 11 of the main square); it is then the triangular parts of these two
rhombs which contain different quantities of odd numbers. On the other
hand, for the order n = 4t + 3 and the following n = 4(t + 1) + 1, the inner
squares will differ, the second having one more border, while the number
of cells in the triangular parts of each rhomb will be the same (Fig. 110
& Fig. 111, with, first, t = 2 and n = 4t + 3, then t = 3 and n = 4t + 1,
thus with the two consecutive orders 11 and 13).
In short, of three consecutive odd orders 4t+1, 4t+3, 4t+5, the rhombs
of the first two contain largest inner squares of equal size, but surrounded
by, respectively, t and t + 1 borders; the largest inner squares for the
last two orders will be of different sizes, namely 2t + 1 and 2(t + 1) + 1
respectively, but will leave triangles of the same size in each of their
rhombs, both inner squares being surrounded by t + 1 borders. This is
exemplified by our three figures.
39
Even if the cells of the four lateral rows of the rhomb are cut by its sides, they
will contain a number since their centre is within the rhomb.
Separation by parity 69
Fig. 109 Fig. 110 Fig. 111

Remark. All that appears from the text of B. Although this structure is
essential since we are later on to differentiate between these squares,
A says nothing about it.
§2. The partially empirical construction of B
Placing the numbers in the inner square is easy (B.21iii): we shall
2
put n 2+1 (odd) in the central cell, then fill the borders successively with
the subsequent smaller and larger numbers, just as we did with bordered
squares, but here employing only odd numbers. The largest inner square
will thus be occupied by two sequences of consecutive odd numbers on
2
either side of n 2+1 , and it will display the characteristics of a bordered
square.
43 83 81 51 47
73 55 69 59 49
77 65 61 57 45
37 63 53 67 85
75 39 41 71 79
Fig. 112
Remark. If we wish, as with other bordered squares, to start in the
outer border of this inner square and place ascending sequences of
smaller numbers (that border being at least of order 5), we shall put
70 The partially empirical construction of B
the smallest number above the lower left-hand corner cell; this number
will be 4t2 + 1 if n (≤ 9) has the form 4t + 1, but 4(t + 1)2 + 1
for the form 4t + 3. Next we shall put the subsequent odd numbers
alternately around the corner, then in the middle and corner cells,
finally alternately on both sides of the opposite corner, just as we did
for usual bordered odd-order squares.
This is exemplified in Fig. 112, of order n = 11. Since n has the form
4t + 3 with t = 2, the smallest term placed on the outer border will be
37.
At this point B’s methodical construction stops and gives way to plac-
ing by trial and error. That is used for the whole part outside the in-
ner square, and thus applied to the odd numbers not yet placed and all
the even numbers. The configurations of two of his three examples are
known.40
He thus first constructs the squares of order 5 and 7 after filling their
inner square of order 3. For order 5 (B.22ii), he tells us to put in the
(upper) corners the even pairs 2 and 4, or 2 and 6, or 2 and 8, or 2 and 12,
then their complements; he next eliminates the differences by means of
the remaining even and odd numbers. There are various possibilities, he
says: we shall obtain numerous figures, which would lengthen the book
if we were to report them. We shall return to that at the end of this
chapter.
34 48 4 5 6 42 36
10 24 37 39 3 22 40
12 9 19 33 23 41 38
43 49 29 25 21 1 7
30 15 27 17 31 35 20
32 28 13 11 47 26 18
14 2 46 45 44 8 16
Fig. 113
For order 7 (B.22iii, Fig. 113), he writes the even pair 24, 22 in the
corners of the first border, then their complements, and makes up the
required sum by means of the remaining odd numbers. Finally, he places
the four still available odd numbers in the rhomb’s corners, thus 5 and
40
His figure for order 5 is missing, as are all other figures, but those for orders 7 and
9 are fully described in the text.
7 in two consecutive corners, and searches for six even numbers adding
up to 170 for the upper row, then four others adding up to 84 for the
(our) left-hand column. It is just the same path followed in his empirical
construction of odd-order bordered squares.
8 80 78 76 75 12 14 16 10 8 80 78 76 75 12 14 16 10
67 22 64 62 61 26 28 24 15 22 64 28 24
69 55 32 52 51 36 34 27 13 23 63 61 31 27
71 57 47 38 45 40 35 25 11 53 35 49 39 29
73 59 49 43 41 39 33 23 9 73 57 45 41 37 25 9
5 19 29 42 37 44 53 63 77 17 43 33 47 65
3 17 48 30 31 46 50 65 79 55 19 21 51 59
1 58 18 20 21 56 54 60 81 58 18 54 60
72 2 4 6 7 70 68 66 74 72 2 4 6 7 70 68 66 74
Fig. 114 Fig. 115

But proceeding thus will not lead him to a general method as it did
with bordered squares. He merely suggests (B.23) a simplification for
filling cells outside the inner square, which he illustrates in his third
example (n = 9): we are to construct a (usual) bordered square having
the same order and transfer to correspondingly situated empty cells of
the target square the numbers meeting the condition of parity —provided
that the odd numbers have not already been used for the inner square
(Fig. 114 et Fig. 115).
The remaining empty cells are filled as before, by calculating the dif-
ference with the sum due and choosing numbers accordingly. B constructs
in this way Fig. 116.
8 80 78 76 75 12 14 16 10
26 22 64 79 69 1 28 24 56
30 42 23 63 61 31 27 40 52
32 77 53 35 49 39 29 5 50
73 15 57 45 41 37 25 67 9
36 11 17 43 33 47 65 71 46
44 62 55 19 21 51 59 20 38
48 58 18 3 13 81 54 60 34
72 2 4 6 7 70 68 66 74
Fig. 116
72 Filling the rhomb according to A
Remark. This auxiliary construction is of limited use: considering how

the method for odd-order bordered squares distributes the numbers by
parity, it will enable us to fill with even numbers only the upper and
lower cells between the diagonals.
§3. Filling the rhomb according to A.
A fills the cells of the inner square just as B does (A.II.8). However,
for the remainder of the rhomb, his filling leaves no room for trial and
error —although that is not made evident by his treatment since he merely
enumerates the occupants and tells us where to put them (A.II.9–A.II.11).
213
11 189 207
7 35 173 183 203
3 31 51 165 167 179 199
201 75 159 155 153 83 87 79 25
205 181 141 95 135 133 103 99 85 45 21
209 185 169 145 125 107 121 111 101 81 57 41 17
211 187 171 163 149 129 117 113 109 97 77 63 55 39 15
9 33 49 69 89 115 105 119 137 157 177 193 217
5 29 65 127 91 93 123 131 161 197 221
1 147 67 71 73 143 139 151 225
223 195 175 61 59 47 27
219 191 53 43 23
215 37 19
13
Fig. 117
In order to make clear the procedure and memorize it, we shall refer
to the method taught earlier for odd-order bordered squares, with its
alternate movement placing the successive numbers in two contiguous
rows. Here the rhomb will be considered as a square and its lateral rows
as borders, the starting point, with 1, being the cell following the middle
of the rhomb’s side, or else, if preferred, the cell next to the lower corner
of the inner square (on the left for us, on the right for the author). We
shall place the odd numbers alternately in the upper halves of the two
contiguous left-hand lateral rows until we arrive at the two corner cells,
which we leave empty whereas we fill the other two, first the one below,
and resume the alternate placing from the second one, but this time in
the right-hand lower halves, until we reach the sides of the inner square.
We do the same for the next oblique borders. The placement will end
with the two cells on the diagonals of the rhomb and next to the inner
square, which we shall fill, following the same movement, first the one
on the bottom then the one on the right. Finally, we shall place the
complements of the numbers already put in —considering, of course, the
borders of the main square. From Fig. 117 & Fig. 118 (n = 15, thus
order 4t + 3, and n = 17, thus order 4(t + 1) + 1) the perfect regularity of
the procedure appears. As we have seen (p. 68), the triangular parts of
the rhombs are the same for these two consecutive orders, and thus will
contain the same smaller odd numbers.
277
11 253 271
7 35 237 247 267
3 31 51 229 231 243 263
79 223 219 215 213 87 91 95 83
265 197 107 191 187 185 115 119 111 93 25
269 245 201 173 127 167 165 135 131 117 89 45 21
273 249 233 205 177 157 139 153 143 133 113 85 57 41 17
275 251 235 227 209 181 161 149 145 141 129 109 81 63 55 39 15
9 33 49 73 101 121 147 137 151 169 189 217 241 257 281
5 29 69 97 159 123 125 155 163 193 221 261 285
1 65 179 99 103 105 175 171 183 225 289
207 67 71 75 77 203 199 195 211
287 259 239 61 59 47 27
283 255 53 43 23
279 37 19
13
Fig. 118
Remark. It is important to note, once again, that this direct filling does
not appear clearly from the instructions given in A. As already said,
we are told where to put, cell by cell and order by order, beginning
with the inside border in the main square, the smaller odd numbers
taken (as he usually does) in descending succession and, opposite, the
74 Placing the even numbers around the rhomb
complements. The regularity of the filling would have been better

revealed by considering, as we did, the oblique rows of the rhomb
and starting from the outer ones with the smaller numbers taken in
ascending succession. As to B, although he has —unlike A— described
the structure of the rhomb and, in particular, explained how it can be
seen as an oblique square with its own rows, he does not use it to
place the odd numbers. Obviously, an earlier theory was incompletely
transmitted or applied.
§4. Placing the even numbers around the rhomb
A. Situation after filling the rhomb
The treatment for orders n = 4t + 1 and n = 4t + 3 has so far been the
same. But it now becomes necessary to differentiate between these two
orders, as well as the particular case n = 5 (A.II.12). Of course, before
placing the even numbers, we are to consider the present situation. That
is, we must determine, for each border, both the number of cells still
empty in each row and the sum this row already contains. We shall then
know its excess or deficit relative to the sum due for the cells already
filled, and the number of cells at our disposal to eliminate the difference
and complete this border’s rows. Now A teaches all that, even if rather
too concisely.
Fig. 119 Fig. 120

1. Determining the number of remaining empty cells
If the order is n = 4t + 1 (Fig. 119), the situation as described by
A is the following (A.II.13): There remain as empty cells in the first
border, which surrounds the inner square, the four at the corners and
eight adjacent to them, two on each side; in the second (border), the
empty cells are (the four) at the corners and twenty-four cells adjacent to
them. It will always be like that: each border has, excepting the corner
cells, 16 more (empty cells) than the preceding border. In short, A tells
us that the first incomplete border comprises 4 + 8 empty cells, the next
(second) border 4 + 8 + 16, and so on by adding each time 16 to the
number of empty cells in the previous border.
If the order is n = 4t+3 (Fig. 120), the situation is as follows (A.II.16):

From the first border, which surrounds the inner square, there remain as
empty the four corner cells; from the second border, the corner cells
(and) 16 cells adjacent to them, 4 on each side; from the third border,
the corner cells (and) 32 empty cells adjacent to them. And so on always:
each border has, excepting the corner cells, a quantity of 16 more than
the one before. The first incomplete border thus comprises 4 empty cells,
the second 4 + 16, and so on by adding each time 16.
Written in the form of a table, the number of empty cells in each
border p cut by the rhomb, starting with the innermost (p = 1), is:
p=1 p=2 p=3 p=4 p=5
n = 4t + 1 4+8 4 + 24 4 + 40 4 + 56 4 + 72
n = 4t + 3 4 4 + 16 4 + 32 4 + 48 4 + 64
Fig. 121
In our terms, this means that:
(1) In a square of order 4t + 1, with an inner square of order 2t + 1
completely filled and thus with t borders partly filled, the pth border
starting from the inner square contains 16p − 4 empty cells, 4p in each of
its rows (p = 1, . . . , t).
(2) In a square of order n = 4t + 3, with again an inner square of order
2t + 1 and thus t + 1 borders partly filled, the pth border contains 16p − 12
empty cells, 4p − 2 in each of its rows (p = 1, . . . , t + 1).
2. Determining the sum required
We know that since the magic sum for a square of order n is Mn =
n2 +1 2
n· the sum due for m cells in one row is m · n 2+1 (A.II.15).
2 ,
Let us examine the previous placement of odd numbers. We shall
disregard the inner square, since each of its rows makes the sum due,
and thus focus on the rows belonging to incomplete borders. From the
arrangement already carried out, we know that, for the cells already filled,
the upper rows in each incomplete border will display an excess over the
sum due, as also the left-hand columns (the right-hand ones in the original
text). The opposite rows, being filled with complements, will show a
deficit of equal amount.
Let us thus consider the incomplete borders starting, as before, from
the border surrounding the inner square. Let us designate by ΔpL, 4t+1 the
excess of the upper line belonging to the pth border if the order of the
main square is n = 4t + 1, and by ΔC, p
4t+1 that of the left-hand column
of the same border; likewise, ΔpL, 4t+3 and ΔC, 4t+3 will be the excesses of
p
the upper line and of the left-hand column in the pth border when the
main square has the order n = 4t + 3. A gives all necessary information
about the excesses or deficits for the orders n = 4t + 1 and n = 4t + 3.
But with the concision of expression and in the absence of a table the
general relation is by no means evident.
(1) Consider first the order n = 4t + 1. From the instructions given
in A.II.14 we know the first elements and may then draw up the table
displaying the differences presented by the lines and columns according
to the number p of the border (we include n = 5):
p=1 p=2 p=3 p=4 p=5
n = 5 ( t = 1) 12 10
n = 9 ( t = 2) 20 18 36 34
n = 13 (t = 3) 28 26 52 50 76 74
n = 17 (t = 4) 36 34 68 66 100 98 132 130
n = 21 (t = 5) 44 42 84 82 124 122 164 162 204 202
Fig. 122
From this table appears the rule of formation for the differences with
increasing order n = 4t + 1 (thus increasing t) and increasing size of the
border p:
ΔL,
1
4t+1
= 8t + 4
ΔL,
2
4t+1
= 16t + 4
ΔL,
3
4t+1
= 24t + 4
ΔL,
4
4t+1
= 32t + 4
generally
ΔL,
p
4t+1 = 8pt + 4.
As for the columns of the same borders, their differences happen to be

less than those of the lines by 2, and we shall thus have
ΔC,
1
4t+1
= 8t + 2
ΔC,
2
4t+1
= 16t + 2
generally
ΔC,
p
4t+1 = 8pt + 2.
We may now return to the text in A.II.14 and note that its indications
(with values given for n = 9, n = 13, and being told how to infer those for
n = 17) are indeed sufficient for drawing up the above table. Remember,
too, that in the Arabic text the columns in excess are the right-hand ones.
After placing the odd numbers as indicated, the first upper row of the
square of 9 is in excess over its sum due, required for it, by 20 (ΔL,
1
9
=
8 · 2 + 4), the lower one is in deficit by 20; the first right-hand row is in
excess over its sum due by 18 (ΔC, 1
9
= 8·2+2), and the left-hand one is in
deficit by 18. The second upper (row) is in excess by 36 (ΔL, 2
9
= 16·2+4),
and the lower one is in deficit by the same amount; the second right-hand
(row) is in excess by 34 (ΔC, 2
9
= 16 · 2 + 2), and the left-hand one is
in deficit by the same amount. For the square of 13, the first upper
(row) is in excess by 28 (ΔL, 1
13
= 8 · 3 + 4), and the lower one is in
deficit by the same amount; the first right-hand (row) is in excess by
26 (ΔC, 1
13
= 8 · 3 + 2), and the left-hand one is in deficit by the same
amount. The second upper (row) is in excess by 52 (ΔL, 2
13
= 16 · 3 + 4),
and the lower one is in deficit by the same amount; the second right-hand
(row) is in excess by 50 (ΔC, 2
13
= 16 · 3 + 2), and the left-hand one is
in deficit by the same amount. The third upper (row) is in excess by 76
(ΔL,3
13
= 24 · 3 + 4), and the lower one is in deficit by the same amount;
the third right-hand (row) is in excess by 74 (ΔC, 3
13
= 24 · 3 + 2), and the
left-hand one is in deficit by the same amount. Similarly for the others.
(In the square of) 9, the second (row) has 16 more than the (row)
before (ΔL, 2
9
− ΔL,1
9
= ΔC, 2
9
− ΔC, 1
9
= 8 · 2). (In the square of) 13, the
first (row) has 8 more than the first (of the square) of 9 (ΔL, 1
13
− ΔL,1
9
=
ΔC,
1
13
− Δ C, 9
1 = 8), then each row has 24 more than the (row) before
(ΔL,i
13
− Δ L, 13
i−1 = Δi
C, 13
− ΔC, 13
i−1 = 8 · 3). (In the square of) 17, the
first (row) has 8 more than the first of (the square of) 13 (ΔL, 1
17
−
L, 13 C, 17 C, 13
Δ1 = Δ1 − Δ1 = 8), then each row 32 more than the (row)
before (ΔL, i
17
−Δ L, 17
i−1 = Δ C, 17
i −ΔC, 17
i−1 = 8·4). Then each row will always
have more than the one before in the same manner (ΔL, i
4t+1
− ΔL, 4t+1
i−1 =
C, 4t+1 C, 4t+1
Δi − Δi−1 = 8 · t).
(2) Consider now the order n = 4t + 3. The instructions given in A.II.17

enable us to set out and extend the table displaying the differences pre-
sented by the lines and columns according to border and order.
p=1 p=2 p=3 p=4 p=5

n = 7 ( t = 1) 4 2 20 18
n = 11 (t = 2) 4 2 28 26 52 50
n = 15 (t = 3) 4 2 36 34 68 66 100 98
n = 19 (t = 4) 4 2 44 42 84 82 124 122 164 162
Fig. 123
From this table appears, as before, the rule of formation for the differences
with increasing order n = 4t + 3 and increasing size of the border p:
ΔL,
1
4t+3
=4
ΔL,
2
4t+3
= 8(t + 1) + 4
ΔL,
3
4t+3
= 16(t + 1) + 4
ΔL,
4
4t+3
= 24(t + 1) + 4
generally
ΔL,
p
4t+3 = 8(p − 1)(t + 1) + 4.
As for the columns of the same borders, their differences are again smaller
by 2 than those of the lines:
ΔC,
1
4t+3
=2
ΔC,
2
4t+3
= 8(t + 1) + 2
generally
ΔC,
p
4t+3 = 8(p − 1)(t + 1) + 2.
The text in A.II.17 now becomes clear. As before, it gives us all values
for the first two orders, n = 7 and n = 11, and tells us how to determine
those of the subsequent ones.
In this class of squares, the first upper row has an excess of 4 over
its sum due (ΔL, 1
4t+3
= 4), and the lower one has a deficit of the same
amount; the first right-hand (row) has an excess of 2 over its sum due
(ΔC,
1
4t+3
= 2), and the left-hand (row) has a deficit of the same amount.
The second upper row of the (square of) 7 has an excess of 20 over its
sum due (ΔL, 2
7
= 8 · 2 + 4), and the second right-hand (row an excess of)
C, 7
18 (Δ2 = 8 · 2 + 2). In the square of 11, the second upper (row) has an
excess of 28 over its sum due (ΔL, 2
11
= 8 · 3 + 4), the second right-hand
C, 11
(row an excess of) 26 (Δ2 = 8 · 3 + 2), the third upper (row an excess
L, 11
of) 52 (Δ3 = 16 · 3 + 4), the third right-hand (row an excess of) 50
(ΔC,
3
11
= 16 · 3 + 2), and all the opposite (rows) have a deficit equal
to the excess. (Square of) 15: the second upper (row) exceeds the first
(upper row) by 32 (ΔL, 2
15
− ΔL,1
15
= 8 · 4), the third the second by 32
L, 15 L, 15
(Δ3 − Δ2 = 8 · 4). And so on for the others: the excess increases

each time by 8 ΔL, i
4t+3
− ΔL, 4t+3
i−1 = ΔC,
i
4t+3
− ΔC, 4t+3
i−1 = 8(t + 1) .
B. Rules for placing the even numbers
We now know, for each row of an incomplete border, the (even) num-
ber of empty cells it contains and its difference with the sum due. We are
then to proceed as follows.
(i) First, we must, by means of a preliminary placement, eliminate the
difference in each row; that is, the cells already occupied by odd numbers
and those about to receive even numbers during this preliminary place-
ment must display their sum due, thus, for the order n, as many times
n2 +1
2 as the number of occupied cells.
(ii) This preliminary step must be carried out in a uniform manner so
that it may be applied to all squares of the same kind of order, namely
n = 4t + 1 or n = 4t + 3.
(iii) It must be carried out with the least possible number of cells in
order to be applicable to the smallest orders, namely n = 9 and n = 7,
respectively (order 5 has a treatment of its own).
(iv) It must settle the question of the corner cells, each common to two
rows.
(v) It must leave a number of still empty cells divisible by 4.
(vi) The remaining even numbers must form groups of four, or pairs of,
numbers consecutive or in arithmetical progression, for this will permit
completion of the treatment by means of neutral placements.
Note that this procedure is not specific to this type of square for, as
we have seen, the same principles underlay the filling of bordered squares
of even orders (placing even numbers only makes no difference). It is
therefore clear that the construction of squares with separation by parity was
discovered once the construction of bordered squares had been not only found
but fully mastered.
Seven rules will be used for filling the rows: five concern the prelim-
inary placement, whereby the excess or deficit is eliminated, while the
other two serve to complete by neutral placements the rows thus equal-
ized. We shall begin with these last two. The first has already been used,
but with consecutive numbers, for common bordered squares of the two
types of even order.
α n2 + 1 − (α + 2) n2 + 1 − (α + 4) α+6 =⇒ 2(n2 + 1)
n2 + 1 − α α+2 α+4 n2 + 1 − (α + 6) =⇒ 2(n2 + 1)
Fig. 124
Rule I. Consider a sequence of four consecutive even numbers

α, α + 2, α + 4, α + 6. If we place the extreme terms in a
row and the middle ones in the opposite row, then put the
complements facing them, not only will the sum due be found
in two opposite cells, but also in the four cells in each row
(Fig. 124).
A describes, or rather hints at, this neutral placement as follows

(A.II.20): If you take four (even, consecutive) small numbers, the

sum of
the first and the second has a certain deficit n2 + 1 − (2α + 2) , and the

sum of the third and the fourth a deficit smaller by 8 n2 + 1 − (2α + 10) .
Then adding the first and the fourth you will find thatthey have a deficit
of half the sum of the two deficits n2 + 1 − (2α + 6) ; adding similarly
the second and the third gives the same result. (. . . ) The knowledge of
this is necessary to you, for you will use it constantly.
This equality of the deficits of extreme and middle terms still holds
if the four numbers form another arithmetical progression, or form two
pairs obeying the same arithmetical progression, that is:
Rule II. Consider two pairs of numbers α, α + 2s, β, β + 2s.

Then their arrangement as above and the placement of the
complements will produce neither excess nor deficit (Fig. 125).
α n2 + 1 − (α + 2s) n2 + 1 − β β + 2s =⇒ 2(n2 + 1)
n2 + 1 − α α + 2s β n2 + 1 − (β + 2s) =⇒ 2(n2 + 1)
Fig. 125
As we read in A.II.31, you put the first of the first pair and the second
of the second pair on one side, the second of the first pair and the first of
the second pair on the facing side, and opposite to each its complement.
Before using these neutral placements, we shall have to eliminate the
differences subsequent to placing the odd numbers. This will be done
by means of the five rules below. But let us first establish, as A does
(A.II.18), the list of the even numbers to be placed, which will make
them easier for us to use. For the order n = 2k + 1, there are 2k2 + 2k,
2
namely k(k + 1) smaller numbers (less than n 2+1 ) and their complements,
thus k(k + 1) larger numbers. Since k(k + 1) is even, all these numbers,
smaller and larger, may be arranged as pairs of consecutive even numbers
as in Fig. 126, where we have also numbered these pairs starting from the
highest smaller numbers; the jth pair of smaller numbers then has the
2 2
form n −8j+3
2 , n −8j+7
2 , with j admitting the values 1, 2, . . . , 12 k(k + 1).
We shall try, as far as possible, to equalize the rows without breaking
these pairs of numbers since that will facilitate later neutral placements
by means of Rules I and II. Remember, too, that since the difference with
the sum due is the same (but with opposite signs) in two opposite rows,
we may carry out the equalization for just one row.
n2 −8j+3 n2 −8j+7 n2 −5 n2 −1
2 4 | ... | 2 2
| ... | 2 2
n2 +8j−1 n2 +8j−5 n2 +7 n2 +3
n2 − 1 n2 − 3 | . . . | 2 2
| ... | 2 2
1
2
k(k + 1) | ... | j | ... | 1
Fig. 126
The subsequent Rule III arises from the limitation of Rule II to just two
terms.
α n2 + 1 − (α + 2s) =⇒ n2 + 1 − 2s
n2 + 1 − α α + 2s =⇒ n2 + 1 + 2s
Fig. 127
Rule III. With a number α written in one row then α+ 2s in

the opposite row, and next, facing them, their complements,
the row of α will contain a deficit of 2s and the opposite row
an excess of 2s (Fig. 127).
It appears that by means of this rule we can eliminate any even dif-
ference, thus any difference which may occur here. But this will have to
be used only if absolutely necessary since we are to maintain, as far as
possible, groups of four, or two, consecutive even numbers to complete
the placement. This explains why the subsequent rules will throughout
keep the pairs of consecutive numbers unbroken.
α n2 + 1 − (α + 2) =⇒ n2 + 1 − 2
n2 + 1 − α α+2 =⇒ n2 + 1 + 2
Fig. 128
Rule IV. Any pair of consecutive numbers of which the

terms are written in two opposite rows will, after placing of
the complements, produce a difference of 2, which will be a
deficit in the row containing the lesser element of the pair
(Fig. 128).
Rule IV is merely a particular case, with s = 1, of Rule III. That is

why A mentions them together (A.II.21): When you write the first small
number, or any small number, on one side (our α) and the subsequent
number on the other, opposite side (α + 2), and you write the two large
numbers which are their complements opposite to them, the side where
the first small number has been written will be less than its sum due by
2, whereas the other side will be in excess by 2. If you write some small
number on one side (α) and you write on the other, opposite side the
third small number counted from this one (α + 4), the side containing
the first small number will be less than its sum due by 4. And so on:
whenever you increase the distance between these two by one number,
the deficit is always increased by 2.
α n2 + 1 − (α + 4) =⇒ n2 + 1 − 4
α+2 n2 + 1 − (α + 2)
n2 + 1 − (α + 6) α+6
n2 + 1 − α α+4 =⇒ n2 + 1 + 4
⇓ ⇓
n2 + 1 − 4 n2 + 1 + 4
Fig. 129
Rule V. Four consecutive numbers placed around the border

will, after putting the complements, produce a difference of 4,
which will be a deficit in the line and the column containing
the lesser pair (Fig. 129).
This rule is stated, not in the theoretical part like the others, but
among the examples, when excesses must be systematically reduced by
4 (A.II.29) —indeed, it too is merely a particular case of Rule III. You
take groups of four small numbers and put the first one in an upper row,
the third in the opposite row below, the second on the right (for us, on
the left) and the fourth on the left, and you put opposite to each of these
four its complement.
Rule VI. The jth pair placed on the same side will, after
putting the complements, produce a deficit of 8j − 4 on the
side of that pair and an excess of the same amount in the

opposite side (Fig. 130).
n2 − 8j + 3 n2 − 8j + 7
=⇒ n2 + 1 − (8j − 4)
2 2
n2 + 8j − 1 n2 + 8j − 5
=⇒ n2 + 1 + (8j − 4)
2 2
Fig. 130
This is indeed what A asserts, except that he expresses it in a less
general way, considering one pair after the other (A.II.19): You will find
that if the last small number and the preceding one are placed on one side
(case j = 1), and opposite to them their complements, the side containing
the two small numbers will be less than its sum due by 4, and the other
more by 4; placing the next two numbers (case j = 2), the differences will
be 12 and 12, and with the next two numbers 20 and 20, then 28, 36, 44,
52, 60 and so on always till (the placement of) 4 and 2.
Rule VII. If the jth pair is written in the corners, it will

produce the same difference as in the previous case, but also
a difference of 2 in the other two rows, which will be a deficit
in the row containing the lesser element (Fig. 131).
n2 − 8j + 3 n2 − 8j + 7
=⇒ n2 + 1 − (8j − 4)
2 2
n2 + 8j − 5 n2 + 8j − 1
=⇒ n2 + 1 + (8j − 4)
2 2
⇓ ⇓
n2 + 1 − 2 n2 + 1 + 2
Fig. 131
This is indeed what is found in the text (A.II.22), which thus also
repeats the previous rule: If four numbers are put in the corners, each pair
in consecutive corners and their complements diagonally opposite, and if
you put the two small numbers consecutively on the upper side, this side
will be less than its sum due by 4 if they are the last two (j = 1), by 12
if they are the two previous (numbers), and so on always, with regular
additions of 8, until you reach 2 and 4. The right-hand side will always
have, relative to its sum due, an excess of 2, or a deficit of 2, without any
augmentation and diminution.
The author also adds, after this last of the equalization rules: You
must understand all this: it belongs to what you need for writing the
even numbers in this class (of squares).
Once the equalization has been effected, completing the arrangement
is easy: as we have seen, it amounts to placing groups of four, or pairs of
numbers, as neutral placements. The text mentions that when treating
the examples (A.II.31, our rules I and II): If there are remaining empty
cells, it can be only four facing four, eight facing eight, twelve facing
twelve (and so on); you will equalize them with groups of four by means
of the still available sequences of four numbers (. . . ) of which each pair
is in progression: you put the first of the first pair and the second of the
second pair on one side, the second of the first pair and the first of the
second pair on the facing side, and opposite to each its complement.
Remark. The magic sum being odd and the quantity of odd numbers
placed in each row also being odd, thus their sum odd, the differences
to be eliminated are always even. Now by applying once, twice or
at most three times Rules IV and/or VI, which both use consecutive
pairs, we can eliminate differences of the form 8u, 8u±2, 8u±4, 8u±6,
that is, any difference which might occur. There may be, though, two
obstacles.
(i) The first arises when there are not enough empty cells for the
number of required applications. In such a case we shall use Rule III,
by means of which two even, non-consecutive numbers (α, α + 2s, with
s = 1) will eliminate the difference; but then we shall have to apply it
a second time in order to use the other two terms of the two broken
pairs. See example below, in the equalization of the first border for
the order n = 4t + 1.
(ii) The second obstacle arises when the difference may be eliminated,
but with fewer cells than the quantity required to leave a number of
empty cells divisible by 4. This will prove to be a minor difficulty, for
we shall just use more cells than strictly necessary (one application
may only partly eliminate the difference, or even increase it). See
example below, in the equalization of the other borders’ lines for the
order n = 4t + 1.
In all other cases we shall be able to apply these rules to the least
possible number of cells, and so as to leave a number of empty cells
divisible by 4.
C. Case of the order n = 4t + 1 (with t ≥ 2)41

Since the bottom lines are to receive the complements, we shall con-
sider only the upper ones, thus those in excess; likewise, we shall consider
the left-hand columns, also in excess. As we have seen (p. 75), the lines
and columns of the pth border (p = 1, . . . , t, counting from the inner
square already filled) comprise 4p empty cells. Making up for the differ-
ences must thus be effected with four cells for the rows of the first border
whereas we may use eight cells for the rows of the other borders, this
being applicable to the outer border of the lowest order n = 9 as well.
1. First border
(i) The excess in the upper line is Δ1L, 4t+1 = 8t + 4 and there are four
empty cells, so that, as said, we are to eliminate it by means of two pairs
of numbers. We shall put first (A.II.24i) the largest pair (j = 1) in the
corners, namely
n2 − 5
in the left-hand corner
2
n2 − 1
in the right-hand corner.
2
See Fig. 132, for n = 9, with 38 and 40; Fig. 133, for n = 13, with 82 and
84; Fig. 134, for n = 17, with 142 and 144. The sum of the two numbers
we have just placed being n2 − 3, which has a deficit of 4 relative to the
sum due, there remain in the upper row an excess of 8t and two empty
cells.
(ii) Since this cannot be eliminated using a single pair of consecutive
numbers, we shall use Rule III, which enables us to eliminate this excess
by means of two numbers displaying a difference of 8t. So we shall place
the lowest of the smaller numbers, 2, in the top row and 2 + 8t in the row
opposite; as the text puts it for this latter number (A.II.24ii): considering
the new excess of the upper row (8t), take its half, and count after 2 as
many small even numbers as this half, and put the number reached in
the lower row. The complement of 2 + 8t, namely n2 + 1 − (2 + 8t), being
placed, like 2, in the top part of the border, their sum will cancel the
excess. See Fig. 132, for n = 9 (t = 2), with 2 and 18; Fig. 133, for
n = 13 (t = 3), with 2 and 26; Fig. 134, for n = 17 (t = 4), with 2 and
34.
41
We shall consider the particular case of order 5 at the end.
(iii) Consider the columns. The left-hand corner cells are now occupied
by the smaller of the numbers placed in (i) and the complement of the
other, thus by
n2 − 5 n2 + 3
, ,
2 2
the sum of which is n2 − 1; the initial excess on the left-hand side,
ΔC,
1
4t+1
= 8t + 2, is thus reduced to 8t. Since there are only two empty
cells available, we must eliminate this excess by Rule III. Let us put 4 in
the left-hand column and, in the right-hand one, 4 + 8t, that is, the 4tth
even number after 4 (A.II.24iii). By doing that, not only shall we have
eliminated the excess of 8t since the left-hand column now contains 4 and
n2 + 1 − (4 + 8t), but we shall also have finished placing the two pairs
previously broken, thereby avoiding the problem of using subsequently
two single numbers. In the first border’s columns of Fig. 132 (n = 9) are
thus placed 4 and 20, in Fig. 133 (n = 13) 4 and 28, in Fig. 134 (n = 17)
4 and 36.
2. Other borders
The excesses in the upper rows are ΔpL, 4t+1 = 8pt + 4, and there
are 4p empty cells. Direct elimination of the excesses by Rule VI would
be possible (putting j = pt + 1), but unsuitable since the number of
cells left empty would not be divisible by 4; elimination by two complete
(unbroken) pairs is not possible either. We shall therefore eliminate the
excesses by means of eight cells, which is applicable to the smallest order
n = 9. Likewise, we shall employ eight cells in the left-hand columns to
cancel their excesses, ΔC,p
4t+1 = 8pt + 2. First, in steps i and ii below,
four cells will be used to reduce the excess in all columns to a uniform
quantity. Then, step iii will do the same for the lines, using two of the six
cells still available. The next two equalization steps, applied together to
lines and columns, will use four cells to remove the remaining differences
(below, 4, steps α and β).
(n) (n)
(i) We first look for two (consecutive) even numbers, αp and αp + 2, to
be placed in the corners of the left-hand columns and adding up to σpA, n
(A for ‘angular’), with
σpA, n = (n2 + 1) − (8p − 4), (∗)
where the subtractive quantity thus depends on the border, and not on
(n)
the order. Since σpA, n = 2αp + 2, we find
(n2 + 1) − (8p − 4) − 2 n2 − 8p + 3
αp(n) = = ,
2 2
and the two numbers will be

n2 − 8p + 3 n2 − 8p + 7
, , (†)
2 2
with the lesser element put in the upper left-hand corner and the larger
in the lower left-hand corner. Let us apply that to the first orders.
— Second border (p = 2; A.II.25i). For n = 9, n = 13, n = 17, we find
(n) (n)
from (∗) for σ2A, n and the corresponding α2 , α2 + 2:
σ2A, 9 = 82 − 12 = 70 (= 34 + 36),
σ2A, 13 = 170 − 12 = 158 (= 78 + 80),
σ2A, 17 = 290 − 12 = 278 (= 138 + 140).
— Third border (p = 3; A.II.26i), for n = 13, n = 17:
σ3A, 13 = 170 − 20 = 150 (= 74 + 76),
σ3A, 17 = 290 − 20 = 270 (= 134 + 136).
— Fourth border (p = 4; A.II.27i), for n = 17:
σ4A, 17 = 290 − 28 = 262 (= 130 + 132).
See Fig. 132, 133, 134.
(ii) The initial left-hand excesses, ΔC,
p
4t+1 = 8pt + 2, have now changed,
for there are in the corners of the pth left-hand column two numbers of
the form (†), the sum of which is (n2 + 1) − (8p − 4); the excess remaining
in the pth left-hand column is then 8p(t − 1) + 6. Let us put, according
to Rule VI,
8p(t − 1) + 6 = 8j − 4,
whence 8j = 8p(t − 1) + 10. Taking j = p(t − 1) + 1, we shall write in the
pth left-hand column (p ≥ 2) the pair
n2 − 8[p(t − 1) + 1] + 3 n2 − 8[p(t − 1) + 1] + 7
, , (††)
2 2
which then leaves an excess of 2 in each left-hand column and four cells
to complete the equalization.
The text explains as follows, symbolism apart, the placing of the pairs
(††). We are to put in each (for us left-hand) column two consecutive
smaller even numbers the sum of which is less than their sum due by the
(initial) excess of the column minus a quantity (depending on the border,
not on the order) 8p − 2. That is,
σpC, n = (n2 + 1) − [(8pt + 2) − (8p − 2)]. (∗∗)

(n) (n) (n)

Write βp , βp + 2 for the two numbers; since 2βp + 2 = σpC, n , they are
immediately determinable. Indeed,
(n2 + 1) − [(8pt + 2) − (8p − 2)] − 2 n2 − 8[p(t − 1) + 1] + 3

βp(n) = = .
2 2
Let us apply this to the first borders.

— Second border (p = 2; A.II.25ii). For n = 9 (t = 2), n = 13 (t = 3),
(n) (n)
n = 17 (t = 4), we find from (∗∗) for the σ2C, n and thus β2 , β2 + 2
following values:
σ2C, 9 = 82 − (34 − 14) = 62 (= 30 + 32),
σ2C, 13 = 170 − (50 − 14) = 134 (= 66 + 68),
σ2C, 17 = 290 − (66 − 14) = 238 (= 118 + 120).
— Third border (p = 3; A.II.26ii), for n = 13 (t = 3), n = 17 (t = 4):
σ3C, 13 = 170 − (74 − 22) = 118 (= 58 + 60),
σ3C, 17 = 290 − (98 − 22) = 214 (= 106 + 108).
— Fourth border (p = 4; A.II.27ii), for n = 17 (t = 4):
σ4C, 17 = 290 − (130 − 30) = 190 (= 94 + 96).
See Fig. 132, 133, 134.
(iii) Consider now the upper rows. Because of i, their excesses have
changed. Indeed, one number has been placed in each pth left-hand corner
while the right-hand one is occupied by a complement. These numbers,
n2 − 8p + 3 n2 + 8p − 5
, ,
2 2
adding up to n2 − 1 = (n2 + 1) − 2, the excesses in the upper rows have
been reduced from 8pt + 4 to 8pt + 2, and there are 4p − 2 empty cells.
Putting then, according to Rule VI,
8pt + 2 = 8j − 4,
we shall take j = pt and thus place the ptth pair of smaller numbers
n2 − 8pt + 3 n2 − 8pt + 7
, († † †)
2 2
within the pth upper row, which then leaves an excess of 6 in each upper
row and four cells to complete the equalization.
The text expresses the choice of these numbers in the same way as
before. We are to put in the upper rows two consecutive even smaller
(n) (n)
numbers, γp and γp + 2, the sum of which is less than their sum due
by the (initial) excess of the upper row minus 8 (thus minus a quantity
independent of border and order). That is, keeping analogous symbols as
before,
σpL, n = (n2 + 1) − [(8pt + 4) − 8], (∗ ∗ ∗)

(n)
giving indeed, since σpL, n = 2γp + 2,
(n2 + 1) − [(8pt + 4) − 8] − 2 n2 − 8pt + 3

γp(n) = = .
2 2
From this can be inferred the following numerical values:
— Second border (p = 2; A.II.25iii), for n = 9 (t = 2), n = 13 (t = 3),
n = 17 (t = 4):
σ2L, 9 = 82 − (36 − 8) = 54 (= 26 + 28),
σ2L, 13 = 170 − (52 − 8) = 126 (= 62 + 64),
σ2L, 17 = 290 − (68 − 8) = 230 (= 114 + 116).
— Third border (p = 3; A.II.26iii), for n = 13 (t = 3), n = 17 (t = 4):
σ3L, 13 = 170 − (76 − 8) = 102 (= 50 + 52),
σ3L, 17 = 290 − (100 − 8) = 198 (= 98 + 100).
— Fourth border (p = 4; A.II.27iii), for n = 17 (t = 4):
σ4L, 17 = 290 − (132 − 8) = 166 (= 82 + 84).
See Fig. 132, 133, 134.
74 50 52 161 94
58 78 62 64 7 145 155 90 112
60 66 82 2 3 23 137 139 151 144 84 104 110
68 4 47 131 127 125 55 59 51 166 102

34 26 28 77 46
153 113 67 107 105 75 71 57 17
30 38 2 3 69 71 64 40 52
157 141 117 97 79 93 83 73 53 29 13
32 4 23 63 61 31 27 78 50
159 143 135 121 101 89 85 81 69 49 35 27 11
73 53 35 49 39 29 9
5 21 41 61 87 77 91 109 129 149 165
75 67 57 45 41 37 25 15 7
1 37 99 63 65 95 103 133 169
1 17 43 33 47 65 81
142 119 39 43 45 115 111 123 28
62 55 19 21 51 59 20 86 168 167 147 33 31 19 26 88
42 80 79 13 11 18 44 80 108 106 163 25 15 92
36 56 54 5 48 76 120 118 9 96
Fig. 132 Fig. 133

130 82 84 277 158
94 134 98 100 11 253 271 154 196
96 106 138 114 116 7 35 237 247 267 150 184 194
108 118 142 2 3 31 51 229 231 243 263 256 144 172 182
120 4 79 223 219 215 213 87 91 95 83 286 170
265 197 107 191 187 185 115 119 111 93 25
269 245 201 173 127 167 165 135 131 117 89 45 21
273 249 233 205 177 157 139 153 143 133 113 85 57 41 17
275 251 235 227 209 181 161 149 145 141 129 109 81 63 55 39 15
9 33 49 73 101 121 147 137 151 169 189 217 241 257 281
5 29 69 97 159 123 125 155 163 193 221 261 285
1 65 179 99 103 105 175 171 183 225 289
254 207 67 71 75 77 203 199 195 211 36
146 288 287 259 239 61 59 47 27 34 148
140 176 174 283 255 53 43 23 152
136 192 190 279 37 19 156
132 208 206 13 160
Fig. 134
3. Recapitulation
First, as seen in i and iii, the upper corners of the pth border (p ≥ 2)
are occupied by
n2 − 8p + 3 n2 + 8p − 5
, .
2 2
Next, as seen in ii, we have written within the left-hand pth column the
pair
n2 − 8[p(t − 1) + 1] + 3 n2 − 8[p(t − 1) + 1] + 7
, ,
2 2
n−1
which, since t = 4 , may also be written as
n2 − 2pn + 10p − 5 n2 − 2pn + 10p − 1
, .
2 2
Finally, as seen in iii, we have placed within the pth upper row the pair
n2 − 8pt + 3 n2 − 8pt + 7
, ,
2 2
n−1
n2 − 2pn + 2p + 3 n2 − 2pn + 2p + 7
, .
2 2
With this and what we have found for the first border, we can set up
Fig. 135 (∗ are cells for complements). Although it gives the quantities
imposed for just five borders, it clearly shows how the terms belonging
to the same oblique line can be formed using arithmetical progressions
n2 +35
∗
2
n2 +27
∗
2
n2 +19
∗
2
n2 +11
∗
2
2n + 2
n2 −1
∗
2
2n
∗
n2 −4n+11
∗
2
n2 −6n+13
n2 −4n+7
n2 −5
∗
2
2
n2 −8n+11 n2 −8n+15
n2 −6n+29 n2 −4n+15
n2 −4n+19
n2 −6n+9
n2 −13
∗
2
2
n2 −10n+13 n2 −10n+17
n2 −8n+39 n2 −6n+25
n2 −21
∗
2
2
n2 −10n+49 n2 −8n+35
n2 −29
∗
2
2
n2 −10n+45
n2 −37
∗
2
Fig. 135
—the only exceptions being found in the first border. This is sufficient
for constructing any square of order n = 4t + 1 (t > 1); indeed, what
remains to complete its rows is straightforward, as we shall see.
4. Completing the construction

Except for the rows of the first border, which are completely filled
(and equalized), there remains in each row a number of cells divisible by
4 and, for each border, uniform excesses of 6 in the upper rows and of 2 in
the left-hand columns (Fig. 136) —as clearly stated in A.II.28. In order
to have an equalization valid from the smallest order (here n = 9), we
shall use four cells and thus place two numbers in each row. We further
know that the numbers still available form pairs of, or groups of four,
consecutive even numbers; indeed, the only pair which had been broken
has been placed.
Eliminating the remaining excesses and, for p ≥ 3, filling the remain-
ing empty cells will be carried out in three steps.
(α) Let us put the four even consecutive numbers δ, δ+2, δ+4, δ+6 around
each border, beginning in its upper line and turning (for us) towards the
left (A.II.29). With the complements in place, that will change (according
to Rule V) the former excesses and deficits of ±6 to ±2 and the former
excesses and deficits of ±2 to ∓2 (Fig. 137).
⇒ +6 ⇒ +2
⇒ −6 ⇒ −2
⇓ ⇓ ⇓ ⇓
+2 −2 −2 +2
Fig. 136 Fig. 137
In Fig. 138 (n = 9) we have thus placed, starting in the upper row,
the groups of four numbers 6, 8, 10, 12 (p = 2); in Fig. 139 (n = 13)
the groups 6, 8, 10, 12 (p = 3) and 30, 32, 34, 36 (p = 2); in Fig. 140
(n = 17), the groups 6, 8, 10, 12 (p = 4), 22, 24, 26, 28 (p = 3), 42, 44,
46, 48 (p = 2).
(β) Take (according to Rule IV) any two pairs of consecutive available
numbers, ε, ε + 2 and ζ, ζ + 2, and place each in a pair of opposite rows,
with the lesser terms on the side of the excess. With that done for all
borders, and the complements written in, all remaining differences will
be eliminated (A.II.30).
In Fig. 138 (n = 9), we have thus placed 14, 16 and 22, 24 (p = 2);
in Fig. 139 (n = 13), 14, 16 and 18, 20 (p = 3), then 22, 24 and 38, 40
(p = 2); in Fig. 140 (n = 17), 14, 16 and 18, 20 (p = 4), then 30, 32 and
38, 40 (p = 3), finally 50, 52 and 54, 56 (p = 2).
Remark. We have therefore, in this last example, employed all numbers

from 6 to 56, continuously except for the numbers 34 and 36, already
used with the broken pair for the first border. Likewise in the other
examples: the choices just fill up the gaps left with the placements
imposed.
34 26 28 6 77 72 14 66 46
30 38 2 3 69 71 64 40 52
32 4 23 63 61 31 27 78 50
8 73 53 35 49 39 29 9 74
75 67 57 45 41 37 25 15 7
70 1 17 43 33 47 65 81 12
60 62 55 19 21 51 59 20 22
24 42 80 79 13 11 18 44 58
36 56 54 76 5 10 68 16 48
Fig. 138
74 50 52 6 160 14 161 154 42 126 124 48 94
58 78 62 64 30 7 145 155 136 22 146 90 112
60 66 82 2 3 23 137 139 151 144 84 104 110
8 68 4 47 131 127 125 55 59 51 166 102 162
158 32 153 113 67 107 105 75 71 57 17 138 12
152 157 141 117 97 79 93 83 73 53 29 13 18
159 143 135 121 101 89 85 81 69 49 35 27 11
20 5 21 41 61 87 77 91 109 129 149 165 150
54 134 1 37 99 63 65 95 103 133 169 36 116
114 132 142 119 39 43 45 115 111 123 28 38 56
100 40 86 168 167 147 33 31 19 26 88 130 70
72 80 108 106 140 163 25 15 34 148 24 92 98
76 120 118 164 10 156 9 16 128 44 46 122 96
Fig. 139
(γ) We have now completed the first two borders around the central
square and are left in the subsequent borders (p ≥ 3) with numbers of
empty cells all divisible by 4. Since the rows of these borders display
neither excesses nor deficits, we may fill them, according to rules I and
II, with pairs of even numbers η, η + 2 and ϑ, ϑ + 2, of which we shall
write the extreme terms (η, ϑ + 2) on one side and the middle terms on
the other. The complements will provide the sum due for four cells, and
these neutral placements will be continued until the whole square is filled,
as stated in A.II.31.
130 82 84 6 280 14 274 58 277 230 228 64 66 222 220 72 158
94 134 98 100 22 264 30 11 253 271 258 74 214 212 80 154 196
96 106 138 114 116 42 7 35 237 247 267 244 50 238 150 184 194
8 108 118 142 2 3 31 51 229 231 243 263 256 144 172 182 282
278 24 120 4 79 223 219 215 213 87 91 95 83 286 170 266 12
272 262 44 265 197 107 191 187 185 115 119 111 93 25 246 28 18
20 252 269 245 201 173 127 167 165 135 131 117 89 45 21 38 270
86 273 249 233 205 177 157 139 153 143 133 113 85 57 41 17 204
275 251 235 227 209 181 161 149 145 141 129 109 81 63 55 39 15
202 9 33 49 73 101 121 147 137 151 169 189 217 241 257 281 88
200 40 5 29 69 97 159 123 125 155 163 193 221 261 285 250 90
92 122 242 1 65 179 99 103 105 175 171 183 225 289 48 168 198
102 166 236 254 207 67 71 75 77 203 199 195 211 36 54 124 188
186 164 56 146 288 287 259 239 61 59 47 27 34 148 234 126 104
180 128 140 176 174 248 283 255 53 43 23 46 240 52 152 162 110
112 136 192 190 268 26 260 279 37 19 32 216 76 78 210 156 178
132 208 206 284 10 276 16 232 13 60 62 226 224 68 70 218 160
Fig. 140
In Fig. 139 (n = 13), 42, . . . , 48 and the pairs 54, 56 & 70, 72 complete
the outer border (50, 52 and 58, . . . , 68 have been used for the prelimi-
nary placement); in Fig. 140 (n = 17), the two upper lines and the ones
opposite have been filled with the groups 58–64, 66–72, and 74–80, re-
spectively, the columns with 86–92, 102–104 & 110–112 (separated pairs),
and 122–128, respectively.
A represents the square of order 9 (a 17), and his differs from ours
only by its right-to-left orientation and the rotation of the inner square
of order 3. The square of order 9 found in B (b 49) is of course different;

as we know (p. 71), it was constructed by another method (except for the
inner square) using, in part, trial and error.
D. Case of the order n = 4t + 3 (with t ≥ 1)

We shall again consider only the rows in excess, that is, the upper
lines and the (for us) left-hand columns.
We already know (pp. 75 & 77–78) that the number of empty cells in
each row of the pth border is 4p − 2 and that the excesses are ΔpL, 4t+3 =
8(p − 1)(t + 1) + 4 and ΔpC, 4t+3 = 8(p − 1)(t + 1) + 2. In order to be left
with a number of cells divisible by 4, we might first consider whether the
placement of a single pair in the corners would suit. This turns out to be
possible for equalizing the lines, as shown by A, first for p = 1 (A.II.32)
then for p ≥ 2 (A.II.33). For the equalization of the columns, of which
two corner cells are now occupied, four cells are available in any case.
1. First border
The excess is Δ1L, 4t+3 = 4 and there are just two empty cells. Now
this suits the situation of Rule VI as described in A.II.19: the largest pair
of smaller even numbers will eliminate the difference. Thus we shall put
in the upper corners of the first row
n2 − 5 n2 − 1
, ,
2 2
with the larger in the right-hand corner, thus in the column with a deficit.
Since the two columns of the first borders then also contain the sum due,
this single placement will equalize all first borders (A.II.32).
Remark. This pair had already been used for the first borders in the case
n = 4t + 1; but there it could not complete the equalization.
In Fig. 141 (n = 7) we have thus placed in the upper corners of the
first border 22, 24; in Fig. 142 (n = 11), 58, 60; in Fig. 143 (n = 15), 110,
112.
2. Other borders
Equalization of the other borders will be carried out in three steps. A
single one (below, i) will suffice for all lines, but two are needed for the
columns, of which the differences have now changed with the occupation
of the corner cells.
(i) Considering, with our symbolism, that the excess in the upper rows
is ΔpL, 4t+3 = 8(p − 1)(t + 1) + 4, we shall put, according to Rule VI,
8(p − 1)(t + 1) + 4 = 8j − 4.
It then appears that ΔpL, 4t+3 will be eliminated if we choose
j = (p − 1)(t + 1) + 1 ;
so let us write in the corners of the pth upper row (p ≥ 1) the correspond-
ing pair, that is,
n2 − 8[(p − 1)(t + 1) + 1] + 3 n2 − 8[(p − 1)(t + 1) + 1] + 7

, , (∗)
2 2
of which we shall put the lesser term in the column with an excess.
This is what A does in A.II.33, when he says to put in the upper
(n)
corners of the pth border a pair of (consecutive) even numbers, say κp
(n)
and κp + 2, adding up to σpA, n = (n2 + 1) − ΔpL, 4t+3 : this indeed gives
(n)
(∗). For, with σpA, n = (n2 + 1) − [8(p − 1)(t + 1) + 4] = 2κp + 2, we find
(n2 + 1) − [8(p − 1)(t + 1) + 4] − 2 n2 − 8[(p − 1)(t + 1) + 1] + 3

κp(n) = = .
2 2
As stated by A at the end of A.II.33, we are then left with filling the
incomplete lines, for p ≥ 2, with neutral placements.
In Fig. 141 (n = 7), we have thus placed in the upper rows 14, 16
(p = 2, thus σ2A, 7 = 50 − 20 = 30); in Fig. 142 (n = 11) 46, 48 (p = 2,
σ2A, 11 = 122−28 = 94), then 34, 36 (p = 3, σ3A, 11 = 122−52 = 70); in Fig.
143 (n = 15) 94, 96 (p = 2, σ2A, 15 = 226 − 36 = 190), then 78, 80 (p = 3,
σ3A, 15 = 226 − 68 = 158), finally 62, 64 (p = 4, σ4A, 15 = 226 − 100 = 126).
Remark. Formulas (∗) are in fact also valid for p = 1.
(ii) Since the corners of the columns are now occupied, their number of
empty cells is 4 (p = 2) or a multiple of 4 (p ≥ 3). Meanwhile, the initial
excesses in the left-hand columns, ΔpC, 4t+3 = 8(p − 1)(t + 1) + 2, have
changed since the corners are now occupied by the first number in (∗)
and the complement of the other, thus by
n2 − 8[(p − 1)(t + 1) + 1] + 3 n2 + 8[(p − 1)(t + 1) + 1] − 5

, ,
2 2
the sum of which is (n2 + 1) − 2; therefore, there remains on the pth

left-hand side an excess of 8(p − 1)(t + 1), to be eliminated now with four
numbers.
To do that, as said in A.II.34, we shall look for a pair of large numbers

such that their sum exceeds their sum due by an amount which, when
added to the excess of the right-hand row, equals the deficit of the sum
of two small numbers. With 8(p − 1)(t + 1) the (new) excess of the (our)
(n) (n)
left-hand column, Λp and Λp − 2 two larger numbers (in descending
(n) (n)
sequence) adding up to ΣC, n
p , and μp and μp + 2 two smaller numbers
adding up to σpC, n , the above condition becomes
C, n
Σp − (n2 + 1) + 8(p − 1)(t + 1) = (n2 + 1) − σpC, n .
(n) (n)
From this results, since ΣC,
p
n = 2Λ
p − 2 and σp
C, n = 2μ
p + 2, the
following expression for determination of the required numbers:
2(n2 + 1) − 8(p − 1)(t + 1)

Λ(n) (n)
p + μp = = n2 + 1 − 4(p − 1)(t + 1).
2
Remarks.
(1) Here the text is not only very concise, but imprecise. Knowing the
(n)
sum of the greater larger number, Λp , and the lesser smaller number,
(n)
μp , seems to leave their individual values indeterminate. As a matter
(n)
of fact, to Λp can be attributed a determinate value. For since we
have placed in the upper corners of the first border (p = 1) the two
largest smaller numbers
n2 − 5 n2 − 1
, ,
2 2
the opposite corners are occupied by their complements, namely the
two smallest larger numbers, that is
n2 + 7 n2 + 3
, .
2 2
Now for the required pairs of larger numbers we shall just take the
subsequent pairs, namely
n2 + 15 n2 + 11 n2 + 23 n2 + 19
, ; , ; ... .
2 2 2 2
That is, generally, the pairs of larger numbers to be put in the left-hand
columns of the pth border will be
n2 + 8p − 1 n2 + 8p − 5
, , (∗∗)
2 2
the sum of which is n2 + 8p − 3 = (n2 + 1) + 8p − 4.
(2) Note also that this unusual intervention of ‘large numbers’ could have
been avoided by telling us to put on the other side of the pth border
the pair of small numbers
n2 − 8p + 3 n2 − 8p + 7
, ,
2 2
that is, the pair which had been elsewhere (for the order n = 4t + 1)
placed in the left-hand corners.
(3) The two large numbers being thus determined, the pair of smaller
numbers to be put with them in the left-hand columns can be found
in the usual way. Since the excess in the pth left-hand column has now
increased to
8(p − 1)(t + 1) + 8p − 4 = 8[(p − 1)(t + 1) + p] − 4,
we may, by Rule VI, put j = (p − 1)(t + 1) + p, and thus take as the
pair of smaller numbers
n2 − 8[(p − 1)(t + 1) + p] + 3 n2 − 8[(p − 1)(t + 1) + p] + 7
, . (∗ ∗ ∗)
2 2
Once these four numbers —two larger and two smaller— have been placed,
all excesses in the left-hand columns are eliminated.
In Fig. 141 (n = 7), we have thus placed in the left-hand columns 30,
32 & 10, 12 (p = 2); in Fig. 142 (n = 11) 66, 68 & 42, 44 (p = 2), then
70, 72 & 26, 28 (p = 3); in Fig. 143 (n = 15) 118, 120 & 90, 92 (p = 2),
then 122, 124 & 70, 72 (p = 3), finally 126, 128 & 50, 52 (p = 4).
34 113 36
70 46 7 97 107 48 52
72 66 58 3 23 89 91 103 60 56 50
26 68 105 43 83 81 51 47 17 54 96
14 45 16 28 109 93 73 55 69 59 49 29 13 94
30 22 3 37 39 24 20 111 95 87 77 65 61 57 45 35 27 11
32 41 19 33 23 9 18 5 21 37 63 53 67 85 101 117
43 35 29 25 21 15 7 42 1 75 39 41 71 79 121 80
10 1 27 17 31 49 40 44 62 119 99 33 31 19 64 78
12 26 47 13 11 28 38 74 115 25 15 76
34 5 36 86 9 88
Fig. 141 Fig. 142

62 213 64
126 78 11 189 207 80 100
128 122 94 7 35 173 183 203 96 104 98
50 124 118 110 3 31 51 165 167 179 199 112 108 102 176
52 70 120 201 75 159 155 153 83 87 79 25 106 156 174
72 205 181 141 95 135 133 103 99 85 45 21 154
209 185 169 145 125 107 121 111 101 81 57 41 17
211 187 171 163 149 129 117 113 109 97 77 63 55 39 15
9 33 49 69 89 115 105 119 137 157 177 193 217
5 29 65 127 91 93 123 131 161 197 221
90 1 147 67 71 73 143 139 151 225 136
92 114 223 195 175 61 59 47 27 116 134
130 219 191 53 43 23 132
146 215 37 19 148
162 13 164
Fig. 143
3. Recapitulation
First, as seen in 1, we have placed in the upper corners of the first border
n2 − 5 n2 − 1
, .
2 2
Next, we have put in the upper corners of the subsequent borders, ac-
cording to (∗),
n2 − 8[(p − 1)(t + 1) + 1] + 3 n2 − 8[(p − 1)(t + 1) + 1] + 7
, ,
2 2
which, since t = n−34 , may also be written as
n2 − 2(p − 1)n − 2p − 3 n2 − 2(p − 1)n − 2p + 1
, .
2 2
Finally, we have placed in the left-hand columns, as seen in (∗∗), the pairs
of larger numbers
n2 + 8p − 5 n2 + 8p − 1
,
2 2
as well as, according to (∗ ∗ ∗), the pair of smaller numbers
n2 − 8[(p − 1)(t + 1) + p] + 3 n2 − 8[(p − 1)(t + 1) + p] + 7
, ,
2 2
n−3
n2 − 2(p − 1)n − 10p + 5 n2 − 2(p − 1)n − 10p + 9
, .
2 2
n2 −8n−9
∗
2
n2 −6n−7
∗
2
n2 −4n−5
∗
2
n2 −2n−3
∗
2
n2 −1
∗
2
n2 −5
∗
2
n2 −2n−15
n2 −2n−11
n2 −2n−7
n2 +11
n2 +15
∗
2
2
n2 −4n−25
n2 −4n−21
n2 −4n−9
n2 +19
n2 +23
∗
2
2
n2 −6n−11
n2 −8n−41 n2 −6n−35
n2 −6n−31
n2 +27
n2 +31
∗
2
2
n2 −8n−13
n2 −8n−45
n2 +35
n2 +39
∗
2
Fig. 144
With all these relations, we may set up Fig. 144, which represents the
placements required for the first five borders. As before (Fig. 135), that
enables us to fill the cells for any order of this type since the way to form
the terms becomes evident. In that case, each row will then display the
sum due for the number of cells filled.
4. Completing the construction

With the differences now eliminated and the quantities of still empty
cells being all divisible by 4, we can complete the rows with neutral place-
ments (A.II.34, end). In Fig. 145 (n = 7), a single group of four numbers
2, . . . , 8 completes the square; in Fig. 146 (n = 11), 2, . . . , 16 and 30, 32
& 38, 40 (p = 3), next 18, . . . , 24 (p = 2); in Fig. 147 (n = 15), 2, . . . , 24
and, for the columns, 54, . . . , 60 and the pairs 66 & 68, 74 & 76 (p = 4);
then 26, . . . , 40 and 82, . . . , 88 (p = 3); finally, 42, . . . , 48 (p = 2).
34 2 118 116 8 113 10 110 108 16 36
70 46 18 102 7 97 107 100 24 48 52
72 66 58 3 23 89 91 103 60 56 50
26 68 105 43 83 81 51 47 17 54 96
14 2 46 45 44 8 16 28 109 93 73 55 69 59 49 29 13 94
30 22 3 37 39 24 20 111 95 87 77 65 61 57 45 35 27 11
32 41 19 33 23 9 18 30 5 21 37 63 53 67 85 101 117 92
43 35 29 25 21 15 7 90 42 1 75 39 41 71 79 121 80 32
10 1 27 17 31 49 40 84 44 62 119 99 33 31 19 64 78 38
12 26 47 13 11 28 38 40 74 104 20 115 25 15 22 98 76 82
34 48 4 5 6 42 36 86 120 4 6 114 9 112 12 14 106 88
Fig. 145 Fig. 146

62 2 222 220 8 10 214 213 212 16 18 206 204 24 64
126 78 26 198 196 32 11 189 207 34 190 188 40 80 100
128 122 94 42 182 7 35 173 183 203 180 48 96 104 98
50 124 118 110 3 31 51 165 167 179 199 112 108 102 176
52 70 120 201 75 159 155 153 83 87 79 25 106 156 174
54 72 205 181 141 95 135 133 103 99 85 45 21 154 172
170 209 185 169 145 125 107 121 111 101 81 57 41 17 56
211 187 171 163 149 129 117 113 109 97 77 63 55 39 15
168 9 33 49 69 89 115 105 119 137 157 177 193 217 58
60 82 5 29 65 127 91 93 123 131 161 197 221 144 166
66 142 90 1 147 67 71 73 143 139 151 225 136 84 160
158 140 92 114 223 195 175 61 59 47 27 116 134 86 68
152 88 130 184 44 219 191 53 43 23 46 178 132 138 74
76 146 200 28 30 194 215 37 19 192 36 38 186 148 154
162 224 4 6 218 216 12 13 14 210 208 20 22 202 164

Fig. 147
The squares of orders 7 and 11 are both constructed by A (Fig. a 20,

a 19); they differ from ours merely by an insignificant permutation of the
numbers in the outer border, irrelevant for a bordered square, and the
rotation of the inner square of order 3. In B, the square of order 7 is
constructed empirically (B.22, Fig. b 47); it does not differ by much from
that of A —sufficiently though to show that he used neither A’s direct
method for filling the rhomb nor the method he himself (later) suggests
and which we have described at the beginning of this chapter.
E. Particular case of order 5
As observed by A (A.II.13), the general construction for order n =

4t + 1, represented in our figure 135, does not apply to the smallest order,
with t = 1 (it will appear that the four even numbers in the corners
also occupy cells within the border). A therefore constructs his square in
another way (A.II.23), with the result as in Fig. 148 (we have changed the
orientation and rotated the inner square). Here again, A merely indicates
where to put each number.
2 16 25 18 4 a 25 b
6 7 21 11 20 c 7 21 11
23 17 13 9 3 23 17 13 9 3
12 15 5 19 14 d 15 5 19
22 10 1 8 24 26–b e 1 f 26–a
Fig. 148 Fig. 149
The figure in B (not preserved) must have been similar. But his
commentary (B.22ii) deserves some attention. He remarks that if we place
2 in a corner, the next corner (horizontally or vertically) must be occupied
by the (smaller) number 4, 6, 8 or 12.42 In this way we shall obtain, he
says, numerous figures (this gives eleven possibilities). As a matter of fact,
taking into account the cases where 2 is no longer necessarily in a corner
(Fig. 149), there are altogether the twenty-one possibilities represented
in Fig. 150.
Remark. Counting the variations of c, d, e, f in their rows and the eight
aspects the inner 3 × 3 square may take (by inversion or rotation), we
obtain 21 · 4 · 8 = 672 configurations for this square of order 5.
42
These indications are repeated by a 12th-century author, al-Kharaqı̄ (C.1 below,
following the extracts from B).
a b 26 − b c d e f
1 2 4 22 6 12 8 10
2 2 4 22 8 10 6 12
3 2 6 20 4 16 8 12
4 2 6 20 8 12 4 16
5 2 14 12 6 22 10 18
6 2 14 12 10 18 6 22
7 2 18 8 10 22 12 20
8 2 18 8 12 20 10 22
9 2 20 6 12 22 16 18
10 2 20 6 16 18 12 22
11 4 6 20 2 16 8 14
12 6 10 16 8 12 4 24
13 6 22 4 14 18 16 24
14 8 2 24 4 6 10 12
15 8 10 16 4 14 6 24
16 8 16 10 4 20 12 24
17 12 4 22 2 6 10 18
18 14 8 18 4 6 10 24
19 14 18 8 4 16 20 24
20 16 8 18 2 6 14 22
21 16 18 8 6 12 22 24
Fig. 150
Chapter IV. Composite magic squares
Here the construction of a larger square is reduced to that of its indi-
vidual parts, mostly squares, each of which is itself magic. The possibility
of constructing a square in such a way depends of course on the divisibility
of its order.
I. What is usually meant by composite square is a square divided into
even or odd subsquares each of the same order, filled one after the other
completely by a continuous sequence of numbers in magic arrangement.
Since the sums in the subsquares are different, but they form an arith-
metical progression, these subsquares must be arranged in such a way
that the main square will itself be magic.
II. If the main square can be divided into subsquares of the same even
order ≥ 4, they can be filled individually with pairs of complements in
magic arrangement. Since these subsquares are all of the same size and
display the same sum, they can be arranged in any manner.
III. A less conventional possibility is to divide the main square into un-
equal parts, square or even rectangular, but here again filled with pairs
of complements. Each part is to display its sum due, which of course will
vary according to its dimensions.
Such are the three types of disposition we shall examine in turn. A
discusses the last two ( A.II.44–A.II.54; general description in A.II.44 and
A.II.46), B the first two (B.24–B.26). Thus the construction of composite
squares in these three forms can be said to be well established in the tenth
century. In fact, it might be much older, as an attempt to apply to larger
squares the magic arrangement already found for smaller orders.
§1. Subsquares having same size and unequal sums

Suppose our main square has a composite order of the form n = r · s,
with r, s ≥ 3 since a magic square of order 2 is not possible. We can
therefore divide our square into r 2 squares of order s which will be filled
each with a continuous sequence of s2 numbers arranged according to
some method suitable for the order s, the r 2 squares thus filled being
themselves placed according to some arrangement suitable for the order
r. If r = s, we shall obtain a different arrangement by exchanging the
rôles of r and s. Likewise, another decomposition of n will lead to another
arrangement or pair of other arrangements.
Composite magic squares 105
The smallest square thus constructible is that of order 9, with nine

subsquares of order 3 (Fig. 151); the arrangement of the subsquares and
their filling is of course as for order 3.43
11 18 13 74 81 76 29 36 31
16 14 12 79 77 75 34 32 30
15 10 17 78 73 80 33 28 35
56 63 58 38 45 40 20 27 22
61 59 57 43 41 39 25 23 21
60 55 62 42 37 44 24 19 26
47 54 49 2 9 4 65 72 67
52 50 48 7 5 3 70 68 66
51 46 53 6 1 8 69 64 71
Fig. 151
B constructs the square of order 9 (B.24). He mentions other examples
of divisions of an order n = 3k, namely k2 subsquares of order 3 arranged
as required for order k or nine subsquares of order k arranged as for order
3. His examples are: n = 15 = 5 · 3, thus twenty-five squares of order 3
arranged as for order 5, or nine squares of order 5; n = 21 = 7 · 3, thus
forty-nine squares of order 3, or nine squares of order 7.
This procedure is also applicable to even-order squares. Indeed, ac-
cording to B again, any number divisible by 4 will have its square divided
up into squares in two manners: one time according to the quantity of
the square of 4, thus in sixteen squares, another time according to the
square quantity of the number dividing the square in question sixteen
times (B.25i). In other words, a square of evenly even order n = 4k ad-
mits of at least two arrangements (if k > 2 and k = 4): sixteen squares of
order k, and k2 squares of order 4. The examples he gives are the square
of order 12, divisible into sixteen squares of order 3 or nine squares of
order 4; the square of order 20, divisible into sixteen squares of order 5
or twenty-five squares of order 4. We have illustrated the first example in
Fig. 152 and Fig. 153, but B merely explains the way to construct them
(B.25iv, B.25iii).
Remark. As already said, A makes no mention of the possibility of a
43
The authors at that time apparently did not consider the possibility of turning or
inverting the subsquares (see, however, note 601).
106 Subsquares having same size and unequal sums
division into same subsquares with different magic sums. Only B at-
tests to such knowledge in the tenth century (even though with the
restriction of one of the two factors being 3 or 4).
2 9 4 119 126 121 92 99 94 65 72 67
7 5 3 124 122 120 97 95 93 70 68 66
6 1 8 123 118 125 96 91 98 69 64 71
101 108 103 56 63 58 11 18 13 110 117 112
106 104 102 61 59 57 16 14 12 115 113 111
105 100 107 60 55 62 15 10 17 114 109 116
47 54 49 74 81 76 137 144 139 20 27 22
52 50 48 79 77 75 142 140 138 25 23 21
51 46 53 78 73 80 141 136 143 24 19 26
128 135 130 29 36 31 38 45 40 83 90 85
133 131 129 34 32 30 43 41 39 88 86 84
132 127 134 33 28 35 42 37 44 87 82 89
Fig. 152
17 30 27 24 129 142 139 136 49 62 59 56
28 23 18 29 140 135 130 141 60 55 50 61
22 25 32 19 134 137 144 131 54 57 64 51
31 20 21 26 143 132 133 138 63 52 53 58
97 110 107 104 65 78 75 72 33 46 43 40
108 103 98 109 76 71 66 77 44 39 34 45
102 105 112 99 70 73 80 67 38 41 48 35
111 100 101 106 79 68 69 74 47 36 37 42
81 94 91 88 1 14 11 8 113 126 123 120
92 87 82 93 12 7 2 13 124 119 114 125
86 89 96 83 6 9 16 3 118 121 128 115
95 84 85 90 15 4 5 10 127 116 117 122
Fig. 153
§2. Subsquares having same size and equal sums

B, after explaining the arrangement with unequal sums, proceeds with
the case where the subsquares display equal sums. He introduces this as
follows (B.26i): It is possible to set up the magic arrangement in squares
with sides having a fourth in a manner more elegant than that we have
explained. It consists in dividing the square (considered) into a (certain)
number of squares in all rows of which the sums will be uniformly the
same. His mention of a manner more elegant indicates which he (or his
contemporaries) preferred —and this might explain the omission of the
previous case by A.
Let n be the order of the main square and m the order of the equal
subsquares. They are then filled one by one with a continuous sequence
2
of m2 smaller numbers together with their complements to n2 + 1. Each
subsquare as a whole will thus display the same sum. The advantage of
this second type of composite square is also practical: since each sub-
square can be filled in the same way and their arrangement is arbitrary,
knowledge of a single (even-order) construction will suffice to fill the whole
square. B is quite explicit about filling the subsquares in any succession
(B.26ii): having divided an empty evenly even square into (for instance)
4 × 4 squares, we arrange in half (of the cells) of one of these squares,
whichever it is, the consecutive numbers taken in natural order from 1 to
8 (. . .) and we complete them by means of the equalizing number for the
(large) square. We move then to any other square and arrange in half
of its (cells) the numbers from 9 to 16, and we complete them by means
of the equalizing number. We proceed in this way until we have finished
with all its squares.
Remark. These subsquares could be mixed, turned or inverted; but here
again (note 43) neither author mentions that. Furthermore, if the
subsquares are filled in the natural sequence, the resulting square will
be pandiagonal.
As examples, B mentions first (B.26iii) the division of a square of
order 12 into nine subsquares of order 4 filled each with eight smaller
numbers and, in the corresponding bishop’s cells, their complements to
n2 + 1 = 145 (we have represented it in Fig. 154); next (B.26iv), the
division of a square of order 8 into four subsquares of order 4.
As to A, after describing the filling method (A.II.48), he gives (with-
out comment) two examples, both for order 12, divided first into four sub-
squares of order 6 then into nine subsquares of order 4 (Fig. a 34 and a 35 in
the translation). Only he mentions (A.II.45) that such a division is possi-
108 Parts having different sizes
ble for certain evenly odd orders (whereas for B the order must be divisible
by 4, as seen above). His examples (no illustrations) are n = 18 = 3 · 6,
thus with nine subsquares of order 6, and n = 30 = 3 · 10 = 5 · 6, with
nine subsquares of order 10 or twenty-five of order 6.
Remark. To be filled in this way, an evenly odd order must have the
form 2t(2s + 1) with t odd = 1. The main square will then contain
an odd number (2s + 1)2 of subsquares each of even order 2t; the two
examples n = 18 and n = 30 are the first evenly odd orders which can
be arranged in this way. As to the subsquares of even orders ≥ 6, they
would be filled by A as bordered squares, since this is the only method
he uses.
1 142 139 8 9 134 131 16 17 126 123 24
140 7 2 141 132 15 10 133 124 23 18 125
6 137 144 3 14 129 136 11 22 121 128 19
143 4 5 138 135 12 13 130 127 20 21 122
25 118 115 32 33 110 107 40 41 102 99 48
116 31 26 117 108 39 34 109 100 47 42 101
30 113 120 27 38 105 112 35 46 97 104 43
119 28 29 114 111 36 37 106 103 44 45 98
49 94 91 56 57 86 83 64 65 78 75 72
92 55 50 93 84 63 58 85 76 71 66 77
54 89 96 51 62 81 88 59 70 73 80 67
95 52 53 90 87 60 61 82 79 68 69 74
Fig. 154
§3. Parts having different sizes

A. Method of the cross
This method consists in filling separately two median lines and two
median columns in a square of order n = 4k+2. That leaves us with even-
order subsquares (Fig. 155) to be filled in one of the known ways. Being
applicable to any square of evenly odd order, this method is therefore
particularly appropriate when division into even subsquares (as seen in
the previous section) is not possible, that is, if the order is twice a prime,
in which case the only possible division would be into four odd-order
squares. The method of the cross is explained clearly, but rather too
concisely, by A (A.II.50–A.II.54). Let us describe it first in general terms.
Fig. 155
Since the order of the main square is 4k+2, it will contain the numbers
from 1 to n2 = 16k2 + 16k + 4 and the cross itself, 16k + 4 of those
numbers. We may choose to fill it with the 16k + 4 numbers in the
middle.44 These numbers, aligned in pairs of complements adding up to
n2 + 1 = 16k2 + 16k + 5, are:
8k2 + 1 8k2 + 2 ... 8k2 + 8k + 1 8k2 + 8k + 2
8k2 + 16k + 4 8k2 + 16k + 3 ... 8k2 + 8k + 4 8k2 + 8k + 3.
It is clear that, in order to obtain the sum due in the cross, we shall
always place pairs of complements in pairs of cells perpendicular to the
axes and, for the central square of the cross, in pairs of cells diagonally
opposite. Considering now the magic sum along the rows of the cross,
it will be convenient to obtain the sum due for a small number of cells
—the least possible in order that this arrangement be applicable to the
smallest order— the remainder then being filled with neutral placements.
Now such a partial equalization can be attained with just twenty cells
(and thus applied to the smallest order) using the last ten of the above
pairs, namely
n2 n2 n2 n2 n2 n2 n2 n2 n2 n2
2 −9 2 −8 2 −7 2 −6 2 −5 2 −4 2 −3 2 −2 2 −1 2
n2 n2 n2 n2 n2 n2 n2 n2 n2 n2
2 + 10 2 +9 2 +8 2 +7 2 +6 2 +5 2 +4 2 +3 2 +2 2 +1
X IX VIII VII VI V IV III II I.
Indeed, the pairs of complements can then be arranged as follows. We

first put (Fig. 156) I and III in the centre of the cross, with their elements
diagonally opposed (A.II.50); then the pairs II, IV, V, VI are written in
the two vertical rows and the remaining ones in the two horizontal rows
44
This is by no means obligatory, we might just as well take the first 8k + 2 and the
last 8k + 2 (their complements). See Les carrés magiques, p. 98 (Russian edition, pp.
109–110), or below, Fig. 163.
as seen in Fig. 156 (and described in A.II.52). With this arrangement the
central 2 × 6 core of the 2 × n cross is equalized, for we find the following
sums:
— diagonally in the central square, n2 + 1 for two cells;
— vertically in the columns, 3(n2 + 1) for six cells;
— horizontally in the lines, 3(n2 + 1) for six cells.
We have thus found the sum due. Our figure 157 applies this to order 14,
with the remainder of the cross filled with neutral placements.
n2 n2
2
+4 2
−3
n2 n2
2
−1 2
+2
n2 n2 n2 n2 n2 n2
2
−7 2
+7 2 2
+3 2
+9 2
−9
n2 n2 n2 n2 n2 n2
2
+8 2
−6 2
−2 2
+1 2
−8 2
+ 10
n2 n2
2
−4 2
+5
n2 n2
2
+6 2
−5
Fig. 156
81 116
115 82
114 83
84 113
102 95
97 100
73 123 122 76 91 105 98 101 107 89 77 119 118 80
124 74 75 121 106 92 96 99 90 108 120 78 79 117
94 103
104 93
85 112
111 86
110 87
88 109
Fig. 157
The four squares of order 6 left will be filled individually with the
remaining smaller numbers and their complements, using the known ar-
rangement for order 6 (Fig. 158; see Fig. 100 & Fig. 33). We may also
divide the rest of the square into subsquares of order 4, which will either
be whole or divided by the cross (Fig. 159).
5 196 2 194 188 6 81 116 23 178 20 176 170 24
7 11 184 181 18 190 115 82 25 29 166 163 36 172
8 182 17 12 183 189 114 83 26 164 35 30 165 171
187 16 179 186 13 10 84 113 169 34 161 168 31 28
193 185 14 15 180 4 102 95 175 167 32 33 162 22
191 1 195 3 9 192 97 100 173 19 177 21 27 174
73 123 122 76 91 105 98 101 107 89 77 119 118 80
124 74 75 121 106 92 96 99 90 108 120 78 79 117
41 160 38 158 152 42 94 103 59 142 56 140 134 60
43 47 148 145 54 154 104 93 61 65 130 127 72 136
44 146 53 48 147 153 85 112 62 128 71 66 129 135
151 52 143 150 49 46 111 86 133 70 125 132 67 64
157 149 50 51 144 40 110 87 139 131 68 69 126 58
155 37 159 39 45 156 88 109 137 55 141 57 63 138 Fig. 158
1 194 191 8 9 186 81 116 183 16 17 178 175 24
192 7 2 193 184 15 115 82 10 185 176 23 18 177
6 189 196 3 14 181 114 83 188 11 22 173 180 19
195 4 5 190 187 12 84 113 13 182 179 20 21 174
25 170 167 32 33 162 102 95 159 40 41 154 151 48
168 31 26 169 160 39 97 100 34 161 152 47 42 153
73 123 122 76 91 105 98 101 107 89 77 119 118 80
124 74 75 121 106 92 96 99 90 108 120 78 79 117
30 165 172 27 38 157 94 103 164 35 46 149 156 43
171 28 29 166 163 36 104 93 37 158 155 44 45 150
49 146 143 56 57 138 85 112 135 64 65 130 127 72
144 55 50 145 136 63 111 86 58 137 128 71 66 129
54 141 148 51 62 133 110 87 140 59 70 125 132 67
147 52 53 142 139 60 88 109 61 134 131 68 69 126

Fig. 159
Since A has no illustration whatsoever of this method, his readers can

hardly have benefited from his brief explanation. He merely mentions
examples of such a division (A.II.46): a square of order 14 with the cross
and four 6 × 6 squares, as in our figure; a square of order 10 with the
cross and four 4 × 4 squares.45 He does not mention the possibility of
constructing in this way the least possible order, n = 6 (Fig. 160); he does
not observe that, generally, whatever the evenly odd order n = 4k + 2, it
can be reduced to a cross and 4 × 4 squares.
1 34 22 15 31 8 18 11 25 27 9 21
32 7 17 20 2 33 22 1 34 31 8 15
11 25 18 21 27 9 17 32 7 2 33 20
26 12 16 19 10 28 14 6 29 36 3 23
6 29 14 23 36 3 24 35 4 5 30 13
35 4 24 13 5 30 16 26 12 10 28 19
Fig. 160 Fig. 161
1 22 34 31 15 8 11 24 28 9 21 18
11 18 25 27 21 9 22 17 35 2 12 23
32 17 7 2 20 33 10 30 1 4 34 32
6 14 29 36 23 3 27 7 33 36 3 5
26 16 12 10 19 28 16 19 6 31 26 13
35 24 4 5 13 30 25 14 8 29 15 20
Fig. 162 Fig. 163

Remark. The method of the cross enables us to reduce the arrangement
for an evenly odd square to that for even squares by first eliminating
two pairs of rows along the central axes. Now the result would be the
same by filling first the outer border, for we would also be left with
even squares. The two constructions are equivalent: the cross with its
four central cells reduces, by transposition, to the outer border with
the four corner cells common to two rows (Fig. 160 and Fig. 161).
Thus the presence of such a method in the tenth century is not really
surprising. By the way, these are just two possible situations for the
pairs of rows, which could also be placed elsewhere (symmetrically) in
the square (Fig. 162).46 As for the filling, as already mentioned (p. 109,
45
There is at the end (A.II.54) an allusion to the application of this treatment to
orders 18 and 22.
46
A mediaeval 14 × 14 example in Les carrés magiques, p. 99 (Russian edition, p.
note 44), it could begin with the first numbers and their complements
in the cross (Fig. 163)
B. Method of the central square

Filling a central square is another way, explained by A, of reducing the
construction of evenly odd squares to that of smaller even-order squares
(A.II.49). Indeed, filling a single evenly odd square in the centre of the
main square will leave even squares and rectangular strips, the latter
being then filled by means of neutral placements.
Consider namely a square of order n = 4k + 2 (k ≥ 2), in the centre
of which we take a square of order 4t + 2 (1 ≤ t < k). The main square
is then divided into the following unequal parts (Fig. 164). The centre
is occupied by the square of order 4t + 2, for which we may choose for
convenience t small. This square being 2(k − t) cells away from the main
square’s sides, it leaves in the corners four squares of the same order.
Whether these squares are evenly even or evenly odd, we know how to
fill them with pairs of complements or reduce them to smaller squares to
be filled in the same way; if need be, as in the case of a 10 × 10 square
containing a square of order 6, we may consider the four corner parts as a
single entity, just as in our preceding figures 160 and 163. Finally, there
remain on the sides, if k = 3t + 1, four rectangles not interfering with
the two diagonals; therefore, we need only fill them with the sum due
horizontally and vertically, which is easy.
Fig. 164
For placing the numbers, we may begin with the corner squares, con-
tinue with the central square, and end with the rectangles. But this
sequence can be varied since each entity filled with pairs of complements
will in any case display its sum due. As for the rectangles, which do
not cut the diagonals, we shall use neutral placements, as explained in
A.II.49: their dimensions, 4t+2 by 2(k −t), make that possible (see Fig. a
37* and Fig. a 38* in the translation). See Fig. 165 (4 × 4 squares filled as
in Fig. 33, and the outer border of the 6×6 square filled as in Fig. 100). If
111).
need be, we can combine a rectangular strip with the one opposite and fill
them together. If the size of the strips permits, we may consider squares
in them, each to be filled separately, which leaves narrower rectangular
strips. Fig. 166 is an example of the this last case.
1 194 191 8 51 55 59 138 142 146 9 186 183 16
192 7 2 193 145 141 137 60 56 52 184 15 10 185
6 189 196 3 144 140 136 61 57 53 14 181 188 11
195 4 5 190 54 58 62 135 139 143 187 12 13 182
107 89 88 110 37 164 34 162 156 38 63 133 132 66
103 93 92 106 39 43 152 149 50 158 67 129 128 70
99 97 96 102 40 150 49 44 151 157 71 125 124 74
98 100 101 95 155 48 147 154 45 42 126 72 73 123
94 104 105 91 161 153 46 47 148 36 130 68 69 127
90 108 109 87 159 33 163 35 41 160 134 64 65 131
17 178 175 24 122 118 114 83 79 75 25 170 167 32
176 23 18 177 76 80 84 113 117 121 168 31 26 169
22 173 180 19 77 81 85 112 116 120 30 165 172 27
179 20 21 174 119 115 111 86 82 78 171 28 29 166

Fig. 165
1 194 191 8 9 186 73 124 183 16 17 178 175 24
192 7 2 193 184 15 123 74 10 185 176 23 18 177
6 189 196 3 14 181 122 75 188 11 22 173 180 19
195 4 5 190 187 12 76 121 13 182 179 20 21 174
57 138 135 64 85 116 82 114 108 86 25 170 167 32
136 63 58 137 87 91 104 101 98 110 168 31 26 169
65 131 130 68 88 102 97 92 103 109 69 127 126 72
132 66 67 129 107 96 99 106 93 90 128 70 71 125
62 133 140 59 113 105 94 95 100 84 30 165 172 27
139 60 61 134 111 81 115 83 89 112 171 28 29 166
49 146 143 56 41 154 77 120 151 48 33 162 159 40
144 55 50 145 152 47 119 78 42 153 160 39 34 161
54 141 148 51 46 149 118 79 156 43 38 157 164 35
147 52 53 142 155 44 80 117 45 150 163 36 37 158

Fig. 166
A does not give any illustration. The text merely mentions possible
divisions (A.II.46): a square of order 14 with a 6 × 6 square in the centre

and eight 4 × 4 squares around it, as in our above figure; a square of order
18 with a central 10 × 10 square and twelve 4 × 4 squares around it (Fig.
167). A does not mention the possibility of constructing in this way the
square of order 10 with an inner 6 × 6 square and 6 × 2 rectangles which
could be combined (nor did he consider the possibility of a cross within
a square of order 6 leaving corner parts which could be combined).47
Fig. 167
47
He does, however, mention the case of 6 × 2 rectangles (A.II.49) although they
cannot be filled separately.
Part II
Translation
Editorial procedure
What is in round brackets has been added by us to clarify the text.
Square brackets indicate interpolations, while angular brackets are occa-
sionally used (if relevant) to indicate lacunas. That is, by the way, only
the case for text A, since this is its first edition and translation. The
same having been done for text B in our edition of it, that is not repeated
in the extracts presented here.
The figures of magic squares, which in the manuscript of A are mostly
put at the end of the section A.II, have in the translation been placed
where they fit. Those with an asterisk are our additions. All follow the
orientation right to left, as in the manuscript but unlike in our commen-
tary in Part I. As to B, the manuscript has omitted all figures, so they
have been reconstructed according to the text’s indications; the dubious
cases, as well as the figures added by us, are pointed out in the footnotes.
Book III of the Commentary of ‘Alı̄ ibn
Muh.ammad al-Ant.ākı̄ on the first part
of the book by Nicomachos of Gerasa
known as Arithmetic

In the name of God the merciful, the compassionate.
Book III of the Commentary of ‘Alı̄ ibn Ah.mad ibn Muh.ammad

al-Ant.ākı̄ on the book48 by Nicomachos of Gerasa known as Arith-
metic, on the mutual relations of numbers as regards their composi-
tion and their decomposition.49
There will be three chapters:
— Chapter I, on the affinity of numbers, their dissimilarity and their
analogies,50 their relations apart from the analogy they display in their
general and particular aspects.
— Chapter II, on the science of the magic square, in two paragraphs:
the first on the science of odd orders, the second on the science of even
orders.
— Chapter III, on the determination of hidden numbers51 .
Chapter I of Book III

(A.I.1) Numbers incomposite are those which no other number than unit
divides; they are also called primes.52
(A.I.2) There is among even numbers no other prime number than 2.
(A.I.3) Numbers composite to one another are those which are all divided
by, besides unit, some other number common to them.53
48
Here we no longer have ‘on the first part of the book’, which indeed did not make
much sense; and al-Ant.ākı̄’s name now appears in full.
49
This specification regards Nicomachos’ book (if al-Ant.ākı̄’s commentary, it could
at most apply to the first chapter).
50
These two terms will be used later on, but more specifically, for numbers which
are relatively prime and proportional, respectively.
51
Numbers thought of by one person and found out by another.
52
(A.I.1) = Elements, Book VII, definition 12. (We shall adopt the numbering
in Heiberg’s Greek text of the Elements; the numbering in Heath’s translation and
Heiberg’s Latin translation facing the Greek text may differ, as here.) The word in-
composite with the meaning of prime corresponds to the Greek ἀσύνθετος, antonym of
σύνθετος, composite; Nicomachos too speaks of πρω̃τος καı̀ ἀσύνθετος ἀριθμός (I.xi.1–
2). The Arabic word employed here for ‘incomposite’, mutabāyin, also means ‘relatively
prime’ (see A.I.40).
53
(A.I.3) = Elements, VII, def. 15 (‘composite to one another’ = σύνθετοι πρὸς
ἀλλήλους, thus numbers with a common divisor = 1). We would also expect the
definition of ‘prime to one another’.
Translation of Text A 121
(A.I.4) Any even number is composite to any even number.
(A.I.5) Any even number when multiplied by an odd number gives an

even result.54
(A.I.5 ) It will be evenly odd.55

(A.I.6) Any even number when multiplied by an even number gives an
evenly even result.56
(A.I.7) Any odd number when multiplied by an odd number gives an

odd result.57
(A.I.7 ) It is called oddly odd.58
(A.I.8) Any even number from which an even number is subtracted leaves
an even remainder.59
(A.I.9) If from an even number an odd number is subtracted the remain-

der will be odd.60
(A.I.10) If from an odd number an odd number is subtracted the re-
mainder will be even.61
(A.I.11) If from an odd number an even number is subtracted the re-

mainder will be odd.62
(A.I.12) Any odd number dividing an even number divides it according

to an even number.63
(A.I.13) If an odd number divides an even number it will also divide its
half.64
54
(A.I.5) = Elements, IX,28; Nicomachos II.xxiv.10.
55
᾿Αρτιάκις περισσός. Elements, VII, def. 9; number of the type 2t (2k + 1), with t, k
natural. But see A.I.14.
56
(A.I.6) ≈ Elements, VII, def. 8; Nicomachos II.xxiv.10. But see A.I.16.
57
(A.I.7) = Elements, IX,29; Nicomachos II.xxiv.10.
58
(A.I.7 ) ≈ Elements, VII, def. 11 (περισσάκις περισσός).
59
(A.I.8) = Elements, IX,24.
60
61
62
63
Statement belonging to the proof of Elements, IX,30 (= A.I.13): περισσὸς γὰρ
ἀριθμὸς ὁ Α ἄρτιον τὸν Β μετρείτω (. . . ) ὥστε ὁ Α τὸν Β μετρει̃ ἀρτιάκις. Became later
an independent proposition (IX,30 in some Arabic versions of the Elements; IX,31 in
the 13th-century mediaeval Latin edition by Johannes Campanus).
64
122 Chapter I of Book III
(A.I.14) Any number with an odd half is evenly odd.65
(A.I.15) Any number of which half of its half is odd is evenly evenly
odd.66
(A.I.16) Any number reducible to unit by halving is evenly even.67
(A.I.17) Any number repeatedly halved without being reduced by halv-

ing to unit but [ some half before it ] with an odd number is evenly evenly
odd.68
(A.I.18) Any number dividing two numbers and dividing itself is the
greatest number dividing these two numbers.69
(A.I.19) Any number dividing two numbers will also divide the greatest
number dividing them.70
(A.I.20) Any number dividing three numbers and dividing itself is the
greatest number dividing the three numbers [ which this number divides
jointly without remainder ].71
Example of 15, 20, 25: the number dividing them all is 5, and 5 divides
itself, it is therefore the greatest number dividing these three numbers.
(A.I.21) For any pair of different numbers the lesser is either part or
parts of the greater.72
65
(A.I.14) = Elements, IX,33 (ἀρτιάκις περισσός μόνον, thus number of the form
2(2k + 1), Arabic zauj al-fard faqat.). Without the μόνον this designation is ambiguous
(see A.I.5 ).
66
Number of the form 22 (2k + 1). But see A.I.17.
67
Number of the form 2n , thus pure power of 2. See Elements, IX,32 (ἀρτιάκις ἄρτιος
μόνον, thus evenly even only).
68
Number of the form 2m (2k + 1) (m ≥ 2). See A.I.15.
69
Absurd. At best, we are to read: ‘Any number dividing another number (‘adadan
instead of ‘adadayn) is, since it divides itself, the greatest number dividing these two
numbers’. Now such a formulation can be traced to Elements, VII,2, where Euclid
teaches how to find the greatest common divisor of two numbers not prime to one
another. In the proof, Euclid dismisses the case in which the lesser of the two numbers
(segments) ΓΔ, AB divides the other exactly, since it would itself be their greatest
common divisor (Εἰ μὲν ο˜υν᾿ ὁ ΓΔ τὸν ΑΒ μετρει̃, μετρει̃ δὲ καı̀ ἑαυτόν, ὁ ΓΔ ἄρα τω̃ν
ΓΔ, ΑΒ κοινὸν μέτρον ἐστίν. καὶ φανερόν, ὅτι καı̀ μέγιστον· οὐδεı̀ς γὰρ μείζων του̃ ΓΔ
τὸν ΓΔ μετρήσει). Similar curious statement in A.I.20.
70
(A.I.19) = Elements, VII,2, corollary.
71
Once again absurd, but reading confirmed by the subsequent example. Originates
from the proof of Elements, VII,3.
72
(A.I.21) = Elements, VII,4; if a and b (a > b) are two natural numbers, b is either
a divisor or a non-aliquot fraction (μέρος ἤ μέρη) of a.
(A.I.22) If a number is some part of another number and another number

is the same part of another number, the sum of the two lesser will be the
same part of the sum of the two greater as the one of its associate.73
Example of 2, 4 and 5, 10: the two lesser (taken together), thus 2 and
5, are the same part of the two greater (taken together), thus 4 and 10,
as was 2 of 4.
(A.I.23) The same holds for parts.74
(A.I.24) If there are four numbers, and the first is a part of the second
and the third the same part of the fourth, then, these numbers being
permutated, the first will be the same part, or the same parts, of the
third as the second of the fourth.75
Such is the case for 2, 4 and 3, 6: the ratio of 2 to 4 equals the ratio
of 3 to 6 and, after their being permutated, the ratio of 2 to 3 will equal
the ratio of 4 to 6.
(A.I.25) The assertion holds as well for parts as for just one. If the
quantity according to which one (number) divides a second number is the
same as that according to which a third number divides a fourth, then,
by permutation, as the first will be to the third, so will the second be to
the fourth.76
(A.I.26) If a number is a part of another number and this very same
part is subtracted from them, the remainders will be to one another as
were the wholes.77
Such is the case for 4 and 8, the former being half of the latter;
subtracting from the two numbers their halves, 4 leaves 2, and 8, 4, which
are also in this ratio: 2, remaining from 4, is half of 4, remaining from 8.
(A.I.27) This remains true for parts.78
73
(A.I.22) = Elements, VII,5; if a = n1 b and c = n1 d, then a + c = n1 (b + d).
74
(A.I.23) = Elements, VII,6; if a = mn
b, c = mn
d, then a + c = m n
(b + d).
75 1 1
(A.I.24) = Elements, VII,9; if a = n b, c = n d (thus a : b = c : d), then a : c = b :
d. Permutation in a proportion (ἐναλλάξ, alternando) means exchanging the middle
terms (A.I.30).
76
(A.I.25) = Elements, VII,10; same case as before, with a = m n
b, c = mn
d, and thus
again a : c = b : d.
77
This should be the analogue of A.I.22, the sum becoming the difference (Elements,
VII,7). That is, with four numbers, if a = n1 b, c = n1 d, with a > c, then a−c = n1 (b−d).
But the example confirms that the author did indeed mean that if a = n1 b, then
(a − n1 a) : (b − n1 b) = a : b.
78
(A.I.27) = (in principle) Elements, VII,8; if a = m n
b and c = m n
d, then a − c =
m
n
(b − d). Analogous to A.I.23.
(A.I.28) This (also) remains true for two numbers from which are sub-
tracted two numbers such that the ratio of the (numbers) subtracted is
the same as the ratio of the wholes: the ratio of the remainders will be
the same as the ratio of the wholes.79
(A.I.29) If there are numbers, then further numbers associated with
them, in the same quantity, and each pair of the second ones is in the
same ratio as each pair of the first ones, they will also be in proportion
by equality of ratio.80
Such is the case for 2, 4, 8 and 5, 10, 20, according to this model:
2 4 8
5 10 20;
for, by equality of ratio, the ratio of 2 to 8 will equal the ratio of 5 to
20. Indeed, the ratio of 2 to 4 equals the ratio of 5 to 10; then, when we
permutate, the ratio of 2 to 5 will equal the ratio of 4 to 10.81 Likewise,
the ratio of 4 to 8 equals the ratio of 10 to 20; then, when we permutate,
the ratio of 4 to 10 will equal the ratio of 8 to 20. Now the ratio of 4 to
10 was equal to the ratio of 2 to 5. So the ratio of 2 to 5 will equal the
ratio of 8 to 20. Therefore, when we permutate, the ratio of 2 to 8 will
equal the ratio of 5 to 20.
(A.I.30) The meaning of ‘permutation’ is that you take (the ratio of) the
antecedent to the antecedent and (of) the consequent to the consequent.82
(A.I.31) For any pair of numbers which multiply each other, one of them
is repeated as many times as there are units in the other.83
(A.I.31 ) This happens also for (aliquot) fractions. Indeed, for any pair of
fractions which multiply each other, one of them will be split up according
to the quantity of parts of 1 of the other.84
79
(A.I.28) = Elements, VII,11; if a : b = c : d, then (a − c) : (b − d) = a : b (a > c,
b > d).
80
(A.I.29) = Elements, VII,14. Equality of ratio (δι᾿ ἴσου, ex æquali) infers that if
we have numbers a, b, c and d, e, f with a : b = d : e and b : c = e : f , we shall have
a : c = d : f ; and likewise for two sequences of more than three terms. The verification
in the numerical example (particular case: a : b = b : c) transposes Euclid’s proof, in
fact.
81
By A.I.24.
82
(A.I.30) = Elements, V, def. 12. Thus, if a : b = c : d (where a and c are the
antecedents and b and d the consequents), permutation will lead to a : c = b : d. We
should have expected this definition in A.I.24.
83
(A.I.31) = Elements, VII, def. 16 (definition of multiplication as a repeated addi-
tion).
84
‘parts of 1’: the denominator of the fraction. A general fraction m n
is expressed
verbally as ‘m parts of n parts of 1’.
(A.I.31 ) In the case of (aliquot) fractions being multiplied by integers,

this can be understood in two ways: either by splitting up the integer
according to the quantity (of parts of 1) of the multiplying fraction or by
repetition of the fraction according to the quantity of the integer.
Examples (for the three cases). You multiply 5 by 6, the result is 30,
which equals the repetition of six five times, or of five six times. For a
fifth by a sixth, you will split up the fifth according to the parts of the
sixth; this gives a sixth of a fifth, thus one part of 30. For a fifth by 5,
you will either split up 5 according to a fifth, which gives 1, or repeat a
fifth according to 5, which gives 1.
(A.I.32) If you multiply a number by two numbers, the ratio of the
products will equal the ratio of the multiplying numbers.85
Thus, if 4 and 5 are multiplied by 6, the two products are 24 and 30,
and the ratio of 24 to 30 will equal the ratio of one (multiplier) to the
other.
(A.I.33) If two numbers are prime to one another, a number dividing
one of them will be prime to the other.86
Thus, 3 and 4 are prime to one another; therefore 2, which divides 4,
will also be prime to 3.
(A.I.34) Likewise, if two numbers are prime to another number and you
multiply them, their product will be prime to this number.87
Such is the case for 3 and 4 relative to 5: their product, thus 12, is
also prime to 5.
(A.I.35) Likewise, the square of each of these two numbers will be prime
to each of them [ and also to 5 ].88
(A.I.36) The same holds if they are raised to the cube: the result of
raising each to the cube will be prime to the other [ (16 is prime to) 3,
and 9 is prime to 4, and each of them is prime to 5 ].89
85
(A.I.32) = Elements, VII,17–18.
86
(A.I.33) = Elements, VII,23; if a and b are prime to one another with a = k · c,
then c and b will be prime to one another. This is the first mention of numbers prime
to one another.
87
(A.I.34) = Elements, VII,24; if a, b are prime to c, then so will ab.
88
(A.I.35) = Elements, VII, 25; if a, b are prime to one another, a2 will be prime
to b, and b2 to a. The example of 3, 4, 5 in A.I.34 has been applied to this case by
a reader, and his gloss has been (partly) copied in A.I.36 and A.I.35 by some later
copyist: 42 is prime to 3, 32 to 4 —and both to 5.
89
If a and b are prime to one another, a3 and b3 will be prime to b and a, respectively.
(A.I.37) Any number is either prime or divisible by a prime number.90

(A.I.38) Any prime number is prime to each number not divisible by
this prime.91
(A.I.39) If the product of the multiplication of any two numbers is di-
visible by a prime number, this prime number will divide one of the two
factors.92
(A.I.40) If two relatively prime numbers are multiplied, the resulting
number will be the least number they divide.93
(A.I.41) If any quantity of numbers beginning from the unit are in con-
tinued proportion, then the ratio of the least to the greatest will have the
quantity of some number among them.94
(A.I.42) If any quantity of numbers beginning from the unit are in con-
tinued proportion, the prime number which divides the last will divide
the number following the unit.95
Such is the case for 1, 2, 4, 8: the prime number dividing 8, which is
the greatest, is 2, and 2 divides 2.
(A.I.43) If any quantity of numbers beginning from the unit are in
continued proportion and that following the unit is a prime, then the
Now Elements, VII,27 proves that if a and b are prime to one another, a2 and b2 , as
well as a3 and b3 , will be prime to one another. Even considering that the text of
A.I.35 and A.I.36 is unclear, it does not seem that such a statement was made.
90
(A.I.37) = Elements, VII, 32.
91
(A.I.38) = Elements, VII,29; if p is prime and does not divide a, a and p will be
prime to one another.
92
(A.I.39) = Elements, VII,30; if p, prime, divides the product ab, it will divide a
or/and b.
93
(A.I.40) = Elements, VII,34 (finding the lowest common multiple of two numbers),
first half of the proof (ed. Heiberg, p. 256: ῎Εστωσαν πρότερον οἱ Α, Β πρω̃τοι πρὸς
ἀλλήλους, καı̀ ὁ Α τὸν Β πολλαπλασιάσας τὸν Γ ποιείτω (. . . ). οἱ Α, Β ἄρα τὸν Γ
μετρου̃σιν. λέγω δή, ὅτι καı̀ ἐλάχιστον); in our terms, the lowest common multiple of
any two relatively prime numbers is their product.
94
(A.I.41) = Elements, IX,11. Better: ‘then the lesser will divide the greater ac-
cording to the quantity of some number among them’. This is our first encounter with
the continued proportion, where the ratio of two consecutive terms remains constant.
Then, if a1 : a2 = a2 : a3 = . . . = an−1 : an , with a1 = 1, that is, if 1, a, a2 , . . . , an−1
form a geometric progression with common ratio a, then an−i : ar (n − i > r) will be
some term in the progression (namely an−i−r ).
95
(A.I.42) = Elements, IX,12; if, in the previous progression, some prime number p
divides an−1 , it will divide a. The subsequent example is a rather poor illustration of
that.
greatest will only be divisible by a number among them.96

(A.I.44) If there are prime numbers in a known quantity, there will exist
a larger quantity of prime numbers.97
Proof of this.
(a) We determine the greatest number which divides them all. It will be
determined by finding the greatest number dividing two numbers among
them, then the number dividing these two numbers —that is, the greatest
number dividing the two numbers, and the third one—, then the number
dividing this number and the fourth one, and so on continuously until we
have determined the greatest number which divides them all.98
(b) Determining the greatest number which divides two numbers neither
equal nor prime (proceeds as follows) —imposing not prime is because
if they were it would not be possible for one number to divide them
both since prime numbers are divided by no number (other than 1), and
(imposing) not equal is because if they were one of them would divide the
other since each one divides itself. If two (numbers) are neither prime nor
equal, you will find the greatest number dividing them by subtracting the
greater from the lesser (!) until there remains less than the lesser, then
subtracting from the lesser the remainder from the greater until there
remains less than the lesser (!), and you will go on, always proceeding in
this way, until reaching a number which cannot be subtracted from the
other because of their equality.99 It will be the least (!) number dividing
them both.
96
(A.I.43) = Elements, IX,13; if, in the previous progression, a = p is prime, an−1
will be (strictly) divided only by the preceding terms of the progression (thus by the
powers of p lower than pn−1 = an−1 ). It is likely that the words in angular brackets
were omitted by a copyist; see, though, A.I.92 and A.I.94.
97
(A.I.44) = Elements, IX,20 (stating and proving that the number of primes is in-
finite). What follows has nothing to do with Euclid’s demonstration, which is perfectly
logical and clear (see below).
98
With the text as it stands we are to look for the highest common divisor of all
these primes, which of course makes no sense since they are primes. If we are, as we
should be, looking for their lowest common multiple, it will just be their product (as,
by the way, inferred from A.I.40). This strange confusion between highest common
divisor (H.C.D.) and lowest common multiple (L.C.M.) would seem to suggest that
we have here an interpolation. The subsequent part b is drawn from Elements, VII,
2, on finding the H.C.D. of two numbers not prime to one another (so-called ‘Euclid’s
algorithm’), and is thus pointless here. (That, too, contains incongruities.)
99
Same expression (for us, odd) in Nicomachos’s example of H.C.D. determination:
ἑβδομάδα δὲ ἀπο ἑβδομάδος οὐ δυνατόν ἀφαιρεθη̃ναι (I.xiii.13); since the steps in Eu-
clid’s algorithm are successive subtractions with remainders, the ‘number which cannot
be subtracted from the other because of their equality’ is the last remainder.
(c) Next you add 1 to what you have found.100 If what arises after adding
1 is a prime number, the assertion is true, for there results another (prime)
number, different from the (previous) ones and larger than them. If the
resulting number is composite, we take101 the prime number which divides
it; it will be another (prime) number, not equal to one of these (prime)
numbers.102 Indeed, it is not possible that it (= the prime number found)
be equal to one of them, for then it (= the composite number) would be,
absurdly, the least number divided by the known primes, so there would
be no other quantity of primes than this one.
(A.I.45) If three numbers in continued proportion are the least of those
which have this ratio, any sum of two of these numbers will be prime to
the third one.103
(A.I.46) If of two numbers each is prime to the other, then the ratio of
the first to the second will not equal the ratio of the second to some other
number.104
(A.I.47) The same holds for any quantity of numbers in continued pro-
portion of which the extreme terms are prime to one another: the ratio
of the first to the second will not equal the ratio of the last to some other
number.105
(A.I.48) For any pair of commensurable numbers multiplying the ratio
of unity to their (greatest) common divisor taken of the one by the other,
the resulting product will be the least number they divide.106
100
Euclid’s demonstration runs (in our terms) as follows. Given the primes
p1 , p2 , . . . , pm and supposing their L.C.M. p1 · p2 · p3 · . . . · pm calculated, we consider
p1 ·p2 ·p3 ·. . .·pm +1. It will be either a prime, evidently then a new one, or a composite
number, thus divisible by some prime P (according to Elements VII,31, omitted from
the Euclidean theorems reproduced by our author). Now P cannot be one of the pi ;
if it were, it would divide not only (by hypothesis) p1 · p2 · p3 · . . . · pm + 1, but also
p1 · p2 · p3 · . . . · pm , therefore their difference, 1, which is absurd. In either case we have
found a new prime, and this can be repeated with any set of given primes.
101
Like the Greek εἰλήφθω (e.g. Heiberg’s ed. of the Elements, II, p. 196,19), this
means that we determine it.
102
So far so good. But what comes now is absurd.
103
(A.I.45) = Elements, IX,15; if α2 , αβ, β 2 are three numbers in geometric pro-
gression in the lowest terms, α2 + αβ, α2 + β 2 , αβ + β 2 will be respectively prime to
β 2 , αβ, α2 (the extreme terms α2 , β 2 must be squares according to Elements, VIII,2
corollary, here A.I.53).
104
(A.I.46) = Elements, IX,16; if a, b are prime to one another (and then also, by
A.I.35, b2 and a), then c with a : b = b : c cannot be an integer.
105
(A.I.47) = Elements, IX,17; in a geometric progression a1 , a2 , . . . , an in which the
extreme terms are prime to one another, there will be no integer c with a1 : a2 = an : c.
106
(A.I.48) ≈ Elements, VII,34, second part of the proof (see A.I.40); if m is the
Thus, 6 and 4 are commensurable, for 2 is commensurable to both; we

take the ratio of 1 to 2, thus a half, of the one, and multiply the result by
the other; this gives 12, which is the least number which 6 and 4 divide.
(A.I.49) Likewise for three, four or any quantity of numbers: you take
the least number which two numbers among them divide and proceed
similarly.107
(A.I.50) If two numbers divide a number, the least number they divide
will also divide this number.108
(A.I.51) For any number divided by a number there will be in the divided
number a part homonymous with the dividing number.109
(A.I.52) If in any quantity of numbers in continued proportion the two
extremes are prime to one another, these numbers will be the least num-
bers in this ratio.110
(A.I.53) If three numbers in continued proportion are the least numbers
in this ratio, the two extremes will be squares. If there are four numbers,
the two extremes will be cubes.111
(A.I.53 ) If there are five numbers, the two extremes will be fourth pow-
ers. If there are six numbers, the two extremes will be fifth powers. If
there are seven numbers, the two extremes will be sixth powers. And so
on invariably, as (for the sequence) 1, 2, 4, 8, 16, 32, 64, 128.112
(A.I.54) If numbers in any quantity are in continued proportion and the
least in this ratio, the two extremes will be prime to one another.113
1
H.C.D. of two composite numbers a and b, m ab will be their L.C.M. Note the omission
of an essential word, for which the example does not really make up.
107
(A.I.49) = Elements, VII,36 generalized. The L.C.M. of several numbers is found
by starting with that of the first two, then finding the L.C.M. of the result and the
third, and so on.
108
(A.I.50) = Elements, VII,35; any number which two numbers a and b divide will
also be divided by their L.C.M. (that is: any common multiple of two numbers is a
multiple of their L.C.M.).
109
(A.I.51) = Elements, VII,37; if d divides a, a has a dth integral part. The word
‘homonymous’ is in Euclid ὁμώνυμος and in Nicomachos, παρώνυμος (I.viii.10).
110
(A.I.52) = Elements, VIII,1; if the two extreme terms of a geometric progression
are prime to one another, this progression is in its lowest terms.
111
(A.I.53) = Elements, VIII,2, corollary. Proposition VIII,2 teaches how to con-
struct any number of terms in continued proportion, say n + 1 terms, such that they
are the lowest having a given ratio, say α : β. These will be αn , αn−1 β, αn−2 β 2 , . . . ,
αβ n−1 , β n , with α, β prime to one another. Euclid’s corollary applies that to the cases
n = 2 and n = 3.
112
Here with eight terms, thus seventh powers.
113
(A.I.54) = Elements, VIII,3; since α : β, the ratio of two consecutive terms, is
(A.I.55) The ratio of any two plane numbers is compounded of the ratios
of their sides.114
Thus for 20 and 6, for the ratio of these two planes is compounded of
the two ratios of their sides; indeed, the sides of 6 are 2 and 3, those of
20 are 5 and 4, the ratio of 2 to 4 is a half, the ratio of 3 to 5 is three
fifths, then the ratio of 6 to 20 will be half of 35 , thus 15 + 10
1 115
.
(A.I.56) If in any quantity of numbers in continued proportion the first
divides the last, it will also divide the second.116
(A.I.57) If between any two numbers there fall numbers in continued
proportion with them, the quantity of numbers falling between them will
be the same as that (of the numbers falling) between any two numbers
having the same ratio as the (original) ones (and all will be in continued
proportion).117
(A.I.58) If two numbers are prime to one another and numbers in any
quantity fall between them and all are in continued proportion, then the
quantity of numbers falling between each of them and the unit will equal
(the quantity) of the numbers falling between the two numbers and all
will be in continued proportion.118
(A.I.59) Between any two square numbers there will fall a number such
that all three are in continued proportion, and the ratio of the square to
the square is the ratio duplicate of that of the side to the side.119
expressed in its lowest terms, α and β must be prime to one another, thus also αn and
βn.
114
(A.I.55) = Elements, VIII,5. The definition of ‘plane number’ is missing: it is
the product of two numbers, called its sides (Elements, VII, def. 17). Thus, here, if
a = bc et d = ef , then ad = eb · fc . A ratio (λόγος = nisba) is said to be ‘compounded
of’ (συγκείμενος ἐκ = mu’allifa min) two ratios when it results from their product
(Elements, VI, def. 5 —interpolated). This term is defined later on (A.I.62 ).
115
Since antiquity, a non-aliquot fraction is frequently converted into a sum of simpler
fractions, preferably aliquot ones. There is all the more reason to do so in Arabic as
only aliquot fractions with denominators from 2 to 10 have their own name, the others
requiring periphrasis (‘m parts of n parts of 1’, see note to A.I.31 ).
116
(A.I.56) = Elements, VIII,7; if, in the geometric progression a1 , a2 , . . . , an (a1 =
1), a1 divides an , it will also divide a2 (as well as all ak ).
117
(A.I.57) = Elements, VIII,8; if a certain number of geometric means falls (ἐμπίπτειν
= waqa‘a) between two numbers a, b, there will be as many between two numbers in
the same ratio as a, b.
118
(A.I.58) = Elements, VIII,9; if between a and b, prime to one another, there are
n − 1 geometric means, the n + 1 terms will have the form a = αn , αn−1 β, αn−2 β 2 ,
. . . , αβ n−1 , β n = b (see note to A.I.53). We may in particular take α (or β) equal to
1.
119 2
(A.I.59) = Elements, VIII,11; a2 : ab = ab : b2 , and ab2 = ab · ab ; Nicomachos
Thus for 4 and 9: between them falls a number, which is 6, and they
are in continued proportion (for) the ratio of 4 to 6 is the same as the
ratio of 6 to 9; and the ratio of 4 to 9 is the same as the ratio of the side
of 4, thus 2, to the side of 9, thus 3, duplicate, for it is 32 of 23 , thus 49 .
(A.I.60) By duplication of a ratio is meant your multiplying one of the
two ratios by the other if there are two, and by itself if there is just one.120
(A.I.61) Between any two cubic numbers there will fall two numbers in
(continued) proportion with them, and the ratio of the cube to the cube
will be like the triplicate ratio of the side to the side.121
(A.I.62) By ‘triplication’ of a ratio is meant the repetition of this ratio
thrice; which is like multipliying that ratio by itself and then the result
again by the same (ratio).
(A.I.62 ) One says sometimes that a ratio is ‘compounded of ratios’ when
these ratios are multiplied together so as to produce a certain ratio.
(A.I.63) If numbers in any quantity are in continued proportion, their
squares will be in such a proportion; likewise, their cubes will also be in
such a proportion; and likewise the further solids.122
(A.I.64) If a square number divides a square number, the side will divide
the side; and if the side of a square divides the side of a square, the square
will divide the square.123
(A.I.65) If the square does not divide the square, neither will the side
divide the side; and if the side does not divide the side, neither will the
square divide the square.124
(A.I.66) Likewise, for any cube which divides a cube, the side will divide
the side; and if the side divides the side, the cube will divide the cube.125
II.xxiv.6 & 8. A ratio is said to be ‘duplicate’ when it is, in our terms, squared, and
‘triplicate’ when cubed. These two terms are explained at the end of this, and the
following, proposition.
120
‘Duplication’, like ‘triplication’ (A.I.62), in fact refers to the multiplication of a
single ratio; otherwise one speaks about ‘composition of ratios’ (A.I.62 ).
121 3
(A.I.61) = Elements, VIII,12; a3 : a2 b = a2 b : ab2 = ab2 : b3 , and ab3 = ab · ab · ab ;
Nicomachos II.xxiv.7 & 9.
122
(A.I.63) = Elements, VIII,13. The powers of the terms will also form a continued
proportion (the ratio itself will become duplicate, triplicate, and so on). The ‘further
solids’ are thus the higher powers.
123
(A.I.64) = Elements, VIII,14; if a2 divides b2 , then a divides b, and conversely.
124
(A.I.65) = Elements, VIII,16.
125
(A.I.67) If the cube does not divide the cube, the side will not divide
the side; and if the side does not divide the side, the cube will not divide
the cube.126
(A.I.68) Between any two similar plane numbers there will fall a number
such that they are in continued proportion, and the ratio of the one
(plane) to the other will be like the duplicate ratio of (corresponding)
sides.127
(A.I.69) Likewise also for two similar solids: there will fall between them
two numbers in (continued) proportion with them, and the ratio of one
solid to the other will be like the triplicate ratio of corresponding sides.128
(A.I.70) If a number falls between two numbers and they are (all three)
in continued proportion, the two numbers will be similar planes.129
(A.I.71) If two numbers fall between two numbers and they are (all four)
in continued proportion, the two numbers will be similar solids.130
(A.I.72) If three numbers are in (continued) proportion and the first is
a square, the third will be a square.131
(A.I.73) If four numbers are in (continued) proportion and the first is a
cube, the fourth will be a cube.132
(A.I.73 ) Likewise, if five numbers are in (continued) proportion and the
first is a fourth power, the fifth will also be a fourth power. And so (on)
126
127
(A.I.68) = Elements, VIII,18. Nicomachos II.xxiv.6. Euclid gives the definition
of ‘similar plane numbers’ at the beginning of Book VII (def. 22): their corresponding
sides are in the same ratio. Thus here, if ab, cd are such that a : c = b : d, aninteger
2
x may be found such that ab : x = x : cd (namely x = ad = bc), and ab = ac =
b 2 cd
d
. There is a (very) slight possibility that the definition omitted here, like that
of ‘plane number’ (above, note to A.I.55), occurred in the missing part of al-Ant.ākı̄’s
commentary.
128
(A.I.69) = Elements, VIII,19. Nicomachos II.xxiv.6. Similar solid numbers are
those having corresponding sides in the same ratio (Elements, VII, def. 22). Here if
abc, def are such that a : d = b : e = c : f , then we may find integralx, y such
that
3 3
abc : x = x : y = y : def (namely x = bdc, y = bdf ), and abc : def = ad = eb =
c 3
f
.
129
(A.I.70) = Elements, VIII,20; converse of A.I.68.
130
(A.I.71) = Elements, VIII,21; converse de A.I.69.
131
(A.I.72) = Elements, VIII,22; if a : b = b : c and a is a square, then c will also be
a square.
132
(A.I.73) = Elements, VIII,23; if a : b = b : c = c : d and a is a cube, then d will
also be a cube. For ad = ( ab )3 .
for the aforementioned proportion when forming (further) solids.133

(A.I.74) If the ratio of two numbers is like the ratio of a square number
to a square number, they will be similar planes.134
(A.I.75) If the ratio of the one to the other is like the ratio of a square
number to a square number and one of them is a square, the other will
be a square.135
(A.I.76) If the ratio of two numbers is like the ratio of a cubic number
to a cubic number and one of them is a cube, the other will be a cube.136
(A.I.77) For two similar plane numbers the ratio of the one to the other
is like the ratio of a square number to a square number.137
(A.I.78) The same holds for any pair of similar solids: the ratio of the one
to the other will be like the ratio of a cubic number to a cubic number.138
(A.I.79) If there are two plane numbers and the ratio of the one to the
other is like the ratio of a square number to a square number, they will
be similar planes.139
(A.I.80) If there are two solid numbers and the ratio of the one to the
other is like the ratio of a cubic number to a cubic number, they will be
similar solids.140
(A.I.81) If any two similar plane numbers are multiplied together, the
result will be a square.141
(A.I.82) If the multiplication of any two numbers produces a square,
they are similar planes.142
133
Higher powers, as in A.I.53 and A.I.63.
134
Proposition added to Elements, VIII,27 by Heron of Alexandria (see the 9th c.
commentary by al-Nayrı̄zı̄’s —Latinized as Anaritius— ed., p. 195,1–4, or Heath’s
translation of the Elements, II, p. 383). It is the converse of Elements, VIII,26 (below,
A.I.77).
135
(A.I.75) = Elements, VIII,24; a2 : b = c2 : d2 implies that b is a square.
136
137 2
(A.I.77) = Elements, VIII,26; if a : c = b : d (note to A.I.68), then ab
cd
= ac · db = ac2 .
138
(A.I.78) = Elements, VIII,27; evident since their sides are a, b, c and ta, tb, tc.
139
Without the first term ‘plane’, it is A.I.74. Found among scholia to Euclid (see
Euclidis Elementa V , p. 398,17–19; with a numerical example).
140
Without the first ‘solid’, it is the second addition by Heron to Elements, VIII,27
(al-Nayrı̄zı̄’s commentary, p. 195,4–7; Heath’s translation, II, p. 383), and the converse
of VIII,27.
141
(A.I.81) = Elements, IX,1. If ab and cd are two similar numbers, thus c = ta and
d = tb, then ab · cd = t2 (ab)2 .
142
(A.I.82) = Elements, IX,2; converse of A.I.81.
(A.I.83) The multiplication of a square number by a square number

produces a square.143
(A.I.84) If you multiply a square number by a number and the result is

a square, the multiplier will also be a square.
(A.I.85) If you multiply a square number by a number and the result is

not a square, the multiplier will not be a square.
(A.I.86) The same holds for the sequence (of higher powers), cube and
all other solids.144
(A.I.87) The multiplication of any two similar solid numbers produces

a cube.145
(A.I.88) If a solid number is multiplied by a solid number and the prod-

uct is a cube, the two solids will be similar.146
(A.I.89) Any cubic number which is multiplied by its side produces a

solid.147
(A.I.90) Likewise, any three numbers when multiplied among themselves

will produce a solid.148
143
Nicomachos II.xxiv.10. This proposition and the next two are appended to the
statement of Elements, IX,2 in Campanus’ 13th-century edition, as corollaries (as he
writes: ex his itaque patens est); they are also found, with proofs, in al-Nayrı̄zı̄’s
commentary (ed., pp. 196,3 – 198,8). Both (and (pseudo-)T . ūsı̄) add a fourth one, that
‘the product of a square by a non-square cannot be a square’ (al-Nayrı̄zı̄, p. 198,9–26).
144
Extension of A.I.83–A.I.85 to higher powers. For the cube, they are the following:
If a cube multiplies a cube, the result will be a cube (Elements, IX,4; Nicomachos
II.xxiv.10). If multiplying a cube by a number gives a cube, this number will be a
cube (Elements, IX,5). If a cube is multiplied by a number and the result is not a
cube, this number will not be a cube (al-Nayrı̄zı̄, p. 199,2–5; Campanus and (ps.-)T . ūsı̄,
appended to the statement of IX,5). Further: If a cube is multiplied by a non-cubic
number, the result will not be a cube (al-Nayrı̄zı̄, pp. 198,29 – 199,2; Campanus and
(ps.-)T. ūsı̄, appendix to IX,5).
145
The quotient of two similar solid numbers is indeed a cube (A.I.78), but not their
product: abc · def = t3 (abc)2 is cubic only if abc is itself a cube. The author (or his
source) has extended without distinction the case of two similar plane numbers, in
which quotient and product are squares (A.I.77, A.I.81).
146
Again an erroneous analogy (see A.I.82): from abc · def = β 3 it does not follow
that the corresponding sides will be in the same ratio.
147
Probably an explanation or a gloss to A.I.86, specifying that the term solid (mu-
jassam, στερεός), usually applied to three-dimensional bodies, and in arithmetic to
products of three factors, will also be applied to products of more than three factors.
The initial likewise in A.I.90 alludes to this twofold meaning of ‘solid’.
148
(A.I.90) is Elements, VII, def. 18.
(A.I.91) If there are numbers in continued proportion beginning with

the unit, the third from the unit will be a square, and each third one in
what follows —that is, (the number reached) by leaving out each time
one (number)— will be a square. The fourth will be a cube, and so each
time you leave out two (numbers) and take the third number this will be
a cube.149
(A.I.91 ) Likewise, the fifth will be a fourth power, then each time you
leave out three (numbers) and take the fourth, it will be a fourth power.
Then, likewise, the sixth will be a fifth power, then each time you leave
out four (numbers) and take the fifth, it will be a fifth power. The seventh
will be a sixth power. And so on by increasing each time by 1 what is left
out for attaining the number: for each time you increase by 1 the spacings
there is an increment of one spacing for the powers.150 Which is made
evident in the table of doubling the unit for the evenly even number.151
(A.I.92) If numbers in any quantity (beginning with the unit)152 are in
continued proportion and that following the unit is a square, all will be
squares; likewise if it is a cube, all will be cubes.153
(A.I.92 ) The same holds for higher powers belonging to this proportion.
The reason is that these other numbers, which come subsequently in the
proportion, are multiples of it; so the very same pertains to them.154
(A.I.93) If what follows the unit is not a square, there will be among
them no other square number than the third from the unit,155 then the
fifth from it, then the seventh and (so on) according to what we have
said previously156 , and always the one will be a square and the other not
149
(A.I.91) = Elements, IX,8; Nicomachos II.xx.5. Let the numbers 1, a1 , a2 , a3 , a4 ,
. . . be in geometric progression, thus of the form 1, a, a2 , a3 , a4 , a5 , a6 , a7 , . . . ; the
numbers with even exponents (or subscripts) will be squares, those with exponents 3k,
cubes.
150
Generalization of A.I.91.
151
Either an extensive list of consecutive powers of 2 was given in the now lost part
of the work, with perhaps some applications —like finding perfect numbers (see below,
A.I.94) and maybe summing on the chessboard, an indeed widely known problem at
that time— or we have here an early reader’s observation.
152
Same omission in A.I.94, but in both cases the context makes up for it. See note
to A.I.43.
153
(A.I.92) = Elements, IX,9; the common ratio of the progression is the second
term.
154
Generalization of the preceding; if the second term is ak , all the others will be kth
powers.
155
Here again (see A.I.91), including the starting number.
156
Above, A.I.91.
(alternately). Likewise, if that following the unit is not a cube, there will
be among them no other cubic number than the fourth from the unit, and
then so on (leaving out two terms) according to what we have explained
previously.157
(A.I.94) For any quantity of consecutive terms in double ratio158 (begin-
ning from the unit) which are added, including the unit, and their sum is
a prime number, multiplying this prime number by the last of the even
numbers added159 will produce a perfect number, according to what we
have illustrated at the beginning of the present book.160
(A.I.95) If there are numbers in continued proportion and the amount
of the first is subtracted from the second and the last, the ratio of the re-
mainder from the second to the first will be like the ratio of the remainder
from the last to the sum of all the numbers preceding it.161
This is seen with 2, 4, 8, 16: if we subtract from the second and from
the last the amount of the first, it leaves, from the second, 2, and from
the last, 14, and the ratio of 2, remaining from the second, to 2, the first,
is a ratio of identity, and the same holds for the ratio of 14 to the sum of
the numbers preceding it, namely 2, 4, 8, for their sum is 14.162
(A.I.96) Any number of which we multiply a fourth by itself, then by
four, produces (a result) equal to the product by itself of half the num-
ber.163
157
(A.I.93) = Elements, IX,10. The reference at the end is again to A.I.91.
158
The ‘double’ (not: ‘duplicate’, A.I.59) ratio is the ratio 2 : 1.
159
Specifying even, as does our text (not Euclid’s) is useless, for it cannot be other-
wise.
160
(A.I.94) = Elements, IX,36. The text’s reference may only be to its missing Book
I; perhaps it was given in correlation with the table of powers of 2 mentioned above
(note 151).
The definition of ‘perfect number’ (τέλειος ἀριθμός) is given in Elements, VII, def. 23
(it is a number equal to the sum of its divisors, excluding itself but including the unit),
and this proposition explains how to find them: if 1 + 2 + 22 + · · · + 2n−1 , thus 2n − 1, is
prime, then 2n−1 (2n − 1) will be a perfect number; their law of formation is also taught
by Nicomachos, I.xvi.4, after indicating which are the first four perfect numbers (6,
28, 496, 8128; I.xvi.2–3). It is likely that Euclid’s formula covers all perfect numbers.
Euler has demonstrated that all even perfect numbers must have Euclid’s form, and it
is today admitted (though still unproved) that odd perfect numbers do not exist.
161
(A.I.95) = Elements, IX,35; a2a−a 1
1 −a1
= a1 +aa2 n+···+a n−1
. This gives the sum of
terms in geometric progression with a the first term and r the common ratio; indeed,
n−1
a1 + a2 + · · · + an−1 = a + ar + ar 2 + · · · + ar n−2 = a · r r−1−1 .
162
The example with the quotient equal to 1 may not be the most illuminating one.
163
(A.I.96). ( a4 )2 · 4 = ( a2 )2 ; al-Khwārizmı̄, Algebra, p. 15. The subsequent relations
are related to this one. They are quite banal except, perhaps, when expressed verbally.
(A.I.97) Any number of which we multiply a third by itself, then by 9,

produces (a result) equal to the product of the number by itself.164
(A.I.98) Any number of which we multiply a sixth by itself, then by
4, produces (a result) equal to the product by itself of a third of the
number.165
(A.I.99) If you multiply it by 36, the result will equal the product of the
number by itself.166
(A.I.100) Any number of which we multiply a tenth by itself then by 4
produces (a result) equal to the product of its fifth by itself.167
(A.I.101) And so on continuously: (the result of) the multiplication of
the fraction by itself, then by 4, will equal the square of the double of
this fraction, whatever it is.168
(A.I.102) If you multiply a part of a number by itself, then by its square,
(the result) will equal the square of the number itself.169
(A.I.103) If you multiply a part of a number by itself, then by the num-
ber (denominating the part), (the result) will always equal the product
of the number by its part.170
(A.I.104) Any number of which a fourth is multiplied by itself, then by
4, then by 4, will produce (a result) equal to the product of the number
by itself.171
(A.I.105) If each of two numbers is divided by the other, the product of
the two quotients will always be 1.172
(A.I.106) If each of two numbers is divided by the other, the sum of
the products of each (number) by itself will equal the product of the two
164
(A.I.97). ( a3 )2 · 9 = a2 .
165
(A.I.98). ( a6 )2 · 4 = ( a3 )2 .
166
(A.I.99). ( a6 )2 · 36 = a2 .
167 a 2
(A.I.100). ( 10 ) · 4 = ( a5 )2 .
168
Generally, the relations A.I.96, A.I.98, A.I.100 may be written as ( ak )2 · 4 = (2 ak )2 .
169
(A.I.102). ( ka )2 · k2 = a2 , as in A.I.97, A.I.99.
170
(A.I.103). ( ka )2 · k = a · ( k1 a).
171
(A.I.104). [( a4 )2 · 4] · 4 = a2 .
172
(A.I.105). ab · ab = 1; see al-Khwārizmı̄, Algebra, p. 46; Abū Kāmil, Algebra, fol.
27r (Latin version, lines 1354–1356); employed several times (lat. 1423–1425, 2192–
2194, 2529–2531, 2611). Quite banal using symbolism. For Abū Kāmil’s Algebra we
refer to the excellent reproduction of the Arabic manuscript; modern editors refer to
its foliotation (Weinberg’s original translation, based on a Hebrew manuscript, did not
but was later reprinted by Chalhoub with the Arabic text).
numbers multiplied by the sum of the two quotients.173

(A.I.107) If a number is divided by a number and you multiply the
quotient of the division by itself, the result will be equal to the quotient
of the division of the square of the dividend by the square of the divisor.174
(A.I.108) If a number when divided by a number gives a result having a
(square) root, the root of this last number will equal the quotient of the
root of the dividend by the root of the divisor.175
(A.I.109) If a number is divided by two numbers, the ratio of the results
of the two divisions will equal the (inverse) ratio of the two divisors.176
(A.I.110) If two numbers are divided by a number, the ratio of the two
results will equal the ratio of the two numbers.177
(A.I.111) If two numbers are divided the one by the other, the result
of the division will equal the result of the division of the square of the
dividend by the product of the two numbers, that is, the dividend and
the divisor [ taken as a single number ].178
(A.I.112) If two numbers are divided the one by the other, the result of
the division will equal the quotient of the division of each of the squares
of the two numbers by their product.179
(A.I.113) If two numbers are divided the one by the other and the quo-
tient is multiplied by the dividend, this will equal the product of the
square of the quotient by the divisor.180
(A.I.114) If two numbers are divided the one by the other, the prod-
uct of the quotient by the dividend then by the divisor will equal the
square of the dividend; you may also say that the product of the divi-
dend and the divisor when divided by the divisor, and the quotient being
173
(A.I.106). a2 + b2 = ab ( ab + ab ); Abū Kāmil, Algebra, fol. 25v (lat. 1279–1284) and
28 –29r (lat. 1467–1471).
v
174 a2
(A.I.107). ( ab )2 = √ ; Abū Kāmil, Algebra, fol. 47v (lat. 2604–2608).
175
a b2
(A.I.108). b
= √b ; Abū Kāmil, Algebra, fol. 19v (lat. 946–948). The specifica-
a
tion ‘having a root’ means that the root is rational (which is irrelevant for the formula
anyway).
176
(A.I.109). ab : ac = c : b.
177
(A.I.110). ab : cb = a : c.
178 2
(A.I.111). ab = aab ; Abū Kāmil, Algebra, fol. 26r (missing in the Latin version).
The bracketed words must have been added in the margin of an earlier copy, probably
in order to remove the subsequent imprecision (A.I.112: ‘the result of the division’).
2
179 +b2
(A.I.112). ab + ab = a ab (equivalent to A.I.106); Abū Kāmil, Algebra, fol. 26v
(lat. 1321–1325).
180
2
(A.I.113). ab · a = ab · b.
then multiplied by the dividend, (the result) will equal the square of the
dividend.181
(A.I.115) If two numbers are divided each by the other, the product of
the squares of the quotients will equal the square of the unit.182
(A.I.116) For any number divided by a number, if the dividend is mul-
tiplied by some number and this (resulting) number is divided by the
divisor, the quotient of the division will equal the product of the multi-
plier of the dividend by the quotient of the first division.183
(A.I.117) If a number is divided into two parts having a (square) root,
and the one being increased by its root and the other being lessened by
its root the (two results) are equal, there will be equality of both with
the product of their roots.184
(A.I.118) If a number is divided into two parts and the two numbers
—namely the first number and one of the two parts— are divided by the
other part, one of the two quotients will always exceed the other by 1.185
Thus, if you divide 10 into 2 and 8, then divide 10 and 8 by 2, the
results of the division are 4 and 5, and the one exceeds the other by 1.
(A.I.119) If a number is divided into two parts, and then the number is
divided by each part186 and the quotients are multiplied, their product
will equal the product of the dividends divided by the product of the
divisors.187
Thus if you divide 10 by 5, the result is 2, and if you divide 6 by 2,
the result is 3; the product of 6 and 10 is 60, and the product of 3 and 2
181

(A.I.114). ab · a · b = a2 , or a·b · a = a2 .
182
a 2 b 2 b
(A.I.115). b · a = 1 (see above, A.I.105); Abū Kāmil, Algebra, fol. 46v
2
(lat. 2562–2563). This odd ‘square of the unit’ originates from the Arabic expression
found in Abū Kāmil’s text.

183
(A.I.116). a·c
b
= c ab ; Abū Kāmil, Algebra, fol. 54r –54v (lat. 2803–2809).
184
(A.I.117). Suppose a1 , a2 are squares (for the results to be rational),
with a1 > a2
√ √
and (more generally than above) a1 − m a1 = a2 + m a2 ; then since a1 − a2 =
√ √ √ √ √ √ √ √
m( a1 + a2 ), whence m = a1 − a2 we have a1 −m a1 = a1 −( a1 − a2 ) a1 =
√ √
a1 a2 . Mentioning division of the number into two parts (our a1 , a2 ) makes sense
only in Abū Kāmil’s problem where this identity is used; see his Algebra, fol. 61v (lat.
3181–3191) —where the identity is not restricted to m = 1 as here.
185
(A.I.118). If a = a1 + a2 , then aa2 − aa12 = 1; Abū Kāmil, Algebra, fol. 63r (lat.
3309–3315).
186
Read, as in the critical apparatus: ‘If two numbers are divided the one by a number
then the other by another number’.
187
(A.I.119). ab · dc = ac
bd
; Abū Kāmil, Algebra, fol. 64v , 66r (lat. 3393–3399, 3489–
3495). The example confirms the corrected interpretation.
is 6, and the product of 5 and 2 is 10, and the division of 60 by 10 gives

6.
(A.I.120) If the sum of two numbers is divided by each of them and this
(sum) is multiplied by each of the two quotients, then (the sum of) this
will equal (the sum of) the divisions of the square of their sum by each
of them.188
(A.I.121) If two numbers are divided the one by the other and the quo-
tient is multiplied by their sum, then this will equal the product of the
division of their sum by the divisor (multiplied) by the dividend.189
[ This will also equal the product of the result of the division of their sum
by the dividend (multiplied) by the divisor. ]190
Thus if you divide 8 by 2, this gives 4, then you multiply 4 by 10,
which is their sum, this gives 40; then you divide 10 by 2, this gives 5,
which you multiply by 8, this gives 40.
(A.I.122) If two numbers are each divided by their difference, the prod-
uct of the sum of this by their difference will equal their sum.191
(A.I.123) If two numbers are multiplied together and the root of the
result is taken, this will equal the product of the root of one of the two
numbers by the root of the other.192
(A.I.124)193 If of two numbers one is divided into any number of parts,
the product of the two numbers will equal the (sum of the) products of
2 2
188
(A.I.120). a+ba
(a + b) + a+b
b
(a + b) = (a+b)
a
+ (a+b)
b
. Banal formula, awkward
verbal formulation.
189
(A.I.121). ab (a + b) = a+b
b
· a.
190
This is a former gloss originating with the lacuna; it has, however, exchanged the
rôles of a and b.
191
(A.I.122). a−b a b
+ a−b (a − b) = a + b.
192
√ √ √
(A.I.123). ab = a · b; Abū Kāmil, Algebra, fol. 19r (lat. 898–900). It would
have been appropriate to put A.I.108 and A.I.123 together.
193
A.I.124 – A.I.133 are propositions 1–10 from Book II of Euclid’s Elements. Since
they correspond to algebraic identities, they are often reformulated without their orig-
inal, geometric form, so that Euclid’s segments of straight line (εὐθει̃αι, lineæ) become
numbers. See propositions PE1 –PE10 in Book A of the Liber mahameleth (with adapted
demonstrations) or al-Karajı̄’s Badı̄‘, fol. 23r –26r (propositions II,5–10). Campanus
also adapts them in an appendix to Elements, IX,16. (He had concluded Elements,
II,10 with Hec autem et omnes premisse veritatem habent in numeris sicut in lineis,
omitting, though, to mention their later adaptation.) In al-Nayrı̄zı̄’s commentary, II,
1–4 are reformulated after IX,16, while II,1–5 contain numerical illustrations. Finally,
a similar Byzantine adaptation (by Barlaam, early 14th c.) of II.1–10 is preserved: see
Euclidis Elementa V , pp. 725–738.
the undivided number by each part of the divided number.194

(A.I.125) If a number is divided arbitrarily into two parts, the (sum
of the) products of the number by each of the two parts will equal the
product of the number by itself.195
(A.I.126) If a number is divided (arbitrarily) into two parts, the product
of one of the two parts by the other plus the square of one of the two
parts will equal the product of the whole number by the number raised
to the square.196
(A.I.127) If a number is divided arbitrarily into two parts, the square
of the whole number will equal the (sum of the) squares of the two parts,
plus twice the product of the two parts.197
(A.I.128) If a number is divided into two equal parts and into two dif-
ferent parts, the product of the two different parts of the number, plus
the square of the number between the two parts, will equal the square of
half the number.198
(A.I.129) If a number is divided into two equal parts and another number
is added to it, the product of the whole number with the addition by the
addition, plus the square of half the number before the addition, will
equal the square of half the number with the addition.199
(A.I.130) If a number is divided arbitrarily into two parts, the square
of the whole number plus the square of one of the parts will equal the
product of the whole number by this part, twice, plus the square of the
other part.200
Thus if you divide 10 into 7 and 3, and multiply 10 by itself and 3 by
itself, (the sum) gives 109; and this equals twice the product of 10 by 3,
plus the square of 7, which is the other part.
194
(A.I.124) = Elements, II,1; if a and b are two numbers, with b = b1 + b2 + · · · + bn ,
then a · b = a · b1 + a · b2 + · · · + a · bn .
195
(A.I.125) = Elements, II,2; if a = a1 + a2 , then a · a1 + a · a2 = a2 .
196
(A.I.126) = Elements, II,3; if a = a1 + a2 , then a1 · a2 + a22 = a · a2 .
197
(A.I.127) = Elements, II,4; if a = a1 + a2 , then a2 = a21 + a22 + 2a1 · a2 .
198
(A.I. 128) = Elements, II,5; if a = 2a = a1 + a2 (a1 > a2 ), then a1 · a2 + (a −
a2 ) = a2 . Considering that, as in Book II of the Elements, a is a segment of straight
2
line cut first at its midpoint and then elsewhere, the ‘number between the two parts’
corresponds to the segment between the two points of section. The text is ambiguous,
for qism renders both τμη̃μα and τομή.
199
(A.I.129) = Elements, II,6; if a = 2a and b are two numbers, then (a + b)b + a2 =
(a + b)2 .

200
(A.I.130) = Elements, II,7; if a = a1 + a2 , then a2 + a22 = 2a · a2 + a21 .
(A.I.131) If a number is divided arbitrarily into two parts and the

amount of one of the parts is added to it, the square of the whole number
with the addition will equal four times the product of the whole number
by the part added, plus the square of the other part.201
(A.I.132) If a number is divided into two different parts and into two
equal parts, the (sum of the) squares of the two different parts will equal
twice the square of half the number, plus twice the square of the number
between the two parts [ which are between the two places of the two
parts ].202
Thus you divide 8 into 2 and 6, and into 4 and 4; then the (sum of
the) squares of the two different parts is 40, which equals twice the square
of 4 plus twice the square of 2, which is between the two parts.
(A.I.133) If a number is divided into two equal parts and another number
is added to it, the square of the whole number with the addition, plus the
square of the addition, will equal twice the (sum of the) squares of half
the number and of half the number with the addition.203
For instance, dividing 8 into 4 and 4, and adding 2 to it, the (sum of
the) squares of 10 and of 2 will equal twice the square of 4 plus twice the
square of 6, for their sum is 104.
(A.I.134) If a number is divided into two parts, the ratio of one of the
parts to the other equals the ratio of their product to the square of the
other; likewise, the ratio of the square of one of them to their product
equals the ratio of the one to the other; and the ratio of the whole number
to each of the two parts equals the ratio of the product of the whole
number and this part, to the square of this part; and the ratio of the
square of the whole number to the square of one of the parts equals the
duplicate ratio of the whole number to this part.204
201
(A.I.131) = Elements, II,8; if a = a1 + a2 , then (a + a2 )2 = 4a · a2 + a21 .
202
(A.I.132) = Elements, II,9; if a = a1 + a2 = 2a (a1 > a2 ), then a21 + a22 =
2a2 + 2 (a − a2 )2 . These last words are to clarify the ambiguity mentioned above
(note to A.I.128).
203 2 2
(A.I.133) = Elements, II,10; if a = 2a and b are two numbers, then (a + b) + b =
2 a2 + (a + b)2 .
204 a1 ·a2 a2 a2
If a = a1 + a2 , then a1
a2
= a2
, 1
a1 ·a2
= a1
a2
, a
ai
= a·ai
a2
, a2
= ( aai )2 . Not only
2 i i
is this proposition out of place but it also repeats earlier identities (A.I.107, A.I.111);
furthermore, as in A.I.117, there is no need for a to be the sum of a1 and a2 .
Chapter II
Science of the magic square
It contains two paragraphs.
§1. Science of odd-order (squares)205
(A.II.1)206 The treatment for this kind consists in drawing square areas
in a quantity equal to the product of a number by itself, in which you
put numbers from 1 to the quantity of these (small) squares so that the
content [ the quantity of these ] [ that is, of the numbers ] in each row is
identical to that in (any) other row.207
(A.II.2)208 Some people begin by placing these numbers according to the

succession of the natural order, from 1 to the number of squares in the
figure where they wish to construct the magic square. Then they move
its numbers —(which are) always in excess in some rows and in deficit in
the rows opposite— and arrange all the rows in a certain manner. This
is a method presenting difficulty for the beginner.
Other people proceed with that in a different, and easier, way.209
(A.II.3)210 For the case of odd (orders, they do it) as follows.211 They
put (Fig. a 1) the median number of these numbers [ which the placement
has reached ]212 in the centre of the square. (Next) they place, diagonally
205
The study of odd-order squares actually begins in A.II.3; what follows here is a
general introduction.
206
Definition (imprecise) of magic squares.
207
In an earlier copy the lacuna has been filled with two glosses (now part of the
text), the second of which clarifies the first.
208
On the difficulty of constructing ordinary magic squares; each one is constructed
individually using the natural square of the same order, thus a square filled with the
natural numbers taken in order.
209
Constructing successive borders, the description of which follows.
210
Construction of the first magic square of odd order, that of order 3.
211
Some words or expressions constantly used in constructing bordered squares must
now be defined. If the numbers to be placed are 1, 2, . . . , n2 , with n odd, the median
2
number will be n 2+1 , those preceding it are the small numbers, the others the large
ones. To each small number a corresponds a large number n2 + 1 − a, its comple-
ment. (Remember that in bordered squares, numbers complementing one another face
each other in the same border, vertically, horizontally or, in the case of corner cells,
diagonally.) In the subsequent description, the numbers are placed beginning with the
median in the central cell, then —with the small numbers taken in descending sequence
and thus their complements in ascending sequence— in the successive borders.
212
This gloss alludes to the natural square, although we have finished with it.
144 Science of the magic square
to its left,213 (above,) the number preceding it and, diagonally opposite

to it, the number following it. Then they put the small number preceding
that small number underneath it, and opposite to it the large number
following that large number, (namely) above the number which follows
that median number. Then they place the small number preceding the
second (small number) in the right-hand corner cell and, opposite to it
in the corner, the large number following the second large number. Next
they place in the (cell) following (that of) this large number the small
number which precedes the third, and opposite to it the large number
following this third large number. With this, the square of 3 by 3 is
completed, and it will have the following aspect.
4 9 2
3 5 7
8 1 6
Fig. a 1
(A.II.4)214 If the square is 5 by 5 (Fig. a 2), you put in the central cell
the median number of 25, namely 13.215 Then you place the number
preceding 13, thus 12, in the cell (next) diagonally, where you had placed
4 in the square of 3; opposite to it diagonally, you place [ this figure ] the
number following 13.216 Next you place the number preceding 12, thus
11, where you had placed 3 in this figure, and you place opposite to it the
large number following 14, thus 15, in the place of 7 in this figure. You
place the number preceding 11 in the (upper) right-hand corner cell of
the inner square of 3, and you place opposite to it, in the cell where 8 had
been put, the large number following 15, thus 16. Then you place in the
cell following the (upper) right-hand corner [ thus 10 ] the large number
following 16, thus 17, then opposite to it the number following 10 [ which
precedes it ]217 , thus 9. With this treatment, the square in the centre of
213
Here and throughout: to the left of the reader.
214
Construction of the magic square of order 5, first the inner 3 × 3 square.
215
Whereas the previous figure was inserted at the appropriate place in the text, the
others (all but one: Fig. a6) appear at the end of the study of magic squares. This
choice is the author’s: their presence there is announced. This is hardly pedagogical
and makes the text of little use to the reader.
216
The bracketed words had once been written by a reader below Fig. a1 and came
to be incorporated in the text here. His purpose was to make clear to what the words
‘this figure’, occurring twice thereafter, were referring.
217
following is indeed used both for the ascending sequence and the descending one
(in the latter case we render it by preceding), sometimes specified by ‘before it’ or ‘after
it’; not here, whence the reader’s addition.
the square of 5 by 5 is completed, and there remain, from the square, 16

(empty) cells and, from the numbers, 16 (available) numbers.
6 8 23 24 4
7 12 17 10 19
5 11 13 15 21
25 16 9 14 1
22 18 3 2 20
Fig. a 2
You then place the number preceding 9, thus 8, in the upper cell next
to the left-hand corner cell, above 12, and 18 in the cell next to the left-
hand corner in the lower row, underneath 16.218 You put 7 in the cell next
to the left-hand corner cell, underneath it, next to 12, and 19 opposite to
it, in the right-hand row, next to 10. You put 6 in the upper left-hand
corner, above 7, and, opposite to it, in the (other) corner of the diagonal,
20. You put 5 in the middle left-hand cell, underneath 7, and you put 21
in the cell opposite to it, on the right-hand side, underneath 19. You put
4 in the cell of the right-hand upper corner, above 19, and opposite to
it diagonally, 22. You put 3 in the lower middle cell, underneath 9, and
opposite to it in the upper row 23, above 17. You put 2 in the second
cell219 of the lower row, underneath 14, and 24 opposite to it in the upper
row, next to 4. You put the remaining (number) 1 in the second cell
of the right-hand row, above 20, and 25 opposite to it on the left. The
square of 5 by 5 is (now) completed.220
8 12 10 45 46 48 6
11 18 20 35 36 16 39
9 19 24 29 22 31 41
7 17 23 25 27 33 43
47 37 28 21 26 13 3
49 34 30 15 14 32 1
44 38 40 5 4 2 42
Fig. a 3
218
As in the text B, s.aff may refer to ‘row’ as well as to ‘border’ (note 505).
219
Counted from right to left, as it should be.
220
We put the figures where they should be (see note 215). We shall even add some
(marked with an asterisk).
(A.II.5)221 If the numbers of the square are more than 5 by 5, (you

proceed) thus (Fig. a 3). Put the median number in the central cell and
its neighbours as were the corresponding ones in the square of 3 by 3,
and of 5 by 5, in these same two squares. When there remains the border
surrounding this (latter) square with numbers left, you deal with the
remaining small and large numbers.222 You take the first two of them223
—which will always be even if the sequence (of numbers) proceeds from
1 by successive increments of 1— and you put the smaller (12) in the cell
next to the upper left-hand corner and, opposite to it, in the (cell) next
to the lower left-hand corner, the larger even number (38). Take next the
two numbers following these two numbers, which will always be odd.224
Put the small one (11) in the cell below the left-hand corner in the left-
hand row and the large (39) opposite to it in the right-hand row. Take
next the two subsequent numbers, which will (always) be even. Put the
small one (10) on the top, next to the cell where the even number before
it had been put, and the large one (40) opposite to it on the bottom.
Next take the two subsequent numbers, which will always be odd. You
put the small one (9) on the left, next to the cell where the odd number
preceding it had been put, and the large one (41) opposite on the right.
10 16 14 12 75 76 78 80 8
15 24 28 26 61 62 64 22 67
13 27 34 36 51 52 32 55 69
11 25 35 40 45 38 47 57 71
9 23 33 39 41 43 49 59 73
77 63 53 44 37 42 29 19 5
79 65 50 46 31 30 48 17 3
81 60 54 56 21 20 18 58 1
74 66 68 70 7 6 4 2 72
Fig. a 4
221
Construction of bordered squares of larger odd orders.
222
In constructing the border of order 7 we shall indicate the numbers (in brackets).
Afterwards (A.II.6), the author will switch from this particular to the general mode of
filling.
223
On either side of the numbers already used, thus one ‘small’ and the corresponding
‘large’.
224
This mention of parity serves no purpose other than, perhaps, to draw attention
to the fact that the even numbers will mostly occupy the horizontal rows and the odd
ones, the vertical rows.
(A.II.6)225 Next, in like manner, you put the odd numbers, the small
numbers on the left and the large ones, which are their complements,
opposite to them on the right, and the small even (numbers) on the top,
and, opposite to them on the bottom, the large (numbers) which are their
complements, until you reach on all (four) sides the middle cell.226 Once
you have done that, take the two numbers following the last two of the
numbers placed, which will always be even. Put the small one in the cell
of the left-hand upper corner and the large one opposite to it diagonally,
in the right-hand cell. Take then the two subsequent numbers, which
will always be odd. Put the small one in the left-hand middle cell and
the large one opposite to it, on the right. Take then the two subsequent
numbers, which will (always) be even. Put the small one in the cell of
the upper right-hand corner and the large one opposite to it diagonally,
in the lower left-hand cell. Take then the two subsequent numbers, which
will always be odd. Put the small one in227 the lower middle cell and
the large one opposite to it, in the upper middle cell. Deal next with the
remaining numbers.228 [ After that ]229 Place then the small even numbers
in the lower row and each of their complements in the upper row, opposite
to its associate, and place the small odd (numbers) in the right-hand
row and each of their complements in the left-hand row, opposite to its
associate, until you have finished with all the numbers. Then you will
see the (previously) empty square filled with numbers, all meeting this
(required) condition.
By following the filling method for the squares of 3 and 5 and always
placing the remaining (numbers) as for them, this method will lead you
to the filling of all odd squares, whatever their (order), provided you fill
(first) the central (parts) and then deal with the outer parts as you did for
the square of 5 relative to the square of 3 —as long as the arrangement is
with numbers taken in their (natural) sequence with successive increments
of 1.230
225
More general description.
226
Without filling it. We shall thus have reached 9 and 41 for order 7, 11 and 71 for
order 9 (Fig. a4).
227
In this section A.II.6 we have explicitly marked two sizeable lacunas, which make
the text meaningless, to show how little attention readers paid to this part. (These
lacunas can hardly be attributed to our copyist, who has reread his copy, as is seen
from marginal additions in his hand.)
228
Starting from the middle cells just filled.
229
Some reader wanted to emphasize the steps. See also A.II.33, A.II.34.
230
The placing of successive terms would be the same with numbers in any other
arithmetical progression.
12 20 18 16 14 113 114 116 118 120 10
19 30 36 34 32 95 96 98 100 28 103
17 35 44 48 46 81 82 84 42 87 105
15 33 47 54 56 71 72 52 75 89 107
13 31 45 55 60 65 58 67 77 91 109
11 29 43 53 59 61 63 69 79 93 111
115 97 83 73 64 57 62 49 39 25 7
117 99 85 70 66 51 50 68 37 23 5
119 101 80 74 76 41 40 38 78 21 3
121 94 86 88 90 27 26 24 22 92 1
112 102 104 106 108 9 8 6 4 2 110

Fig. a 5
(A.II.7)231 You may construct these squares in another way, namely by

putting all odd numbers in the central part of the successive squares and
the even ones in the outer parts.
The construction consists in this. You construct a square resulting
from the multiplication of an odd number by itself, as you did before.
Then you consider (first) the odd numbers from 1 to the last of those
which will be in this square; you arrange them inside the square so as to
give them the shape of a rhomb within the main square, leaving in the
main square (groups of empty) cells with the shape of triangles having
each the same number of cells. Then you write there the even numbers
from 2 to the last to be found in these squares, and in such a way that
the sums be equal everywhere. Thus doing, the odd numbers will be in
the larger square within a rhombic figure and the even numbers will be
surrounded by this (rhombic figure) and all four sides of the square, as in
this figure (Fig. a 6).232
ev
en
en
ev
odd
ev
en
Fig. a 6
en
ev
231
Bordered odd-order squares with separation of the numbers by parity, general
description (without describing the structure of the rhomb and its largest inner square,
unlike in B.21i–ii).
232
This is the only other figure inserted in the text (see note 215). It is understood
that the squares will be bordered, as before.
(A.II.8)233 The way to place the (odd numbers) is as follows. You take
1 and the last term belonging to this square, namely its largest number,
then 3 and the (odd) number preceding this largest one, and so on always
until you reach their middle. You put the median term in the central cell
of the square. You put its two adjacent odd terms in the cells where you
had put, of the nine terms which were in the square of 3 by 3, the two
adjacent to the median.234 You do the same for the remaining numbers
until you complete the square of 3 by 3, if this is what you are dealing
with, and you do the same for the square of 5 by 5, or 7 by 7, or 9 by
9, if this is what you are dealing with. You will always do this until you
complete the whole square contained by the inner rhomb.
(A.II.9)235 Once you have done that, take the two odd terms reached,236
and put the small one in the middle cell of the first border237 on the left
and its complement opposite to it in the first border on the right, the
next small number on the bottom in the middle cell of the first border,
and opposite to it on the top its complement.238 Take then the two terms
reached and put the small one on the bottom, to the left of the middle
cell, and, opposite to it on the top, its complementary term. Put then
the next small term on the left, just above the middle cell, and opposite
to it on the right its complement. Then put the next small number in
the middle cell of the second border on the left and, opposite to it in the
second border on the right, its complement. Then put: the small number
following that term in the middle cell of the second border on the bottom
and opposite to it its complement; the following small term next to the
middle cell in the first border on the top and opposite to it in the first
border on the bottom its complement; the following small number in the
first border on the right, next to the middle cell, below it, and opposite
to it on the left its complement. Once you have done that for these two
squares (of order 7 and 9), you will have reached 1 and the last (odd)
term and the placement will have been performed in the desired way. In
233
Arranging the odd numbers by pairs of complements; filling the largest square
within the rhomb.
234
In all subsequent inner squares of order 3 the arrangement is, relative to that seen
in Fig. a1, inverted around the descending diagonal.
235
Filling the remainder of the rhomb with the odd numbers left, here for the 7 × 7
and 9 × 9 squares. The square of order 5 will be treated separately (A.II.23).
236
Reached without being placed. For the orders 7 and 9 they will be respectively 15
and 35, 15 and 67.
237
The borders are counted from the inner square already filled.
238
This would also be applicable to the square of order 5, in which case we would
have finished placing the odd numbers.
the case of the square of 9, your treatment is then completed.239

77
71 69 3
45 27 31 61 63 23
39 37 3 9 29 39 37 47 53 73
9 23 21 31 41 7 15 25 49 41 33 57 67 75
7 15 33 25 17 35 43 81 65 35 45 43 17 1
49 19 29 27 1 59 51 21 19 55
11 13 47 11 13 79
5 5
Fig. a 7* Fig. a 8*
(A.II.10)240 In the case of the square of 11, and of 13, do the same until
the above situation is attained.241 Then you write the small number,
following the last of the small numbers already placed, in the first border
on the bottom, in the third cell from the middle (cell) on the left, and
opposite to it its complement. Then you put the following small number
in the first border on the left, in the third cell above the middle (cell),
and its complement opposite to it on the right. Then you put the subse-
quent small term in the second border on the bottom, next to the middle
cell,242 and its complement opposite to it on the top. Then you put the
subsequent small term next to the middle cell in the second border on the
left, and its complement opposite to it on the right. Then you put the
small term reached in the middle cell of the third border on the left, and
opposite to it its complement. Then you put the subsequent small term
in the middle cell of the third border on the bottom, and opposite to it on
the top its complement. Then you put the subsequent small term next to
the middle cell in the second border on the top, and opposite to it on the
bottom its complement. Then you put the subsequent small number in
the second border on the right, next to the middle cell, and opposite to it
on the left its complement. Then you put the subsequent small (number)
239
From this it may be inferred that the part of the rhomb outside the inner square
comprises the same number of cells for these two consecutive orders.
240
Filling the remainder of the rhomb for both the 11 × 11 and 13 × 13 squares.
241
With 21 (Fig. a9*) and 101 placed last. Since for orders 11 and 13 there is the
same quantity of cells to be filled with odd numbers outside the inner square, we give
just one illustration.
242
Only one adjacent cell still empty. Same in what follows.
in the first border on the top, in the third cell from the middle (cell), and
opposite to it on the bottom its complement. Then you put the subse-
quent small number in the third cell from the middle (cell) in the first
border on the right, and opposite to it on the left its complement. Once
you have done that, you will have placed the odd numbers in these two
squares,243 to the last, in the centre (of the squares).244
113
107 97 7
103 91 89 23 3
17 47 51 81 83 43 105
13 29 49 59 57 67 73 93 109
11 27 35 45 69 61 53 77 87 95 111
117 101 85 55 65 63 37 21 5
121 79 71 41 39 75 1
19 31 33 99 119
15 25 115
Fig. a 9*
(A.II.11)245 If you are dealing with the square of 15, or of 17, you do
the same as you did for the square of 13 until you reach the third bor-
der.246 Then you put the subsequent small number in the first border
on the bottom, in the fourth cell from the middle one, and its comple-
ment opposite to it; then the subsequent small number in the first border
on the left, in the fourth cell from the middle one, and opposite to it
its complement;247 then the subsequent number in the second border on
the bottom, in the third cell from the middle one, and opposite to it its
243
Of orders 11 and 13.
244
‘in the centre’: meaning, as in A.II.7, in the central (rhombic) part. Note the
tedious cell-by-cell approach where a general method could have been described (see
Commentary).
245
Filling the remainder of the rhomb in the 15 × 15 and 17 × 17 squares. General-
ization.
246
We shall have placed the pairs up to and including 29, 197. In Fig. a10*, the
numbers to be placed now form the outer border of the rhomb. From what follows
could be inferred the general way of filling the lateral triangles of the rhomb.
247
Here and below: exact situation relative to the middle cell not specified.
complement; then the subsequent number in the second border on the

left, in the third cell from the middle one, and opposite to it its com-
plement; then you put the subsequent number in the third border on
the bottom, next to the middle cell, and opposite to it its complement;
then the subsequent number in the third border on the left, next to the
middle cell, and opposite to it its complement; then the subsequent num-
ber in the fourth border on the left, in the middle cell, and opposite to
it its complement; then the subsequent number in the fourth border on
the bottom, in the middle cell, and opposite to it its complement; then
the subsequent number in the third border on the top, next to the mid-
dle cell, and opposite to it its complement; then the subsequent number in
213
207 189 11
203 183 173 35 7
199 179 167 165 51 31 3
25 79 87 83 153 155 159 75 201
21 45 85 99 103 133 135 95 141 181 205
17 41 57 81 101 111 109 119 125 145 169 185 209
15 39 55 63 77 97 121 113 105 129 149 163 171 187 211
217 193 177 157 137 107 117 115 89 69 49 33 9
221 197 161 131 121 93 91 127 65 29 5
225 151 139 143 73 71 67 147 1
27 47 59 61 175 195 223
23 43 53 191 219
19 37 215
13
Fig. a 10*
the third border on the right, next to the middle cell, and opposite to
it its complement; then the subsequent number in the second border
on the top, in the third (cell) from the middle one, and opposite to it
its complement; then the subsequent number in the second border on
the right, in the third (cell) from the middle one, and opposite to it its
complement; then the subsequent number in the first border on the top, in
the fourth (cell) from the middle one, and opposite to it its complement;
then the subsequent number in the first border on the right, in the fourth
(cell) from the middle one, and opposite to it its complement. Once you
have done that, you will have finished with the odd numbers for these two
squares in the desired way. If you wish to proceed with (squares) above
these, continue placing (the numbers) step by step in the same way.
(A.II.12)248 At this point you will find that the squares are divided into
classes requiring each a distinct treatment for the arrangement of the even
numbers in the corners. There are (those of order 5,) 9, 13, 17, and so on
by adding 4; and (those of order) 7, 11, 15, 19, and so on by adding 4.249
(A.II.13)250 The situation for that of 5 and those of the same kind (is as
follows). There remain as empty cells in the first border, which surrounds
the inner square, the four at the corners and eight adjacent to them, two
on each side; in the second (border), the empty cells are (the four) at
the corners and twenty-four cells adjacent to them. It will always be like
that: each border has, excepting the corner cells, 16 more (empty cells)
than the preceding border (Fig. a 11*).
1st 2nd 3rd 4th 5th
(n = 4t + 1)
4+8 4 + 24 4 + 40 4 + 56 4 + 72
Fig. a 11*
For the square of 5 only, one cannot proceed as for the others of its
kind.251
(A.II.14)252 For the other squares belonging to the (same) class (the
situation is the following). After placing the odd numbers as indicated,
the first upper row of the square of 9 is in excess over its sum due, required
for it, by 20, the lower one is in deficit by 20;253 the first right-hand row is
248
For placing the even numbers outside the rhomb, we are to consider three categories
of odd order n = 2k + 1, namely n = 5, n = 4t + 1 (t = 1), n = 4t + 3.
249
The square of order 5 belongs to the first kind for calculating the number of empty
cells (A.II.13) but not for filling them (A.II.23). This is not very clearly explained in
the text (note 251).
250
Quantity of cells remaining empty in squares of orders n = 4t + 1 (see Fig. a8*).
251
The square of order 5 stands alone as the third of the three kinds for placing
the even numbers. This sentence might be expected at the end of A.II.12, but it is
evidently connected to the subsequent one.
252
Excesses in the upper and right-hand rows after placing the odd numbers (order
n = 4t + 1).
253
‘required for it’ might be a reader’s addition. The concept of sum due will be
explained below.
in excess over its sum due by 18, and the left-hand one is in deficit by 18.
The second upper (row) is in excess by 36, and the lower one is in deficit
by the same amount; the second right-hand (row) is in excess by 34, and
the left-hand one is in deficit by the same amount. For the square of 13,
the first upper (row) is in excess by 28, and the lower one is in deficit by
the same amount; the first right-hand (row) is in excess by 26, and the
left-hand one is in deficit by the same amount. The second upper (row)
is in excess by 52, and the lower one is in deficit by the same amount;
the second right-hand (row) is in excess by 50, and the left-hand one is
in deficit by the same amount. The third upper (row) is in excess by 76,
and the lower one is in deficit by the same amount; the third right-hand
(row) is in excess by 74, and the left-hand one is in deficit by the same
amount. Similarly for the others (Fig. a 12*).
1st 2nd 3rd 4th 5th
n=5 12 10
n=9 20 18 36 34
n = 13 28 26 52 50 76 74
n = 17 36 34 68 66 100 98 132 130
n = 21 44 42 84 82 124 122 164 162 204 202
Fig. a 12*
(In the square of) 9, the second (row) has 16 more than the (row)
before. (In the square of) 13, the first (row) has 8 more than the first (of
the square) of 9, then each row has 24 more than the (row) before. (In
the square of) 17, the first (row) has 8 more than the first of (the square
of) 13, then each row 32 more than the (row) before. Then each row will
always have more than the one before in the same manner.
(A.II.15)254 ‘Excess over its sum due’ means (the following). The re-
quired amount for a cell equals the quantity of the median [ the middle
of the totality of these numbers ]255 . Thus each cell of the square of 5 has
a sum due of 13. Therefore you will add the odd numbers in each row
and divide the result by the number of cells filled with numbers; if the
quotient is less than the median, (the sum) in the cells will be less than
their sum due by the product of this deficit and the (number of) cells;
analogously if (the quotient is) in excess.256 So the subsequent placing
254 2
Definition of the sum due for m cells filled in a square of order n (namely m· n 2+1 ).
255
Former gloss explaining ‘median’.
256
Simpler: take the difference between the sum in the cells filled and the multipli-
cation of their quantity by the median number.
of the even numbers in the empty cells of each row must be such that it
compensates the deficit, if any, or falls short by the amount of the excess,
if any. We shall show this in the appropriate place, when discussing the
matter of the even (numbers).257
(A.II.16)258 The situation for the (square of) 7 and (the others) of this
class (is as follows).259 From the first border, which surrounds the inner
square, there remain as empty the four corner cells; from the second
border, the corner cells (and) 16 cells adjacent to them, 4 on each side;260
from the third border, the corner cells (and) 32 empty cells adjacent to
them. And so on always: each border has, excepting the corner cells, a
quantity of 16 more than the one before (Fig. a 13*).
(n = 4t + 3) 1st 2nd 3rd 4th 5th

4 4 + 16 4 + 32 4 + 48 4 + 64
Fig. a 13*
(A.II.17)261 In this class of squares, the first upper row has an excess of
4 over its sum due, and the lower one has a deficit of the same amount;
the first right-hand (row) has an excess of 2 over its sum due, and the
left-hand (row) has a deficit of the same amount. The second upper row
of the (square of) 7 has an excess of 20 over its sum due, and the second
right-hand (row an excess of) 18. In the square of 11, the second upper
(row) has an excess of 28 over its sum due, the second right-hand (row an
excess of) 26, the third upper (row an excess of) 52, the third right-hand
(row an excess of) 50, and all the opposite (rows) have a deficit equal
to the excess. (Square of) 15: the second upper (row) exceeds the first
(upper row) by 32, the third the second by 32. And so on for the others:
the excess increases each time by 8 (Fig. a 14*).262
1st 2nd 3rd 4th 5th
n=7 4 2 20 18
n = 11 4 2 28 26 52 50
n = 15 4 2 36 34 68 66 100 98
n = 19 4 2 44 42 84 82 124 122 164 162
Fig. a 14*
257
A.II.18–A.II.22.
258
Quantity of cells remaining empty for orders n = 4t + 3.
259
See Fig. a7*, a9*, a10*.
260
Two on either side of a corner cell.
261
Excesses in the upper and right-hand rows after placing the odd numbers (order
n = 4t + 3).
262
The differences of the excesses are multiples of 8.
(A.II.18)263 Thus, writing the even (numbers) in the (rows) must be

done so as to equalize them, and you must then know how to proceed in
searching for numbers (the sum of which presents) the required excess or
the required deficit.264
The way to determine this is (first) to associate 2 and the last even
term, then 4 and the next (term) preceding (the last), and so on until you
reach the two terms which are the medians of these even numbers (Fig.
a 15*).
n2 −8j+3 n2 −8j+7 n2 −5 n2 −1
2 4 | ... | 2 2
| ... | 2 2
n2 +8j−1 n2 +8j−5 n2 +7 n2 +3
n2 − 1 n2 − 3 | . . . | 2 2
| ... | 2 2
Fig. a 15*
(A.II.19)265 Once you have done that, you will find that if the last small
number and the preceding one are placed on one side,266 and opposite
to them their complements, the side containing the two small numbers
will be less than its sum due by 4, and the other more by 4; placing
the next two numbers, the differences will be 12 and 12, and with the
next two numbers 20 and 20, then 28, 36, 44, 52, 60 and so on always
till (the placement of) 4 and 2. Now you will notice that the borders
which require placing even numbers in them have indeed, on two of their
sides, this succession of excesses or deficits, namely 4, 12, 20, 28, 36 and
so on always.267 Thus, placing the excess of these even numbers where
the deficit is and writing opposite to each (placed) number its comple-
ment among the large (numbers), you will have equalized the borders on
two sides, and there will remain (the) two (other) sides requiring to be
equalized, which you will do with the remaining even numbers.268
(A.II.20)269 If you take four (even, consecutive) small numbers, the sum
of the first and the second has a certain deficit, and the sum of the third
263
Associating the even numbers by pairs, with their complements, from the smallest
pair of small numbers.
264
The subsequent rules of equalization by means of even numbers are in A.II.19–
A.II.22.
265
Deficits resulting from putting one of these pairs of small numbers in a row (with
an excess of the same amount appearing opposite). This is Rule VI in our commentary,
pp. 82–83.
266 2 2
The ‘last small (even) number’ is n 2−1 and the one before, n 2−5 .
267
As seen in A.II.14 (Fig. a12*) and A.II.17 (Fig. a14*) for the horizontal rows.
268
It will equalize completely pairs of opposite rows in two cases (A.II.32, A.II.33).
269
Rule of neutral placements, which provide exactly the sum due for four cells: of four
consecutive (even) numbers, the two extremes are put in one row and the two middle
and the fourth a deficit smaller by 8. Then adding the first and the fourth
you will find that they have a deficit of half the sum of the two deficits;
adding similarly the second and the third gives the same result.
For instance, the sum of the last two small numbers is less than their
sum due by 4, and the sum of the two preceding numbers less by 12;270
thus, adding the last and the first, and the third and the preceding num-
ber, will produce a deficit of 8, which equals half the sum of 12 and 4. The
knowledge of this is necessary to you, for you will use it constantly.271
(A.II.21)272 (i) When you write the first small number, or any small
number, on one side and the subsequent number on the other, opposite
side, and you write the two large numbers which are their complements
opposite to them, the side where the first small number has been written
will be less than its sum due by 2, whereas the other side will be in excess
by 2. (ii) If you write some small number on one side and you write on
the other, opposite side the third small number counted from this one,
the side containing the first small number will be less than its sum due
by 4. (iii) And so on: whenever you increase the distance between these
two by one number, the deficit is always increased by 2.273
For instance, if 4 is put on one side and 6 on the other, opposite side,
and opposite to them their complements, the side containing 4 will be
less than its sum due by 2 and the side containing 6 will be more by 2.
If you write 8 instead of 6 and do the same, the side containing 4 will be
less by 4 and the side of 8 more by 4. And whenever you increase the
distance between these two, its increment (will vary) accordingly.
(A.II.22)274 If four numbers are put in the corners, each pair in consec-
utive corners and their complements diagonally opposite, and if you put
the two small numbers consecutively on the upper side, this side will be
ones in the opposite one (Rule I in our commentary) —this remains true for two pairs
of even numbers having the same difference (Rule II). But this is unfortunately (in
view of its considerable application) not explained clearly.
270 2 2 2 2
See A.II.19; indeed, n 2−1 + n 2−5 = n2 + 1 − 4, n 2−9 + n −13 2
= n2 + 1 − 12.
271
See examples in A.II.31, A.II.33, A.II.34.
272
With an even number put on one side and some other even number on the other,
and facing them their complements, the side of the smaller one will display a deficit
equal to their difference, the other an excess of same amount (Rule III).
273
The text has here two versions of the same sentence, one in the margin, both in
the copyist’s hand. Some early reader may have tried to correct the number in the
text (unsuccessfully, with 6 instead of 4, both wrong).
274
If a pair of smaller numbers is put in the corners, it will produce the same difference
as before in two opposite rows (A.II.19), but a difference of 2 in the other pair of rows,
which will be a deficit in the row with the lesser number (Rule VII).
less than its sum due by 4 if they are the last two, by 12 if they are the
two previous (numbers), and so on always, with regular additions of 8,
until you reach 2 and 4. The right-hand side will always have, relative to
its sum due, an excess of 2, or a deficit of 2,275 without any augmentation
and diminution.
You must understand all this: it belongs to what you need for writing
the even numbers in this class (of squares).
Examples of treatments for all that we have explained
(A.II.23)276 Treatment for the square of 5 by 5 (Fig. a 16). You put the
odd numbers in the inner square of 3 as we have explained. Those remain-
ing are 1, 25, 3, 23. You put 1 in the lower middle cell and 25 opposite to
it above, 3 in the left-hand middle cell and 23 on the opposite side, on the
right. Next, you put 2 in the upper right-hand corner and opposite to it
diagonally, in the lower left-hand corner, its even277 complement, namely
24. You put 4 in the upper left-hand corner and opposite to it diagonally,
in the lower right-hand corner, its complement, namely the even number
22. You put 6 on the right side and its complement, 20, opposite to it on
the left. You put 10 and 8 on the bottom, 12 on the right, and you put
opposite to each its complement.278
4 18 25 16 2
20 11 9 19 6
3 21 13 5 23
14 7 17 15 12
24 8 1 10 22
Fig. a 16
(A.II.24) You pass now from 5 to 9, 13, 17 and those of this kind.279
275
Depending on whether it contains the larger small number or not.
276
Construction of the particular magic square of the type n = 4t + 1, the square of
5.
277
No need for this last indication. Same in the next sentence.
278
None of the previous rules applied since this square is a particular case (see
A.II.13).
279
A.II.24–A.II.31 teach how to place the even numbers for the orders n = 4t + 1 (t ≥
2). In this section, how to fill the first border. For order 9 (Fig. a17), once all odd
numbers have been placed, there remain to be placed the even numbers from 2 to 40
and their complements. The first one will be 40 since they are taken in reverse order.
(i) Put the last small even term in the upper left-hand corner of the first
border —that is, the border following the inner square which you have
(completely) filled with odd numbers— and opposite to it diagonally, in
the lower right-hand corner of the first border, its complement. Put the
preceding small term in the upper right-hand corner of the first border,
and opposite to it diagonally, in the lower left-hand corner, the term
which is its complement.280 [ Once you have done that, you will find that
the excess of the upper row is 16, the excess of the right-hand row 16,
thus the excess of the upper row equals the excess of the right-hand row. ]
(ii) Put then 2 in the upper row; consider the excess of the upper row281 ,
take its half, and count after 2 as many small even numbers as this half,
and put the number reached in the lower row. Put opposite to each of
these two numbers its complementary term. (iii) Put 4 on the right side;
then count after [ this, thus ] 4 as many small even numbers as half of the
excess, and put the number reached in the left-hand row.282 Put opposite
to each of these two numbers its complementary term.
Once you have done that, all the sides of the first border will be
equalized in this kind of square.
(A.II.25)283 (i) Put then the two small terms the sum of which is less
than their sum due by 12 in the right-hand corners of the second border,
with the lesser one above; put in the diagonally opposite corners, on the
left, their complements. (ii) Then look for the pair of small numbers
which have a sum less than their sum due by (an amount equal to) the
excess of the second right-hand row less 14; put them in the second border
on the right and, opposite to them on the left, their complements. (iii)
Then look for the pair of small numbers which have a sum less than their
sum due by (an amount equal to) the excess of the upper row less 8; put
them in this row and, opposite to them below, their complements.284
280
We shall thus put 40 and 38 for order 9 (Fig. a17), 84 and 82 for order 13 (Fig. a
18*). As said in what follows, with the five cells filled in the top row (and in the right-
hand column), the excess will become 221 − 205 = 16. Now this is true for order 9:
for order 13, with seven cells filled, the excess will be 619 − 595 = 24. The subsequent
sentence must be an early reader’s observation, as the similar one in A.II.35. See also
the gloss in A.III.25 (note 441).
281
The excess before placing 2; we apply here the rule seen in A.II.21.
282
The manuscript repeats this sentence twice, with the bracketed words omitted the
second time.
283
First steps for filling the second border. On all this, see Commentary, pp. 86–89.
284
See Fig. a17 and a18*. The numbers to be inserted later are in smaller characters.
46 66 14 72 77 6 28 26 34
52 40 64 71 69 3 2 38 30
50 78 27 31 61 63 23 4 32
74 9 29 39 37 47 53 73 8
7 15 25 49 41 33 57 67 75
12 81 65 35 45 43 17 1 70
22 20 59 51 21 19 55 62 60
58 44 18 11 13 79 80 42 24
48 16 68 10 5 76 54 56 36
Fig. a 17
(A.II.26)285 (i) Put then the pair of small numbers which have a sum
less than their sum due by 20 in the right-hand corners of the third
border, with the lesser one above; put, opposite to them diagonally, in
the corners of the third border on the left, their complements. (ii) Then
look for the pair of small numbers which have a sum less than their sum
due by (an amount equal to) the excess of the third right-hand row less
22; put them in this row, and, opposite to them in the third left-hand row,
their complements. (iii) Then look for the pair of small numbers which
have a sum less than their sum due by (an amount equal to) the excess
of the third upper row less 8; you put them in this row and, opposite to
them in the third lower row, their complements.
(A.II.27)286 (i) Put then the pair of small numbers which have a sum
less than their sum due by 28 in the right-hand corners of the fourth
border, with the lesser one above, and opposite to them diagonally on
the left their complements. (ii) Then look for the pair of small numbers
which have a sum less than their sum due by (an amount equal to) the
excess of the fourth right-hand row less 30; put them in this row and,
opposite to them in the fourth left-hand row, their complements. (iii)
Then look for the pair of small numbers which have a sum less than their
sum due by (an amount equal to) the excess of the fourth upper row less
8; put them in this row and, opposite to them below, their complements.
Proceed always likewise: for the right-hand row, with successive addi-
tions of 8, and for the upper one next to it, with deficits of 8 and 8 again.
285
First steps for filling the third border.
286
First steps for filling the fourth border.
94 48 124 126 42 154 161 14 160 6 52 50 74
112 90 146 22 136 155 145 7 30 64 62 78 58
110 104 84 144 151 139 137 23 3 2 82 66 60
162 102 166 51 59 55 125 127 131 47 4 68 8
12 138 17 57 71 75 105 107 67 113 153 32 158
18 13 29 53 73 83 81 91 97 117 141 157 152
11 27 35 49 69 93 85 77 101 121 135 143 159
150 165 149 129 109 79 89 87 61 41 21 5 20
116 36 169 133 103 95 65 63 99 37 1 134 54
56 38 28 123 111 115 45 43 39 119 142 132 114
70 130 88 26 19 31 33 147 167 168 86 40 100
98 92 24 148 34 15 25 163 140 106 108 80 72
96 122 46 44 128 16 9 156 10 164 118 120 76
Fig. a 18*
(A.II.28)287 Once you have done that, the excess of each upper row over
its sum due will be 6, the excess of each right-hand row over its sum due
will be 2, and the number of remaining empty cells will be four in each
second row, eight in each third row, twelve in each fourth row, and so on
always by successive additions of four.
(A.II.29)288 You will now turn your attention to the remaining numbers.
You take groups of four small numbers and put the first one in an upper
row, the third in the opposite row below, the second on the right and
the fourth on the left, and you put opposite to each of these four its
complement; (you proceed likewise) until you have done this for each
border.
(A.II.30)289 Once you have done that, you consider a pair of small num-
bers; put the first in some upper row, the second below (in the same
border), and opposite to them their complements. Then take another
pair; put the first on the left side of the same border, the second on
287
The remaining excesses are now uniformly 6 in the top rows and 2 in the right-hand
columns (same deficits opposite), to be eliminated by filling four cells.
288
Reducing (by application of our Rule V, or twice A.II.21ii) these differences to
±2.
289
Eliminating all these differences by means of Rule IV (A.II.21i).
the right, and opposite to them their complements. Do the same for all
borders.
Once you have done that, each row and its opposite will be equalized
and none will display any excess.
(A.II.31)290 If there are remaining empty cells, it can be only four facing
four, eight facing eight, twelve facing twelve (and so on); you will equalize
them with groups of four by means of the still available sequences of four
numbers in progression, in the way explained at the beginning of the
section for four numbers of which each pair is in progression:291 you put
the first of the first pair and the second of the second pair on one side,
the second of the first pair and the first of the second pair on the facing
side, and opposite to each its complement.
(A.II.32) Treatment for (the squares of) 7, 11, 15, 19 and the like.292
You put the last of the small even terms in the still empty upper
left-hand corner of the first border, which surrounds the inner square
which you have filled with odd numbers, and opposite to it diagonally,
in the lower right-hand corner of the same border, its complement. Put
the preceding small term in the upper right-hand corner of the same
border and, opposite to it diagonally, in the lower left-hand corner, its
complement.
Once you have done that, you will have equalized the first border for
all squares of this kind.293
(A.II.33)294 Then consider [ after that ]295 the amount of the excess of
each upper row and look for the pair of small numbers such that their sum
is less than their sum due by the same amount. Put them in the corners
of this row, the lesser on the right, and opposite to them diagonally below
their complements. You complete in this way all the remaining corners.
290
Completing all the rows by means of neutral placements (Rule II; A.II.20).
291
A.II.20 is at the beginning of the section (which opened with A.II.18) on placing
the even numbers outside the rhomb. But here the group of four is divided into two
pairs, a situation not mentioned there.
292
A.II.32–A.II.34 teach how to complete the squares of order n = 4t + 3. First,
complete all first borders by putting the largest pair of smaller numbers in the upper
corners.
293 2
The ‘last of the small even numbers’ is n 2−1 , thus 24 for order 7 (Fig. a20) and
60 for order 11 (Fig. a19).
294
Equalization of all horizontal rows for the borders following the first (see pp. 95–
96).
295
Either the gloss already seen (note 229) or to be inserted after ‘remaining empty
cells’ below (as in A.II.34 —if so then not a gloss there).
Once you have done that, you will have equalized all the upper and
lower rows, and all the remaining empty cells will be four facing four, and
so on by successive additions of 4. You will then equalize each group of
four by means of four numbers in the way we have explained previously.
36 16 108 110 10 113 8 116 118 2 34
50 48 24 100 107 97 7 102 18 46 72
52 56 60 103 91 89 23 3 58 66 70
96 54 17 47 51 81 83 43 105 68 26
94 13 29 49 59 57 67 73 93 109 28
11 27 35 45 69 61 53 77 87 95 111
92 117 101 85 55 65 63 37 21 5 30
32 80 121 79 71 41 39 75 1 42 90
38 78 64 19 31 33 99 119 62 44 84
82 76 98 22 15 25 115 20 104 74 40
88 106 14 12 112 9 114 6 4 120 86
Fig. a 19
(A.II.34)296 There remain the right-hand rows, exceeding their sum due
by 16, 24, 32, 40, 48, and what results from successive (additions of) 8,
and the remaining empty cells are four facing four, eight facing eight,
and so on by successive additions of 4. You then look for a pair of large
numbers such that their sum exceeds their sum due by an amount which,
when added to the excess of the right-hand row, equals the deficit of the
sum of two small numbers. Put then the two small numbers and their
two associates on this side, and put on the left the complements of the
four numbers, each pair opposite to its complements.
Once you have done that, you will have equalized each side of this pair
of sides, and the remaining empty cells [ after that ] will be four facing
four, and so on by successive additions of 4. You will then equalize each
group of four by means of four subsequent numbers as we have explained
previously.
(A.II.35) Example of the treatment for the right-hand side of the square
of 7. You will find that the excess of the upper row is 20 and the excess
of the right-hand row, 18. Putting the two numbers (which have a sum)
less than their sum due by 20 in the two upper corners, with the lesser
296
Equalization of all vertical rows for the borders following the first (see pp. 96–98).
one on the right, and, opposite to them below, their complements, you
will find that the excess on the right-hand side is 16. Consider then the
two large numbers with an excess of 12. [ For if you add 12 to 16, which
is the right-hand excess, this gives an excess of 28, equal to the deficit
of the sum of two small numbers. ]297 You then put these two numbers
on the right, as also the two small numbers which have a sum less than
their sum due by 28, and you put on the left, opposite to each one, its
complement (Fig. a 20).
16 8 44 45 46 2 14
18 24 39 37 3 22 32
20 9 23 21 31 41 30
7 15 33 25 17 35 43
40 49 19 29 27 1 10
38 28 11 13 47 26 12
36 42 6 5 4 48 34
Fig. a 20
The treatment is the same for all other (squares of this kind).
The squares have been constructed under the condition of a progres-
sion of the numbers from 1 to the last by additions of 1 and in such a
way that the odd are inside. You will rely on them to infer what we have
not described after we have completed the account of the (construction
of squares of) even (orders), with the help of God Most High.298
§2. Science of even-order (squares)

(A.II.36)299 The treatment for even(-order) squares (depends on) their
three categories: evenly even, evenly odd and evenly evenly odd.300
297
This sentence, which when removed leaves a perfectly continuous text, might well
be an addition.
298
The purpose of this last sentence is just to mark the end of a section.
299
Stating the three categories of even-order squares.
300
‘Evenly even’, thus of the form 2n (n ≥ 2, see A.I.16) or 2m (2k + 1) (m > 2, k
natural, see A.I.6); ‘evenly odd’, thus of the form 2(2k + 1), see A.I.14 but see A.I.5 ,
such as 6, 10, 14, . . . ; ‘evenly evenly odd’, thus of the form 22 (2k + 1), see A.I.15
but see A.I.17, such as 12, 20, 28, . . . . In later treatises, the distinction between the
first and the last of the three categories will disappear, simple general methods being
applicable to either one. By the way, this distinction between three categories of order
will occur only once (A.II.40-A.II.42; see A.II.44), since filling the border of the square
of order 8 is a particular case of the method applicable to the other orders divisible
by 4. The author will now examine one by one the even orders beginning with the
The first is (the square of order) four; indeed, the first even (order) is
2 and it is not possible to place in it numbers in magic arrangement.
(A.II.37)301 So let us dismiss it and turn our attention to the square (of
order) 4 (Fig. a 21*302 ). You place (the numbers as follows). One of its two
medial numbers (8) in its centre and the other (9) in its corner according
to the bishop’s move on the chessboard, namely the third [ diagonally from
it ]303 . Then, the number preceding the small median (7) next to the large
median, on its right304 , and the (number) following the large median [ its
complement305 ] (10) in the bishop’s cell as well, next to the small median.
Then, the number following the small median (6) underneath the large
median, and that following the large (median) [ its complement306 ] (11)
in the bishop’s cell. Then, the number following the small median (5)
in the lower right-hand corner and its complement (12) in its bishop’s
cell; the number following the small median (4) in the upper right-hand
corner and its complement (13) in its bishop’s cell; the number following
the small median in the cell next to (that of) the large number which
you have reached [ thus 3 ]307 , and its complement, thus 14, in its bishop’s
cell; 2 in the cell underneath (that of) 13, and 15 in its bishop’s cell; 1
underneath 14 and 16 in its bishop’s cell. This square is finished.
9 7 14 4 5 10 15 4 1 12 13 8
6 12 1 15 11 8 1 14 15 6 3 10
3 13 8 10 2 13 12 7 4 9 16 5
16 2 11 5 16 3 6 9 14 7 2 11
Fig. a 21* Fig. a 22 Fig. a 23
smallest.
301
Square of order 4 —particular case: since there is no magic square of order 2, as
just stated, the square of order 4 will be filled as a whole without considering its outer
border separately.
302
We have added this figure, to correspond with the subsequent instructions. The
manuscript has Fig. a22 (same, but turned around the ascending diagonal), used in
all subsequent even-order squares. Fig. a23 is also found in the manuscript but not
mentioned in the text.
303
From the text, it is clear that ‘third’ refers to a corner, not a cell.
304
As seen by the reader (note 213). The subsequent instructions are no longer
applicable to Fig. a22.
305
Complement of 7 (‘adad) and not of the ‘large median’.
306
Again, complement of 6.
307
It is its place in the Arabic text which makes this indication of dubious authenticity.
(A.II.38)308 As to the treatment for the evenly odd square, it is as

follows. You draw the square, and then put the two medians of the
numbers in the central square, where you had put the two medians of the
previous square, the subsequent (numbers) where were their analogues,
(and so on) until you have finished with the inner square, as you had done
with the odd (orders).309 You are then left with the outer (part of the)
square and the remaining numbers.
(A.II.39)310 Once you have done that, you take the two terms you have
reached;311 you put the small one —which will always be even if the
sequence (of the numbers) proceeds from 1 to the last by successive addi-
tions of 1— on the left underneath the corner cell, and its complement in
the facing right-hand side, opposite to it. Next you take the two numbers
following them;312 you put the small one, which is odd, on the bottom,
the large one in the facing upper side, opposite to it. Then you always
continue placing in this way313 the small even numbers alternately on the
right and on the left —but with the beginning on the left— and you place
the small odd (numbers) alternately on the top and on the bottom —but
with the beginning on the bottom— until you attain on the right-hand
side the small even number which precedes the even number correspond-
ing to the denomination of the side of the square [ with an odd number ],
this occurring always in the cell completing half the side.
Once you have written it there, put the next small odd number again
in this same side, thus the right-hand one, and opposite to it on the left
its complement. Place then the preceding small number, namely that
corresponding to the denomination of the side [ thus if the denomination
of the side is 6, it will be the number 6; if it is 10, it will be the number
10 ]314 in the upper left-hand corner, and opposite to it diagonally, in the
(lower) right-hand corner, its complement. Put then the small odd num-
308
Construction of squares of evenly odd orders, beginning with the inner square of
4.
309
Meaning: we fill first the inner square of order 4, just as, for odd-order squares,
we filled first the inner square of order 3; see A.II.4, A.II.5 (and A.II.8).
310
General construction of the borders for evenly odd squares.
311
That is, the next small number in descending sequence, and the next in ascending
sequence for the large numbers; thus, 10 and 27 for order 6 (Fig. a24), 18 and 83 for
the border of order 10 (Fig. a25, omitting the 14 smaller and 14 larger numbers kept
for the border of order 8).
312
The next in each sequence, thus 9 and 28, and 17 and 84, respectively.
313
Part of the account might be missing.
314
Hardly the author’s since he would (presumably) have put it before, at the first
occurrence of ‘denomination’.
ber preceding this number in the upper right-hand corner, and opposite
to it diagonally, in the lower left-hand corner, the number which is its
complement.
6 36 2 34 28 5
10 15 20 25 14 27
29 21 18 11 24 8
30 12 23 22 17 7
4 26 13 16 19 33
32 1 35 3 9 31
Fig. a 24
Once you have done that, you place the remaining small even numbers
alternately on the right and left sides, beginning on the left, until you
attain 4. In this way the completion of the right-hand and left-hand cells
is attained. You place (then) the small odd numbers (following the one
placed in the corner) alternately on the top and on the bottom, beginning
on the bottom. You put opposite to each number its complement, until
you reach 3. Once you have done this, you put 3 on the bottom, and
opposite to it on the top its complement; 2 on the top, and opposite to
it on the bottom its complement; 1 on the bottom, and opposite to it
on the top its complement. In this way you will have finished what you
wanted.
10 100 2 98 5 94 88 15 84 9
18 26 74 73 72 71 31 32 25 83
85 77 38 68 34 66 60 37 24 16
14 23 42 47 52 57 46 59 78 87
89 79 61 53 50 43 56 40 22 12
90 21 62 44 55 54 49 39 80 11
8 20 36 58 45 48 51 65 81 93
95 82 64 33 67 35 41 63 19 6
4 76 27 28 29 30 70 69 75 97
92 1 99 3 96 7 13 86 17 91
Fig. a 25
(A.II.40)315 The treatment in the (case of the) evenly even and evenly
evenly odd is as follows. You place the two medians and the numbers
next to them in the square inside these squares until you complete it, as
we have explained for the odd and evenly odd (orders).316
(A.II.41)317 If your treatment is for the square of 8, you take the num-
bers attained. Put two consecutive small (numbers) on the top, excluding
the corner cell, and put their complements opposite to them on the bot-
tom; then you put four consecutive small (numbers) on the bottom and,
opposite to them on the top, their complements.
8 56 55 54 53 13 14 7
59 20 50 16 48 42 19 6
5 24 29 34 39 28 41 60
61 43 35 32 25 38 22 4
3 44 26 37 36 31 21 62
2 18 40 27 30 33 47 63
64 46 15 49 17 23 45 1
58 9 10 11 12 52 51 57
Fig. a 26
(A.II.42)318 If your treatment is for (an order) higher than 8, you have
as (quantity) of cells in each row, excepting the corner cells, ten cells, or
fourteen cells, or eighteen cells, and so on always by successive additions
of 4 [ according to what you have put in the square of 8 ]319 . Now you
are to know that, for any group of four consecutive numbers facing four
(complementary) numbers, if you put on the same side the first and the
last of the four small numbers and the two middle of the four large ones,
and opposite to them on the other side their complements, you will have
equalized both sides.320 Thus if what remains, excepting the corner cells,
is ten cells, you will equalize four on the top with the four facing them
315
Even orders at least divisible by 4; first, filling the inner square.
316
In A.II.4–A.II.5 and A.II.38.
317
Filling the horizontal rows of the border for order 8 (particular case); see Fig. a
26, with the descending sequence from 14 and ascending from 51.
318
Filling the inner part of the horizontal rows for orders n = 4k, k ≥ 3.
319
This must be a gloss to the next sentence (indeed, a neutral placement has com-
pleted the two vertical rows).
320
This arrangement has already been mentioned for a sequence of even numbers
(A.II.20), and used several times (A.II.31, A.II.33, A.II.34).
on the bottom by means of four of the numbers you have reached after
filling the inner square and their complements.321 Likewise, if fourteen
(cells) remain, you will equalize eight with eight. And always like that
until you are left with six cells on the top and six on the bottom.
Once you have done that, put, of the small numbers you have reached,
two consecutive numbers on the top and, opposite to them on the bottom,
their complements; then put, of the small numbers you have reached, four
consecutive ones on the bottom and, opposite to them on the top, their
complements.
12 132 131 130 129 17 18 19 125 124 22 11
135 32 106 37 110 116 27 120 24 122 31 10
9 40 48 96 95 94 93 53 54 47 105 136
137 107 99 60 82 88 56 90 59 46 38 8
7 36 45 64 69 74 79 68 81 100 109 138
6 111 101 83 75 72 65 78 62 44 34 139
140 112 43 84 66 77 76 71 61 102 33 5
141 30 42 58 80 67 70 73 87 103 115 4
3 117 104 86 63 57 89 55 85 41 28 142
2 26 98 49 50 51 52 92 91 97 119 143
144 114 39 108 35 29 118 25 121 23 113 1
134 13 14 15 16 128 127 126 20 21 123 133
Fig. a 27
(A.II.43)322 Once you have done that, put the small number you have
reached —which [ is always even ]323 , for this succession, corresponds to
the denomination of the side— in the upper left-hand corner cell and,
opposite to it diagonally, in the (lower) right-hand corner, its complement.
Put then the subsequent small number, which is odd, in the upper right-
hand corner and, opposite to it diagonally, in the left-hand corner, its
complement. Put then the subsequent small number on the right and
its complement on the left, and the subsequent small number on the left
and its complement on the right. Once you have done this, you will have
321
For order 12 (Fig. a27), we begin with 22 and 123, respectively.
322
Filling now the corners and the two vertical rows.
323
The subsequent words make this information useless.
equalized these two sides as well.324

What you are left with (as empty cells in the vertical rows) is four
cells facing four, or eight facing eight, or twelve facing twelve, and so on
with successive additions of 4. Of these, you equalize each group of four
cells with the four opposite as we have explained,325 that is, putting (for
each group) on the (same) side the first and the last small and the two
large middle, and their complements opposite to them on the other side,
until you have finished with all (empty cells).
Fig. a 28* Fig. a 28 *
(A.II.44)326 As for the division of the even (order squares), it proceeds

as follows.327 You draw the main square, which arises from the multipli-
cation of an even number by itself. Then its inner part is divided into
four squares such that the side of each equals half the side of the main
square;328 or it is divided into sixteen squares such that the side of each
of these squares equals a fourth of the side of the main square;329 or it
is divided into nine squares such that the side of each equals a third of
the side of the main one;330 or it is divided into one square and other
squares, some of which arise from the multiplication of 8 by 8, others
from the multiplication of 6 by 6, others from the multiplication of 4 by
4.331 Then you will arrange the numbers in all the cells of the main square
324
The four cells thus filled in each vertical row make their sum due.
325
In A.II.42 (and note 320).
326
A.II.44–A.II.54 are about composite squares.
327
All subsquares must be of even order and filled with pairs of complements (filling
described in A.II.48).
328
Main order n = 4k, k ≥ 2. Example in Fig. a34.
329
Main order n = 8k, k ≥ 2.
330
Main order n = 6k, k ≥ 2. Example in Fig. a35.
331
See e.g. Fig. a28*. But not arranged in any arbitrary manner: a configuration like
that of Fig. a28 * will not produce the required sum in the main diagonals.
from 1 to the end of their quantity in such manner that in the (main)
square the sums calculated are everywhere the same, and that in each
inner square considered by itself the sums calculated are everywhere the
same.332 This (arrangement) is possible for the evenly even and evenly
evenly odd orders.333
(A.II.45)334 As for the evenly odd orders, it is not possible to divide the
square into four parts and to arrange in it the numbers in such a way as
to satisfy the above condition; indeed, it is absolutely impossible for the
evenly odd number [ found in it ]335 to have an integral fourth, without
fraction [ now, by convention, a fraction could not possibly be put in any
of these squares ].336 That is why this (kind of division) is impossible
for evenly odd (orders), (that is) for the specific case of division (of this
order) by two.
(On the other hand,) for an evenly odd order having a third or some
part other than a half, as (for the orders) 18 or 30 or the like, it will
be possible to divide it according to the part found in it, other than the
half, with the resulting quantity being even —let us leave out 2 which,
as noted by us before, offers no possibility337 . Arranging then there the
numbers, it will lead to what we have explained.
Thus, the square of 18 is divisible into nine squares, each with size 6
by 6, (where the numbers will be arranged) in such a way that the sums in
the main square are the same and also in each of these (smaller) squares.
The square of 30 is divisible into nine squares, each with size 10 by 10,
and also in twenty-five squares, each with size 6 by 6, and also into four
squares in its four corners, each with size 12 by 12, and, separated, nine
squares, each with size 6 by 6,338 where the numbers will be arranged
in such a way that the sums will be everywhere the same, in the main
(square) as well as in each of the small ones.
332
In the manner to be explained in A.II.48.
333
And for some evenly odd orders, but not for all, as will be explained right now.
334
Case of an evenly odd order (thus divisible by 2 only) with the odd factor being
composite (the identical even-order subsquares will be filled with pairs of complements
adding up to n2 + 1 with n the order of the main square). First, the division into an
even number of odd-order subsquares will be dismissed.
335
Early reader’s addition prompted by the lacuna.
336
The four subsquares will be of odd order, whereas filling a subsquare with comple-
ments requires its order to be even, thus the order of the main square to be divisible
by 4. This statement was misunderstood by the reader.
337
Repetitive, perhaps an addition.
338
Thus with different subsquares (see A.II.44, Fig. a28*).
(A.II.46)339 It is (also) possible with these two orders340 as well as with

other evenly odd orders to carry out a division into different parts and to
place in them the numbers so as to be led to the constancy of the sums
in compliance with the condition.341
Fig. a 29* Fig. a 30*
Fig. a 31* Fig. a 32*

Thus (with the square of) 10 (Fig. a 29*), it is possible that there
might be in its four corners four squares each with size 4 by 4 displaying
individually equal sums everywhere, with the sums in the whole square
being (themselves also) everywhere the same.342 It is likewise possible
with (the square of order) 14 (Fig. a 30*) that there might be in its four
corners four squares of size 6 by 6, each displaying sums everywhere the
same; it is moreover possible that there might be in the centre of this
square a square of size 6 by 6 surrounded by eight squares of size 4 by 4,
each displaying individually same sums (Fig. a 31*343 ). It is again possible
with the square of (order) 18 (Fig. a 32*) that there might be in its centre
339
Division into parts of different sizes (here: squares and rectangular strips).
340
Namely 18 and 30.
341
Namely, constancy of the sums in the main square and in its various parts, as will
be specified below.
342
How to fill the central cross will be explained in detail further on (A.II.50–A.II.53).
343
Or with the 2 × 4 strips flanking the corner 4 × 4 squares: the author seems to
avoid considering strips within squares.
a square of size 10 by 10 surrounded by twelve squares of size 4 by 4

displaying also same sums. And the same for all such squares.
(A.II.47)344 As for the odd(-order) squares, it is impossible that one of

them, being divided into squares —with the order of each being odd—
might then satisfy this condition. Indeed, a square of odd order neces-
sarily has a single median number; so if one places in the main square a
certain quantity of squares all of the (same) odd order and supposed to
display same sums, the median number of each will have to be that of the
main square in order that you might perform the equalization by means
of the remaining numbers.345 [ Now of the numbers (to be placed) none
may ever occur two or three times ] [ only in a single place ]. Now, whether
you place the median in all (these) squares or you omit their places al-
together and you put consecutive numbers [ in such a way as to omit the
place of each pair of medians ] [ a pair of complementary numbers ], this
will not meet the condition of magic square and it will not lead to (the
situation) we described for the preceding odd orders346 and the main sum
will be wrong because of the repetition of some (numbers) and (in the
second case) the omission of others.347
344
Impossibility of dividing odd-order squares into equal odd-order subsquares con-
taining identical magic sums.
345
Since each subsquare must contain the same magic sum, their central cells should
all contain half the sum of two complements, thus the median number of the main
square.
346
A.II.3–A.II.6.
347
The text here is corrupt because of early readers’ attempts to make sense of it.
The argument is that, whether we write the (same) median in all central cells or leave
them all empty, and then fill the remainder with pairs of complements, the constructed
square will not meet the magic condition. In fact, without even considering the central
cells, filling the borders of subsquares of odd order m < n with pairs of complements
for the main square will not produce the same sums in the outer border (Fig. a33*).
6 8 n 2 − 2 n2 − 1 4
7 n2 − 6
5 n2 − 4
n2 1
n2 − 3 n2 − 7 3 2 n2 − 5
Fig. a 33*
Consistent with his initial restriction to bordered squares, the author considers only
squares filled by means of complements. He therefore does not consider the possibility
60 82 88 56 90 59 24 118 124 20 126 23
64 69 74 79 68 81 28 33 110 115 32 117
83 75 72 65 78 62 119 111 36 29 114 26
84 66 77 76 71 61 120 30 113 112 35 25
58 80 67 70 73 87 22 116 31 34 109 123
86 63 57 89 55 85 122 27 21 125 19 121
6 136 142 2 144 5 42 100 106 38 108 41
10 15 128 133 14 135 46 51 92 97 50 99
137 129 18 11 132 8 101 93 54 47 96 44
138 12 131 130 17 7 102 48 95 94 53 43
4 134 13 16 127 141 40 98 49 52 91 105
140 9 3 143 1 139 104 45 39 107 37 103

Fig. a 34
69 74 79 68 29 114 119 28 61 82 87 60
75 72 65 78 115 32 25 118 83 64 57 86
66 77 76 71 26 117 116 31 58 85 84 63
80 67 70 73 120 27 30 113 88 59 62 81
21 122 127 20 53 90 95 52 13 130 135 12
123 24 17 126 91 56 49 94 131 16 9 134
18 125 124 23 50 93 92 55 10 133 132 15
128 19 22 121 96 51 54 89 136 11 14 129
45 98 103 44 5 138 143 4 37 106 111 36
99 48 41 102 139 8 1 142 107 40 33 110
42 101 100 47 2 141 140 7 34 109 108 39
104 43 46 97 144 3 6 137 112 35 38 105

Fig. a 35
(A.II.48)348 If you wish to place the numbers in some of the squares you
have divided into squares not separated by strips insufficient to contain
of subsquares with different magic sums, though this was known at the time (B.24).
348
How to fill a square of even order n completely divided into equal even-order
subsquares of orders m = 2. (The case of rectangular strips remaining to separate
the subsquares will be examined later on.) For a subsquare, the two medians are the
a complete square (proceed as follows). Take the two medians of the

numbers which are to occupy this square and put them in one of the
squares into which you have divided it, at the place of the two medians
of the 4 by 4 square, and then (proceed) according to this (known way
of) filling to the end of this square. Take then the two terms you have
attained, attribute them the rôle of medians and put them in some other
square among these, and proceed as before. (Do) always the same until
you have finished with what you wanted (Fig. a 34, Fig. a 35).349
(A.II.49)350 If (now) you have divided such a square into squares sep-
arated by strips insufficient to contain a complete square and the sepa-
rations are in the sides without interfering with the diagonal (proceed as
follows).
Fig. a 36*
— If the separation is 6 by 4 (Fig. a 36*), you will place in each group of
four cells with a group of four cells below it eight complementary terms
(Fig. a 37*); with them, each row will equalize its conjugate, as we have
explained at the beginning of this section.351 You proceed like that until
you have completed the six (rows).
— If the separation is 6 by 8, you will treat the four in two steps, and
the six as here above.352
It is not possible that the square be divided into parts with the sepa-
ration being odd (in dimensions) because of what we have explained;353
largest small number and the least large number to be placed in it, as stated below.
349
Since each subsquare will present the same sum, they can be dealt with in any
sequence. In the two figures given by the author, the first square filled is the upper
left-hand one, starting from the medians 72, 73 of the main square; the numbers
immediately following (54, 91 and 64, 81 respectively) will be the medians of the next
square considered.
350
Squares containing non-square parts, first rectangles surrounding a central square;
how to fill these rectangles.
351
Above, A.II.42.
352
Considering two strips like that of Fig. a37*.
353
The case of odd-order subsquares having been dismissed (A.II.47), the sides of the
as a matter of fact, the separation will always be 6 by 4, or 6 by 8, or 2

by 4, or 2 by 6, or 2 by 8, and so on always.354 If the separation is 2 by 4
(Fig. a 38*) and is situated in the sides without meeting the diagonal, the
treatment for its equalization will be as we have explained previously.355
α1 α4
β1 β4
γ1 γ4 α1
γ2 γ3 α2
β2 β3 α3
α2 α3 α4
Fig. a 37* Fig. a 38*
(A.II.50)356 This (separation) may be in the middle and thus meet the
diagonal. There appears then in the centre of the (main) square a square
of size 2 by 2. This for instance occurs in the (square of) 10 by 10 when
are placed in its corners four squares each of size 4 by 4, or in the square
of 14 when are placed in its four corners four squares each of size 6 by
6.357
(You will proceed in this case) as follows. You place the two medians
of the whole set of numbers358 in two diagonally opposite corners of the
square with size 2 by 2 in the centre of the main square (Fig. a 39*).
Omitting two numbers after the two medians,359 you take the next two
and place them in the two opposite corners left. At this point, you will
find that the row containing the two small numbers has a deficit of 3
relative to its sum due [ one relative to that parallel to it ], that the row
parallel to it has an excess of 3, that the row containing one of the two
small (numbers) and the larger of the two large (numbers) has an excess
strips are necessarily even.

354
The separations 6 × 4 and 6 × 8 have just been treated; a pair of rectangles 2 × 6
may be reduced to the case 6 × 4 by combining them; a strip 2 × 8 is reducible to
two 2 × 4, to be discussed now. Note that Fig. a37* can be reduced to Fig. a38* by
removing a 4 × 4 square.
355
Using a neutral placement.
356
Reducing a square of evenly odd order n to a central cross of size 2 × n and
even-order squares; how to fill the central square of the cross.
357
These examples have been mentioned in A.II.46 (see Fig. a29* and a30*). What
follows is generally applicable to evenly odd squares.
358
Thus the two medians of the main square.
359
One on either part.
of 2, and thus that (the row) parallel to it has a deficit of 2.
+3 −3
⇑ ⇑
n2 n2 ⇒
2
+3 2
+2
n2 n2 ⇒
2
+1 2
−2 −2
Fig. a 39*
(A.II.51)360 If then what remains in the cross is, in its four directions,361
2 by 6 (Fig. a 40*), you will have to equalize the (central) square by means
of two cells on each side;362 in this way, the central part 6 by 2 of the
cross will be equalized and there will remain, to complete the cross, on
the four sides, (strips of size) 2 by 4 which you will equalize in the manner
explained.363
• •
• •
• • × × • •
• • × × • •
• •
• •
Fig. a 40*
(A.II.52)364 (In order to equalize the central square) you proceed as

follows. After putting in the centre the four numbers which we have
360
How to equalize the four remaining branches of the cross.
361
On each side of the 2 × 2 central square.
362
That is, by means of two pairs of cells on each side of the 2 × 2 central square.
363
This is applicable to order 14, as in the figure, and generally to orders n = 8k + 6;
indeed, after filling the 6 × 2 central part of the cross we are left with, in each branch,
a number of cells which is a multiple of 4 (unlike the orders n = 8k + 2, treated in
A.II.53). But, as said, we are first to eliminate the differences displayed by the central
square.
364
As told in the previous section, we must now equalize the central square, already
filled (A.II.50), by means of two pairs of cells in each branch. Our addition, in brackets,
of the numbers alluded to and of Fig. a41* will make the text less abstruse.
indicated, you consider the row with a deficit of 3 (Fig. a 39*). Put in the
pair of cells adjacent to it, on one of the two sides, the lesser of the two
2
numbers you have omitted after (placing) the two medians (thus n2 − 1)
and the large number following the two large numbers you have written
2
in the centre (thus n2 + 4); and put in the two cells adjacent to it (= the
central square), on the other side, the lesser of the two small numbers
2
following the small number you have reached (thus n2 − 4) and the larger
of the two large numbers following the large number you have reached
2
(thus n2 + 6); and put opposite to each its complement. Once you have
done that, you will have equalized these six (pairs of) cells.
n2 n2
2
−3 2
+4
n2 n2
2
+2 2
−1
n2 n2 n2 n2 n2 n2
2
−9 2
+9 2
+3 2 2
+7 2
−7
n2 n2 n2 n2 n2 n2
2
+ 10 2
−8 2
+1 2
−2 2
−6 2
+8
n2 n2
2
+5 2
−4
n2 n2
2
−5 2
+6
Fig. a 41*
Next, you examine the row in deficit of 2. Put on one of the two sides,
in the two cells adjacent to it, the least of the terms you have reached
2
( n2 − 6),365 then the larger of the two large (numbers) you have reached
2
( n2 + 8);366 and (proceed) in like manner (as in the previous case) for the
other side.367 Put opposite to each number its complement. Once you
have done this, you will have equalized these six (pairs of) cells.
At that point, you will equalize the remaining (cells) on the (four)
sides, of (size) 4 by 2 on each side, according to what I have explained to
you.368
(A.II.53)369 (But) if what is left (in the cross), excepting the central
365 2
The lowest term already placed (as complement) being n2 − 5.
366 2
Indeed, the complement of n2 − 6 has not yet been placed.
367
As seen for the vertical branch. We shall thus place the lower of the next small
2 2 2 2
pair ( n2 − 7, n2 − 8) and the higher of the next large pair ( n2 + 9, n2 + 10).
368
Case of order 14 (Fig. a40*), each remaining part of the cross receiving one neutral
placement.
369
Case of order 10, leaving on each side of the central (2 × 2) square a group of four
square, is on each side 2 by 4 (Fig. a 42*), you will place that which will
equalize the row in deficit of 3, and the same for the row in deficit of 2,
in four cells on (just) one of the two sides (of the central square): you do
not need to make a repartition on (all) four of its sides.370
Once you have done that, you will equalize each group of four cells in
the two remaining sides as I have explained.
• •
• •
• •
• •
× × • • • •
× × • • • •
Fig. a 42*
(A.II.54)371 You will proceed likewise for a remainder of 2 by 8, or of 2

by 10.372 You will (first) equalize the centre:373 when there remain eight
(cells on each side), you will equalize on two sides only, each time with
four cells; when there remain ten (cells), you will equalize on the four
sides, each time with two cells. (Then) you will equalize what remains,
(consisting of strips of size) 2 by 4, as I have explained to you before.
It will be for you, after examining the squares which we have filled and
becoming acquainted with how to place the numbers and arrange them,
to deduce from it, with the help of God and his valuable assistance, what
I have omitted to explain; thinking about it will lead you to the purpose,
if such is the will of God Most High.374
Here are, grouped together, the illustrations for what we have ex-
plained about the odd and even (orders).375
empty cells.
370
It should be said that we are also to place in each of the two adjacent rows the
complements of the four cells just filled.
371
Method of the cross for the next two orders.
372
Disregarding the central square, thus case of, respectively, order 18 (with equaliza-
tion of the central square on two of its sides only) and 22 (distributing the equalization
among its four sides). In fact, this distinction is superfluous since a neutral placement
does not require consecutive cells.
373
That is, eliminate the differences displayed by the 2 × 2 central square.
374
If referring to the method of the cross: extension to orders n = 8k + 2 and
n = 8k + 6. If not, usual conclusion (see note 298).
375
As already said (note 215), here are found all figures (without asterisks) (excluding
a6 but including a1, repeated).
Chapter III
Determining the hidden number
(A.III.1) The utility of this chapter lies in exercising the determination
of what is unknown for numbers when it is not easy. We have already
expounded on this in sufficient manner by way of first elements and rudi-
ments.376 I have divided it into two parts: the first on the determination
of hidden numbers; the second on the determination of hidden names.377
(A.III.2)378 The determination of hidden numbers is divided into two
main parts: the knowledge of the quantity of the hidden number and
the knowledge of its quality, which is the knowledge of its attributes. Its
attributes are divided into two categories, namely the specific and the in-
cidental. The specific are what has preceded in our present book:379 even
and odd and their (sub-)species: evenly even, evenly odd, evenly evenly
odd,380 (then) odd prime, odd composite, and the third one, intermediate
between those two, namely that which is intrinsically composite but, rel-
ative to another, prime and incomposite381 . The incidental (attributes)
are like being perfect, deficient, abundant,382 linear, plane, solid383 .
(A.III.3) Quantity is divided into two categories: it is either single, like
three, four, or compound (additively), in which case it is divided into
376
Maybe an allusion to the numerical identities and the propositions from Book II
of the Elements seen in the first part (A.I.96–A.I.133).
377
A.III.6–A.III.33, A.III.40–A.III.41 and A.III.34–A.III.39 (note 488), respectively.
378
The coming section A.III.2, mainly on the quality of number, has no bearing on
the subject treated here. A.III.3, on the quantity of number and on the categories
of numerical quantities, is also irrelevant. Both are also in the treatise on the same
subject by al-Kindı̄ (see p. 15) —hereafter designated by K. They must thus be an
early addition to their common source by a glossator who misunderstood the meaning
of ‘hidden number’ in this context, taking it to refer to the determination of a quantity
in general.
379
Whereas our author here refers to his treatise, K at this same place refers to some
other work of his own.
380
See A.I.6 & A.I.16, A.I.5 & A.I.14, A.I.15 & A.I.17. See also A.II.36.
381
A.I.1 for the first; composite and relatively prime numbers are not really defined in
the present text. Nicomachos I.xi.1–xiii.1 explains the three cases, respectively called
πρω̃τος καı̀ ἀσύνθετος ἀριθμός, δεύτερος καı̀ σύνθετος ἀριθμός, καθ᾿ ἑαυτὸν δεύτερος
καı̀ σύνθετος προς ἄλλον δὲ πρω̃τος καı̀ ἀσύνθετος.
382
Numbers called, respectively, τέλειος, ἐλλιπής, ὑπερτελής; Nicomachos I.xiv–xvi.
See p. 12 and note 160.
383
Numbers called, respectively, γραμμικός, ἐπίπεδος, στερεός; Nicomachos II.vi.1.
The last two are the products of two and three factors, respectively, whence these
geometric qualifications. Without being defined, these two denominations occur in the
first part (A.I.68–A.I.71, A.I.77–A.I.82).
two categories: either (compound of) same, like two plus two, three plus
three, and beyond that, or (compound of) different, like two plus three.
It384 is again divided into two categories: it is either simple or composite.
The simple is divided into two categories:385 it is either multiplied, or
increased by a fraction. The composite is divided into two categories: it
consists either of a single species or of two species. That which consists
of a single species is like that increased by a fraction, aliquot or not; that
which consists of two species is like the composite increased by a fraction,
aliquot or not.386
(A.III.4) The knowledge of all that we have mentioned will be realized
after we have completed the means to know the hidden quantity.387 In-
deed, if we know its quantity, we shall know to which species the number
belongs;388 [ therefore knowing, by their quantity, hidden numbers, we
shall thereby know to which types of correlation they belong ]389 since
for any number which is known all its properties and attributes, and its
species, will be known.390 Consequently, our aim must be to seek the
means existing for (finding) the quantity of the number [ since thereby
one finds the complete knowledge of this number ]391 using what the one
who wishes to find out the hidden number has noticed [ by means of its
multiplications or its divisions, or both together ]392 . So let us put first
a preliminary which will guide our research in this chapter.
(A.III.5) So we say that any number whatsoever has to any number a
certain ratio, whichever it is. This means that if a number is multiplied by
some number, whichever the multiplier —plane or solid—, or is divided,
whichever the divisor, and the same is done with the unit, then the result
arising from the augmentation or the diminution of the number will be
to the original number as the result arising from the unit, subject to the
very same treatment, will be to the original unit.
384
Referring to the quantity ‘(compound of) different’, according to K.
385
Lacuna filled using the text K.
386
We chose to leave the text as it is.
387
Both texts have ‘the simple quantity’ instead of ‘the hidden quantity’, perhaps an
alteration by an early reader trying to make a link with the previous section.
388
At least we shall know its above-mentioned specific attributes.
389
Gloss —early one, also in K— by a reader trying to establish a link with the
aforesaid kinds.
390
If that were so, number theory would be considerably simplified. K confirms
that we shall know not only the parity of a number but also if it is perfect, deficient,
abundant.
391
This —also in K— is an unnecessary repetition.
392
Gloss alluding to what will follow.
182 Determining the hidden number
(A.III.6) After putting this as a preliminary, let us start with the expla-
nation of what we intend to expound [ which is the determination of these
hidden numbers ]. We say then that the way to determine them (is as fol-
lows). He who wants to discover the number in question orders the person
to apply to it what we have explained [ what we have mentioned ]393 —
multiplications or divisions— and he himself does the same with the unit.
Then what arises from his aforesaid treatment of the unit will result for
each unit of the hidden number. The person is then ordered to subtract
from what has resulted to him from the hidden number the number which
has resulted to him394 from the same treatment applied by him to the
unit [ division, multiplication, or both together, whichever the treatment
applied to it ]395 ; you count how many times the person has subtracted
this from the result until it is eliminated. You take then 1 for each time,
and what results will be the number which was hidden.396
(A.III.7) Example of this. We suppose the hidden number to be 3. We

order the person to double it, which gives 6. For our part, we double
1, which gives 2. We order the person to subtract 2 from what he has,
repeatedly, until all that he has is eliminated. The (number of) times
there has been subtraction is 3. Therefore we shall say: ‘You have held
3’.
The reason for that is the following. When the person has carried
out a certain treatment [ repeated multiplications, or divisions, or both
of them ] [ that is, on the number ]397 , you did yourself the same to the
unit. Now the ratio of the unit to the hidden number equals the ratio
of the result of your treatment applied to the unit to the result of his
393
The reader refers to his own addition (see previous note).
394
This time the reference is to the asker. With such ambiguities it is not surprising
that the text has been abundantly glossed. For the asker the text mostly uses the
second person, sometimes the first plural.
395
Useless repetition. Also in K.
396
In our terms, the person, following the instructions of the asker, forms
a1 · a2 · . . . · an
x = A · x.
b1 · b2 · . . . · bm
The asker knows the factor A; in order to determine x, he will just ask the person how
many times A is contained in his result, the person obtaining this either by division
or —which better serves the purpose of the method, which is to confuse him— by re-
peated subtractions of A. This approach (applying to the required number a succession
of multiplications and divisions without remainders), which is explained in detail in
A.III.5 and A.III.6, is in fact employed only in A.III.7–A.III.13 and A.III.17–A.III.21.
397
Two separate glosses, as it seems. One clarifies the word ‘treatment’, the other
states the object of its application.
own treatment applied to the hidden number.398 Therefore if his result

is divided by the result of your same treatment applied to the unit, there
will result precisely what he has hidden.
However, he who (wishes to) find out (the hidden number) must in-
crease the (number of) multiplications and divisions in order to keep the
person in ignorance of the determination procedure [ the fact of accumu-
lating the things possibly intensifies their secrecy ]399 . I shall give you
two examples which you will take as model to know how to proceed for
determining this, with the help of God.400
(A.III.8) You order the person to multiply what he has by some num-
ber you choose, for instance you order him to multiply it by four, then
to double the result and to multiply that by three. For your part, you
do the same with one. Then your whole treatment applied to one [ this
treatment ]401 gives twenty-four. You order him then to subtract from
the whole its sixth. Next, after your subtraction of the sixth, you order
him to eliminate the whole with 20s, and you take one for each twenty
he subtracts. Then what the person has hidden is the sum of your count-
ing.402
(A.III.9) You may order him to subtract half of what has remained after
the (subtraction of the) sixth, and you take one for each ten (subtracted):
it comes to the same.403
(A.III.10) You may order him to subtract, after the half, a fifth of the
398
As seen in A.III.5.
399
This is really repetitive. Addition of the same kind in K.
400
Let us now suppose that the multiplicative factor A is divisible by c1 , c2 , . . . , ck .
We shall then multiply A by one or several of the factors

1 1 1
1± , 1± , ... , 1 ± ,
c1 c2 ck
which will modify the number to be subtracted at the end and also add to the person’s
confusion. There are several examples of it, but the author announces only two; our
A.III.8–A.III.13 may be considered as two groups of three problems. Note, too, that
the ‘multiplications and divisions’ thus include adding and subtracting fractions of the
numbers involved.
401
Gloss —inserted in an inappropriate place— to clarify ‘the same’ here above.
402
Here the number x is successively changed, according to the requirements, into
4 · x, 2 · 4 · x, 3 · 2 · 4 · x, the result on 1 being 24. The two results are then reduced each
by a sixth of it. Since 24(1 − 61 ) = 20, it is this number which the person will have to
subtract repeatedly. Thus, if x = 4 (A.III.8a), 80 will be the number obtained by the
person, and 20 that which the asker will tell him to subtract.
403
That is, we shall end with the same result. The person calculates 3 · 2 · 4 · x(1 −
1
6
)(1 − 12 ), and, since 20(1 − 12 ) = 10, he will repeatedly subtract 10. Thus, if x = 4
(A.III.9a), 40 will be the number he obtains.
remainder, and you take yourself one for each eight.404

(A.III.8a) Thus if the person holds 4. He multiplies it by 4, which gives
16. Double it, this makes 32.405 Multiply the result by 3, this gives 96.
Subtract from it its sixth, 80 remains. If he eliminates the remainder
with 20s, the (number of) subtractions will be 4, and this is what had
been hidden.
(A.III.9a) Likewise if he subtracts half of what has remained to him,
there remains forty; you take one for each 10, and 4 results.
(A.III.10a) Likewise if he subtracts from 40 its fifth, there remains 32;
for your part, you subtract from 10 its fifth, which leaves 8; if he eliminates
32 with 8s, 4 results.
(A.III.11) This determination may be (even better) concealed by telling
the person, when he arrives at the subtraction of the fifth: ‘Add to what
remains to you its eighth; and you subtract from the result 9 repeatedly’.
The result will be 4, which is the hidden number.406
(A.III.12) If you wish to conceal even better its secrecy, you will tell
him: ‘Add to your result its eighth, and to the result its ninth’. You will
then take one for each 10.407
(A.III.13) You may (further) tell him: ‘Add to your result its tenth’.
You will take one for each 11. The result will be the hidden number.408
(Other) chapter 409

(A.III.14) You may order the person to add to it what he wishes and
to subtract from it what he wishes; then to multiply the (sum or the)
remainder by 5, and to double the result; and to add to it some number
you choose. You keep for your part these additions in mind. Next you
order him to subtract the tens, the hundreds, the thousands, if there are.
For your part, you subtract from what you have kept in mind tens, and
404
The person thus calculates 3 · 2 · 4 · x(1 − 61 )(1 − 12 )(1 − 15 ) and will subtract
10(1 − 15 ) = 8. If x = 4 (A.III.10a), 32 will be his result before the subtractions.
405
The instructions are for the person.
406
In this problem and the next two, involving more (now additive) quantities will
make the procedure more obscure. Here 3 · 2 · 4 · x(1 − 61 )(1 − 12 )(1 − 51 )(1 + 18 ), to be
eliminated by subtractions of 9.
407
3 · 2 · 4 · x(1 − 16 )(1 − 12 )(1 − 15 )(1 + 18 )(1 + 19 ), with subtractions of 10.
408
3 · 2 · 4 · x(1 − 61 )(1 − 12 )(1 − 15 )(1 + 18 )(1 + 19 )(1 + 10
1
), with subtractions of 11.
409
Here we also find, applied to the required number, additions and subtractions of
other quantities.
its hundreds and thousands. You will be left just with the same (number
of) units as he.410
Example of this. The person holds 4. He adds 5 to it, which gives
9. He subtracts 3 from it, which leaves 6. He multiplies it by 5, which
410
These explanations are confused, as is the subsequent example. No allusion is
made to a later example (A.III.24). The person must calculate an expression of the
form

. . . a3 a2 (a1 x + b1 ) + b2 + b3 . . . ,
where the ai , bi are chosen by the asker. The result will thus be of the type Ax + B,
with A and B known by the asker. After telling the person to subtract B, he will ask
him to remove A as many times as possible.
The brevity of this ‘(other) chapter’ is surprising, for we have here one of the most
common forms of the problem of hidden numbers. We find it already in the early
Middle Ages (which clearly indicates its antique origin) and quite commonly in the
late Middle Ages. Most frequently, the required number appears in the final result
multiplied by some power of 10 to facilitate the subtractions. In the case of several
required numbers (all but the first < 10), each will appear as multiplied by a different
power of 10 and thus occupy in the answer a determined decimal place. Here are some
characteristic examples (see our Récréations mathématiques, ch. xv).
(1) If we wish to know how many coins somebody has in his purse, we tell him to
double the number and then to add 5; next to multiply the result by 5 and add
10; to multiply finally the whole by 10 and subtract from it 350. The number of
hundreds will be the number of coins in the purse. Indeed,

10 5(2x + 5) + 10 − 350 = 100x.
The numbers we add here serve no purpose other than to confuse the person and
may be varied at will since in the end their sum is subtracted. A similar problem is
found in Bede (above, p. 15).
(2) We may also determine who, in a group of persons sitting in a circle, holds a ring.
We ask one of the participants, who knows in whose possession it is, to mentally
count the number of people between him and the holder. He is then to apply to this
number the above sequence of operations. See A.III.36 (powers of 10 not involved).
(3) We may also determine in which hand the ring is held by calculating

10 5(2x + 5) + 10 + m − 350 = 100x + 10m,
with m = 1 (left hand) or 2. The hundreds thus indicate the person and the tens,
the hand.
(4) A further game consists in finding the exact location of the ring: hand, finger,
phalanx. To do this we shall require to be calculated

10 10 5(2x + 5) + 10 + m + f + p − 3500,
with the person asked to take m = 1 or m = 2 according to whether the ring is in
the left or right hand, while f indicates the finger (counting from the little one) and
p, the phalanx (from the palm). Since the above expression reduces to
1000x + 100m + 10f + p,
we know at once from the number of thousands, hundreds, tens, units where the
ring is.
produces 30. He doubles it, which makes 60. Next you order him to add
to it 5, then 5, then 5, then 5, then 5, for instance, or, if you wish, more
or less (times) than that, (of the) same or another. Next, you order him
to subtract the hundreds and the tens, and for your part you do the same
with what you have kept in mind. You will be left with 5, as will he.
But if you have taken five an even number of times, you will be left with
nothing, and neither will he.
Other kind of this 411

(A.III.15) You order the person to multiply what he keeps hidden by 3,
then to take half the result, to multiply what remains by 3, and to take
half the result. [ Next, he examines (the result). If there is a fraction, it
will be multiplied by 4. ]412 Then he eliminates the result with 9s. For
your part you take, for each of their subtractions [ of the nines ]413 , 4, and,
for what remains according to this computation which does not complete
9, you will take one for each two and a fourth. The result will be what
was hidden.414
Other kind of this

(A.III.16) You order the person to add to what he has its half —if (the
number) has an (integral) half— or its third, or its fourth, or whatever
part it has. He then multiplies it by the number homonym with this
part.415 Next you add to the denominator of the fraction this same frac-
411
Multiplications and divisions of the required number; the results may involve frac-
tions. A.III.15 & A.III.16 are in K.
412
Interpolation, not in K (from what follows it is clear that the fraction should be
kept).
413
To clarify the meaning of ‘their’.
414
The person is thus to multiply the hidden number x twice by 23 ; the divisor is then
4. According to whether x is of the form 4k, 4k + 1, 4k + 2 or 4k + 3, the results found
will be, respectively, 9k, 9k + 94 , 9k + 18
4
or 9k + 27
4
. The person must then subtract 9
as many times as possible, thus obtaining k and, if there is one, indicate the remainder.
Multiplying k by 4 and counting 1 for each 2 + 14 = 94 contained in the remainder, the
asker will know x. We could also multiply the remaining fraction by 4 and eliminate
the result by subtractions of 9. This must be the meaning of the interpolation.
This too is a common mediaeval problem, certainly of antique origin. In its simplest
form, there are two multiplications by 3 but just one division by 2; then each 9 removed
is worth 2, and the presence or absence of a final remainder will determine whether or
not to add a unit, and will thus indicate the parity of the hidden number (Récréations,
problems 181–182). These two forms of the problem are found in early mediaeval times
(Bede, see p. 15).
415
That is (as we shall translate below) the denominator of this (aliquot) fraction.
See A.I.51.
tion (taken) of itself,416 then you multiply the result by this denominator.
You then order him to subtract (from his own result) the number result-
ing from (the multiplication of) the denominator, and you keep in mind
how many times there has been a subtraction. [ They will equal the ratio
of one to this hidden number. ] You multiply them by the denominator
of the fraction; the result will be the hidden number.417
Example of this. The person keeps hidden 3. He adds to it its third,
which gives 4; he multiplies it by 3, which gives 12. For your part, you
add to the denominator (of the fraction), thus to 3, its third, which gives
4; then you multiply it by 3, which gives 12. You order the person to
eliminate his result with 12s. This gives 1. This is a third of the hidden
number. Then you multiply it by the number of the part; this gives 3.
You will thus tell him: ‘You have held 3’.418
Another kind of this 419

(A.III.17) Order the person to multiply the hidden number by 4, to
subtract half the result, to multiply the remainder by 4, to subtract half
the result, to multiply the remainder by 4, to subtract half the result, then
to eliminate the remainder by 8s. There will result the number itself.420
(A.III.18) You may order him to add to the remainder a fourth of it

416
Simpler (this is the asker!): he adds 1 to the denominator.
417
The asker tells the person to divide his number x by one of its divisors, which the
person declares, say k, and to add the result to x; he himself carries out the same two
operations on k. They thus obtain, respectively,
x + k1 x k + k1 k = k + 1.
The results must then be multiplied by k. These two expressions then become
x(k + 1) k(k + 1).
The second, known by the asker, must be subtracted as often as possible from the first.
Since x is of the form k · t, the remainder 0 will be attained after t subtractions, and
the asker, who now knows its two factors, can calculate the hidden number.
418
The author has really chosen the least illuminating example, that of a prime
number, so that the number declared equals the hidden number. K takes x = 30,
k = 2 (t = 15).
419
Once again, multiplications and (integral) divisions of the hidden number. The
first of these problems resembles A.III.15, after which we meet again the situation seen
in problems A.III.8 to A.III.13 —without any reference to them in the text.
420
The hidden number x is multiplied three times by 4, and each time half the result
is removed —thus, more simply, x is multiplied by 23 (as done by K, who considers
three consecutive doublings). The number of times the person can subtract 8 will
be the required number. In order to further conceal the procedure, the author will
propose to add to, or subtract from, the final result a fraction of itself; the quantity to
be repeatedly subtracted will be modified accordingly.
and to eliminate the result by 10s. There will result the number itself.421
(A.III.19) You may order him to add to it, after the fourth, a fifth of
the whole, and to eliminate the result by 12s.422
(A.III.20) You may increase its secrecy by telling him, after the addition
of a fifth (of it): ‘Subtract from the result a fourth of it’. For your part,
you will take one for each 9.423
(A.III.21) You may order him to subtract from what has remained a
ninth of it, and for your part you will take one for each 8.424
In short: Order him to subtract from this what you want, then to
eliminate the remainder by means of the quantity remaining from the
number (which you have obtained). The result will be precisely the hidden
number.
Other kind of this 425

(A.III.22) You order the person to halve what he has; you hold for your
part one, which you then double. You ask him whether the result contains
a fraction or not. If not, you order him to halve it a second time, and for
your part you double a second time what you have. You ask him whether
the result contains a fraction or not. If not, you order him to halve it a
third time, and for your part you double the number a third time. (The
person goes on) always like that until he is left with nothing;426 for your
part, you double every time he halves. If there has never been a fraction,
he has held an evenly even number, which will be precisely what you
have. If he announces one time that he has in his hand427 a fraction,
you will order him to ignore it and you will for your part hold in the
other hand as much as the sum you have (obtained) by doubling.428 You
will do the same each time the person announces that he has a fraction.
When nothing remains from the hidden number,429 you will add the two
421
8x (1 + 14 ) = 10x (subtractions of 10).
422
10x (1 + 15 ) = 12x (subtractions of 12).
423
12x (1 − 14 ) = 9x (subtractions of 9).
424
9x (1 − 19 ) = 8x (subtractions of 8, as in A.III.17).
425
Reduction of the hidden number by divisions.
426
In fact, he reaches 1 (the subsequent 21 would be disregarded).
427
That is, in the result he has obtained.
428
One hand records the sum of the successive doublings of the unit, the other the
sum of the numbers attained when a fraction occurs. A hand can represent a number
belonging to two consecutive decimal orders; but holding numbers ‘in the hand’ is often
meant figuratively.
429
Again, 1 and then disregarding the final fraction.
numbers (contained) in your hands [ and you will take half the sum ]; this
will be precisely equal to what was hidden.430
If you order the person to subtract a third of what he has,431 you
will multiply the unit by 3, and the rest of the treatment will be like the
previous case of halving. Likewise if you order him to subtract a fourth
of it, you will multiply the unit by 4, and the rest of the treatment will
be like for halving. Likewise (for the case of another fraction): you will
multiply the unit by its denominator, and the rest of the treatment is
identical.432
Other kind of this

(A.III.23) You order the person to multiply what he keeps hidden by 3
and to subtract (from the result) its half. When he has done that, you ask
him whether the (remaining) half contains a fraction or not. If it does,
you order him to complete it, and you take for your part one. Then you
order him to multiply his result by 3, and to subtract its half. When he
430
The person must repeatedly divide the hidden number by 2, while, at the same
time, the asker doubles the unit for each division performed. If the person comes upon
a fraction, he will disregard it, but must declare its presence; the asker then keeps in
mind the number he has already attained. Adding up at the end the number obtained
by the doublings and the numbers kept in mind, he will obtain the required number.
(a) Let us begin with the simple case where x is a pure power of 2, say 2n . The
first division leaves 2n−1 , and the asker obtains 2; the second leaves 2n−2 , and the
asker obtains 22 . Continuing the procedure to the end, thus until the nth division,
the remainder will be 1, while the asker will have obtained 2n . Since the next division
gives 12 , the procedure must stop. (In the particular case when x = 1, the procedure
already stops before the first division, and the number will be the initial number, thus
1.)
(b) If now x is not a pure power of 2, it may be represented as

n
x= ak 2k with ak = 0 or 1,
0
since any natural number may be represented as a sum of powers of 2 in only one way.
The procedure described in fact reconstitutes this representation, the terms in question
being then those having the coefficient 1.
Let, for example, x be the sum of three terms, say x = 2α +2α+β +2α+β+γ (β, γ = 0).
The αth division by 2 gives 1 + 2β + 2β+γ , and, with 12 occurring in the next division,
the asker will retain 2α since such is the number he has reached; the (α + β)th halving
gives 1+2γ , and thus for the asker the quantity retained 2α+β ; finally, the (α+β +γ)th
division gives 1, and the final number of the asker will be 2α+β+γ . Therefore, the sum
of the quantities retained and the final number will be 2α + 2α+β + 2α+β+γ . The
interpolation is by a reader who thought that the number retained was the one next
reached, after noting the presence of a fraction.
431
Thus he is to retain its 23 .
432
All this is a very strange attempt to generalize the preceding case.
has done that, you ask him whether there is in the half a fraction or not. If
it does, you order him to complete it, and you take for your part 2, which
you add to the first unit (if any433 ). Next you order him to eliminate his
result by 9s, and you take for your part 4 for each 9 he subtracts. When
he says that what remains is less than 9, reject it without taking it into
account. You will (then) add the (number of) times there has been a
subtraction to the units you have taken when fractions occurred during
halving. If there was none, you will take (only) the fours taken when the
nines were subtracted. The result will be precisely what he had taken (in
mind). Understand that and think about it. If there was a fraction in
the second halving but none in the first, you will take 2 for the second
and add it to the (number of) times there has been subtraction.434
Other kind of this
(A.III.24) You order the person to add 2 to what he holds, to double
the result, to add 5 to it, to multiply it by 5, then by 10, and to subtract
from it 450. For each 100 in the remainder you will take one. The result
will be what the person has held.435
Example of this. The person holds 4. He adds 2 to it, which gives 6;
he doubles it, which gives 12; he adds 5 to it, which gives 17; he multiplies
it by 5, then (the result) by 10, which gives 850; he subtracts 450 from
it, which leaves 400. You take one for each 100; it gives 4, which is what
he has held.
433
This is clarified in the last sentence (which could be an early reader’s gloss).
434
The person must multiply his number twice by 32 , and at the end declare the
number of subtractions of 9, for each of which the asker will count 4. If there is in the
first multiplication a fraction, it will be rounded up, and the asker will add 1 to his
result. If there is a fraction in the second multiplication, the same will be done, but
the number added will be 2. K has the same problem.
The origin of this is the representation of x as a multiple of 4. If x = 4k, there will
be no fraction and x will be transformed into 9k; 9 being subtracted k times, k will be
multiplied by 4. If x = 4k + 1, we shall obtain first 6k + 23 , which becomes 6k + 2, and
1 is retained, then we shall have 9k + 3, from which we subtract 9. If x = 4k + 2, we
find 6k + 3 then 9k + 4 + 12 , and we shall add 2 to 4k since the fraction occurred in the
second step. Finally, if x = 4k + 3, we shall have to add 3 to the result since there was
a fraction in each step: the first multiplication gives 6k + 4 + 12 , rounded up to 6k + 5,
and the second, 9k + 7 + 21 .
This is just another form of A.III.15: what is counted there at the end is counted
here during the reckoning.
435
We are thus calculating

10 5 2(x + 2) + 5 − 450 = 100x + 450 − 450,

so that the number of hundreds is indeed the required quantity. We have discussed
this case in A.III.14.
Determination of two hidden numbers 436

(A.III.25) If you wish to determine this, order the person to, succes-
sively, add these (two numbers) and announce their sum, raise their
sum to the square, multiply the sum of the two numbers by the greater
number, multiply the lesser by a number lower than the sum of the two
numbers by a quantity of your choice, add the two products resulting
(from the multiplications) of the greater by the sum and of the lesser by
(the number) which is lower than their sum by a quantity of your choice.
You then subtract this from the square of their sum, next divide the re-
mainder by the difference between the sum and the number lower than
it [ that one which you have multiplied by the lesser number ]437 . The
result will be the lesser number; subtracting it from the sum will leave
the greater number.438
Example of this. The person hides for instance 6 and 7. When you
add them, their sum will be 13, the square of which is 169. You multiply
the greater, thus 7, by the sum, thus 13; this gives 91. Then you multiply
the lesser, thus 6, by some number of your choice less than it [ that is, less
than the sum ]439 , as if you multiply it by 10; this gives 60. You add 60
and 91, which gives 151; you subtract it from 169, which leaves 18. You
divide this by 3, which is the difference between 13 and 10; this gives 6,
which is the lesser number. When you subtract it from the sum, thus 13,
there remains 7, which is the greater number.440
436
The person (not always distinguished from the asker!) must carry out various
operations with these two numbers.
437
This must be a gloss: there is no need for clarification and the ‘you’ is inappropriate
here.
438
Let the two numbers be x and y, the author supposing x > y (though this is not
necessary here; K, who has this same problem, calls the unknowns ‘first’ and ‘second’,
but without any implication of relative size). The person must calculate for himself
the two quantities (x + y)2 and x(x + y). He has to declare —now or at the outset—
the value of x + y (our text fails to mention that, but it is found in K, with the value
declared at the beginning, as in our addition): knowing this sum is necessary for the
asker to be able to fix the value of a mentioned hereafter. Indeed, the asker tells him
to multiply y by a number a < x + y and to calculate the expression

(x + y)2 − ay + x(x + y)
,
x+y−a
equivalent to
(x + y)(x + y − x) − ay (x + y)y − ay
= = y,
x+y−a x+y−a
which will enable the asker to know x too since x + y has been declared. It appears
from the formula why a must be less than x + y.
439
Gloss which the ambiguity of the text renders necessary. Same in A.III.25 .
440
K has an example with x = 4, y = 5, a = 3.
(A.III.25 ) You may order the person to add the two numbers, to mul-
tiply their sum by 4, to announce the result, then to multiply it [ that is,
the sum of the two hidden numbers, thus 13 ]441 by 6, and to announce
again the result. Next, you subtract the first of the two results from the
second and you keep the result in mind. Then you subtract the four from
the six, which leaves 2. Then you divide what you have kept in mind
by 2. The result will be the sum (of the two numbers). One of the two
numbers will then come out as before. You subtract it from the sum; this
leaves the greater number.442
(A.III.26) You may add the two numbers then multiply the sum by one
of the two numbers and keep the result in mind.443 Then you multiply
the other number by the sum less one. You add the two products to the
sum of the two numbers, and you take the (lesser) integral root closest to
the result; indeed, it will always be equal to the sum of the two numbers.
From the whole, you subtract the square of the sum of the two numbers;
the remainder from the whole will equal one of the two numbers.444 Then
you subtract the number (just) remaining from the sum; this will leave
the other number [ the remainder of the number will equal the number
which had been multiplied by the sum ].445
(A.III.27) You may order the person to add the two numbers, to raise
the sum to the square, and to announce the result; then, to raise each
one to the square, to add them and to announce again the result; then,
to subtract the sum of the two squares from the square of the sum. You
take half the remainder; the result will equal the product of one of the
two numbers by the other. Next, you tell him: ‘Multiply the product of
the two numbers by twice the other’446 , or thrice it, or any multiple you
441
Typical interpolation, not separating general instructions from numerical example.
442
Banal modification of the previous problem: instead of letting the person declare
the sum x + y, the asker will tell him to declare successively the results of 4(x + y)
and 6(x + y), of which he himself will divide the difference by 2.
443
Here there is not even any mention of a ‘person’.
444
The subsequent gloss specifies which one.
445
We calculate x(x + y), then y(x + y − 1), and add their sum to x + y. We then
take the integral root by default of the whole (which thus presupposes knowledge of
some integral squares). Since the radicand of

(x + y) + x(x + y) + y(x + y − 1)
reduces to x + y + x + xy + yx + y 2 − y = (x + y)2 + x, which is less than the integral
2
square immediately above (x + y)2 , namely (x + y + 1)2 = (x + y)2 + 2x + 2y + 1, the

integral root by default will equal the sum x + y. Raising this sum to the square and
subtracting the result from the radicand will give x, then y by subtracting x from the
known sum.
446
Read: ‘by twice one of them’.
wish447 , and the person will announce the result. Then you divide the
result by the multiple —by which you have ordered him to multiply one
of the two numbers— (multiplied) by the product of the two numbers;
the result will be the number multiplied.448
(A.III.27 ) You may order the person to multiply the product by half
one of the two numbers, or a third of it, or a fourth of it, or some other
(aliquot) part of it. He will announce the result. You then divide the
result by the product; this gives the (other) part of the number. You
multiply it by the denominator; this gives the number multiplied.449
(A.III.28)450 You may order the person to hold a number equal to it,
then to add to the second what he wishes of the first, then to add to the
remainder of the first (something taken) from the second equal to what
there is in it, and to multiply the remainder of the second by the result
in the first [ add it to it ] [ equal to what remains of it ]451 and to divide
this by what there is in the second. [ The quotient will then always be
equal to what there is in the first. ]452 Then he adds to the result of the
division the remainder of the second, and he announces the result. It will
always be twice the hidden number.453
Example of this. The person keeps hidden 5. So, when he takes a
second number equal to it, it will be 5. Then, when he adds to this
447
The ‘you wish’ refers to the asker.
448
The person is asked to compute (x + y)2 then x2 + y 2 and to declare half their
difference (the text says to declare them individually). Since
(x + y)2 − (x2 + y 2 )
= xy,
2
the asker knows the product of the two unknowns. He then asks for one of the two
numbers, say x, to be multiplied by some factor k he has chosen (here integral, see
next problem) then the result by the calculated product xy and for this last result, thus
kx2 y, to be declared. Since the asker already knows xy and k, he will divide kx2 y by
their product; he will thus find x (the ‘number multiplied’), whence y since he knows
xy.
449
Same, with as multiplier a fraction k1 and x divisible by k (thus x = k · t). The
quantity calculated will be k1 x · xy = t · xy, and dividing it by xy will give t, whence
k · t = x and y.
450
Although this problem belongs to the section on the determination of two numbers,
a single one is sought; our author has either misplaced or misunderstood the problem.
K, who has the same problem, indeed mentions that just one hidden number is required.
451
The first gloss refers to the previous addition (adding ‘what there is in it’), and the
second gloss clarifies the ‘it’ of the first gloss (the two ‘it’ are equal quantities, namely
our x − m below).
452
This must be an early reader’s observation: it is not part of the determination.
453
The person is told to represent twice the hidden number, say x. He must first
transfer a quantity m (m < x) from the one to the other. He obtains successively
(latter 5) what he wishes from the first five, for instance adding to it 3, 2
will remain from the first and the second will become 8. If he adds from
the second, where there is 8, to the first, where there is 2, as much as it
contains, the first will become 4 and the second, 6. When he multiplies
6 by 4, it gives 24. When he divides it by what there is in the second [ of
the two ], thus 6, this gives 4. If the person adds it to the second, thus 6,
it gives 10, double of the five which was the hidden number.454
(A.III.29) You may order the person to hold in his right hand what he
wishes and in his left what he wishes.455 Then he adds 2 to what he
has in his right hand, doubles that and adds 5 to the result. Then he
multiplies this by 5, and adds to it what he has in his left hand. Then
he multiplies the result by 10, subtract from this 450, and announces the
remainder. You then take, for each 100, one; what results is what he had
in his right hand. Next you take, for each 10 in the remainder, 1; what
results is what he had in his left hand.456
(A.III.30) To this belongs also (what follows).457 You order the person
to hold an even and an odd. You order him (successively) to raise each
to the square and to add the results; to double the odd, to add one to it,
and to multiply the result by the even; to add the whole and to take the
nearest (lesser) integral root of it; indeed, the root will always be equal to
the (sum of the) two numbers. After that, the person subtracts from the
(previous) sum the square (of the approximate root); the remainder will
always be the even. He then subtracts it from the (approximate) root;
the odd will remain.458
x x
x−m x+m
2(x − m) 2m.
The product of the two results is then divided by the second, which is itself added to
the quotient. The result is declared. Now it is twice the required number since
2m · 2(x − m)
+ 2m = 2x.
2m
454
K has an example with x = 4, m = 3.
455
It may be two numbers signified by the fingers (note 428) —or objects of which
we are to determine the quantities.
456
With x in the left hand and y in the right hand, the sequence of operations leads
to
10 5 2(y + 2) + 5 + x − 450 = 100y + 10x.

It is a variant of our case 3 (note 410), except that here x is not limited to the pair 1,
2 but may be any integer less than 10.
457
The three subsequent problems also involve two unknowns.
458
Let the two unknown numbers be x = 2m and y = 2n + 1. The person computes
(A.III.31) You may order the person to multiply one of the two numbers
by their sum, then to raise this number to the square and to announce the
result. Then, he subtracts from the product the square. There will then
remain the product of one of the two numbers by the other. You divide
it by the root of the square number; it will give the other number.459
(A.III.32) You may order the person to add the two numbers and to
announce the square of their sum; you take yourself always a fourth of it.
Then you order him to multiply one of the two numbers by the other and
to announce the amount of the product. You subtract this from a fourth
of the square of the sum of the two numbers and you take the root of the
remainder. You subtract the result from half the root of the square (of
the sum) of the two numbers. The remainder will be the lesser number.
You subtract it from the root of the larger square;460 there will remain
the greater number.461

(2m)2 + (2n + 1)2 + 2m 2(2n + 1) + 1 ,
and declares the result (the text fails to mention this last point). Since this expression
reduces to
2 2
2m + (2n + 1) + 2m < 2m + (2n + 1) + 1 ,
the integral root by default (taken by the person or the asker) will give, as in A.III.26,
the sum of the two numbers. Raising it to the square and subtracting the result from
the first sum leaves 2m = x; the sum of the two numbers being known, we can find
2n + 1 = y. But the condition of different parity for the two required numbers is
irrelevant; for any x, y, we have

x + y < x2 + y 2 + x(2y + 1) < x + y + 1,
and we may thus compute x2 + y 2 + x(2y + 1) − (x + y)2 = x.
459
The person computes x(x + y) and x2 , and declares the latter result (so x will be
known), then he divides their difference by x. We shall obtain y since
x(x + y) − x2
= y.
x
This is an application of the identity in Elements II.3 (A.I.126 in this text).
460
The asker knows two squares, the lesser of which is a fourth of the other.
461
Let the two required numbers be x and y, with x > y. The person must compute
the two quantities (x + y)2 and xy, and declare them. The asker can then compute

(x + y)2 x−y
− xy = ,
4 2
whereby he will find, since he also knows x + y,
x+y x−y
− = y, then x.
2 2
The occurrence of these two identities is not surprising. We find them in use since
Mesopotamian times to solve quadratic systems (and equations) as well as in Greek
and mediaeval elementary algebra; see our Introduction to the History of Algebra, pp.
11–16, 19–23, 130 (pp. 11–16, 19–23, 135 in the original French edition). The first
identity is Elements II.5 (A.I.128 in this text).
Determination of three numbers

(A.III.33)462 Here the person holds three equal numbers. Then he adds
to the second something known from the first and something known from
the third. Then he adds to the first a quantity equal to what remains
in it, taken from the median; there will then remain in the median, al-
together, (a quantity) equal to what he has added to it from the third
plus twice what he has added to it from the first. He subtracts, from the
median again, what he had added to it from the third, and he halves the
remainder. This gives what he had added to it from the first.463
Determination of hidden names 464
(A.III.34) A name will necessarily consist of two letters, of three letters,

or more.465 If it consists of more than three letters, determine to begin
with the first two letters, then the two or three remaining letters [ then
three or two letters, then three ].466
(A.III.35) Here is the explanation of the numerical value of the Arabic

letters, taken in alphabetical order,467 for this section (Fig. a 43).468
462
No mention of an asker here.
463
This recalls A.III.28 (see below). Starting with three equal numbers, say x each,
we modify them by means of two quantities taken from the first and the third, say m1
and m3 (m1 , m3 < x), as follows
x x x
x − m1 x + m1 + m3 x − m3
2(x − m1 ) 2m1 + m3 x − m3
2m1
This determines m1 (x and m3 may be determined using the results found last in the
other columns).
K has the same problem and an example (x = 4, m1 = 3, m3 = 2). This problem
is followed there by (our) A.III.28, and, in solving it, K refers to the former. This
confirms their correlation.
464
The asker must find a name, known to the person, by determining its individual
letters through their numerical values.
465
In Arabic non-accentuated vowels do not appear as letters.
466
It is assumed that we now know how to determine two or three hidden numbers.
So we shall break up the number into groups of two or three letters to be determined
each in turn by one or other of the methods taught from A.III.25 on.
467
And not the order of the usual alphabetical numerical system (p. 8).
468
K has an application of a similar table (determining letters by determining num-
bers) and maybe there was a similar application in our text (the ‘other kind’ of the
next heading suggests a lacuna here). But see next note.

a b t th j h. kh d dh r z s sh s. d. t.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

z. ‘ gh f q k l m n h w la y
17 18 19 20 21 22 23 24 25 26 27 28 29
Fig. a 43
Other kind of this

(A.III.36) Let a circle be drawn, on which are found the Arabic letters
appearing in this figure (Fig. a 44).469
y a b
la t
w th
h j
n
.
h
m kh
l d
k dh
q r
f z
gh
‘ s
z. sh
t. d. s.
Fig. a 44
Then you order the person to count on the circle from a to the first of
the name’s letters. Then he doubles this, and adds to it an even number,
which you choose. Then you order him to halve this and to count from
one of the circle’s letters, any one he chooses,470 up to the end of the
number remaining to him from the halving, and to announce the letter
to which this number has led him. Then you count the interval between
the letter where the person has started to count and the letter which he
has attained. The result is half the interval of letters from a to this letter
469
There are around the circle 29 signs, namely the 28 letters of the Arabic alphabet
and the ligature representing the two letters l and a combined, thus the same signs
as in the previous table. Fig. a44 is just an aid to counting the intervals between
letters, although this can be directly found from Fig. a43 of which this was perhaps
the purpose.
470
In fact, the asker must choose it (as in K) —or at least he must be informed of
the choice.
(first of the name), taken twice,471 plus half the even number you ordered
him to add. Subtract from it half the even number added. There will
remain the number from the letter a to the letter to which the person has
counted;472 you count then this same number from a, and you will reach
this letter. Using this same means you will determine all other letters of
the name.473
Example of this. You wish to know the first letter of the (hidden)
name Muh.ammad. You order the person to count from a to m. You find
24.474 You double it, this gives 48. Next you add to it, say, 2. This gives
50. Then you halve it; this gives 25. You count these twenty-five from the
letter b, for instance;475 the counting leads to the letter w. You subtract
from it 1, which is half the even number you had added; the remainder
is 24. You count them from the letter a. The counting leads you to the
letter m. Then you know that its first letter is m.
This method is somewhat lengthy. We know another way, less de-
manding than this one, which we are going to explain.476
(A.III.37) You begin in that by ordering the person to announce the
number of letters in the name. If it consists of three letters, you order
him to add the second and third letters and to announce the sum of
this, which you call the ‘first’; then to add the first and the third and
to announce the sum of this, which you call the ‘second’; then to add
471
The ‘twice’ corrects the previous error (‘half’). Same oddity in K.
472
This refers to the person’s first counting.
473
The asker tells the person to count the distance from a to the first letter of the
name, say distance δ, to double it and to add to it an even number which he proposes,
say 2α. The quantity kept in mind by the person will then be 2δ + 2α. The asker tells
him to take its half and to declare, starting with a letter he imposes, what letter he
reaches when counting this quantity. The asker will now know two letters distant by
δ + α; since he knows α, he can find δ, which will give him the distance between a and
the first letter of the name.
Note that any letter other than a may be the starting point, and the order in which
the letters are arranged around the circle (or along a line) may be fixed arbitrarily.
Thus, in his similar problem, K uses one letter less (the ligature la) and, on his circle,
a different order; accordingly, the values found in his numerical example are different
from those here below, although the required letter is also m.
474
Counting the starting letter. Result also readily inferred from Fig. a43. From
that moment, the asker (according to the verbal forms) is at work —playing thus both
rôles; accordingly, he has just obtained the number he was to seek.
475
This time without including the starting letter. Result also readily inferred from
Fig. a43. This discrepancy between the starting points suggests the use of Fig. a43.
476
In all subsequent cases each letter is given its numerical value according to the
usual alphabetical numerical system (see Fig. 5, p. 8).
the first and the second and to announce the sum of this, which you call
the ‘third’. Next you add the three sums and take half the result. You
subtract from it the first sum; it leaves the first letter. Then you subtract
from it the second sum; it leaves the second letter. Then you subtract
from it the third sum; it leaves the third letter.477
(A.III.37 ) If the name consists of four letters, you add the three letters
(which are) the second, the third, the fourth, omitting the first, and you
call the result ‘first’, according to the designation of the letter omitted.
Then, likewise, you add the first, the third, the fourth, omitting the sec-
ond, and you call the result ‘second’, here too according to the designation
of the letter omitted. Then, likewise, you add the first, the second, the
fourth, omitting the third, and you call the result ‘third’, according to
the designation of the letter omitted. Then you add the first, the second,
the third, and you call the result ‘fourth’, according to the designation of
the (letter) omitted. Then you take a third of the whole sum. Then you
subtract from it the ‘first’; it leaves the first letter. Then you subtract
the ‘second’; it leaves the second letter. Then you subtract the ‘third’;
it leaves the third letter. Then you subtract the ‘fourth’, it leaves the
fourth letter.
(A.III.37 ) You proceed likewise if the name consists of five letters; you
will take a fourth of the sum. If the name consists of six letters, you will
take its fifth [ that is, a fifth of the sum ]. Likewise, you will always take
the ratio which is that (of 1) to the whole quantity of letters less one. The
reason for that is the following. You add each time the quantity of letters
of the whole (name) less one; it is therefore necessary to take the ratio
of one to this number, (the amount of this ratio being) taken from their
sum. Then you subtract from it the sums; there will remain precisely the
letters omitted.478
477
Let the three letters be x1 , x2 , x3 . The person declares three sums of two letters,
each time mentioning the missing letter; with our symbolism, this gives
x 2 + x 3 = s1
x 1 + x 3 = s2
x 1 + x 2 = s3 .
The asker then calculates 12 (s1 + s2 + s3 ); since this equals x1 + x2 + x3 = S, he knows
the sum of the three letters and will determine each individually by subtracting from
S each sum of corresponding index. See subsequent example, in A.III.37a.
478
This is explained very concisely. Let it be a name with n letters xi , and S = xk .
The person will calculate, and successively declare, the n sums

sj = xk .
k=j

The asker then computes sk which, since each xj is omitted once, equals (n − 1)S.
Other kind of this479

(A.III.38) You order the person to raise the first and the second to
the square. You keep the sum in mind. [ Then he adds the first and
the second. ]480 Then he doubles the second and adds 1 to the result. He
multiplies this by the first letter, and adds this to the sum of their squares.
You take its nearest (lesser) integral root; this will always be equal to the
sum of the first two letters. Next you subtract the square of this result
from the sum;481 there will always remain the first letter. You subtract
it then from the (above approximate) root; there will always remain the
second letter.482
Other kind of this
(A.III.39) The person adds the first and second letters and always mul-
tiplies the sum by 3; you keep in mind the result. Next, he multiplies the
first by 4, the second by 3, and adds the whole. You subtract from it the
product of the sum by 3 which you had kept in mind. The remainder is
the first letter; the complement to the sum will be the second letter.483
Examples of this
(A.III.37a) First (method). For instance, you wish to determine Zayd.484
You add the second and the third, which gives 14; it is the ‘first sum’.
He is then able to find one by one each letter by means of
1
x j = S − sj = sk − sj .
n−1
Solving such linear systems by adding the equations has been known since antiquity,
and will be widely used in the Middle Ages by the Pisan Leonardo Fibonacci (and by all
who took inspiration from his works, thus the majority of mediaeval mathematicians);
see our Introduction to the History of Algebra, pp. 26–27, 103–117 (pp. 26–27, 106–122
in the original French edition).
479
A.III.38–A.III.39 teach another way to determine the letters, but by pairs (we
may thus be left with finding a single hidden number).
480
This gives the result we are now to calculate. Same instruction, though, in the
example (A.III.38a).
481
This is the sum of which the root has been taken.
482
Let x, y be two letters of a noun (the ‘first’ and the ‘second’ in the text). The
asker tells the person to compute x2 + y 2 and x(2y + 1), and to declare their sum. He
then computes its nearest integral root. Now since

(x + y)2 < x2 + y 2 + x(2y + 1) = (x + y)2 + x < (x + y + 1)2 ,
x + y will be known. By subtracting its square from the above radicand (declared), the
asker will know x then y. See example in A.III.38a; this method has in fact already
been used (A.III.30).
483
Few will be fooled by that! The person computes, and declares, 3(x + y) and
4x + 3y; from their difference the asker will know x and, from the first result declared,
x + y, whence y. Example in A.III.39a.
484
z, y, d = 7, 10, 4 according to the numerical values of the letters. In all three
Then you add the first and the third, which gives 11; it is the ‘second
sum’. Then you add the first and the second, which gives 17; it is the
‘third sum’. Next you add the sums, which gives 42. You take half of this,
which is 21. You subtract from it the ‘first sum’; there remains 7, which
is the first letter. Then you subtract the ‘second sum’; there remains
10, which is the second letter. Then you subtract the ‘third sum’; there
remains 4, which is the third letter.485
(A.III.38a) Example of the second method, again for Zayd.486 You raise
the first two letters to the square; their sum will be one hundred and forty-
nine. [ Next you add the first and the second, which gives seventeen. ] You
double the second, which gives twenty. You add one to it, which makes
21. Then you multiply it by 7, which gives 147. You add it to the sum
of their two squares, thus to 149, which makes 296. You take the nearest
(lesser) integral square root, which is 17.487 Then you subtract the square
of 17, thus 289, from the sum, thus (from) 296. The remainder is 7, which
is the first letter. Then you subtract this from the sum of the two letters;
the remainder is 10, which is the second letter.
(A.III.39a) Example of the third method, again for Zayd. You add the
first two letters, which gives 17. You multiply it by 3, which gives 51,
which you keep in mind. Then you multiply the first, thus 7, by 4, and
the second by 3, and you add the whole. This gives 58. You subtract
from it fifty-one; this leaves 7, which is the first letter. Then you subtract
this from the sum; this leaves 10, which is the second letter.
Do the same for whichever determination you wish of numbers and
names, and you will succeed, with the help of God Most High.488
Here are examples of tables for
the determination of hidden numbers
(A.III.40) The first of them,489 a stairway-like figure, enables you to
determine hidden numbers from 1 to any desired (number), according to
examples the ‘person’ is, once again, forgotten.

485
Incidentally, we would obtain the same result with a different set of values at-
tributed to the letters, such as with that in Fig. a43 (z, y, d = 11, 29, 8).
486
Or rather, in this and the next example, for determining its first two letters.
487
It should be mentioned here that we have thus found the sum of the two letters
(this omission may explain the presence of the gloss; but see note 480).
488
Note the allusion, right at the end, of the possibility of thus also determining
hidden numbers. We now return to hidden numbers (and a ‘person’ reappears).
489
There are only two, though the Arabic has the plural (instead of dual). Same for
‘examples of tables’ in the heading.
that illustration (Fig. a 45).490
1 2 4 8 16 32 64 128
1 1 5 5 21 21 85
2 14 14 14 14 14
4 4 20 52 52
8 24 24 120
16 16 80
32 96
64
Fig. a 45
If you wish to operate with this table (you will do what follows).
Order the person to hold a number less than any of the numbers in the
first line —and not to hold a higher number.491 Then the person adds
to the number its half, and you ask him whether there is in it a fraction
or not. If there is, you order him to double the fraction and you yourself
hold the first number belonging to the first of the rows of this number,
among the odd (numbers).492 Then you order him to add to it again its
half, and you ask him whether there is in it a fraction or not. If there
is, you order him to double the fraction and you yourself hold the second
number. Then you order him to add to it its half, and you ask him
whether there is in it a fraction. If there is, you order him to double the
fraction and you yourself hold the third number. You continue invariably
to proceed like that until you reach the last number of this (vertical)
row. Then you subtract from the sum of what you have held in your
hands the first number, written at the top, in the line of the evenly even
(numbers).493 The remainder will be the number the person had kept
hidden. When, on the occasion of one or the other announcement, there
is no fraction, you will not hold the number in the column in question
and omit it, whichever this number. If there has been no fraction at all,
you will know that the person has held the first number of the (vertical)
490
The table ends with 128, but could indeed be continued (see below).
491
To mean, since the top row contains successive powers of 2: he will hold x with 2t <
x ≤ 2t+1 . The quantities in the column of the greater are used for the determination.
492
The second line contains odd numbers, and it is the only one.
493
The subtraction will be carried out —if need be, repeated— if the quantity ob-
tained is greater than the head of the column.
row in question.494
Example of that. The person keeps hidden 13, belonging to the verti-
cal (row) of sixteen. When he adds to it its half, thus six and a half, the
sum will be 19 and a half. There is in it a fraction. You yourself hold,
because of the fraction, the first number, among the odd ones, thus 5,
and you order him to complete the fraction; there results 1, which gives
him 20. Then you order him to add to twenty its half, which makes 30.
There is no fraction in it; you thus omit the second number. Then you
order him to add to 30 its half. The result is 45. There is no fraction
in it; you thus omit the third number. Then you order him to add to
45 its half. There results 67 and a half. There is a fraction in it; you
yourself then hold the last number, thus 8. So you will have (altogether)
13, and there are no more numbers in the (vertical) row considered. At
that point, you will know that the person had held 13.
Operate likewise, just as explained above, for all other (numbers),
small or great, and you will succeed, if such be the will of God Most
High.495
(A.III.41) This is another table enabling you to determine the hidden
494
The asker tells the person to choose a number and to declare the column (see note
491). The person multiplies the number by 32 ; if a fraction occurs, he rounds it off to 1
and declares its presence to the asker, who retains the first number in the column. The
operation is repeated with the results obtained, and at any occurrence of a fraction it
is rounded off and its presence declared, the asker retaining the corresponding number
in the column. When the end of the column is attained, the asker adds up the numbers
retained and subtracts from that sum the number at the head of the column as many
times as possible. The remaining number will be the number thought of. As said in
the text, no occurrence of a fraction means a pure power of 2.
495
Here is the author’s example and one we add:
13 + 6 + 12 = 19 + 12 5 125 + 62 + 12 = 187 + 12 85
20 + 10 = 30 188 + 94 = 282
30 + 15 = 45 282 + 141 = 423
45 + 22 + 12 = 67 + 12 8 423 + 211 + 12 = 634 + 12 120
635 + 317 + 12 = 952 + 12 80
953 + 476 + 12 = 1429 + 12 96
1430 + 715 = 2145.
The sum of what is retained in the first case indeed gives 13, and in the second case
381, which leaves 125 after subtracting twice the number at the head of the column.
To extend the table, we put at the head the next higher power of 2 and below the
previous power of 2 (here 256 and 128). We then apply the above algorithm to the
bottom number increased by each of the lesser powers of 2, taken in decreasing order
(thus here 128 + 64, 128 + 32, 128 + 16, up to 128 + 1). With this we shall be able to
fill successively, from the bottom, the cells still empty (here with, respectively, 64, 224,
80, 120, 180, 142, 85).
numbers, odd and even ones (Fig. a 46). You operate in it by ordering the
person to eliminate the hidden number by means of, according to your
choice, odd or even (numbers) having their amounts (written) in the top
row.
er
n
sev
ht
th
rgi
r
tw
e
mb
ten
six
fou
fiv
eig
ree
en
o
ma
nu
1 1 1 1 1 1 1 1 1 1
2 2 2 2 2 2 2 2 2
3 3 3 3 1 3 3 3 3
4 1 4 4 4 4 4 4
5 2 5 5 1 1 5 5 5
6 1 6 6 2 6 6
7 1 2 7 1 3 1 7 7
8 2 3 1 8 2 8
9 4 2 9 1 1 3 1 9
10 1 3 10 2 4 2
11 2 1 4 11 1 3 5 3 1
12 2 5 12 4 2
13 1 3 6 13 1 1 1 5 3
14 2 4 14 2 2 6 4
15 1 15 1 3 3 7 5
16 1 1 2 16 4 6
17 2 2 3 17 1 1 5 1 7
Fig. a 46
For the odd ones, it will be by 3s, then 5s, then 7s, and the person
will announce each time the (final) remainders. When he has announced
them, you look at the cells where there are (numbers) identical to the
remainders which have been announced [ as being identical ]; then the
number [ hidden placed there ] in the margin [ the place at its beginning ]
[ the row of the number ] equals what he has hidden.496
496
The first gloss fills a short lacuna. Of the last three glosses, the third clarifies
the second, which itself clarifies the first. In the last, ‘of the number’ refers to the
subsequent example, where there is just one number (that is, one remainder = 0).
If the order given by you to him was to eliminate it by means of even

(numbers), order him to eliminate it by 2s, then by 4s, then by 6s, then
by 8s, then by 10s, and to announce, each time it is eliminated, what
remains. Then you look at the cells where there are (numbers) identical
to the remainders which have been announced; the hidden number will
be that inscribed in the margin [ in the row of this number ]. This is the
figure of the table.497
Example of this. The person hides 15. You order him to eliminate it
by means of odd ones, using (first) 3s; there remains nothing. Then you
order him to eliminate it by means of 5s; there remains nothing either.
Then you order him to eliminate it by means of 7s; there remains 1. Then
you look at the cells with identical results, namely no remainder for 3,
nor for 5, and one for 7. You find, facing it,498 in the margin, 15. You
will tell him: ‘You have hidden 15’.
Proceed likewise with the even ones.
With this treatment you may increase at will the number of tables,
and extend the quantity of subtractions, after finishing (the study) of
what we have described, with the help of God Most High.499
Having now explained sufficiently that science, let us put an end to

our book; indeed, the essential of it has been completed, in compliance
with its uses and for any purpose.
Praise be to our Creator and our Educator, for He is worthy of it
and deserves it because of His preeminence and His uncontested predom-
inance.
497
The column headed ‘number’ just repeats (for the even divisors) the column with
‘margin’.
498
Reference to the remainder 1 since the other two cells are empty. Since 15 is the
only number here divisible by 3 and 5, we could have dropped the third step.
499
Only one manuscript page (copy, and probably original) was devoted to the above
table. But such a table could indeed be extended, and easily since the same sequences of
numbers are reproduced in each column. But the table should stop with 105 on the left
(‘margin’), with 120 in the middle (‘number’): after these two numbers, which are the
lowest common multiples of the divisors considered, the same sequences of remainders
recur, and in the same order (unless what the author means is that the list of odd and
even numbers could be extended, thereby increasing their common multiples).
Treatise by Abū’l-Wafā’ al-Būzjānı̄
on magic arrangement in squares
(Extracts)

(B.1)500 You are to know that a magic square consists of numbers, differ-
ent and in arithmetical progression, placed in its cells in such a way that
their sum counted (in the rows) in any manner —vertically, horizontally
and diagonally— produces the same quantity.501
For example: with magic arrangement in the cells of the square of 3,
comprising nine cells, each row, vertically, horizontally or diagonally, will
produce the same quantity. Here are four illustrations. The first (Fig.
b 1) contains the consecutive numbers taken in natural order beginning
with 1; each of the rows, vertically, horizontally and diagonally, makes
15. The second (Fig. b 2) contains the odd numbers beginning with 5 and
ending with 21. The third (Fig. b 3) contains the even numbers beginning
with 8 and ending with 24. The fourth (Fig. b 4) contains the numbers
beginning with 3, (following each other) with the constant difference 5
and ending with 43.502
2 9 4 7 21 11 10 24 14 8 43 18
7 5 3 17 13 9 20 16 12 33 23 13
6 1 8 15 5 19 18 8 22 28 3 38
Fig. b 1 Fig. b 2 Fig. b 3 Fig. b 4

(Edition, p. 141)
(B.2) Useful preliminaries.
(i) You are to know that magic arrangement in even and odd squares is
of two kinds. In the one the numbers are arranged in the square in such
a way that the quantities in each of the odd and even squares containing
one another make the same sums. Thus, the square of 9 encloses the
square of 7, which (itself) encloses the square of 5, which (itself) encloses
the square of 3; if then in each of these squares considered alone the
quantities (obtained) by adding vertically, horizontally or diagonally are
the same, we shall say that this arrangement is ‘regular’.503 But if we
500
Remember that these fragments do not always follow the order in the original
text, for they are arranged here by subject, as in our commentary. That is: general
introduction (B.1–B.2); construction of ordinary magic squares of odd (B.3–B.6) then
even orders (B.7–B.11); construction of bordered squares, namely, after an introduction
(B.12–B.14), odd-order ones (B.15–B.17) then even-order ones (B.18–B.19); construc-
tion of bordered squares with separation by parity (B.20–B.23); composite squares,
first of odd order (B.24) then of evenly even order (B.25–B.26).
501
Note the explicit mention of natural numbers in (any) arithmetical progression.
Both the constructed (magic) square and the constant (magic) sum in the rows are
expressed in Arabic by ‘adad wafq.
502
All figures are missing in the (only) manuscript preserved. There is, however, no
doubt about the reconstruction of almost all of them.
503
Bordered magic squares.
Translation of Text B 209
arranged the numbers in such a way that those in the large(r) square
make the same sum but those in the small(er) square, inside the large
square, do not, we shall say that this arrangement is ‘not regular’.504
(ii) When we wish to place the numbers in the squares according to the
regular arrangement, it will be easy to equalize one border after the other
if we know the magic sum which is to appear in each of the rows.505
If (first) the square is odd and we wish to know the quantity which is
to appear in each of its rows, we take its median number and multiply it
by the side of the square of which we wish to know the magic sum. The
result will be the quantity which is to appear in each row of this square.
Thus, for the square of 9: if we wish to know the magic sum in the inside
square of 5, we multiply its median number, namely 41, by 5; the result
is 205, and we shall say that such is the quantity to appear in each row
of the square of 5. We shall do the same for the other odd squares.
If (second) the square is even, we multiply its equalizing number by
half the side of this square, and the result will be what we are looking
for.506 Thus, for the square of 10: if we wish to know the quantity which
is to appear in each row of the square of 6 inside the square of 10, we take
its equalizing number, namely 101, and multiply it by half of 6, which is
3; this gives 303, and such will be the quantity which is to appear in each
row of the square of 6.
(Edition, pp. 149–150)
504
Ordinary magic squares.
505
For a square of order n filled with the first n2 natural numbers, the magic sum
equals 21 n (n2 + 1) and if the square is bordered, 12 m (n2 + 1) for an inner square of
order m. The quantity n2 + 1 is called below the ‘equalizing number’ (defined in a
previous part in B, for us B.13iii), and n, the order, is simply the ‘side of the square’.
Finally, ‘to equalize’ a row, or a border, means setting up in it the magic sum. As in
the text A, s.aff can mean both ‘row’ and ‘border’.
506
This distinction amounts to dividing by 2 whichever of the two factors is even,
thus n2 + 1 if the order is odd and n if the order is even. In the case of odd orders, the
integer thus obtained is the ‘median’ number, that is, the middle term in the sequence
1, . . . , n2 .
210 Construction of ordinary magic squares
(Construction of ordinary magic squares)

(Odd-order squares)
(B.3)507 Magic arrangement in the square of 3.

(i) So we say that the first number of those we must begin to search for
the square of 3 is that which is to occupy its central cell, which does
not interfere with the four (outer) rows. Then, when we have found the
occupant of the central cell, we shall be left with eight cells, four in the
two diagonals and four in the two (medial) vertical and horizontal rows,
and we shall be left with searching for the occupants of a single border
surrounding the central cell.
So we say that there can be in the central cell only 5. Indeed, the
numbers which are to occupy the cells of this square begin with 1 and
end with 9. Then if we place in the consecutive cells of this square the
sequence of numbers taken in the natural order from 1 and put their
beginning in one of the corners, 5 will naturally appear in the central
cell. The quantity found in each diagonal will be 15, and such is the
quantity required for each row of this square.508
(ii) We must then only search for two other numbers to be placed, the one
in a corner and the other in some middle cell of the vertical and horizontal
rows; for, with the determination of these three numbers which do not
occupy one and the same vertical and horizontal row, it will be easy to
find what must be placed in the third cell in order to complete the row:
it will be the remainder to complete the quantity which is to appear in
each row, namely 15.509
So we say that since 5 appears in the central cell of the square, the
remainder to complete 15 is 10. Now 10 breaks down into two different
parts in four ways, twice into two odd, namely 9 and 1, 7 and 3, and
twice into two even, namely 2 and 8, 4 and 6. But the even parts must
necessarily occupy, of the remaining cells, the corner cells, and the odd
parts the middle cells, and it cannot possibly be otherwise. Admitting it
507
As said (note 500), our B.3 to B.11 are about the construction of ordinary magic
squares, first for odd orders (B.3–B.6), then for even ones (B.7–B.11).
508
Considering a square of any order n filled in with the n2 numbers placed in nat-
ural sequence (so-called natural square), the sum appearing in each of the two main
diagonals will be equal to the magic sum to be found in the magic square having the
same order and filled with the same n2 numbers.
509
The square of order 3 will be easily completed once we have determined what to
put in the central cell, any corner cell and a third one, either in the middle of a lateral
row or, also, in the next corner. This is repeated in B.5.
would not be impossible that one of the odd be in a corner, suppose 9

occupies it; therefore 1 will be in the (diagonally) opposite cell in order
to complete 15. We put in any empty cell any arbitrary number, even or
odd, and complete the other cells; from this will result that numbers are
repeated, or cells are left empty, without any number, or that numbers
exceed 9. Now all of that is impossible. From this it follows that the even
(numbers) will be in the corners and the odd (ones) in the middle cells.
(iii) Then we put 1 in one of the middle cells; so 9 will be facing it in
order to complete 15. We are to put in one of the (two) corners, next to
the cell containing 1, an even number.510 There are four, two large and
two small.511 But it is not possible to put one of the small ones next to 1,
for the remainder to complete 15 would be larger than 9. Thus we shall
place there 8 or 6, and complete the remaining cells.512
(B.4) With our explanations513 have appeared the methods to find how
to obtain the magic arrangement for the table of nine (cells).514 But it
is also possible to place these numbers in the square of 3 by means of
displacements, as follows.515
3 2 1 2 9 4
6 5 4 7 5 3
9 8 7 6 1 8
Fig. b 5 Fig. b 6
(i) We place the consecutive numbers, taken in the natural order from
1 on, in the successive cells of this square (Fig. b 5). Considering the
content of the two diagonals and of the two middle rows, we move (first)
the content of each corner towards the corresponding knight’s cell in a
same direction and move (then) in the opposite direction what was in
this last cell towards the next corner, that belonging to the row where
the move initiated.516 This will make the magic arrangement in this
510
According to ii.
511
Relative to the median element. Same designations in A (note 211).
512
Briefly: since with 1 the only possibility to reach 15 is with 6 and 8, 1 must be in
a middle cell with these two numbers as its neighbours.
513
Seen in B.3.
514
The ‘methods’ must refer to the steps in the previous reasoning.
515
Generally, the purpose of these ‘displacements’ is to obtain a magic square by
moving the elements in a natural square of the same order.
516
‘Opposite direction’: moving e.g. 1 along the right-hand column and then on to
the left, 8 will be moved to the right.
square (Fig. b 6). (. . . )517

(ii) When we place the consecutive numbers taken in the natural order
from 1 in this square, the content of each of the two diagonals is 15 and
that of the two middle vertical and horizontal rows is 15 (Fig. b 7). The
reason is that each of these rows contains three numbers in arithmetical
progression, that is, with equal differences, and it is known that a property
of such a progression is that (the sum of) its two extremes equals twice
the median.518 So we are left with equalizing the rows surrounding the
central cell.519 If we move the content of each cell of the border around
the central cell towards the next cell —so that 1 takes the place of 2,
2 that of 3, 3 that of 6, and so on with the others— the result will be
as in this illustration (Fig. b 8). It will then be easy for us to equalize
their content. Indeed, it so happens that the sum of what is in any pair
of corner cells of one row and in the middle cell of the opposite row is
15; indeed, one may verify that (the sum of) 4, 8, 3 is 15, and the same
for the others. Thus if we move the occupant of each middle cell of the
border towards the opposite cell, the rows will be equalized (Fig. b 9).
3 2 1 2 1 4 2 9 4
6 5 4 3 5 7 7 5 3
9 8 7 6 9 8 6 1 8
Fig. b 7 Fig. b 8 Fig. b 9

(B.5) A way without displacement is the principle for finding the magic
arrangement in other odd squares, and it is the following.520 We place
the median number in the central cell of the square. Then we place 1
in the middle of one of the rows surrounding the central cell. We place
2 in its knight’s cell and 3 in its queen’s cell, which is the knight’s cell
of 2. Generally, after filling three cells of this square one of which is the
central cell, the second, one corner, and the third, one middle (in a lateral
517
The text omitted here is our B.5.
518
Thus each of the four rows passing through the centre displays the required sum.
519
This equalizing of the border will now be effected in two steps, the first of which
is to exchange the rôle of the middle rows and the diagonals.
520
This is how a bordered square will later be constructed: filling the central cell and
then, one after the other, half the number of cells of the surrounding borders (the other
half being filled with the complements to n2 + 1); there is thus no use of the natural
square. For the square of order 3, though, the result will be just the same as before:
apart from rotations and inversions, there is only one possible arrangement with the
first nine numbers, as seen in B.3.
row), or with the (last) two belonging to a same (lateral) row, whichever
it is, we shall (be able to) complete the remaining cells. This will be done
either by subtracting from 15 what is in two cells of the same row and
placing the remainder in the third cell, or by subtracting from 10 what is
at one end of a row and placing the remainder at the other end.
(B.6) Other way for setting out the magic arrangement in the square of
5, by displacement.521
(i) It is possible for us to set out the magic arrangement in this square
(of order 5) by means of displacements and applying certain principles,
by starting from the consecutive numbers (written) in this square from 1
on in the natural order.522 The numbers in the two diagonals make then
the magic sum, as seen in this illustration (Fig. b 10).
(ii) Next, leaving all these numbers (of the diagonals) in place, (thus)
leaving their cells unchanged, we move the numbers belonging to the
inner square of 3, except those belonging to the two diagonals, towards
their respective bishop’s cells and conversely following a same direction —
or (also) following two different directions, each element of a pair moving
towards the other—523 so that it will be as in this illustration (Fig. b 11).
Next, we move what is in the remaining cells in the border towards the
opposite cells, in the order and without inversion, so that (the result) will
be according to this figure (Fig. b 12).
5 4 3 2 1 5 12 3 2 1 5 12 24 23 1
10 9 8 7 6 10 9 20 7 18 11 9 20 7 18
15 14 13 12 11 15 22 13 4 11 16 22 13 4 10
20 19 18 17 16 8 19 6 17 16 8 19 6 17 15
25 24 23 22 21 25 24 23 14 21 25 3 2 14 21

(iii) Other displacement method. It is (also) possible for us to displace
the numbers which are in the inner square (Fig. b13)524 along two parallel
521
‘Other way’: in the original text, construction of the bordered square of order 5
has just been taught (our B.15).
522
Thus the natural square (note 508). As we shall see, all the displacements are
performed between opposite (horizontal and vertical) rows.
523
This looks like an early reader’s remark which later became incorporated into the
text (the ‘same direction’ mentioned alluded, as in B.4i, to that of the rotation).
524
Thus the four numbers previously moved to their bishop’s cell.
diagonals so that 14 comes into the cell of 2 and 2 into that of 14, 8 comes
into the cell of 20 and 20 into that of 8, 18 comes into the cell of 6 and 6
into that of 18, 12 comes into the cell of 24 and 24 into that of 12. The
diagonal containing 12, 6, 24, 18 is thus parallel to the diagonal containing
8, 2, 20, 14 (Fig. b14).525 Next, we displace the content of the remaining
cells in the border towards the cells of the opposite rows. After we have
made these moves, the situation will be as in that figure (Fig. b15).526
5 4 3 2 1 5 4 3 14 1 5 23 22 14 1
10 9 8 7 6 10 9 20 7 18 11 9 20 7 18
15 14 13 12 11 15 2 13 24 11 16 2 13 24 10
20 19 18 17 16 8 19 6 17 16 8 19 6 17 15
25 24 23 22 21 25 12 23 22 21 25 12 4 3 21

(Even-order squares)
(B.7) The first of the even squares in which one can set out a magic
arrangement is that of 16 (cells). For though the first (even) square
actually is that of 4 (cells), it is not possible to have in it a magic sum
except in the diagonals. Indeed, if we place in it the numbers from 1 in
the natural order, they end with 4, and the content of each of the two
diagonals will be 5, which is the required quantity for each row of this
square;527 it will not be possible, however, for a sum equal to the content
of the two diagonals to appear in the vertical and horizontal rows. Indeed,
a corner cell is common to (three) rows, vertical, horizontal and diagonal;
then if we disregard it, there will remain in each row a single cell, and
the content of any (of these cells) ought to be equal to that of the other,
which is impossible since by hypothesis the numbers in a square with a
magic arrangement must be different.528
(Edition, p. 142)
(B.8) Magic arrangement in the square of 4.
525
All numbers moved thus flank the (for us) ascending diagonal. (The descending
one might just as well have been chosen.)
526
The ‘principles’, common to both methods, alluded to in i must thus be the
exchanges of elements between symmetrical rows.
527
See note 508 (or the formula in B.2ii).
528
As stated in the definition of magic square (B.1).
(i) It has appeared clearly that the square of 3 is the foundation of the
magic arrangement in odd squares.529 We shall show that for even squares
the foundation is, likewise, the square of 4, after showing how the numbers
are arranged in it.530
(ii) We say that if we place the consecutive numbers taken in the natural
order beginning with 1 in the square of 4 (Fig. b 16), it so happens that
the content of each of the two diagonals is 34, and such is the quantity
which is to be found in each row. Thus we are left only with equalizing
the four rows surrounding the four central cells and the four medial rows,
(two) vertical and (two) horizontal.531
(iii) The equalization of (the first ones) is made easy using the remarkable
natural property which has appeared to us:532 the content of any two
corner cells belonging to a same row of this (natural) square, when added
to the numbers in the two middle cells of the opposite row, gives again 34.
Therefore, if we move the content of the two medial cells of each (lateral)
row into the two medial cells of the opposite row, the content in the four
rows surrounding the central square will be equalized, with their quantity
being 34, as in this figure (Fig. b 17).
4 3 2 1 4 15 14 1 4 14 15 1
8 7 6 5 5 7 6 8 9 7 6 12
12 11 10 9 9 11 10 12 5 11 10 8
16 15 14 13 16 3 2 13 16 2 3 13

(iv) We are left with equalizing the four middle rows, vertical and hori-
zontal. Their equalization is once again easy:533 we move the content of
the second cell into the third and that of the third into it, the content of
the fifteenth cell into the fourteenth and that of the fourteenth into it, the
content of the eighth into the twelfth and that of the twelfth into it, and
we displace the content of the fifth into the ninth and that of the ninth
529
When borders are constructed successively around it. This is our B.12–B.13, but
it precedes B.8 in the manuscript.
530
This is the first even-order square considered. Like the square of 3 for odd-order
squares, the square of 4 plays a fundamental rôle in the construction of bordered squares,
with borders being constructed successively around it. This will be shown in (our)
B.18–B.19.
531
This will be done in iii and iv, respectively, using two properties (in fact, one) of
the natural square.
532
The author will return to this property (below, vii, 1).
533
The author will return to this property (below, vii, 2).
into it. Then the equalization of all the rows of this square is realized
vertically, horizontally and diagonally, as in this figure (Fig. b 18).
(v) There appears in this square by nature, besides the magic arrangement
we were searching for, a nice equalization, namely that the sum of the
content of any two middle cells of the border around the central square
and that of the two opposite cells also gives 34: thus 12 and 8 added to 9
and 5, and 14 and 15 added to 3 and 2. It also appears that when we add
the content of two cells adjacent to any corner cell to the content of the
two cells adjacent to the opposite corner cell, this gives also 34; indeed, if
we add 12 and 15 to 2 and 5, and if we add 14 and 9 to 8 and 3, it gives
34 (in both cases). It further appears that if we add the content of the
four cells gathered together in any corner, it gives also 34; thus 1, 15, 12
and 6 make 34.534
(vi) (Other way for the) magic arrangement in the square of 4 by dis-
placement. If we wish that, we leave the numbers in the middle cells of
the border where they are and move the content of each corner into the
opposite one. (Next) we displace the content of each of the four central
cells, in the centre of the square, once again to its opposite, that is, to its
queen’s cell. There will result for the magic arrangement in the square
all that has appeared in the first figure. Here is the figure of it (Fig. b
19).535 (. . . )536
13 3 2 16
8 10 11 5
12 6 7 9
1 15 14 4
Fig. b 19
(vii) Those of the remarkable natural properties concerning the magic
sums occurring in the square of 4 which we have chosen to take as a
fundamental principle useful for its (construction) are the following.537
(1 ) We have found that the numbers in the four corner cells, starting with
534
The first two are not unique to the magic square of order 4 since they are also
found in the natural square; furthermore, they appear in any magic square of order
4, however constructed. On all this, see Les carrés magiques, pp. 191–193 (Russian
edition, pp. 203–205).
535
It differs from that of Fig. b18 merely by a rotation.
536
The part omitted here presents another construction resulting in the same square.
537
The ‘fundamental principle’ is thus the two properties of the natural square used
in iii and iv and repeated here. As before (note 526), they concern exchanges between
symmetrical rows of the natural square.
1, equalize the numbers of the four opposite middle cells belonging to the
two horizontal rows and adjacent to the central cells, and (also) the four
middle cells belonging to the two vertical rows and adjacent to the four
central cells when they are considered for numbers taken consecutively
from 1. (. . . ) (2 ) We have (in addition) found that the (content of the)
four cells in the centre equalize (that of) the four middle cells belonging
to the two (middle) vertical rows, and likewise equalize the (content of
the) four middle cells belonging to the two (middle) horizontal rows.538
(. . . )
(viii) Then, if we choose to make it a principle to use for setting out the
magic arrangement in this square, that is possible. It will lead us to a
configuration in which the numbers in the square make the magic sum in
the vertical and horizontal rows, in the two diagonals, in the four corners,
in the opposite middle cells in the border around the central square, and
in the four central cells.539 But there will be no other magic sum than
these.540
(B.9) General way of searching for complete magic arrangement in the

square of 4.
(i) This is the figure we have found in our predecessors.541 It is possible
to arrive at the magic arrangement in this square by means of methods
without displacement showing regularity and elegance. We can make
it the foundation for finding the magic arrangement in the other even
538
These two properties, used in iii and iv, respectively, actually reduce to a single
one: for any two symmetrically placed rows in the natural square of order 4, the two
extremes in one of them plus the two middle in the other make the magic sum 34.
539
Generally, in Fig. b18, the magic sum is found in the groups of cells having the
same sign in Fig. b20 —which also applies to the natural square of Fig. b16 since the
numbers there form the same groups.
◦ † † ◦
— ∗ ∗ —
— ∗ ∗ —
◦ † † ◦
Fig. b 20
540
This is an odd statement: in v it was said that the magic sum is also found in
each quadrant and in the four cells enclosing opposite corner cells. In B.9i another
construction will lead to a further property.
541
The figure here should be that of b23, to be constructed now.
squares.542 You may here begin putting the numbers in any of the cells,
starting, in your placing the numbers, with 1, and continuing according
to the succession of the numbers until all cells are filled.543 Any (set of)
four cells will make the magic sum if the arrangement follows a uniform
rule. This is what is called ‘full and complete magic arrangement’.544
(. . . )
(ii) When we wish to place the numbers in this square, we must start
their arrangement with 1. We place it in one of the corners. We put 2
and 3 in the cells adjacent to the opposite corner cell and put 4 in that of
the four central cells which is diagonally adjacent to the initial corner cell,
according to that figure (Fig. b 21). We may notice that in this placement
there is one cell belonging to each of the (four) groups into which we have
divided the cells of the square, namely:545 one of the corner cells, which
contains 1; one of the four central cells, which contains 4; one of the (four
middle) cells of the (lateral) vertical rows, which contains 3; one of the
(four middle) cells of the (lateral) horizontal rows, which contains 2. If
3 were at the place of 2 and 2 at the place of 3, this would (also) be
possible. When we wish to place in four other cells the numbers following
that which has been reached, we place them in the order reverse to that
adopted in the previous figure, and in the symmetrical (cells). The result
will be as in that figure (Fig. b 22).546
(iii) We have, in order to finish placing of the numbers in the remaining
cells, two ways.547 The first is to put in each empty cell the complement,
relative to the equalizing number of this square, of what is in the third
cell diagonally from it, that is, its bishop’s cell. Thus, if we wish to know
what will appear in the second cell, we consider its bishop’s cell, namely
the twelfth. We find 3 (there). We subtract it from the equalizing number
of this square, thus 17; the remainder is 14, which we place in the second
cell. Likewise, we place in the third cell 11, and so on for the remaining
542
See B.26.
543
Thus the numbers are placed directly, without resorting to the natural square.
This is the method ‘without displacement’. Concerning the arbitrary choice of the
initial cell, see Commentary, pp. 28–29.
544
The three properties seen in B.8v are thus extended, the last now being verified
for any square of four cells.
545
In the passage omitted by us before ii, the cells of the square are divided into four
groups of four cells each: those in the corners, those in the centre, the medial in the
lateral vertical rows and the medial in the lateral horizontal rows.
546
Thus how the first four numbers are placed determines where we shall put the
next four, whereby (see iii) the whole square is determined.
547
We omit the second, which is banal (enumerating the cells where to put the se-
quences 9, . . . , 12 and 13, . . . , 16).
cells. The result will be as in this figure (Fig. b 23). (. . . )548
1 8 1 8 11 14 1
4 5 4 10 5 4 15
3 3 6 3 16 9 6
2 2 7 13 2 7 12
(B.10) Other way to set out the magic arrangement in the square of 6,
by displacement.549
(i) It is possible to set out the magic arrangement in the square of 6 by
methods of displacement showing regularity and elegance550 for whoever
wishes that his treatment be made easier.
(ii) To do that, we place the consecutive numbers from 1 in the natural
order (Fig. b 26). We find that in the two diagonals appears ipso facto
the magic sum, which equals what must be in the rows of this square.
Therefore, we shall leave their numbers unchanged.551 As a result, in the
square of 4 within this square, the four central cells and the four corner
cells will maintain their numbers in their position.
11 10 9 8 11 9 28 8
17 16 15 14 23 16 15 20
23 22 21 20 14 22 21 17
29 28 27 26 29 27 10 26
Fig. b 24 Fig. b 25
(iii) When we wish to finish arranging the numbers in the cells of the
square of 4 within the square considered by means of displacements (we
proceed as follows). We move 9 to the cell of 10, 10 to the cell of 27, 27
to the cell of 28, and 28 to the cell of 9.552 We do the same with the four
548
In the passage omitted here, the author explains how to proceed when 1 is put in
any other cell; see our commentary, p. 29.
549
‘Other way’: construction of the bordered square has already been taught (our
B.18).
550
The very same words were found in B.9i.
551
This is characteristic of the construction of ordinary magic squares in the 10th
century —and also, at least for odd orders, its weakness (see Commentary, pp. 23–24).
552
We have added for convenience the two figures b24 and b25.
remaining cells of the border, that is, we move 20 to the cell of 14, 14
to the cell of 23, 23 to the cell of 17, and 17 to the cell of 20. We have
finished with the numbers of the square (of 4) inside the square of 6.553
(iv) Next, we displace the numbers in the inner cells of the border. We
begin with the two middle cells, and displace their (content) to the two
opposite cells as follows: we move 3 to the cell of 34, 34 to the cell of 4,
4 to the cell of 33, and 33 to the cell of 3. Next, we move, following the
same pattern, 13 to the cell of 18, 18 to the cell of 19, 19 to the cell of 24,
and 24 to the cell of 13. Thus doing we shall have displaced the (contents
of the) inner (middle) cells of the border.554 Next, we move 2 to the cell
of 35, 35 to the cell of 32, 32 to the cell of 5, and 5 to the cell of 2. Next,
we move 7 to the cell of 25, 25 to the cell of 12, 12 to the cell of 30, and
30 to the cell of 7.555 With this the arrangement of the numbers in this
square by displacement is finished. This is the figure of it (Fig. b 27).
6 5 4 3 2 1 6 32 34 33 5 1
12 11 10 9 8 7 25 11 9 28 8 30
18 17 16 15 14 13 13 23 16 15 20 24
24 23 22 21 20 19 19 14 22 21 17 18
30 29 28 27 26 25 12 29 27 10 26 7
36 35 34 33 32 31 36 2 3 4 35 31
Fig. b 26 Fig. b 27
(v) It is possible to arrange the numbers in the square of 6 using many
methods of displacement.556 We have reported what is easy to understand
by the beginner, who might infer others. Here is the figure of it.557
(B.11) Magic arrangement in the square of 8 by displacement.

553
These displacements conform to those of the other rows (of order 6 and 8), but
not to the transformation seen in B.8. These rows will therefore not be equalized here.
554
In short: the contents of a pair of middle cells and their opposite are exchanged,
then the terms of one of them are inverted.
555
In short: diagonal exchange of the contents of opposite pairs of cells, then two
opposite terms return to their original row.
556
The author presented just one ‘method of displacement’ for the order 4 and two
for the order 5.
557
Judging by the size of the blank space, there should have been two figures; though
not according to the text (hā instead of humā).
(i) It is indeed possible to arrange the magic sums in this square, as well
as in the larger squares, by means of methods of displacement analogous
to what we have seen for the square of 6.558 It is certainly suitable that
we explain here its arrangement in order that the treatment of the others
be made easier for the student. If we wish that, we (begin by) writing
the consecutive numbers from 1 in the square (Fig. b 28).
8 7 6 5 4 3 2 1
16 15 14 13 12 11 10 9
24 23 22 21 20 19 18 17 15 51 53 52 14 10
32 31 30 29 28 27 26 25 42 22 20 45 19 47
40 39 38 37 36 35 34 33 26 38 29 28 35 39
48 47 46 45 44 43 42 41 34 27 37 36 30 31
56 55 54 53 52 51 50 49 23 46 44 21 43 18
64 63 62 61 60 59 58 57 55 11 12 13 54 50
Fig. b 28 Fig. b 29
(ii) Next, we carry out their displacement in the square of 6 within the
square of 8 as we carried out the displacement of the numbers in the
square of 6.559
We begin their displacement with the numbers in the square of 4. We
move 20 to the cell of 21, 21 to the cell of 44, 44 to the cell of 45, and 45
to the cell of 20. Next, we move likewise 35 to the cell of 27, 27 to the
cell of 38, 38 to the cell of 30, and 30 to the cell of 35.
Next, we move likewise 12 to the cell of 53, 53 to the cell of 13, 13 to
the cell of 52, and 52 to the cell of 12. We also move 26 to the cell of 31,
31 to the cell of 34, 34 to the cell of 39, and 39 to the cell of 26. Next,
we move 11 to (the cell of) 54, 54 to the cell of 51, 51 to the cell of 14,
and 14 to the cell of 11, and we move 18 to the cell of 42, 42 to the cell
of 23, 23 to the cell of 47, and 47 to the cell of 18. Then the quantities
of the square of 6 within the square of 8 are all the same, the sum of the
content of each row being 195. Here is the figure of it (Fig. b 29).
(iii) Next, we move 5 to the cell of 4 and 4 to the cell of 5, (but) leave
60 and 61 in their place. Next we move 2 to the cell of 63, 63 to the cell
558
B.10. Note the allusion to an extension to higher orders, confirmed by the next
sentence and iv.
559
Indeed, the procedure is as seen in B.10, both for the inner square of order 4 and
for the border of order 6.
222 Construction of bordered magic squares
of 7, 7 to the cell of 58, and 58 to the cell of 2.560 Next we move, by

analogy, 3 to the cell of 62, 62 to the cell of 6, 6 to the cell of 59, and 59
to the cell of 3. Then the sum in all inner vertical rows is the same, and
in accordance with the quantity to be found in each row, namely 260, and
the first and eighth horizontal rows also make the same sum, (for) they
contain the quantity 260.
Next, we move 25 to the cell of 33 and 33 to the cell of 25, (but) leave
32 and 40 in their cells. Next we displace the (contents of the) cells of the
first and eighth vertical rows by analogy with the displacement performed
for the (contents of the) cells in the first and eighth horizontal rows.561
For we move 49 to the cell of 16, 16 to the cell of 9, 9 to the cell of 56,
and 56 to the cell of 49. Next, we move likewise 41 to the cell of 24, 24
to the cell of 17, 17 to the cell of 48, and 48 to the cell of 41. Then the
quantities of all rows of this square are equalized. Here is the figure of it
(Fig. b 30).
8 63 62 4 5 59 58 1
49 15 51 53 52 14 10 16
41 42 22 20 45 19 47 24
32 26 38 29 28 35 39 33
40 34 27 37 36 30 31 25
17 23 46 44 21 43 18 48
9 55 11 12 13 54 50 56
64 2 3 61 60 6 7 57
Fig. b 30
(iv) We may, by analogy with this reasoning, set out the magic arrange-
ment in any other even squares whatsoever.
(Construction of bordered magic squares)

(B.12) Once a magic arrangement for the first odd square, that of nine
(cells), has been found, it will be easy to attain it for other odd squares,
560
In short: vertical exchange of terms in pairs placed symmetrically, then horizontal
exchange of two of them. Same moves in what follows.
561
As in the previous note, but with the vertical exchanges becoming horizontal ones
and vice versa.
and the same will hold once a magic arrangement for the first even square
has been found.562
Indeed, the square of 3 is within the square of 5; therefore, if we have
attained the (magic arrangement of the) numbers for the square of 3, a
single border around it will be left; it will then be easy to determine its
(magic arrangement), as we shall explain. Likewise, the square of 5 is
within the square of 7; therefore, if we have already attained the (magic
arrangement of the) numbers for the square inside, the remainder will
consist of a single border. The same holds for the square of 7 inside the
square of 9 (and so on). (. . . )
Likewise, once we know the magic arrangement for (the square with)
16 (cells), that for the square of 6 will be easy, and thus (also) for the
square of 8.
(Edition, p. 142)
(B.13) Preliminaries to magic arrangement in even and odd squares.

(i) It has been seen before563 that the square of 5 contains the square
of 3 and that, once we know the magic arrangement in the square of 3,
the magic arrangement in the square of 5 will be made easier, and that
the same holds for all other squares containing one another. Indeed, once
we shall have set out the magic sums in any square, even or odd, setting
them out in the next square in fact amounts to equalizing the numbers
belonging to the border surrounding the inner square. We must therefore
consider the differences between the (number of cells) in (consecutive)
odd and even squares and which numbers will occupy them.564
(ii) We therefore say that the difference between any two square numbers
is equal to the product of the sum of their sides by their difference.565
Therefore, when we have odd or even square numbers, the difference
between any two consecutive ones will be equal to twice the sum of their
sides —since the difference of (consecutive) sides for odd and even squares
is 2. Then if we wish to know the difference between two consecutive odd
or even squares, we add their sides and double the sum; the result will be
562
As already said (note 520), to construct (empirically) bordered squares we start
with the smallest inner square.
563
B.12.
564
These two questions, concerning the number of cells in a given outer border and
the numbers to put therein, will now be settled (ii, iii). Those chosen for each row,
which depends on the magic sum, will be examined later (B.15–B.16).
565 2
n − m2 = (n − m)(n + m), with n − m = 2 for consecutive squares having the
same parity.
the number of cells between the two squares and (thus) the quantity of
numbers which are to be found in them.566 (Then) if we have set out the
magic arrangement in the first square, we shall be left with equalizing the
numbers in its border, that is, in the cells forming the difference between
the two squares.
(iii) This being so, if we have a square and wish to set out in it the
magic arrangement, we shall add the side of this square to the side of the
square just preceding, and the result will be the quantity of consecutive
numbers starting with 1 which are to occupy half the cells of the border,
the other half being occupied by their associated numbers, that is, their
complements to the equalizing number of this square.567 The meaning of
our term ‘associated number’ is that two numbers are associated when
their sum equals the equalizing number. The ‘equalizing number’ itself
is the quantity of the sum of the two extremes (of the sequence) of con-
secutive numbers to be put in the squares, such as 82 for the square of 9;
it equals twice the median number for odd squares and (the sum of) the
two median numbers for even squares.568
(iv) Once we shall have arranged these numbers in the cells of the (outer)
border in the way we shall explain, we shall add once again the side of this
(smaller) square to the side of the next smaller square; the result will be
the quantity of numbers which are to occupy the cells of half the border
around the third square. Likewise, when we add the side of the third
square to the side of the fourth square, the result will be the quantity
of numbers which are to occupy the cells of half the border around the
(fourth) square within. (We proceed in the same way) until we reach, for
the cells, the single (central cell) in the case of odd squares.569
(v) Example. We wish to know the numbers which are to occupy the
various borders of a square of 9, to 81 (cells), and the total quantity of cells
belonging to (the border of) a large(r) square around the (next) small(er)
566
Thus here n2 − (n − 2)2 = 2 [n + (n − 2)] = 4n − 4 is the number of cells in the
border surrounding the square of side n − 2.
567
Since the outer border has 4n − 4 cells, half of it will contain the numbers from 1
to 2n − 2, and the other half their complements, that is, the numbers from (n2 + 1) − 1
to (n2 + 1) − (2n − 2) since the complement of ai is (n2 + 1) − ai . These ‘associated
numbers’, or pairs of ‘complements’ will now be defined, as will the term ‘equalizing
number’ (already known to us from B.2, but B.2 comes after B.13 in the original text).
568 2 2
The median is 12 (n2 + 1) for odd squares, the two medians are n2 and n2 + 1 for
even squares.
569
If the inner border has the order m < n, it will contain m + (m − 2) numbers and
their complements.
square.570 We add the side of 81 to the side of the next smaller square
among the odd numbers, thus 49; this gives 16. So we shall say that such
is the quantity of half the cells of the border around the square of 7, that
the numbers which are to occupy this half are the consecutive numbers
from 1 to 16 and that the other half of the cells of this border will be
occupied by the numbers associated with them, (thus) their complements
to the equalizing number for the square of 9, namely 82, which will be
the consecutive numbers from 66 to 81.
Likewise, we add 7, the side of 49, to 5, the side of 25, which gives 12;
so we shall say that such is half the number of cells of the second border,
belonging to the square of 7 around (the square of) 25 (cells), that half of
the cells will be occupied by twelve consecutive numbers, (namely) from
17 to 28, and that the other half of the cells in this border will be occupied
by their associated numbers, from 54 to 65.
Likewise, we shall add the side of 25 to the side of 9; this gives 8,
which is half the number of cells in the border around the square of 3;
the numbers which are to occupy them are eight consecutive numbers,
(namely) from 29 to 36, and the numbers for the remaining cells in this
border are again eight numbers, beginning with 46 (and the next) to 53.
Likewise, if we add the side of 9 to the side of 1, this gives 4, and this is
half the number of cells around the single (central cell); the beginning of
the (smaller) numbers is 37, their end 40, the beginning of the associated
numbers is 42, and their end 45.
In this way the two numbers (40 and 42) have been simultaneously
attained (by starting) from the extreme terms, namely 1, the smallest
number to be placed in the square, and 81, the largest to be placed in it.
There remains the median number (namely 41), which we shall place in
the central cell of this square.
(vi) For the even numbers, the way to obtain (the quantity) of cells in the
border and the numbers which will occupy them follows what we have set
forth for the odd squares.571 There is, however, a difference when we reach
the square of 4, for the twelve (cells) around the four (central cells) will
not be occupied by the numbers attained by (considering) the difference,
and the numbers will not be arranged in the border surrounding the four
(central cells) as they were in the other borders of larger squares, even or
odd. Indeed, the magic arrangement in the square of 4 differs from that
570
Keeping in mind that the outer border will receive the smallest number (thus also
the largest).
571
The only difference will be explained now: since there can be no magic square of
order 2, the inner 4 × 4 square has to be arranged as a whole.
in the other squares, as we shall explain.572

(B.14) Other way for determining the numbers which are to occupy the
cells in the border.573
If we wish to know this, we (first) place the median number, which is
the number to occupy the central cell in an odd square. Then we take
four (pairs) of associated numbers, on both sides of the median number.
These are, together with the median number, the numbers which are to
occupy the square of 3. They are, for the case of the square of 7,
21 22 23 24
29 28 27 26,
the median number being 25. Next we take, on the sides of the two
numbers we have attained, eight (pairs of) associated numbers, namely
13 14 15 16 17 18 19 20
37 36 35 34 33 32 31 30,
which are the numbers to occupy the border around the square of 3
belonging to the square of 5. Next we take, on the sides of the two
numbers we have reached, twelve (pairs of associated) numbers, namely
1 2 3 4 5 6 7 8 9 10 11 12
49 48 47 46 45 44 43 42 41 40 39 38,
which are the numbers to occupy the cells of the border around the square
of 5 belonging to the square of 7. We (thus) came on one side to 1, the
smallest number to be placed in this square, and on the other to 49, the
largest number to appear in it.574
In the case of even squares, we take two numbers, on either side of
the two median numbers —which are, for the (case of the) square of 8,
32 and 33—, thus 31 and 34. Then we take, on the sides of 34 and 31,
six (pairs of) associated numbers which follow the numbers taken first,
namely
572
See the (above) fragments B.8–B.9.
573
Starting now from the central cells (with the medians), and determining the oc-
cupants of the successive borders. Since from one border to the next the number of
cells increases by 8, each sequence contains four additional numbers. From now on
the author will present the sequences by pairs of complements, aligned vertically, to
facilitate placing.
574
Compare with the last sentence in B.13v; here the numbers are just taken in
reverse. A also fills the square from within.
25 26 27 28 29 30
40 39 38 37 36 35,
which are the numbers to occupy, together with the four numbers men-
tioned previously, the square of 4, namely the 16 cells in the centre of the
square of 8.575 Next we take, on the sides of the two numbers we have
reached, ten (pairs of) associated numbers, namely
15 16 17 18 19 20 21 22 23 24
50 49 48 47 46 45 44 43 42 41,
which are the numbers to occupy the border around the square of sixteen
(cells). Next we take, on the sides of the two numbers we have reached,
fourteen (pairs of associated) numbers, which are the numbers to occupy
the border around (the square of) 36 (cells), namely
1 2 3 4 5 6 7 8 9 10 11 12 13 14
64 63 62 61 60 59 58 57 56 55 54 53 52 51.
The numbers we have reached are on one side 1 and on the other 64,
which is the number equal to the quantity of cells in the square where we
wish to set out the magic arrangement.
(Odd-order squares)
(B.15) Setting out the magic arrangement in odd squares, starting with
that of 5.576
(i) When the consecutive numbers are in natural order, beginning with 1
or not,577 we take, for the odd squares, the median number, from which
we always subtract 4.578 The remainder will be the number with which
we shall begin filling the square of 3 located inside the square of 5. We
arrange in the square of 3 the consecutive numbers, beginning with the
number remaining (from the subtraction), which we put in one of the
middle cells of the border, (then) we proceed in accordance with what we
have described for the square of 3;579 indeed, we put what follows it in its
575
Thus we have, as before, regular addition of four pairs. But it would have been
simpler to say that we are to take seven numbers below 32 and seven above 33: it has
already been explained that the 4 × 4 square is to be filled as a whole. Indeed, see
B.18ii, B.19ii.
576
We already know how to arrange the inner 3 × 3 square.
577
The words ‘or not’ may be an earlier reader’s addition: throughout, the numbers
will start with 1.
578
B.14.
579
B.3–B.5.
knight’s cell, and fill the remaining cells of the square of 3, the result being
as in this figure (Fig. b 31). Then the quantities (in the square) inside the
square of 5 are equalized and each of its rows contains 39, which is the
result of multiplying 13 by 3, the side of the square.580
10 17 12
15 13 11
14 9 16
Fig. b 31
(ii) We must (now) explain the equalization of the border by means of
the numbers remaining among the set of those which are to appear in the
square of 5, in such a way that the content of each row of the square of
5 makes, with what will be put at its extremities, the magic sum. Now
this is easy for us considering the numbers already found which we have
arranged in this figure (Fig. b 31). Indeed, when we look at this figure,
we find that the quantities in each pair of opposite cells in the border are
equal to the equalizing number; such are 10 and 16, 9 and 17, 14 and 12,
(15 and 11,) since each pair in these opposite cells makes 26. What we
have found in this square thus shows us that, for the remaining cells in
the border, the content of each pair of opposite cells must again be equal
to the equalizing number.581 Therefore we shall place in two lines (of
associated numbers) the numbers remaining among the set of numbers to
be arranged in this square, namely
1 2 3 4 5 6 7 8
25 24 23 22 21 20 19 18,
which are the numbers with which must be made the equalization of the
whole border. This being so, the equalization must first take place in the
diagonals; for the equalization in the diagonals affects the equalization in
the whole border, and, when a further row has been equalized, equalizing
the remaining rows will hardly present us with any difficulty.582
(iii) Therefore we place in two of its consecutive corner cells two numbers
580
According to the formula taught in B.2ii.
581
This placing of complements in opposite cells (horizontally, vertically, diagonally
for corner cells) ensures a uniform increment of the magic sum by n2 + 1, as required.
582
Since we no longer use the natural square and therefore do not keep its diagonals
unchanged as before, we are first to seek two numbers to fill the corner cells, each
of which is common to two rows (iii). Then we shall proceed with equalizing the
two upper rows (iv) and the lateral columns (v). The author will finally discuss other
possibilities for the corner cells and thus for the lateral rows (vi–vii), and then consider
permutations of the numbers within the rows (viii).
among those of the two (conjugate) lines, say 4 in the first and 6 in the
second; in each of the two opposite corners we put the complement to
the equalizing number. Then, with what we have put in the corners, the
diagonals (of the square of 5) are equalized, each containing 65. We are
left with equalizing the vertical and horizontal (inner) rows in such a way
that, after we have put in the extremities of each vertical and horizontal
row a pair of associated numbers among those written in the two lines,
there will be (in these rows), with what we have (already) placed, here
too 65.
(iv) Example of that. We wish to equalize the vertical rows.583 Their
equalization is attained by putting in the empty cells of the first (hori-
zontal) row, where there are three cells, three numbers the sum of which
make, together with the two numbers in the first and second corners, 65.
Now the quantity (still) required for that row is 55. Then we are to seek
among the set of the remaining associated numbers which we have writ-
ten in the two lines, thus (among) twelve numbers, three numbers with
sum 55. Considering the two lines, we find that such are 24, 23, 8: if we
place them in the empty cells of the first horizontal row, it will contain 65.
Putting in the opposite row the numbers associated with them, there will
be in the twenty-second cell 2, in the twenty-third 3, in the twenty-fourth
18, and the fifth horizontal row will also contain 65, as follows, in this
figure (Fig. b 32).
6 8 23 24 4 6 8 23 24 4
10 17 12 7 10 17 12 19
15 13 11 5 15 13 11 21
14 9 16 25 14 9 16 1
22 18 3 2 20 22 18 3 2 20
Fig. b 32 Fig. b 33
(v) After looking for three numbers adding up to 41, which is that by
which the first vertical row will be equalized, then placing them in its
empty cells and putting their associates in the opposite cells, each at
the other extremity of the row considered, we shall have equalized all
the vertical and horizontal rows left.584 Doing that by considering the
(remaining numbers in the) two lines, we find for the three numbers 19,
21, 1. Putting them in the empty cells will fulfil our purpose, and the
583
The three medial columns.
584
Thus the two lateral columns and the three medial lines.
figure will be the following (Fig. b 33).

(vi) One should not think that the two numbers which we have (at first)
put in the two consecutive corner cells are there by chance; many pairs of
numbers taken in the two conjugate lines and put in these two corner cells
would not permit us to complete the magic sum in this square. Thus,
putting in these two corners, instead of 4 and 6, 1 and 4, or 1 and 2, or 4
and 5, or 6 and 7, we shall not find in the two conjugate lines numbers the
sum of which would, together with the (occupants of the) two corner cells,
make 65. (On the other hand,) putting 1 and 3, or 3 and 7, or 2 and 8, or
4 and 6, or 5 and 7, or 6 and 8, it will be possible to find three numbers
such that their sum will make, together with the occupants of the corners,
65. So let us decide to put the number preceding the side of the square
in the first corner and that following the side of the square in the second
corner: this pair will enable us to equalize the row and (therefore) it will
be possible to find the other numbers for the other rows.585
(vii) The situation being as we have described, we are now to place the
(pairs of) numbers mentioned above together with the numbers to be
put in the two vertical and horizontal (lateral) rows for those wishing (to
know) the (various) magic arrangements in the square of 5; thus doing,
they will realize how many, and of how many kinds, there are. The figure
of the table has been given above.586
(viii) It already appears that each of these configurations diversifies into
twice six kinds. Indeed, of the three numbers which, in the preceding
figure (Fig. b 33), we have placed in the first horizontal row —thus 24,
23, 8—, 24 has been put in the second cell of the row; now, it could have
been possible to place there 23 or 8. Likewise, what has been put in the
third cell might have been placed in the second or in the fourth. In this
way arise six combinations for the first horizontal row, and we shall have
similarly six combinations for the first vertical row, between the first and
the third corner cells. This makes twelve configurations.587 Multiplying
them by the configurations in the table will make . . . configurations.588
585
Relating these numbers to the order of the border helps to memorize them. This
is also in agreement with the (single possible) arrangement of the 3 × 3 square.
586
There is in the manuscript, a few lines above, a sizeable blank space, probably
intended for a table indicating some or all possibilities. The author will now consider,
for a given configuration, permutation of its elements within the row.
587
Sic, instead of 36. The error cannot be attributed to the author, and some reader
of an earlier copy must have altered the text, misled by the ‘twice six kinds’ above.
588
Space left blank (indeed, whatever was written there cannot have made sense with
the previous ‘twelve’). We do not know whether the original text had the correct
number, 360, since we do not know if the author knew all ten possible configurations
(ix) We conceive then that the magic arrangement may be set out in the
square of 5 as we have explained (namely in many ways). And if in the
square of 5 we can set out the magic arrangement according to (all) these
configurations, one may imagine how the situation will be for the squares
with sides larger than 5. But, for our part, we shall restrict ourselves to
a small number of configurations among this set, usable by the student
learning it and interested in a construction method.
(B.16) Magic arrangement in the other odd squares.

(i) We have explained in what precedes how to determine the numbers
to be arranged in the successive borders of odd and even squares,589 and
we have (afterwards) explained how to set out the magic arrangement in
the square of 5590 . We are now to add an example of a square with side
larger than 5 in order that this may become easier for the student.
(ii) We consider a square with side 9. We begin with the median number,
thus 41 since the square (of 9) is 81, and we put it in the central cell.
Next, we take on both sides of it four (pairs of) associated numbers.591
We place the smallest of them in one of the middle cells of the border
surrounding the central cell, and we put the other numbers taken one
after the other in the remainder of the nine cells according to (one or the
other of) the ways described for the square of 3.592 Then the quantities
in the rows of this square will be the same, each of them containing 123.
(iii) Next, we take eight numbers on the two sides of the two numbers we
have reached, namely 37 and 45, and we set them in two lines, according
to this figure:
29 30 31 32 33 34 35 36
53 52 51 50 49 48 47 46.
We put the fourth, thus 32, in the first corner of the border and the
sixth, thus 34, in its second corner, in accordance with what we have
for a bordered 5 × 5 square.

589
B.13–B.14.
590
B.15.
591
What comes now just follows (and extends) the instructions given in B.14.
592
Whichever the way, the result is the same since there is, apart from inversions and
rotations, just one configuration of the square of order 3 (as seen in B.3). In B.21iii,
the author will again speak of ‘methods’ for filling the 3 × 3 square. See also above,
note 514.
described for the square of 5.593 We place their associates in the two
opposite corners. (Next) we look, among the set of the twelve (remaining)
associates, for three numbers making the sum 139: indeed, the numbers
in each row of the considered square, thus the square of 5 within the
square of 9, must all make the same sum; since it appears that each
of the two diagonals contains 205, there must also be 205 in each of
the vertical and horizontal rows; now the row containing 32 and 34 is
in deficit, relative to 205, by 139. Having then sought, in the set of
numbers remaining in the two conjugate lines, three numbers with sum
139, (then) put them in the empty cells of the row of 32 and 34, and
in the opposite cells their associated numbers, all (inner) vertical rows
will be equalized: their quantities will all make the same sum, the total
in each row being 205. Now, looking at the two (conjugate) lines, we
find in them three numbers with sum 139, namely 51, 52, 36.594 After
inscribing them in the row containing 32 and 34, and their associated
numbers in the opposite cells of this square, each of the (inner) vertical
rows will make 205. We are then to seek three other numbers among the
remaining ones such that their sum, together with what has been put in
the first and third corners, make 205. Seeking, among the set of these
(pairs of remaining) associated numbers, three numbers with sum 125,
we find 49, 47, 29, with the associates 33, 35, 53. We put the (first) ones
in the empty (right-hand) cells of the square, and their associates in the
opposite cells. Then the rows of the square considered, that of 5 by 5,
are equalized.
(iv) It is now necessary to equalize the quantities in the subsequent bor-
der, surrounding the square of 5. We take twelve consecutive numbers on
both sides of the two numbers we have reached, namely 29 and 53, and
set them out in two lines in order to facilitate us their arrangement in the
border.595 They are the following twenty-four associated numbers:
17 18 19 20 21 22 23 24 25 26 27 28
65 64 63 62 61 60 59 58 57 56 55 54.
We put in the first corner the sixth of these numbers and in the second
corner the eighth. Then there will be in the first corner 22, in the second
593
The fourth and the sixth of the smaller numbers were put in the corner cells of
the 5 × 5 bordered square. Accordingly, for a border of odd order m, it will be the
(m − 1)th and the (m + 1)th in the list of smaller numbers to be placed; see below, iv
and v.
594
They could also be obtained by taking, from the two conjugate lines, those in the
same place as for the order 5 (B.15ii–v).
595
One wonders why this remark about usefulness was not made before.
corner 24, in the third corner 58, in the fourth corner 60. There thus
appears in each of the two diagonals 287, and the very same will have to
be in each row. So we look, among the set of associated numbers, for five
numbers making the sum 241 in order to have, with what we have put
in the two (upper) corners, 287. We find that it is possible to put in the
row of 22 and 24 five numbers with this quantity as their sum, and these
are 64, 62, 61, 26, 28. We put their associates in the opposite cells, each
at the other end of the row considered. We likewise find for the row of
22 and 58 five numbers making the sum 207, which are 17, 19, 59, 57,
55, and we put their associates in the opposite cells. The (sums of the)
quantities which are in the square of 7 inside the square of 9 appear then
to be the same, each of its rows containing 287.
10 16 14 12 75 76 78 80 8
15 24 28 26 61 62 64 22 67
13 27 34 36 51 52 32 55 69
11 25 35 38 45 40 47 57 71
9 23 33 43 41 39 49 59 73
77 63 53 42 37 44 29 19 5
79 65 50 46 31 30 48 17 3
81 60 54 56 21 20 18 58 1
74 66 68 70 7 6 4 2 72
Fig. b 34
(v) We are now to look for the numbers to put in the border of the
square of 9 among the remaining numbers, that is, (the following) sixteen
numbers (and their associates):
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
81 80 79 78 77 76 75 74 73 72 71 70 69 68 67 66.
We then put, in accordance with what precedes,596 the eighth number in
the first corner and the tenth in the second, and we put in the two opposite
corners their two associated numbers. There will then be, in each of the
two diagonals, 369, which is the quantity to be in each row of this square.
So we are now to look for seven numbers making the sum 351. Such are
80, 78, 76, 75, 12, 14, 16. Putting them between 8 and 10, the sum will
be 369, and putting their associates in the opposite cells there will be,
596
See note 593.
together with the two numbers 72 and 74, 369. Likewise, we look among
the remaining associated numbers for seven numbers making, together
with the two numbers 8 and 72, the sum 369. We find for this 67, 69, 71,
73, 5, 3, 1. Putting them in the row of 8 and 72, and their associates in
the opposite cells, the whole content of the rows of the square, vertically,
horizontally and diagonally, will be 369. This is the illustration of it (Fig.
b 34).
(B.17) Magic arrangement for all odd squares.

We have (thus) set up the magic arrangement in the square of 9.
We must now explain a method to improve the student’s skill and (also)
intended for those who prefer to save themselves the trouble of working
out which numbers to arrange in the square.597
Let the square intended for magic arrangement be the square ABGD,
and let the middle cells of the vertical and horizontal rows of this square
be those forming the two rows ET. and ZH . (Fig. b 35). We place in the
central cell the median number. Starting from any one of its corners, say
corner A, we shall put the consecutive numbers from 1 on both sides of
A, alternately in the cells of the two rows enclosing it —excepting their
middle cells— until half the cells of the two rows are filled.598 It then
appears that the odd numbers are in one of the two rows and the even
in the other. Next, we put the number reached in the middle cell of row
AG,599 the next in the second corner, the next in the middle cell of row
GD, the next in the fourth corner. We put the subsequent numbers again
in the cells of the two rows enclosing the fourth angle, in the cells of those
597
This will be a general method for filling in any bordered square of odd order: we
shall be able to place the numbers taken in their natural sequence directly, without
any computation or choice by trial and error. Starting from a corner cell, we alternate
the numbers in its adjacent side rows until we reach a middle cell, where we put the
next number. Those following are written, respectively, in the upper corner of the
initial row, in the middle cell of the (contiguous) opposite row, in its (still available,
the other being for a complement) corner, from which we start alternating as before,
with the even numbers in the row opposite the one containing even numbers. This will
come to an end when half the border cells have been filled, the other half being for
the complements. The inner borders are filled in the same way with the subsequent
numbers.
598
The cells filled in successively may be adjacent or not; if not, corner and middle
cells must be excepted. Note that ‘half the cells’, here and below, means in fact m−3 2
for a border of order m, thus the number of cells between a corner and a medial cell.
599
Provided that we began with 1 in line AB (that connecting the first and the second
corners according to their subsequent enumeration).
(inner) rows which have no number at their other extremity, until, here
again, half of the cells of these two (lateral) rows are filled. It appears
again that the odd (numbers) are in one of the two rows, namely that
facing the row which contains odd, and the even in the next row. We
shall have thus arranged the numbers in half the cells of the border of
the square. So when we have put the associate of each of the numbers
(already) placed in this border in the opposite cell of the same (vertical
and horizontal) row, according to what we have explained for the other
squares600 , the whole border will be filled with numbers.
D T. B
H
. Z
G E A
Fig. b 35
We must (next) put the numbers following them, taken in natural
order, in the border adjoining that in which we have finished arranging
the numbers, in the way described, (then proceed further with the next)
until we reach the central cell, whereupon our aim will be achieved.601
(Even-order squares)
(B.18) Magic arrangement in even squares larger than 4.

(i) It has been said before that magic arrangement in other even squares is
made easier after completing the arrangement in the square of 4 because of
its location within the other even squares, just as the presence of a square
of 3 within the other odd squares made easier our magic arrangement for
600
B.15–B.16.
601
A passage (not appearing in these extracts —ed., p. 159, lines 429–433, 442–444)
informs us that for each border 3, 5, 7, 9 the author gave three further examples
of configurations, and also two squares of order 11, one constructed as here and the
other combining various inner configurations. There is indeed space for it left in the
manuscript. Note that his examples for order 3 must have involved inversions or
rotations since only one configuration is possible.
the (higher-order) odd squares.602 The method for even (order squares)
is similar to the method for odd (ones).603
(ii) If we wish (to do) that, we subtract 7 from half the square of the
number considered. The remainder will be the first of the numbers to
appear in the central square, that with side 4.604 Next, we arrange in it
sixteen numbers, beginning with the number left (from the subtraction)
and (taking) the numbers in succession until we fill the square. Next, we
arrange the numbers in the border surrounding this square, (and so on)
until we are done with all the borders.
(iii) Let the first of the squares in which we wish to place the numbers be
that of 6. We want to arrange in it the numbers in such a way that the
quantities appearing in each of the rows are all the same. We subtract 7
from half the square of 6, thus from 18; the remainder is 11. We put it,
as well as the subsequent numbers, in the square of 4 within the (square
of) 6, until all the cells are filled with (these) sixteen numbers, the first
of which is 11 and the last 26, according to what is in this figure (Fig. b
36).605
14 24 25 11
19 17 16 22
15 21 20 18
26 12 13 23
Fig. b 36
(iv) Next, we set out the remaining numbers in two conjugate lines in the
customary way, as in this illustration:
1 2 3 4 5 6 7 8 9 10
36 35 34 33 32 31 30 29 28 27.
We place two numbers of one of the two lines606 in two consecutive corners
and put their associates in the two opposite corners. (Next) we look for
four numbers such that after putting them in one of the two rows in
the border607 there appears, together with what is at its extremities, the
magic sum, namely 111. Looking for that, we shall find many numbers
602
B.12.
603
Meaning the method of selecting the numbers, not that of placing them.
604 2 2
The numbers to be placed in the central 4 × 4 square are n2 − 7, . . . , n2 + 8.
605
No figure preserved, so we take that of b18.
606
Thus either two smaller or two larger numbers.
607
That with the two numbers just placed or its opposite.
the placing of which satisfies the purpose.608

(v) But we must (first) examine what we shall put in the two corners.609
Among all pairs which may be placed there, we (have chosen to) put in
one of the figures 1 and 2, in the second 9 and 10, in the third 5 and
6. Had we put, instead of them, 1 . . . and 10, it would not suit and
would not (enable us to) complete construction of the border.610 It is
preferable (however) to put in the first corner of this square, among the
associated numbers, that which equals the side of the square in which
we wish to arrange the numbers, and in the second corner the number
preceding it;611 next, we shall seek the numbers to be placed in the inner
cells of the border we consider for placing the numbers. Thus, for the
square considered here, which arises from the multiplication of 6 by 6,
and (where therefore) the number equal to the side is 6, this number will
be 6. After putting it in the first corner and 5 in the second corner and
seeking the numbers to be put in the inner cells, the figure will be as we
have represented in this square (Fig. b 37 and Fig. b 38).612
5 36 34 28 2 6 5 27 9 29 35 6
33 14 24 25 11 4 36 14 24 25 11 1
27 19 17 16 22 10 3 19 17 16 22 34
8 15 21 20 18 29 4 15 21 20 18 33
7 26 12 13 23 30 30 26 12 13 23 7
31 1 3 9 35 32 31 10 28 8 2 32
Fig. b 37 Fig. b 38
(B.19) Magic arrangement in the square of 8.

(i) It is possible to set up the magic arrangement in the square of 8 in
accordance with the ways we have explained previously for odd and even
608
See the list of possibilities in our commentary, pp. 63–64.
609
Considering here the ten smaller numbers.
610
There is obviously a passage missing; indeed, 1, 10 is not impossible (it even offers
six possibilities). Furthermore, the Arabic text (bi-hā and not bi-humā) confirms that
there were more than the two numbers mentioned.
611
Thus, as before (note 593), we shall put in two consecutive corner cells numbers
related to the order m, but here the mth and the (m − 1)th in the list.
612
These are the two possibilities; but the text probably had the first since that is
how the inner square within that of order 8 is arranged later on (B.19iii).
squares, as follows.613
(ii) We subtract 7 from half the square of 8; the remainder is 25. We
place it, together with the subsequent numbers up to 40, in the square of
4 inside the square of 8 by means of one of the methods seen previously
for arranging the numbers in the square of 4, as seen in this figure (Fig.
b 39).614 Then, the sums in this square will be the same.
28 38 39 25
33 31 30 36
29 35 34 32
40 26 27 37
Fig. b 39
(iii) We are (now) to arrange the numbers in the border surrounding this
square in accordance with what we have seen previously for the other
squares.615 We set out ten (pairs of) associated numbers, (the small
ones) beginning with 15 and ending with 24, as follows:
15 16 17 18 19 20 21 22 23 24
50 49 48 47 46 45 44 43 42 41.
We place the sixth of them, thus 20, in any corner of the border around
the square of 4, say the first corner, and the preceding number, thus
19, in the following corner, say in the second corner, and we put their
associates in the opposite corners, just as we did in the previous figure.616
19 50 48 42 16 20
47 28 38 39 25 18
41 33 31 30 36 24
22 29 35 34 32 43
21 40 26 27 37 44
45 15 17 23 49 46
Fig. b 40
613
Meaning: the steps for obtaining bordered squares.
614
In the reconstructed figure, we have chosen the same arrangement as before, thus
that of Fig. b18.
615
That is: list the numbers to be placed; choose the ones for the corner cells; and
complete the sum between them.
616
Fig. b37, for order 6, with the fifth and sixth numbers listed.
Next we seek, among the remaining numbers of the two conjugate lines,
four numbers to complete the row containing 20 and 19, the sum of which
will make, with those two, 195, which is the quantity to be found in each
row of the square of 6 within the square considered. Such are 16, 42, 48,
50; we put their associates, thus 49, 23, 17, 15, in the opposite cells. Next,
we seek four numbers to be placed in the empty cells of the row having
20 and 46 at its extremities, and such as to make, again with those two,
the sum 195. Such are 18, 24, 43, 44; we put their associates, namely
47, 41, 22, 21, in the opposite cells. The square will then be as we have
represented in this figure (Fig. b 40).
7 62 61 60 59 2 1 8
56 19 50 48 42 16 20 9
10 47 28 38 39 25 18 55
11 41 33 31 30 36 24 54
53 22 29 35 34 32 43 12
52 21 40 26 27 37 44 13
14 45 15 17 23 49 46 51
57 3 4 5 6 63 64 58
Fig. b 41
(iv) We must (now) arrange the remainder of the numbers intended for
the square in the border around the square of 6 as we did for the smaller
squares. For that purpose, we write out the set of numbers to be placed,
thus 14 (pairs of) associated numbers, in two conjugate lines as in the
subsequent figure:
1 2 3 4 5 6 7 8 9 10 11 12 13 14
64 63 62 61 60 59 58 57 56 55 54 53 52 51.
We place the eighth of them, thus 8, in any corner we choose, say the first
corner, and the seventh, thus 7, in the second corner. (Then) we seek in
the set of numbers of the two lines six numbers adding up, together with
the two numbers 8 and 7, to 260, which is the quantity to appear in each
row of this square. Such are 1, 2, 59, 60, 61, 62. With these placed in the
empty cells of the first horizontal row, and their associates, namely 3, 4,
5, 6, 63, 64, each in the cell opposite to (that of) its associate, the inner
vertical rows of this square will each make 260. After searching among
the remaining numbers of the two conjugate lines for six numbers to be
written in the empty cells of the first vertical row such that their sum,
240 Particular cases of odd-order squares
together with what is at its extremities, makes 260, and having put their
associates in the opposite cells, we shall obtain a figure like this (Fig. b
41).
(v) With what we have shown here one will come to realize the multitude
of ways permitting to find the magic arrangement in this square and in
larger ones.617
(Particular cases of odd-order squares)

(B.20) Curiosities presented by the magic arrangements in odd squares.
(i) You are to know that this kind of magic square is (indeed) used by
mathematicians, and that it also relies on principles of which we shall
make use and to which we shall resort.618
(Squares with separation by parity)

(ii) Such is the placement of all odd numbers in the centre of the square
and of the even ones in its corners. The square will then look as if had
been constructed inside it a (second) square, with its corners in the middle
of the sides of (the first), occupied by all the odd (numbers) whereas the
even ones are in the four corners, as we have represented it in this figure
(Fig. b 42).
ev
en
en
ev
odd
ev
en
en
ev
Fig. b 42
(B.21) Furthermore, the squares which appear within the (larger) square
are of two kinds.
617
Considering just the possibilities with the first ten numbers in the border of order
8, there are already 128 possibilities (listed in our commentary, pp. 57–58, of which
the present one is the 92nd).
618
Two further arrangements for odd squares will now be presented: one where the
bordered squares display a separation of the numbers by parity (B.20ii–B.23), and the
other concerning composite squares (B.24). Their general use in the 10th century is
attested by the author’s assertion ‘used by mathematicians’.
(i) One is the square with its corners placed in the middle of the sides
of the larger square, with an odd number of cells, and in the cells of
which will be found the odd numbers.619 When we consider the rows
of such (squares), which are (therefore) parallel to the sides having their
extremities placed in the middle of the sides of the large(r) square, we
find that the lateral ones equal the consecutive numbers taken in natural
order beginning with 2. Indeed, one sees that the side of the (oblique)
square within the square of 3 is 2, that the side of the square within the
square of 5 is 3, that the side of the square within the square of 7 is 4,
that the side of the square within the square of 9 is 5, and so on following
the succession of the numbers.620 Therefore, when we wish to know the
side of such an oblique square, we add 1 to the side of the (main) square
and take half the result; this will give the side of the oblique square. As
for their (inner) rows, they are unequal: the (number of) cells in a row
may be equal to the side and less than it by one cell.
(ii) The second kind (of square included in the large square) is the square
in its centre, which follows the model explained (previously) for odd and
even squares.621 But each square (of this kind) appears (inside the main
squares) according to the succession of consecutive odd numbers for a
couple of squares.622 Indeed, in the square of 5 and in the square of 7
appears one and the same square, which is the square of 3; in the square
of 9 and in the square of 11 appears one and the same square, which is
the square of 5; likewise, in the square of 13 and the square of 15 appears
inside the square of 7.
(iii) When we wish to arrange in these (latter squares) the numbers (we
proceed as follows). We put in the central cell of the square the median
number. Next, we subtract 8 from it, and we assign to the result the place
of 1 in the square of nine (cells).623 We place the subsequent consecutive
odd numbers in the square of nine (cells) until its cells are filled, using
619
In an odd-order square, there is one more odd number, and indeed the quantity
of cells found in this rhomb (‘oblique square’, as the text calls it below) exceeds that
of the four corner triangles by one.
620
For an order n = 2k + 1 of the main square, the side rows of the rhomb comprise
k + 1 cells and the others, as will be stated, alternately k and k + 1.
621
It will just be a bordered square (but filled with odd numbers only).
622
The largest square inside the rhomb has the side 2t + 1 for both n = 4t + 1 and
n = 4t + 3.
623
Since the inner square of 3 will contain only odd numbers, we subtract 8 from
the median (instead of 4 in B.15i). The arrangement seen before applies whether the
numbers are consecutive or in some other arithmetical progression.
(one or other of) the methods we have explained previously.624 Next, we

subtract 16 from the number we have put in the place of 1,625 and assign
to the result the place of 1 in the square of 5, and we arrange it with the
subsequent odd (numbers) up to the number we have put (before) in the
place of 1, each with its associate, in the cells of the border around the
square of 3 according to what we have explained for the square of 5, until
the border is entirely filled.626 Next, we subtract 24 from the number
which we have put secondly in the place of 1, and again assign to the
result the place of 1 in the square of 7. Then we go on subtracting, from
the numbers remaining (successively),627 quantities each time increased
by eight (and place the numbers) until the square appearing in the middle,
having its side reaching the side of the oblique square —between two of
the cells of the side of the oblique square— is filled.628
(iv) Having attained this situation, we shall put all even numbers, as
well as the remaining odd numbers starting with 1 up to the number
reached in placing, each with its associates, in two lines, according to the
customary way. Next, we shall equalize the remaining cells according to
the model explained hereafter.
(B.22) (i) Case of the square of 3. The odd (numbers) appear naturally
in the centre of the square and there is in each of its corners an even
number.629
7 21 11
17 13 9
15 5 19
Fig. b 43
(ii) Case of the square of 5. We put the median number, thus 13, in the
central cell and subtract 8 from it, which leaves 5. We place it, as well
624
B.3–B.5.
625
The place of 1 in the square of order 3 just completed.
626
How to construct a bordered square of order 5 is known from B.15.
627
‘remaining’ from the successive subtractions of multiples of 8; they are the numbers
to occupy the position of 1 in the successive borders.
628
The explanation between dashes seems to be a later addition. Indeed, ‘between
two of the cells of the side’ is unclear, see Fig. b44 and Fig. b47.
629
As settled above, B.3ii.
as the subsequent odd (numbers) up to the median number, as well as

their associates, in the square of 3 —thus of nine (cells)— which is in the
centre, according to the arrangement of its numbers already explained,
in accordance with what we have represented in this figure (Fig. b 43).
It is then easy for us to arrange the remaining numbers. Indeed, we
place the even numbers, which range from 2 to 24 —the last of the even
numbers due to appear in this square— in two conjugate lines, according
to the customary way, adding to them the two (pairs of) odd numbers
remaining from the set of odd numbers, according to this model:
2 4 6 8 10 12 1 3
24 22 20 18 16 14 25 23.
We put 1 in any middle cell of the border, 3 in the next, and their as-
sociates in the two opposite cells. Next, we put in the first and second
corners 2 and 4, or 2 and 6, or 2 and 8, or 2 and 12, and we write their
associates in the opposite corners.630 We seek, in the two conjugate lines,
two numbers permitting to complete the row between the first and the
second corners, and two numbers permitting to complete the row between
the first and the third corners, and we fill the remaining, opposite cells
with their associates. We shall (then) obtain numerous figures, which
would lengthen the book if we were to report them; among all these,
there is the following (Fig. b 44).631
4 18 25 16 2
20 7 21 11 6
3 17 13 9 23
14 15 5 19 12
24 8 1 10 22
Fig. b 44
(iii) Case of the square of 7. We put the median number, thus 25, in the
central cell. We subtract 8 from it; this leaves 17. We put it, together
with the subsequent odd (numbers) until reaching 25, in the square of 3
in the centre, and we place their associated odd (numbers) so as to fill
the nine cells according to this figure (Fig. b 45).632
630
With 2 in a corner, the occupant of the next corner must indeed be one of these
four numbers. But there are other possibilities (see list in the Commentary, p. 103).
631
Such was most probably the figure in the original text: it is found in A and in the
fragment C.1 below, which seems here to rely on the present text.
632
Thus the largest square within the rhomb is filled.
19 33 23
29 25 21
27 17 31
Fig. b 45
(Next) we place the remaining odd numbers in two conjugate lines
according to the customary way, as in this figure:
1 3 5 7 9 11 13 15
49 47 45 43 41 39 37 35
and (also), in two conjugate lines, the even ones, namely
2 4 6 8 10 12 14 16 18 20 22 24
48 46 44 42 40 38 36 34 32 30 28 26.
We are (now) to seek the numbers to be put in the border around the
square of 3, namely sixteen numbers: four even numbers, which will ap-
pear in the corners since they belong to the corner triangles, and twelve
odd numbers. We put in two consecutive corners two even numbers, say
24 and 22, and place their associates in the two opposite corners.633 Next,
22 3 39 37 24
41 19 33 23 9
1 29 25 21 49
35 27 17 31 15
26 47 11 13 28
Fig. b 46
we seek among the odd numbers three numbers to be put in the three
intermediary cells of the row of 24 and 22, the sum of which, together
633
Thus the largest pair of (smaller) even numbers is placed in the corner cells of the
first border.
with 24 and 22, make 125; indeed the numbers in each row of this square,
which is 5 by 5, must make a sum equal to the content of (each of) the
two diagonals. We likewise seek three numbers to be put in the three cells
between 24 and 28. We find that the numbers to be put in the row of 24
and 22 are 37, 39, 3, and we find that the numbers to be put in the row
of 24 and 28 are 9, 49, 15. We place opposite to each cell (already filled)
the corresponding associate among the odd numbers. The result will be
what we have represented in this square (Fig. b 46).
36 42 6 5 4 48 34
40 22 3 39 37 24 10
38 41 19 33 23 9 12
7 1 29 25 21 49 43
20 35 27 17 31 15 30
18 26 47 11 13 28 32
16 8 44 45 46 2 14
Fig. b 47
There remain four odd numbers, 5, 7, 45, 43, and twenty even, which
(all) must be placed in the border around the square of 5, which is the
last border. We put the first two of the four odd numbers in two (middle,
consecutive) cells and their associates in the opposite cells, thus complet-
ing the filling of the oblique square in the centre, intended for the odd
numbers. Next, we seek six numbers, adding up to 170, to be put in the
first row, that is, the row containing 5. Finding that is easy for us: tak-
ing three pairs of numbers with their sums making tens —(thus) without
units—634 and placing them on both sides of 5 so as to fill the cells of the
first row, we shall have reached our objective. Seeking this, we find 4 and
6, which we put on either side of 5, then 48 and 42, which we put after 6
and 4, then 34 and 36, which we put in the two corners. Thus the sum
of the content of the row of 5 is 175. After having placed their associated
numbers in the opposite cells, two rows of the border will be filled, and
eight even numbers remain, which are the following
10 12 18 20
40 38 32 30.
We seek among them four numbers with sum 84, to be placed in (the)
four cells of the first vertical row, so as to complete what the magic sum
634
This is a futile simplification.
requires, namely 175. We find 10, 12, 30, 32. Putting them in the first
row, and putting the associated ones in the opposite cells, their aspect will
be as in this square (Fig. b 47), where the quantities in each row vertically,
horizontally and diagonally give the same sum, namely 175, and where
the odd (numbers) appear in the centre, in the oblique square, and the
even in the corner triangles, according to what is in this figure.
(iv) It is possible, following this model, to place the numbers in squares
other than the above ones for those who have acquired practice of this
kind.
(B.23) General arrangement of the magic square in odd squares for this
elegant kind, including other (squares larger) than the preceding ones.635
(i) It is possible to reach more rapidly this type of magic arrangement
using the method without displacement which we have explained in the
first chapter of this book.636 Indeed we shall use, for arranging the num-
bers in the cells of the odd squares, the method we have followed there for
consecutive numbers beginning with 1, under the following restrictions:
— We shall (first) arrange the odd numbers in the square in the middle of
the main square according to the arrangement seen in the previous case.
— We shall (for the remaining part of the square) make a distinction
between the cells allotted to even (numbers) and the cells allotted to odd
(ones). Then, when we attain a cell allotted to an odd and the number
reached is an odd number, we shall put it there; when we attain a cell
allotted to an even and the number reached is even, we shall put it there;
when we attain a cell allotted to an even or to an odd and the number we
have reached is of the other type, we shall disregard the number and the
cell altogether, without writing anything; when we attain a cell allotted
to an odd and we reach an odd number already written in the square
in the centre, it will not be put there since it already occupies another
place among the cells, and we shall disregard the number and the cell
altogether. We shall continue doing this until we reach the first of the
numbers placed in the square in the middle.637
At this point, we shall arrange the still available odd and even numbers
in the remaining cells, until all cells are filled.
635
In fact, the way to be explained now is a rather poor aid in comparison with the
method, avoiding trial and error, described by A.
636
B.16–B.17, indeed in the first chapter of the book (on the construction of magic
squares of odd order).
637
But this limit can only apply to the odd numbers: even ones may still be placed.
(ii) Example. We wish to set out the magic arrangement in a square of 9

with the odd numbers in the oblique square in the centre and the even in
the corner triangles. We begin by putting the median number, thus 41,
in the central cell. Next, we arrange the odd numbers to appear in the
square of 5 within the square of 9, so as to fill all its cells, as explained
previously.638 Next, we begin at one of the corner cells, (as if we were) to
place 1 in the cell next to it; since this cell is allotted to an even number,
we put nothing in it. We put 2 in the cell placed diagonally to it in the
second row.639 We do not put anything in the cell following the first cell
in the first row,640 for it is allotted to an even number. We put 4 in the
third cell of the second row. We do not put anything in the fourth cell of
the first row, for it is allotted to an even number. We put 6 in the fourth
cell of the second row. We put 7 in the fifth cell of the second row, for it
belongs to the middle row of the square,641 (then) 8 in the first corner, 9
in the middle cell of the ninth row,642 and 10 in the second corner. We
continue filling the cells by means of this process as we have explained for
the previously seen figure, about magic arrangement in odd squares,643
until we are finished with the first border.644 Next we turn to the second
border, and we operate in it in the explained manner. Then we complete
the opposite cells by means of the equalizing number. The result will be
according to this illustration (Fig. b 48).
10 16 14 12 75 76 78 80 8
24 28 64 22
27 31 61 63 23
29 35 49 39 53
9 25 45 41 37 57 73
65 43 33 47 17
59 51 21 19 55
60 54 18 58
74 66 68 70 7 6 4 2 72
Fig. b 48
638
B.15, Fig. b33.
639
Here the second row is the bottom one.
640
Here the first row is the first column.
641
And thus belongs to the rhomb.
642
Here the ninth row is the ninth column.
643
See Fig. b34.
644
Here the first border is the outer one (A numbers the borders from within).
Next, we seek among the odd numbers three numbers to be placed in

the row of 22 and 24 making the sum 149, so that adding the numbers
in this row will make 287, which is the quantity to be found in each row
of this square, in accordance with the content of the diagonals. Looking
for these numbers, we shall find that a large quantity of numbers may be
placed in these cells. Among all these we (choose to) put 79, 69, 1. Next,
we seek among the remaining numbers five numbers making the sum 207,
two even and three odd, such that after writing them in the row of 80
and 2, its sum, with the numbers it (already) contains, will be 369, which
is the required quantity. The odd numbers are then 77, 15, 11, and the
even ones 42, 62. Once these are placed, and with their complements in
the opposite cells, the contents of the second and eighth (vertical) rows
will make the magic sum. Next, we seek among the remaining numbers,
that is, twelve even numbers, six numbers to place in the first vertical row
adding up to 216, so that, after adding to them the numbers (already)
there, the sum will be in agreement with the magic sum. We then find,
among the even (numbers), 26, 30, 32, 36, 44, 48. Putting them in the
row of 8 and 72, and placing their associates in the opposite cells, the
sum of the numbers in each row of this square will be the same, namely
369 (Fig. b 49).
10 16 14 12 75 76 78 80 8
56 24 28 1 69 79 64 22 26
52 40 27 31 61 63 23 42 30
50 5 29 35 49 39 53 77 32
9 67 25 45 41 37 57 15 73
46 71 65 43 33 47 17 11 36
38 20 59 51 21 19 55 62 44
34 60 54 81 13 3 18 58 48
74 66 68 70 7 6 4 2 72
Fig. b 49
(Composite squares)
(Odd-order squares)
(B.24) Another way.
(i) It is in fact possible to arrange the numbers in the square of 9 in an-

other elegant way, which consists in dividing this square into nine squares
such that the quantity in the rows of each of them (considered separately)
is the same.
To do that, we divide the square into nine squares in which we place
the numbers from 1 to 9 according to the arrangement we used for the
square of 3 so that each row of it was making 15 (Fig. b 50).
2 9 4
7 5 3
6 1 8
Fig. b 50
Next, we arrange in the square where 1 occurs, which has nine cells,
the first set of nine, from 1 to 9; (then,) in the square where 2 occurs, the
second set of nine, from 10 to 18; (then,) in the third square, the third
set of nine, from 19 to 27. We proceed in this way until we have finished
with all the numbers to be found in the square of 9. Then the quantities
in each of its rows will be the same. Here is the figure of it (Fig. b 51).645
11 18 13 74 81 76 29 36 31
16 14 12 79 77 75 34 32 30
15 10 17 78 73 80 33 28 35
56 63 58 38 45 40 20 27 22
61 59 57 43 41 39 25 23 21
60 55 62 42 37 44 24 19 26
47 54 49 2 9 4 65 72 67
52 50 48 7 5 3 70 68 66
51 46 53 6 1 8 69 64 71
Fig. b 51
(ii) It is of course possible to do that for any odd square having the factor
3, such as the square of 15 and the square of 21. Indeed, the square of 15
is divided into twenty-five squares of 3 by 3 and (also) into nine squares
645
In such a composite magic square the smaller squares, taken in magic arrangement,
are filled each with a continuous sequence of numbers arranged magically. The whole
square is magic, as are the subsquares, each with its own constant sum.
250 Particular cases of even-order squares
of 5 by 5; the square of 21 is divided into forty-nine squares of 3 by 3 and

(also) into nine squares of 7 by 7. It is (thus) possible to set out the magic
arrangement in all (these squares) in the way explained previously.646
(Edition, p. 183)
(Particular cases of even-order squares)

(Composite squares with different sums in the subsquares)
(B.25) Curious and elegant cases presented by the magic arrangement in

even squares.647
(i) You are to know that even numbers are divided into two kinds; one
is (that of the even numbers) divisible by 4, which are the ‘evenly even’
and the ‘evenly evenly odd’, the other (that of the even numbers) not
divisible by 4, which are the ‘evenly odd’.648
A number divisible by 4 will have its square divisible by a square
number containing as many units as the square of 4, thus 16. Therefore
any number divisible by 4 will have its square divided up into squares in
two manners: one time according to the quantity of the square of 4, thus
in sixteen squares, another time according to the square quantity of the
number dividing the square in question sixteen times.
Example(s) of that. 12 is a number that 4 divides three times; its
square, 144, will thus be divided into sixteen squares of nine cells, and it
will also be divided into nine squares of sixteen cells. The same holds for
20, for 4 divides it five times; its square, 400, will (then) be divided into
sixteen squares of twenty-five (cells), and it will (also) be divided into
twenty-five squares of sixteen (cells).
(ii) The situation being as we have described, if we wish to place the
numbers in a square with its side having the factor 4, we shall divide
the square into one of the two kinds of square, and we shall arrange in
these squares the natural sequence of the numbers from 1 according to
the arrangement seen previously for odd and even squares.649 Next, we
646
B.3–B.6, B.15–B.17.
647
Here (B.25) we are taught a construction analogous to the previous one but for even
orders divisible by 4 , with the subsquares, arranged magically, receiving continuous
sequences of numbers arranged magically in each of them. (In B.26, we shall see a
construction in which all subsquares display the same magic sum and can therefore be
arranged as we wish.)
648
For these denominations, see note 300. The last kind, with order 2 (2k + 1), is not
considered here.
649
Numbering the subsquares according to a magic arrangement.
shall arrange, in the first of these squares, numbers equal (in quantity)
to the number of its cells, beginning with 1 and ending with the number
of cells. Then we shall arrange, in the second square, numbers equal (in
quantity) to the number of its cells, beginning with the number which
we have reached in the first square. We shall arrange in this way the
numbers step by step until we have filled all cells of the (large) square.
We shall thus obtain the magic arrangement in all (small) squares as well
as in the large one.
(iii) Example of that, for (the order) 12. Since 4 divides it three times,
its square will be divided into nine squares of sixteen cells. If we wish to
arrange in it the magic sums, we (first) carry out its division into nine
squares of sixteen cells, and we arrange in these squares the numbers from
1 to 9, so that they will be as in this figure (Fig. b 52).
2 9 4 8 11 14 1
10 5 4 15
7 5 3
3 16 9 6
6 1 8 13 2 7 12
Fig. b 52 Fig. b 53
Next we arrange the numbers from 1 to 16 in the first square, from
17 to 32 in the second square, from 33 to 48 in the third square, and we
likewise arrange in each subsequent square sixteen consecutive numbers
until we have finished with all the numbers to appear in the square of 12.
Then all the numbers which are in this square will make the magic sum
vertically, horizontally and diagonally.650
(iv) Now if this square is divided into sixteen squares and the successive
numbers taken in natural order are placed in them according to the ar-
rangement we have explained for the square of 4 (Fig. b 53651 ), then in
each square (of order 3) its sequence of numbers is placed successively,
the quantities appearing in them will make the magic sum.
(Composite squares with same sums in the subsquares)

650
As before (B.24), each subsquare is magic, and their sums form an arithmetical
progression (whence the need for them to be in magic arrangement).
651
Added by us, on the model of b23: there is neither reference to it in the text nor
space left blank in the manuscript.
(B.26) (i) It is possible to set up the magic arrangement in squares

with sides having a fourth in a manner more elegant than that we have
explained. It consists in dividing the square (considered) into a (certain)
number of squares in all rows of which the sums will be uniformly the
same.
(ii) If we wish that, we divide the square by 16.652 The quotient will be
the number of squares in which the considered square is to be divided,
each of these squares having 16 cells. We arrange in half (of the cells) of
one of these squares, whichever it is, the consecutive numbers taken in
natural order from 1 to 8 following one of the arrangements established
in the second chapter for the square of 4,653 and we complete them by
means of the equalizing number for the (large) square. We move then
to any other square and arrange in half of its (cells) the numbers from
9 to 16, and we complete them by means of the equalizing number. We
proceed in this way until we have finished with all its squares. At this
point, the quantities in all the rows of the (large) square make the same
sum, while the quantities in all the rows of each (small) square are also
the same.654
(iii) Example of that, for (the square of) 144 (cells). Its division by 16
gives the quotient 9. So if we divide it into nine squares, they will have
each sixteen cells. Beginning with arranging the numbers from 1 to 8 in a
first square and arranging their associated numbers in the remaining cells,
(each time) in the bishop’s cell, as we have seen in the second chapter,655
and arranging (likewise) in each of its squares sixteen associated numbers,
the (large) square will be magic.
(iv) Example of this in the square of 8, of 64 cells. Its division by 16 gives
4. So if we divide it into four squares, each will have sixteen cells, where
we shall arrange the numbers as prescribed.656
652
Dividing the square of the order.
653
The second chapter (our B.7–B.11, B.18–B.19) is on the construction of even-order
magic squares. Here only B.9 will be used.
654
If the order n is divisible by 4, we can divide the whole square into 4 × 4 squares,
in which we shall put a sequence of eight consecutive smaller numbers, the other cells
being filled with their complements to n2 + 1; this makes the constant sum in each
square the same —namely 2 (n2 + 1).
655
Our B.9.
656
No space for a figure in the manuscript.
Opuscule on the magic square by al-Kharaqı̄ 657
(Extract)
(C.1) On magic arrangement with separation and borders.658

(i) In this kind of magic arrangement the odd (numbers) are in the centre,
separated from the even —taking into account the magic arrangement.
This kind can be carried out only for odd squares.
(ii) If we wish that, we take an odd(-order) square and draw inside it an
oblique square with its corners falling in the middle of the sides of the odd
square. Next, we divide the odd square into its parts and determine the
equalizing number, which we divide by two; we put this half, which is the
median number, in the central cell. Next, we take away from the median
number eight numbers, taken in reverse order of succession; we put the
number reached in the place of 1 for the square of 3 and count, including
it, nine consecutive odd (numbers), including the median number, and
arrange them659 in the square of three in the known way.660 Next, we
subtract, from the number we have put in the place of 1, 16. We put
the number reached in the place of 1 for the square of 5, and count,
including it, the odd (numbers) up to the number which we have put in
the place of 1 before, and count, on the upper side, as many numbers,
starting with that we have reached before; we arrange them in the rows
surrounding the square of three. Next we subtract, from the number we
had put in the place of 1, 24; we take the odd (numbers), starting with
the number reached, including it, together with their associates on the
other side, for the square of seven. We proceed in this way, by successive
increments of 8, and arrange (the odd numbers) in (the borders of) their
squares.661 Next, we place the odd (numbers) remaining on both sides662
in the corners of the oblique square, and arrange the even numbers in and
657
Al-Kharaqı̄’s (d. 1138/9) treatise has been edited and analyzed by us (Herstel-
lungsverfahren, III ).
658
Bordered square with separation by parity as seen in B.20–B.23.
659
The median number has already been placed. Same remark for the example.
660
In al-Kharaqı̄’s treatise, this chapter follows the explanation on how to construct
ordinary and bordered magic squares.
661
Filling the square within the rhomb, as seen in B.21iii.
662
Still unplaced are the smallest and (thus) the largest odd numbers.
254 Opuscule on the magic square by al-Kharaqı̄
next to the corners of the large square, as far as is possible,663 until the
(large) square is filled, with its sums being equal.
(iii) Let us illustrate our explanations by means of the square of five. We
put the median number in the central cell, thus 13. Next we subtract
8 from 13; we reach 5. We count, including it, nine consecutive odd
numbers, including the median, and arrange them in the square of three
within the square of five, beginning with 5 and ending with 21. The square
then has the following aspect.664 Next, we take the even (numbers) from
2 to 24, which are (all) the even of the square (considered), and set them
out in two conjugate lines, and add next to the even terms the odd ones
left from both sides, as in this figure.665 Then we place the remaining odd
in the corners of the oblique square: we place the first two, thus 1 and 3,
in two consecutive corners, and, opposite, their associates, thus 25 and 23.
The square will then take the following aspect.666 Next we place in the
two corners of the first horizontal row 2 and 4, or 2 and 6, or 2 and 8, or 2
and 12.667 Let us stipulate, for our example, that it is 2 and 4. We place
them then in the corner cells of the first horizontal row, and put their
associates facing them (diagonally). Next we seek in the two (conjugate)
lines, among the associated even numbers, two numbers enabling us to
complete, together with the three numbers in the first horizontal row, 65,
which is in our example the quantity to arrange in each row in order to
complete the magic square. We then find 18 and 16. We place them on
both sides of 25, and put their corresponding (complements) opposite.
We seek (now) two other numbers to place in the first vertical row in
order to complete 65. We find 12 and 6, which we place on both sides of
23, and we put the corresponding (complements) opposite. The square
will then have the following aspect (Fig. c 1).
(iv) We trust that for the able student this present section will suffice.
We have supplied in our other writing about the magic square additional
information on this example.668
663
Thus this early 12th-century author did not know about the existence of a method
to avoid trial and error.
664
Presumably as in our figure c1 (missing in the manuscript; indeed, the manuscript
and the copyist are the same as for the text B).
665
Two conjugate rows as in B.22ii (figure missing in the manuscript).
666
It would seem unnecessary to add a figure with just four more numbers, but the
manuscript does indeed leave space here for a figure.
667
This is just what we had in B.22ii. What follows is an accurate enough description
for us to be able to reconstruct the missing figure there.
668
This text not being extant (or not yet discovered) we cannot know what such
an addition might have been —perhaps discussing the other possibilities (‘numerous
Translation of Text C 255
4 18 25 16 2
20 7 21 11 6
3 17 13 9 23
14 15 5 19 12
24 8 1 10 22
Fig. c 1
figures’) alluded to in B.22ii.

Part III
Arabic texts
Editorial procedure
For A, of which this is the first edition, we have filled the lacunas, the
missing parts being then inserted in angular brackets. Former glosses,
now incorporated in the manuscript’s text, are in square brackets. We
have also recorded, in the critical notes, any variants or errors of the
manuscript. None of this applies to the fragments reproduced here from
B since the whole text has already been edited; consequently, lacunas
have just been filled and glosses removed.
As usual in early Arabic mathematical treatises, numerals were origi-
nally expressed verbally in the text and symbolically in figures and tables,
by means of either Arabic numerals or letters (alphabetical numerals, see
Fig. 5, p. 8). In the text of B, the verbal form is still mostly found; in
A, however, copyists have adopted numerical symbols, leaving the verbal
form either inadvertently or when (though not systematically) a prepo-
sition accompanied the numeral. In the edited text, we have throughout
reproduced what appears in the two manuscripts.
In A’s figures Arabic alphabetical numerals were originally used; in
the manuscript, the first six figures use numerical symbols, doubtless
a change attributable to copyists. Here too, we have reproduced what
appears in the manuscript. These figures have been inserted where they
appear in the manuscript (and are therefore left unnumbered by us). For
text B, the figures had to be reconstructed since the copyist has left
only blank spaces; the few cases where reconstruction is uncertain are
mentioned at the corresponding point in the translation. In these figures,
we have throughout used Arabic numerals, but have added some figures
with alphabetical numerals in those cases where the text describes them
this way. These figures have been numbered according to the fragment
they belong to. (Note that this numbering does not correspond to that
258
in the translation since there a few figures have been added.)

{1r }

% '* '
!"#$ &() +, !-. /01 +2
!3 45/6)7 8 9:29 4, "
&;
%

<4=>
, /?

Arabic text of A 261
'
@34:/ +A B/ C @3D$ {1v } 5

'* B '
E !" #$ &() +, &A +, !-. /01 +2
< 4=>
, /?
!3 5/6)7 8 9:29
4 ,
4
F4 G4"/H !I F?, &. 9: !I
&. JKL?,
M, 9, <N 10
'
5 &?, 69: JO JK, 4 , &. /H5 !I
52 9
4,%/:
JO JO '4MP !I
, 9N Q I9

, &. @-. !I !R 9
'
&. @-. !I
/I
&. @-. !I !R

15
/(; /S 5 !I G

&. +2 9
8 (et 3) !3 45/6)7 8 !3 , /6)7 8 cod.

˜
12 4,%/:
JO 4 /:
˜ JO, cod.

+2 '
9
' '
<N T I &B9 &. N &?,4 !3 <N , 4 &. (A.I.1)
U '
;$ 4 V,% <(W$
'
20 (N&B + H4 , /H4. &.
&. !I XY4 (A.I.2)
&B9 Z2 ?U AB N &?, ' &. <N " /HW &. (A.I.3)
4 4 !3 [
/: 6
W2 &.
' 4 ', I
&. VP '

&. V\
W$ (A.I.4)
U ' '
B
9E, 4 Z(S I /I &?, ].9^
&. VP (A.I.5)

9E,
/ 4 (A.I.5)
25

4 ', I
&?, ].9 ^
&. VP '
/
9E, (A.I.6)
U ' '
/I 9E, 4 Z(
S I / I &?, ].9 ^ /I &. VP (A.I.7)
/I <(W$ '
/ 4 (A.I.7)
U ' _ '
B
9E, 4 ! I I
&. 2 V; I
&. VP (A.I.8)
U ' V;I_ (A.I.9)
30 /I 9E, 4 !I I /I &.
&. +2
U ' V;I_ (A.I.10)
B
9E, 4 !I I /I &. /I &. +2
4 !I ' V;I_ (A.I.11)
9E, I
&. /I &. +2
U
{2r } /I
' ' U U ' '

&?, &?,4 , I B
&. &?,4 /I &. VP (A.I.12)
$ &?, ' U ' U U ' (A.I.13)
35 ; 4 ;$4 ,I B
&. &?,4 /I &. P
' ˜
V ,% cod. 20 (N&B &B cod. 21 N &?,4 9E
˜ ˜ ˜
19 , pr. scr. et del. 22
W2
_
cod. 24 /I
cod. 29 V;I V; I saepissime cod.
/HW2
˜ ˜

', I /I ; $ &. VP '
/ (A.I.14)
/
', I /I ; $ ]; $ &. VP '
/ (A.I.15)
' '
/
, I &B9 !` 4;Y , !I !3JKa , 4 &. VP (A.I.16)
U ' '
V, &B9 !` 4;Y , !I !3JKa , 4
/2 ] ; Y,4 &. VP (A.I.17)

/ /
96I /I &. &. , ]; $ 40
' ' ' ' '

b, 4 &?,4 &. /H" 96I W , &?,4 +, 4 &. &?,4 &. VP (A.I.18)
+, 4 &?
' /H" &? &?, ' ' ' '
(N &?,4 & 4 ,I +,4 &. &?,4 &. VP (A.I.19)
' ' ' ' '
M &?,4 &. /H" , I W , &?,4 &. M, &?,4 &. VP (A.I.20)
&? b "
U /HW2 JK4 , 4 !3 &. 45
) 7 8 c 9N 6'MP &N

W( &?, '
&? ' I dc de fc V2
&
4 '
M &. &N &?, '
4 &. /H" !36I JKD, &?,
gU/: 4 2' /?^ ' '
/H" +2 9E, I + H4M ) * +, 4 &. VP (A.I.21)
g/: 2'
+2 gU/: /: [ &. P /: [ &. +2 2 gU/: &. P (A.I.22)
+ H4.9()* + H4'MI ' 7 [
50
(N + H4.9()* +, 4 /H" +2 I g/) 8 b V2 /: &.

, 4 /I +2 &B9 P & g/) 7 8 b
' '

+, 4 /H" +2
c d (N +, 4 &M +H4MI I fe c h d V2
h +2 d 9N & g/) 7 8 b V2 fe h (N +, &M
4
JKD , g/: b &" (A.I.23) 55
G P !R +2 gU/: '

P &. ?,
G ,P
(A.I.24)
!3 g/: g/) 7 8 P G &,' JO' I g/) 7 8 b V2 Z, / +2 gU/:
' ˜
, I e corr. JK4 , 4 JK,
cod. b b &, cod.
˜ ˜ ˜
38 38 /

cod. 45 45
!R 9E , !3 g/: g/) 7 8 V2 G +2 '
+2 9E,
{2v } Z, /
!` M WY" ?,

!` + H , W$ ' I i j h + H ,P
4 ' 4 '
'5 !` ?,
60

WY" M, !` + H , W$ G ,P
G &, W
4
' &B9 g/) 7 8 g/: !I 9 &EN
&?,4 &B9 P (A.I.25)
' U U , &. &?,4 2
&, 4U, , &.
' U
(" V,4 & !I 9E, 4 , I ?,
&. G
Z, / &. !R b &" G &. &B9
_' [ gU/:
g/) 7 8 b V2 (JK2 F , @k8 /: &. +2 &. P (A.I.26)
' ' '
65
V\ V2 !I +2

V\ +2 !I I 4?,
+2 <, (6; $ +, &? +2
4 F ,_ I 6; $ JO' I > ?,
4

P

' U , ?,

!3. ;$4 WY &N !-. <N ?,
4( +2
( +2
4I ?,
];$ (N ?,
+2
+H4a4I +H4 ,
4
70 g/: !I &N &EN (A.I.27)
_ (A.I.28)
W$ G ,P &. (JK2 F I &. b &"
P
!` !I W$ ' I V\ ' '
!I !` V\ WY" 9 !` 9
' '
V\ !` V\ WY"
' '@k8 &. G ,P
JO &. !-. 6 ,
/: &. G ,P
2 I (A.I.29)
' ' + ' [ + '
75 !I JO I + H4 +2 H4, VP W$ !-. +, 4 /: +2 H4, VP P
5 2 9E , W W$

&N !-. de fe c l h + H4 ,P
l h d
de fe c
58 V2 9N cod. 58 g/) 7 8 pr. scr. et corr.
˜ ˜
59 Z, /
˜
add. in marg. 71

b &"
˜
9N add.
˜
74 &. (pr.) &. cod. 75 P ˜ G ,P
cod.
) 7 8 WY" ( !` + H , W$ 9E, '

!` W( 4 4 4 W' W$ !I JO I 80
/0D? !` W( ) 7 8 WY" ?,

!` + H , W$ b +, /0D?
4 '4
/0D? !` ?,
WY" W( ) 7 8 !` + H , W$ G ,P
&, I
' 4

&, I +, 4 /0D? !` /0D? WY" 4( !` ?,
W$ b &"
G ,P &I +, /0D? !` ( WY" /0D? !` ?,
W$ G ,P
4 4
+ H4 , {3r} WYI W()7 8 !` + H4 , WY" /0D? !` ?,
W$

'
85
+
!` H4, W$ G,P &, I +,4 /0D? !` 4( WY" W() 8 !`
7
+, 4 /0D? !` W( ) 7 8 WY" (
4
' ' %
!` !` & !` & & B , 9N V,4 & !3?2 (A.I.30)
!`
' [ '
&I (N&B I /: !I (N&B /0L$ 4 +, 4 &. VP (A.I.31)
[ [ U
90
/: B &?,

/2 ].9 ^
_ ' ' U
/0m +, 4 /0D" VP b ;$ 4 b /)n84
9WE !I (A.I.31)
&B9 +2 /: [ g/:
&, g /: ' ' [
&I (N&B I /: !I (N&B
!-. b ' /0m_ 2' I (A.I.31)
+ H4a4?2 I S; !I
9WE G,
' /0DE
&, o4S; ,%/) n8 2'
&, /0DE ] 4?; $ 2 4I /0L 95
o4S;
' p /0m b',P
W ] 4?; $ V2 9N je \ I i !I c G, b M 2
' _ _ /2' G5 ' 7 /2' XAB
,%/) n8 , I &5 !I XAB V2 W() 8
U + _ _ _ % _
g/: 4 H M, +2 g /: 9N XA B &5 b 9E 4I &W g/: , X() 7 8
&B 9E
U 4I X(_) 7 8, W( ) 7 8 ,%/) n8 2' ', I c !I XAB_ V2
U 7 _
100
&B 9E4I W() 8, X() 7 8 ] 4?; $

' ' ' U p /0m (A.I.32)
o#W !` o#W W$ I +, 4 &. !I &. G,
' b ˜ WY"
81 W$
b&" cod. 82 I cod.
˜
85 WY I W$ cod.
˜
97 9N
˜
partim
eras.
[ +, 4 &? &B WY"

/: !` + H4, /0L
je dh ,9E, 4 (JKS#W2 ' ' ' _
4 [ I 5 !I , /0m WAB ?,
P
105 /: !` (JK2 &B WY" je !` dh WY I

&? ' [ U
' +, 4 &. +2 &B VP P ' (A.I.33)
& I /: &.
[ ' 9N (N&B &?, '
/: &? b &. 4
,I /: [ &. ' (JK2 &B VP
&M ' '
I ?,
MP
U ' '
;$4 M &. (N ?,
&?,4
(N&B G, p /0m /: [ &. &. ' &.
b &"
P (A.I.34)
U' ' I /: [ !I
110
&? b &. (JK2 Z(S o#W \

U ' fd {3v } 9N & (6S#W2 ' ' c h
;$ 4 c &. I &. MP
' ' ' ' (A.I.35)
(JK2 &B VP &. 9N (JK2 &B VP Z, /2 b &"
U
;$ 4 W( )7 8 &.
' ' (W:_ b &" (A.I.36)
115 9N (JK2 &B VP @3D)* +2 Z()n84 2 P
' ' ' ?W Y M /: [ &. '
(JK2 &B VP ?,
&.
W( )7 8 &.
' ' ' ' ' '
&. &?45 2 2 96I &. VP (A.I.37)
' ' ' ' '
&? b &?,4 &. VP &. 96I &. VP (A.I.38)
'
120
' U' [ _ '
&?,4 S#W2 /H4;4I /: !I (N&B /0L $4 +, 4 &. VP (A.I.39)
' ' ' '
+ H4, /0L
+, 4 &? &B &?, 4 &? b I &.
' [ _ '
&? I /: !I (N&B /0L 4$ + 4H, 4 2 +, 4 &. VP (A.I.40)
104 4 ˜ 9E,
,9E, 4 cod. 105 &B ˜ &B VP cod. 108 ,I ˜ ,
cod. 109 ' (N ˜ <N
cod. 110 G, p /0m
˜ G, I ˜ P
/0LI cod. 111 \ cod. 112 (6 #W2 ' ˜
(JK4S#W2 cod. 115
_˜ U
(W: (W: cod.
' '
, &?,4 &. VI 9N (JK2 Z(S
'
W$ I G ,P
@" 5 2 &B9 +2 492 &. G ,P (A.I.41)
V' I
125
JK2 &.
&, /H" &.
'
I G ,P @" 5 2 &B9 +2 492 &. G ,P (A.I.42)
&? &?, ' [ ' '
&B9 !-,4 & 4 J K2 / : p &?, 4 & &?
4( &?, ' ' '
<N !3 4 & &? 'I l h d &B9P
+ H4 , &?,4 , , 9N /H" 130
G ,P
P @" 5 2 &B9 +2 492 &. G ,P (A.I.43)
' ' U'
JK2 &. /H" &?,4 XY4MI &B9 !-,4 &
4 & I &? p (A.I.44)
&. +2 9E, 29M?2 V,% &. G ,P
JK2 &. /H" 9N 2 V,%
b N/,
135
% ' ' /H" &? /S W$ ', (a)

, 9E, 4 : /S 5 6MP N &?,4 &
' ' ' '
b,4 &?,4 & &? @k8 JK2 +,4 &. &?,4 & /H" &? /S W$
& &? '@k8 G &? +, 4 &? ? ' /
H" &? !3. +, 4 &?
/H" &? /S W$ !3: ' U % '
q> &EN Z, / &? &? b &?,4
' '
6MP N &?,4 & 140
' /H" &? 9: 2' I (b)
+ H4a,4 WY > WY4 +, 4 &. &?,4 & '
'
' %
>4 @ +H4 ,P 9 (JO +H4 , WY4 {4r } (JO =/H5 > + H4'
' '
+E
' ' ' U '
2 &. N &?,4 V,% &. &B &. ?4AB (N &?,4
' I + H4a,4 WY2 ,P (JOr ' + '
&?,4 (N&B \ I H4a,4 WY > WY4 (JO
'% 2' I W , &?, ' ' [
+ H4 , WY4 ,P 4 (J
K2 &B VP P /: 145
' ˜
/H" ˜ /H " sæpius cod. 126 JK2 ˜ (JK2 cod. 130 &?, ˜
126 4 &? , cod. 132 &B9 &B 9
I ˜ r
JK2 cod. 138 +, 4 &? +, 4 &?M cod. 143 &. (v. seq.) pr. scr. et del. 144 \
˜ ˜ I
cod.
/H" F ,4_ %, 9E, 4 (N &?, '

4 & /H" &? 9: 9I + Ha, WY2
44

/H" +2
<, 2 F,4 @k8 VI +2 _ ' ' '
VI <,4 !3: VI ' +2
+2
' U U % ' ' ' '
!3: &, q> b V? , [/ , VI +2 VI < ,4 !3: VI
9N b 9E
4I (JO4 WY /: +2 ,;
, +E >4 &. !` !3JKa ,
U ' '
150 ?4AB (N &?,4 &. VI
U p 2 !-. &,4 / , '@k8 (c)
&B ,4 /, Z(: 2 P I &B &B
I JK2 /H" N/H4. /: [ &. Z()n84 ',
P /H) 7 8 Q: ' U ' U
&I &.
' ' ' % '
&? & B , , I "/2 &? b +2U ' Z(S
,I &?,4 &
+E >4 ', &. bM, +2 &B9 W2 /H4. /: [ &. 9E, 4
29M?2 V,% &. &?, ' ' U * U '
155 4 &. VI ) P JK2 &B V'2 9E, 4
V,% &. +2 N/H4. &?, XY4MI
'
!-. &. VI G ,P 5 2 492 &. M, G ,P (A.I.45)
&? &. ' &. (6I JK2 ?( )n84 +, 4 &. VP ' '
G I JKW$
GWY 4MI /: [ &. U' (JK2 &B VP '
P &. (A.I.46)
P
[ '
160 /: &. !` !R WY" !R !` (JK2 W$
' @" 5 2 492 &. G ,P b &" (A.I.47)
VP P G ,P
'
!R !` W$ GWY 4MI /: [ &. U' + H4I/# {4v } &B
+2
[ [
/: &. !` /:p WY"
&? +2 &B9 W$ /0m + H4"/HW2 +, 4 &. VP '
& (A.I.48)
' ' '
Z(S o#W I /: !I (N&B +2 [ 4I P /HW$4
165 VI 9N b +2
'
, &?,4 &.
˜ ' ˜ ˜
146 /H" (post.) VI pr. scr. et supra corr. 149 cod. 149 (JO4 WY (JK4,WY cod.
' ˜ ' ˜ U'
&. cod. 159 cod. 159
˜ ˜
155 pr. scr. et corr. 157 VI cod. 158 &.
[ ˜ [˜
/ : / : &. cod. 160 !R (post.) / : cod. 160 / : / :
˜ ˜
cod. 161 &. &. cod.
˜ &B cod. 165 o#W' ˜
162 &B (JK2 o#W cod.
% + H4 , ' I P

, & B I (JK4I P /HW2 /HW2 (N +, 4 &M h i V2
[
/: !I , /0L I (N&B +2 ]; <N + H , +2
4 &B9 W$

'W &?, ' ' I
?, 4 &. VI <N fd \
' (A.I.49)
b, I G ,P 2 &. ?,
&. M, G ,P
b &"
' ' %
170
, V? I JK2 &.

&EN &?,4 &. VI & B ,

U ' ' ' U '
;$ 4 9N , &?,4 &. VI I &. &?,4 &. (A.I.50)
P
'
&? b &?,4
' ' ' '
&?M <s1 g/: &? &? !I I &. &?,4 &. VP (A.I.51) '
? 175
' @" &B W$ !-. 492 &. G ,P (A.I.52)

VP P G ,P
' ' [ U ' +
JK W$ !-. &. VI JO I /: &. H4I/# +2 &B
'
&. VI G ,P W$ !-. 492 &. M, G ,P (A.I.53)
I &. ?,
G ,P ' + '
I/#

? ,
/2 H
4I/# I JKW$ !-.
'
?E2
180
2 2 _ I &. WA B G ,P

G ,P 2 2 I/# (A.I.53)
5 G ,P _ I &. '5
&. ? G ?" 2 G _?" 2 I/#
l h d &B9P qU >% &EN G?" G?" G?" G?" I/# I
fdl ih jd fi
' '
VP \ I G,P @" W$ !-. 492 &. VI G ,P (A.I.54) 185
[ U ' +
{5r } /: &. H4I/# +2 &B
' ˜
(JK4I (JO cod. 168 <N 9N cod. 169 &?,4 &?, cod. 171 JK2 (JK2 cod. 176 @" 2
˜ ˜ ˜ ˜
167
˜ P '
cod. 177 JO I ˜ <N JO I cod. 179 + H4I/#
˜ I/#
pr. scr. et corr. 179
cod. 176 P
&. add. in marg. cod. 179 I/#I + H4I/#I cod. 183 &B9P &B9I cod. 185 &. &.

˜ ˜ ˜ ˜
˜
pr. scr. et corr. 185 @" 2 cod. 185 \ I I cod.
˜
+2 '92 % /: [ !` (N&B W$ ' + '

I H4S#W2 +,4 &. VP (A.I.55)
'
(6.r^ !3 W$
[ ' +2 ' ' i
/: !` + H4S#W +, 4 &N &B VP W$ I +,4 /0D?P
190 h c de r^ j d i r^ '
(6.r^ !3 W$ +2 '92 %

AB ,r, W( ) 7 8 !` M W$ ];

$ ?,
!` + H , W$
4
_ _
'
$ AB M, +, /0D? !` W W$ 9E
/0D. XAB <N 6; 4 4I
' '

P I G,P @" W$ !-. 492 &. G ,P (A.I.56)
U ' ' [ '
;$ 4 !R &?,4 , I /:p &?,4
' ' '
195 2 &. I 5 2 !`9a, &. (JK 4, Z, 4 +, 4 &. VP (A.I.57)
' &. +2
(JK W$ !-. +, 4 &. VP + H4, Z, 4 &EN +, 4 &? + H4, Z, 4
G? I /: [ &. ' (JK2 &B VP '
&. P (A.I.58)
' '
I &B W$ !-. 492 6MP &.
; I G ,P @" &. (JK 4,
' + '
Z, 4 2 V2 &. +2 &B9 + H4, (JK2 &B VP H4, Z,4 2 &.
200 &B W$ !-. 492 Z4()7 8 /H4;$4 &. +2 +, 4 &? + H4,
' ' '
M !`9a, &. (JK 4, Z,4 , I + H4?, /2 +, 4 &. VP (A.I.59)
U ' ' '
2 ?M^ !` ?M^ WY" Z, / !` Z, / W$ 9E,
4 5 2

/4, /E,
4 5 2 !`9a, i 9N &. (JK 4, Z, 4 ', I t ?,
9E,
P

WY" ?W Y !` ?,
W$ t !` i WY" i !` ?,
W$
205
M,_ ', U ' 2 j 9N & ?W Y ZM^ !` d 9N & ?,
ZM^

6.W$ ?,
9N JKM,_
4
d ˜ d cod. 190 c ˜ c cod. 195 !`9a, ˜ 9a, cod. 195 &. ' ' ˜
190 I &?,4 , I cod. 200
/H4;$4 /H4;4I cod. 201 !`9a, 9,4 cod. 204 !`9a, 9a, cod.
˜ ˜ ˜
!I + H4YWY &B /0L $ 9N WY 4 Y, !3?2 (A.I.60)

{5 } &B G,P JKD, !I Y, ,P /:
v
' '
(6 5 2 &. (JK 4, Z, 4 , I ?E2 &. P (A.I.61) 210
U' '
/4, /E, M2 ?M^ !` ?M^ WY" G?E !` G?E
W$
GM , WY bM, ] 4?; $ 9N WY !I /,4 /E, G 4M, !3?2 (A.I.62)
' /2'
JK 4?, JK4I Z(: 2 @k8 6M2 !I WY bM, /0L $ V2 b
&N G .9 ^ !32 GW$ +2 '92
% WY ' , 4 &I (A.I.62)

2 W$ G ,&BI F? , JKL?, GWY
215
'
I G,P @" &B W$ !-. 492 &. G ,P (A.I.63)
WY bM, . ;$ U ' ', /2
b &" !- 4 JO ?E2 b &" 'WY bM, !-. JO?
/: ( WS +2 b &?, (4I
' ' U' U ' ' '
?M^ &?,4 ?M^ I ?, /2 &. &?,4 Z, /2 &. VP (A.I.64)
' ' ' ' ' ' '
Z, / &?,4 Z, / I Z, / ZM^ Z, / ZM^ &. 220
' ' ' ' ' '

&?,4 @ ?M^ &?,4 ?M^ I Z, / Z, / &?,4 @ (A.I.65)
' ' ' ' p _
Z, / &?,4 Z, / I ?M^ ?M^
' ' ' U' ' ' ' (A.I.66)
&. ?M^ &?,4 ?M^ I ?E2 &?,4 G?E2 VP b &"
' ' ' '
G?E &?,4 G?E I ?M^ ?M^
' ' '
@ ?M^ &?,4 ?M^ I G?E G?E &?,4 @ (A.I.67)
' ' ' ' '
225
G?E &?,4 G?E I ?M^ ?M^ &?,4

' ' '
5 2 !`9a, &. (JK 4, Z, 4 , I + H4JO W Y2 + H4S#W2 VP (A.I.68)
U'
2 ?M^ !` ?M^ WY" /: [ !` (N&B W$ 9E,
4
209 Y , ,P cod. 211 W$ ˜ WY I cod. 212 /,4 /E,
˜ + H4, G ,P 4M, ˜ /,4 /E cod.
G 213
JK4I JK2 pr. scr. et del. 214 &N
˜ ˜ I cod. 215 2 W$ ˜ 2' UW$ cod. ' ˜
223 &. &. cod.
' ˜ 4 227
,I e corr. 227 !`9a, 9, cod.
˜
' U ' (A.I.69)

&. (JK 4, Z, 4 , I ;$ 4 JO
'WY (WS b &"
'
230 /H4# ?M^ !` ?M^ WY" @3DS !` @3DS
W$ (6 52
U'
M2
'
+, 4 &? I 5 2 G 49, +, 4 &. + H4, &. ZI (A.I.70)
W Y2 S #W2 '
JO
'
+, 4 &? I 5 2 G
49, +, 4 &. + H4, &. ZI (A.I.71)
235
W Y2 (
JO W)' *
' I ?U ', /2 JK2 ' '
G 9E,4 5 2 &. M, VP (A.I.72)
'
Z, /2
' U' ' '
Z, / I ?E2 JK2 9E,
4 5 2 &. ?,
VP (A.I.73) '
G?E2
' r B G ,P (A.I.73)
JK2 P 52 {6 } &. WA b &"
U '
240
WY bM, !I b &" ;$

!3 4 2 2 9N X2)7 8 I 2 2
@34WS !I VI N,/"
'
!` Z, /2 &. WY" /: [ !` (N&B W$ +, &. VP '
'4 (A.I.74)
W Y2 S
JO #W2 (JO I Z', /2 &.
'
'
&. !` Z, /2 &. WY" /: [ !` (N&B W$ G ,P I (A.I.75)
245
' [ ' ' % '
Z, /2 /: I Z, /2 (N&B Z, /2
'
!` G?E2 &. WY" /: [ !` (N&B W$ +, &. VP '
[ ' ' 4
(A.I.76)
' % '
G?E2 /: I G?E2 (N&B G?E2 &.
' '
!` (N&B W$ I JO WY2 S #W2 &. (A.I.77)
P
250
' '
Z, /2 &. !` Z, /2 &. WY" /: [
W Y ˜ JO
WYM cod.
˜ Z,
cod. % ˜
229 JO 238 ?,
248 (N&B (JO (corr. ex &B) &B cod.
[ ' ' ' (A.I.78)

/: !` (N&B W$ I + H4JO W Y2 + H4(W)* VP b &"
' '
G?E2 &. !` G?E2 &. WY"
[ '
/: !` (N&B W$ G ,P S #W2
&. (A.I.79)
P
W Y2 S ' ' '
JO #W2 (6I Z, /2 &. !` Z, /2 &. WY"
[ W)' * &.
/: !` (N&B W$ G ,P ( (A.I.80)
P
W)' * (6I G?E2 ' '
255
W Y2 (

JO &. !` G?E2 &. WY"

' [ ' '
, I /: !I (N&B /0L$ 4 + H4JO W Y2 + H4S#W2 +,4 &. VP (A.I.81)
U'
?, /2 /H4;$4
' U' [ _ '
(JO I ?, /2 9E 4I /: !I (N&B /0L $4 +, 4 &. VP (A.I.82)
W Y2 S #W2 '
JO 260
' ' _ ' '

4 Z(S
9E, I Z, /2 &. !I /0L 4$ Z, /2 &. VP (A.I.83)
U'
?, /2
' ' U ' U p
/0L I Z, /2 Z(: I &. !I ?, /2 &. G, /0m (A.I.84)
U '
;$ 4 Z, /2 4I
' ' U' U p /0m (A.I.85)
I Z, /2 /H4. Z(: I &. !I ?, /2 &. G, 265
'
Z, /2 /H4. 4I /0L
( WS ' % G?E '
g95 /: /,5 GY4,/H, b &" (A.I.86)
' [ _ ' '
, I /: !I (N&B /0L $4 + H4JO W Y2 + H4(W)* +, 4 &. VP (A.I.87)
U'
?E2 9E, 4
' ' ' _
G?E2 Z(: I @3D)* {6v } &. !I @3D)* &. /0m (A.I.88) 270
W Y2 + 4H(WS ' '

JO I
˜ G ,P
(corr. ex P
) cod. 254 S ' ˜
#W2 ˜
253 G ,P S#;2 cod. 255 &. &. pr. scr. et
U' U ˜
corr. 265 ?, /2 &. Z,/2 &. cod.
U' ' _ ' '

(W)* /H4;$4 , I ?M^ !I /0L $4 G?E2 &. VP (A.I.89)
' /0L $4_ &. ' (A.I.90)
Z(S I F?, !I
JKL?, M, VP b &"
U'
(W)* 9E, 4
G '
I 5 2 &B9 +2
492 &. G ,P
(A.I.91)
+2
'
275
' ,4_ !3. ' '
, I &B &B
S JK2 G , VP b &?, (2 Z, /2 &B9
p % p , ('MP b &" U?E2 ' '
&? & B d d G"/ J
K2 Z, /
9E, 4 , /2
Z
U'
?E2 b P G
p % p , ('MP '@k8 2 2 JK2 X2) 7 8 b &"
& B j j G"/ (A.I.91)
' ' '@k8 2 2 9 : % P
280 (MP @k8 G?" 2 JK2 W b &" Z, /
G?" Z, W G?" 2 9 : % P X2) 7 8 p & B % h h G"/ p ,
' U U G?"
b, I &? ZM, G 4: !` &B &B JK4I /H &,4 / H,4 b &"

/N= 9N &P @34WS !I &?, U
?, !I &B U
('MP
p
/
&. !I &B9 ] 4?; $ &B !I
G ,P @" 5 2 492 &. G ,P (A.I.92)
285 !-,4 & P
' U' ? ', /2 6?AB ' U'
6?4AB I ?E2 P b &" 4 I ?, /2 JK2 &B' 9
?E2
' ' +2 (A.I.92)
b !I M? @34WS !I WY &N 2 b &"
4I 2 ].
JK2 9E ; a, W$ !I 2 &?, (I !3
4 /: &. bM,
290 4?, M2
U' +E,
&. JK2 XY4MI ?, /2 JK2 &B9 !-,4 & 4 @ I (A.I.93)
'
VI M I 2 !-. Z, W @k8 2 X2) 7 8 @k8 &B9 +2
' G ' Z', /2
' ˜ ˜ '˜ '˜ p
˜
276 (2 2 cod. 276 &?, &?,4 cod. 279 (MP e corr. cod. 280 (MP 2 VP cod. 283
U ˜

cod. 283 &?, &?, pr. scr. et corr. 284 / ˜

cod. 288 +2 ˜

cod. 289 !3 !R cod.
˜
b &" &,U qU >% Z', /2 /H. /: [ ?U ', /2 &B 9E, 4
!-,4 & P 4
'
@k8 &B9 +2 Z, / ' G?E2 '
&. JK2 XY4MI G?E2 /H4. &B9
'
U' ''
2 & 2 Y4, 2 !-. 9M,4 (4I b &" 295
' ; W$ {7r } !-. 492 &. G ,P (A.I.94)

@k8 G,P @" ]?
' _ U ' U _
!I &? b /0m &. JKMA B G ,P &B9 Z2 G?A B
' ' B '
[
!I M2 2 !-. , &. b Z4A I .9(S
&. /:p
, " '
&N
_
!R +2 V;I &B W$ !-. 492 &. G ,P (A.I.95)
' W$ ' I '
300
WY" !I

!I
+2 !` !R _ +2 V2 /H4B
B MI !3
G?A &. Z4AB !` /H4B

' !R +2 , I fi l h d /, 2 V2
V2 /H4B ;
!R +2 d W$ 9E
+ H4a4I , fh /H4B +2
<, d !R +2
' <,
Z4AB !` /0D. ?,
W$ b &" W W$ + H + H , !`
4 4 305
'

f h b Z4AB l h d <N 6MI !3 &.
4 ', I ?,
!I '@k8 M2 !I ?,
_ /0L $_ &. VP '
V2 9E, 4 (A.I.96)
M2 !I &? ];$ /0m

V2 9E, 4 ', I t !I '@k8 M2 !I M,_ /0L $4_ &. VP '

(A.I.97)
M2 !I &? /0m
310
4 ', I ?,
!I '@k8 M2 !I 5&5_ /0L $_ &. VP '
V2 9E, 4 (A.I.98)
M2 !I &? GM ,_ /0m
M2 !I &? /0m V2 9E, 4 ', I ji !I ,p /0m I (A.I.99)
U' ˜ U ˜ U _ ˜
B ˜ JO&.
293 ?, /2 Z, /2 cod. 297 JKMA
cod. 297 &. /I cod. 297 /0m /0L I cod. 298
' ˜ '
303 !R corr. ex !I cod. 307 , I I cod. 313 ji ˜ jh cod.
˜ ˜
I corr. postea in P
4 ', I ?,
!I '@k8 W , !I /0D._ /0L $_ &. VP '
9E, 4 (A.I.100)
_
315 M2 !I WAB /0m V2
V2 ?,
!I '@k8 M2 !I /0DE /0m 9E, 4 qU >% &EN
U
(A.I.101)
2 ,P
P % /0DE b ]? ^ Z', /2

' '
?, /2 !I @k8 M2 !I &? g/: p
V2 P G, /0m (A.I.102)
'
W , &? Z, /2
320 V2 P &? !I '@k8 M2 !I &? g/: G, p /0m (A.I.103)
U U
&, q>% ,%/: !I &? /0m

!I '@k8 ?,
!I '@k8 M2 !I ?,
_ /0L $_ &. VP '
?, 4 (A.I.104)

M2 !I &? /0m V2 9E, 4 ', I
' [ ' _ '
2 /0m I /: !-. (JK2 &B VP @3D ,4 +, 4 &. VP (A.I.105)
U [
325 &, &B 9N /: !I (N&B + H4(W +2 /) n48
' [ ' _ '
I /: !-. (JK2 &B {7v } VP @3D ,4 +, 4 &. VP (A.I.106)
'
!I +, 4 &? &B /0L W2 + H4.9()* M2 !I (JK2 &B VP /0m
U
+ H4(W 9()* +2 /: (4I Z(: 2 '@k8 /: [
/: 2 G, p /0m b', I &. !-. @3D ,4_ &. VP '
+2 (A.I.107)
'
330 Z, /2 (WI +2 /) n48 2 V2 b +2 Z()n84 2 P W , !I @3D
'
4M. 9W &? Z, /2 !-. 9W &?
' _ '

&B I
&B &. /S4I &. !-. @3D,4 &. VP (A.I.108)
9W &?
&B
!-. 9W &?
&B @3D U W2 &? b
4M.
_˜
˜ post ' ˜
314 /0L $4 /0m cod. 316 ?,
hoc add. 4
9E, cod. 324 (JK2 &B VP corr. ex
(N&B cod. 328 2 2 /0m cod.
˜
+2 /) n48 2 W$ ' _ '

+ H4(W I +,4 &. !-. @3D ,4 &. VP (A.I.109)
[ /: [ !` (N&B
335
/: !` (JK4M. + H429W +, 4 &? &B WY"

' ,4_ +, 4 &. VP '
+, 4 &? &B W$ I &. !-. (W (A.I.110)
[ [
/: !` +, 4 &? &B WY" /: !` + HB
) 7 8
4
+2 /) n48 2 ' [ _ '
@3D I /: !-. (N&B @3D ,4 +, 4 &. VP (A.I.111)
' ' V2
+, 4 &? o#W2 !-. 9W &? Z, /2 @3DI +2 /) n48 & 340
U U _
&B &. r?: 4M. 9W 9W !3.
& ' I /: [ . (JK2 &B VP ' _ '
@3D ,4 +, 4 &. VP (A.I.112)
!-
' '
+, 4 &? <?, /2 +2 &B VP @3DI +2 /) n48 W2 @3D +2
/) n48
U '
(6S#W2 !-.
_ ' [ _ '
/: 2 /0L $4 @k8 /: !-. (N&B @3D ,4 +, 4 &. VP (A.I.113) 345
@3D +2 /: 2 Z', /2 /0L W2 b ' I 9W !I @3D +2

U
4M. 9W !I
' [ _ '
/) n48 2 /0m I /: !-. (N&B @3D ,4 +, 4 &. VP (A.I.114)
' '
9W Z, / W2 4M. 9W !I @k8 9W !I @3D +2
U ' ' p %
4M. 9W !-. 4M. 9W 9W o#W2 @3DI GM I GY 5 350
' /) n8 2 /0m_ '@k8

{8r } 9W Z, / W2 9W !I @3D +2 4
U
' [ ' _ '
/0m I /: !-. (JK2 &B VP @3D ,4 +, 4 &. VP (A.I.115)
' [ <?', /2
&B9 Z, /2 V2 /: !I (N&B + 4H(W
_ ' _ '
!I 9W &? /0m , I &. !-. @3D ,4 &. VP (A.I.116)
U
+ H4 29W ˜ 9W cod. 341 &B _˜
(. . . ) add. in marg. 341 r? : V?: cod. 350
˜
336
˜ ˜ ' ˜
@3DI om., 9W2 add. in marg. 350 9W (pr.) @3D cod. 353 <?, /2 corr. ex Z,/2 cod.
355 /) n48 2 P 4M. 9W &? !-. &? b @3DI_ '@k8 P &. '
& !I 9W &? !I /0L b /0L ,4U W2 @3D +2
+2
!` (W /:
(N&B . / HI +,
&) * + H(W, @3D ,_ &. VP '

&B !- 4 4 4 4 (A.I.117)
/0L$ '
4 (JO% 95 I , 4 9WY4I
&B [
/: +2 F ,4_

&B 9E, [
/:
&B !I (N&B
360
' _ ' _ '
&? !3. &?
@3D ,4 @k8 + H4(W, @3D ,4 &. VP (A.I.118)
[ &B ' + [ &B%
/: !-. &,4 /, 4 + H4(W I H4(W /: !-. + H4(W
U U
&, &B
+ H4 , !-. 4( /0D? G(Wp I '@k8 4 > + H4 ,, fe G(W p I b',P
[ / % +2 /Sa I
365 &B9, /: +2 H" (N&B c h @3D
' _ ' _ '
&B VP
+2 !-. &? @3D ,4 @k8 + H4(W, @3D ,4 &. VP (A.I.119)
' ' [ /: 2 /0L $4_ '@k8 + H4(W
o#W2 I /: !I (N&B + H4(W +2
' + H29W +, &? o#W '
+, 4 &? o#W2 !-. 9W2 4 4 W2 + H4(W
U
(JK4M. + H429W
370 j /Sa I d !-. i G(W p I d /Sa I c !-. fe G(W
p I b',P
+ H4 , !I c o#W2 ' i d !I j o#W2 ' ie fe !I i o#W2 '
i <N fe !-. ie (W I /0D.
_ ' ' B @3D ,_ +, 4 &. VP '
/0L $4 @k8 (JK2 &B VP !-. (JKMA (A.I.120)
' '
B Z, /2 @3DI &?,4 b I + H4(W +2 &B VP '
!-. (JKMA !I b
U [˜ U ˜ '
,4 W2 corr. ex W2 cod. 362 / : &B cod. 363 &B &B cod. 366-7 &. VP
˜
356
U
(. . . ) ˜ /: [ &. !-. /: [ @3D ,_ '@k8 &. !-. (N&B @3D ,_ +, &. VP
'
+ H4(W 4 4 4 legenda 369
+ H429W ˜ 9W cod. 370 b',P
˜ P
cod. 372 <N ˜ 9N cod.
'
{8v } (JK2 &B VP 375
_ ' [ _ '
/: 2 /0L $4 @k8 /: !-. (N&B @3D ,4 +, 4 &. VP (A.I.121)
&? !-. (JKMA B @3DI /0L W2 b ' I (JKMA B !I @3D +2
U
' ' U
2 o#W W2 b I ;$ 4 (JK2 9W &? !I 4M. 9W
U
4M. 9W !I 9W &? !-. (JKMA B @3DI +2
/) n48
<N !3 fe !I h G, p /0L I h /Sa I d !-. 4 > G(W p I b',P 380
' p p I '@k8 he /) n48 (6.9()*

, I l !I JK, /0m c /Sa I d !-. fe G(W
he /) n48
' ' _ '
I (JK 4, V; !-. (JK2 &B VP @3D ,4 +, 4 &. VP (A.I.122)
B &?,4 (JK 4, V; !I b ZAB /0L2
(JKMA 4
2
&B & B9, '@k8 /: [ !I (N&B /0L $_ +, &. VP '
[4 4 4 (A.I.123)
'
385

/:
&B !I +, 4 &? &B
&B /0L W2 b I Z()n84

U
'
&B /0m I G ,P @" WI %, (N&B @3D ,4_ +, 4 &. VP '
(A.I.124)
' _
&B VP !I @3D ,4 @ & &? /0L [
W2 /: !I +, 4 &?
+2
U
9W &? WI
' ' _ '
!I &? /0m I Q , 2 ] 4" + H4(W, @3D ,4 &. VP (A.I.125)
'
390
M2 !I &? /0L W2 + H4(W +2 &B VP

U
[ &B /0m I + H(W, @3D ,_ &. VP
' '
/: !I + H4(W ' 4 4 (A.I.126)
'_ '
Z,
& &? !I MP &? /0m &?,4 + H4(W &B Z, /2
' ' ' ' _ '
MP &? Z, /2 I Q , 2 ] 4" + H4(W, @3D ,4 &. VP (A.I.127)
@3DI ];$
˜
377 cod.
o#W2 ' ^ <?', / W2

395 + H4(W ]? + H4(W U
' + Ha, WY2 + H(W, @3D ,_ &. VP '
I + H4M ) * + H4(WI 4 4 4 4 (A.I.128)
% '
+ ,\ Z, / Z2 MS &? (WI , T4)n48 & o#W '
&? +2
'
&? ];$ Z, / W2 + H4(W + H4, (4I &
U
' [ _ '
I /: &. 4M. /, 4 + H4a,4 WY2 + H4(W, @3D ,4 &. VP (A.I.129)
% Z', / Z2 , / , / Z2 'MP &? +2
+ ,\ {9r } + ,\ % o#W '

400 4 4'
Z2 &? ];$ +2 % Z, /(M W2 , / VI &? ];$ +2
+ ,\
U 4
,4 /
' ' ' ' _ '
MP &? Z, /2 I Q , 2 ] 4" + H4(W, @3D ,4 &. VP (A.I.130)
' ' ' &B Z', /2
Z, /2 + H4,/2 @3D b !I MP &? /0L W2 + H4(W
U [
405 /: @3D

^ 6M2 !I j 6M2 !I fe G, p /0m j ? W$ fe G(W
p I b',P
@3D <N !3 u Z', /2 j !I fe /0m ]? ; ?2 b fet
[
/:
' _ '
&B V2 4M. /, 4 Q , 2 ] 4" + H4(W, @3D ,4 &. VP (A.I.131)
&? /0m 2 ?,
W2 , / Z2 'MP &? Z', /2 ' I + H(W
410 U 4 4
[ Z', /2 &, / @3D b !I 'MP
/: @3D 4
' + HM ) * + H(W, @3D ,_ &. VP '
I + H4a,4 WY2 + H4(WI 4 4 4 (A.I.132)
'
% Z, / ]?; ,
+ ,\
4 W2 + H4MS + H4(W <?', /2
&? ];$ +2
% '
(4I +, 4 &M +H4(W +H4, (4I & &? +2 + ,\ Z, / ]?;
415 + H4(W
<?^92 + H4,
% ' ˜ % ' ˜
413 + ,\ Z , / + H4 ,\ + H4 ?,/ cod. 414 + ,\ Z , / + 4H ,\ + H4 ?,/ pr. scr., del., et in
marg. corr. 415 <?^92 ˜ corr. ex Z^92 cod.
9N he + 4HM S + H4(W <?', /2 ' I h h i d (W p l

I V2

+H4(W + H, (I (N +, &M d Z', /2 ]? ;
'
h Z, /2 ]? ; W2
4 4 4 U
[ _ '
/: &. 4M. /, 4 + H4a,4 WY2 + H4(W, @3D ,4 &. VP (A.I.133)
]? ; W2 , / Z', /2 , / Z2 'MP &? +2 % Z', / ' I
+ ,\
U 4 4

+ H4.9()* ,4 / &? ];$ +2 &? ];$ +2 % + H?', /
+ H4 ,\ 4 420
4 W2 d fe ?, /AI d JKI &,4

' _

I l 2 '
, 4 ' h' h G(W
feh b Z4AB I i Z, /2 ]? h Z', /2 ]?
; ;
[ &B W$ ' I + H(W, @3D ,_ &. VP '
/: !` + H4(W 4 4 (A.I.134)
' U ' '
o#W WY"
W$ I ;$ 4 (N/: Z, /2 !` , # 4)n48 &
{9v }
[ ' '
W$ /: !` (N&B WY" (JK2 + ,\ % o#W !` (N&B Z, /2 425
' ' '
&? +2 % o#W
+ ,\ WY" + H(W
4
+2 &B VP !` MP &?
+2 + ,\ % Z', / W$ 9E, @3D b Z', /2 !` @3D
b 'MP
4
U ' 2 @3D b !` 'MP &? WY" + H(W &B Z', /2 !` 'MP &?
4
/4, /E,
˜ & cod. ' ˜

417 +, 4 &M 419 I !-,4 cod. 421 4 W2 ˜ W2 cod.
, 422 feh ˜ bis scr. 427

@3D
˜ U
(pr.) &? cod.
430 Q I9
&. @-. !I !R 9
, 9N
,
'
&. @-. !I
/I
JO &. ' U ' U !I V(? ' (A.II.1)
+2 ?, /2 B9#5 @31/, 9 &N
B !` &B9 +2 U
bM, MA &. JK4I G Y, M2 !I &. /0m
U ' '
I 92 JK2 /#5 VP &. !3. bM, MA B !I 2 9E

4I ? , /
[
435
/: /#W !I
v# !`9, !-. &. &N Z^9, g&Y4, + 2p +A I (A.II.2)
!3
9; &N ? ', /2 &. '
G?M, G4: !` &B9 +2 <?4#
F?, &,U N&. V , 4 '@k8 JK4I Q I9
!I ,4 /, &. @31/, &, /,
4 44
' ' '
440 !-.
9#W Z4AB QWY,4 @k8 6 ;
9#W !I p ;
9#W
g& !-. , 9?^ 4I , &N &B /2%
&N +2 VJK1 /: [ : !-. M(?,4 + 2p +2
bM, +2 T5 &? Z;$ 4 %I /I
9^ !I 2' 9N (A.II.3)
'
/#I !-. V?)n84 Z, / T5 !I Z^9

JK4 !3JKa , !3 &.
2 /H" 4M,4 & &? ,%
, 2 /?^ 4M,4 & &?
W$ +.
445
' 4
)n8 /?^
&? b !-,4 & /?^ &? V?)n84 @k8 /#I !-.
I /H" &? b !-, & % V?)n84
&? 9 4 ' /H" &? ,
,
!R !-, & /?^
&? V?)n84 @k8 T5 &? b !-,4 &
4
@31/, GY ˜
45 cod. 434' M2 &. pr. scr. et del. 434 MAB VAB cod. <?4# v#
˜ ˜ ˜
433 437-8
# v# $ cod. 440 Q WY, ˜ Q,
? cod. 443 T5 ˜ T59 cod.
4 4

!-,4 & /H" &? ,4 / !I ,%
, 4 ,4 / {10r} GY
!3( 4, !I
/?^ '
& &? /H" &N !-,4 (4I V?)n84 @k8 !R /H" &? 450
/H" &? b !-,4 & /H" &? ,%

, 2 /?^ G !-,4

9; &N . 9E
!- , j !I j ? ', /2 b &, '@3 I G
4
h t d
j c u
l f i
T5 GY 4 !I V?)n8 b', I c !I c Z', / P 2' I (A.II.4)

' B +2
M !` &? V?)n8 @k8 fj 9N +, 4 /0D. WA T5 &?

GM?:
?, p G 4: /#I !-. & GY 4 !I fd 9N 2 /?^ /0D.
V?)n8 M Z', /2 !I
455
/0D. M !-,4 & &?

9; bM, ,%
,
9N 2 /?^ /0D. !3 , !-,4 & &? V?)n8 '@k8 /#I . 2 /H"
!-

V?)n8
9; bM, !I M GM?: p
& /H" &? ,%
, G 4: ff
Y,
9; bM, !I u Z^92 !I fc 9N fh !-,4
!-,4 & &? G
'
V?)n8 VB& M, Z, /2 +2 !3(4 ,4 / GY4, !I 2 /?^ ff 460
fi 9N fc !-,4 & /H" &? l I GY _ %

4 & GY4 !I ,
,
,

/H" &? fe 9N !3(4 ,4 / !-,4 & GY4 !I V?)n8 @k8 '
/?^ 9N /0D. !-,4 & &? ,%
, '@k8 fu 9N fi !-,4 &
!I WA B ? ', /2 T5 !I !3 ? ', / V(? &JO '@3 I t 9N 2
U U ' 4
&. fi &. +2
Y4, fi ?, / +2 B
< ,4 WA 465
449 !3( (corr. ex ,4 /) ,4

cod. 449 ,4 / (post.) ˜ ,4
cod. 450 /H" ˜
4 ,4 / ˜ +>4
!R add. (v. supra) et del. 453 V?)n8 M?)n8 cod. 454 M MM cod. 455 /#I /#I cod.
˜ ˜ ˜
%
, '
, / corr. ex Z, / cod.
457 /H " !3 cod. 457 !3 , , cod. 458 ,
˜ ˜ ˜

,
cod. 464 ? ˜

˜
465 &. add. supra cod.
4 !I l 9N JO t !-,4 & &? G YI

!-,4 & !-. GY
,4 / !-,4 & GY 4 !I /0D. 4( fd 9 I /0D , / GY 4,
4 4
{10v } !I G Y, fi G) n8 V5 '
!-,4 & GY 4 ]; !I /0D4
' !` )n8 /0D4 ,4 / GY
ft + >4 ]; !I ,%
, u fd G,B 4,
JO%
, ? W 9 I /0D M? , / !I i G Y, fe G,B !`
470 4 4 4
G Y, u G) n8 /0D$4 T5 GY 4 !I c G Y, de /# ,
!I
4

!I h G
Y, + >4 G,)7 8 +2 ft G)
n8 ,4
94, & GY 4 !I df
!I j G Y, dd /#I !I ,%
, ft 9 I !3( 4 4M? ,4 / GY 4,
I dj !-. ]; ' n8 V5
fu 9 !I JO%
, t G) T5 GY 4
n8 V5 '
475 !I ,%
, fh G) ]; +2 !R GY 4 !I + H4 , G Y,
!R GY 4 !I !I f G , h G,B dh '
+2 Y !` !- . ];
' ' I + >4 ]; '
c !I c ?, /2 Gq> dc /0D$4 !I ,%
,
de 9
&? GY ,I c !I c +2 /H" ? ', / &. G ,P I (A.II.5)

j !I j ? ', /2 +2 /H4# $ P G 4: 9M,4 2 T5 GY 4 !I T5

', / &JO
? T)n8 & ]; ' ' '
<, I + H4?, / b, 4 !I c !I c
480
p % 4 4
& B /H4E M M
4 4I U &. !` &A p . 4I &.

, /, &B +2
QWY P
&, B
(N JK2 + H4 +, 4 &? '
4 GY
4M? /0D4 ,4 / !-,4 & 4 !I (JK2 V4M G YI &B &B
'
& B @k8 /H" / &? !-W /0D , / !-, (I ,%
,
4 4 4 4
(JK2 V4M GY ,I &,U / I (N +, 4 &? +, 4 &N 4M,4 +, 4 &M +, &?
4
'
485
!I ,
, /H4E /0D$4 ]; !I /0D4 ,4 / G)n8 & GY4 !I
%

(N +, 4 &N
B 9M, +, &M +, &? & B '@k8 + >4 {11r } ]; '
4 4 4
4I GY ,_ & GY 4 G,B !` 4M? 6) 7 8 !I (JK2 VM

,I
4 GY
fd ˜ fi cod. 468 V5 ' ˜
˜ /0D$ cod. 469 fd ˜ fi cod. 480 ]; ˜
467 4 Z, / cod. 482 (N
<N cod. 482 +2
˜ ˜
!-. cod. 484 !-W quasi VW sæpius cod.
'
& B @k8 !-W 6) 7 8 !I ,%
,
/H4E MI P & / &?

6)7 8 !I (JK2 VM G YI &,U / I (N +, &N M,4 +, 4 &M +, &?
4 _ 4 4 4 490

MI P & / &? 4I GY
, & GY
G,B !` /0D
U 4 7 / 4
M, 2 !3(4 6) 8 !I H4E
2 /0D4 6) 7 8 !I M M Y, b &" '@k8 (A.II.6)
4 /I &. G

6)7 8 !I M4M
JO%
, !3( 4 6) 7 8 !I /HE +2 6M, , 4
'
!I !3JKa , !3: /HE +2 6M, , 4 2 !-W 6) 7 8 !I JO%
,
4M?
p I I T5 GY
495
+, &? & Sa I b GM?

+, 4 &M 4 !` 6) 7 8 Z4AB
4
(JK2 V4M GY ,I &,U B
(N JKap Y, !3 &. /H4B 9M,4
+ >4 GY 4 !I /#I !-. ,%
, /H4E 4M? /0D4 ,4 / GY 4, !I
(JK2 V4M GY ,I &,U / I (N +, 4 &N +, &? & B '@k8
4M,4 +, 4 &M 4
+,4 &M +, &? & B '@k8 + >4 !I ,%
, / 4 !I
4 H4E /0D$4 T5 GY 500
!3( 4 ,4 / GY 4, !I (JK2 V4M GY ,I B

(N +, 4 &N 9M,4
'
+, 4 &? & B @k8 V5 /0D$4 GY 4 !I /#I !-. ,%
, /H4E 4M?
GY 4 !I (JK2 V4M GY ,I &,U / I (N +, 4 &N 4M,4 +, 4 &M
' 4 !I /H4E V5
2 !` &A. @k8 ,%
, !-. T5 GY T5
' &. V?: I b &?, &. +2
< ,4
]; !I M4M
505
' '
V?: !-. ]; !I :^ g
, &B VP V, , 4 2 V5
' '
!I :^ g
, &B VP 6M, , 4 2 + >4 ]; !I M M
4
/I
9W: &. b @3D$/H4I &. Z4AB !-. !R %, !3: '
/0D$ 4 ];
'
#$ /0D &N ' /4 '
4 !- . 6 MP H;$
4 4 / ^ Z , /
p ' p
(6M> 4I
G
s1
{11v } c j <?, /2 9W: Q, 4 /= GEM5 510
M I V I cod. 492 !3( 4 ˜ /0D4 cod. 494 JO%

, 7 8 ˜ 6: cod. 495
˜
491 ˜ (JO
, cod. 495 6)
˜
/HE corr. ex GY 4 cod. 496 Z4AB b add. et del. cod. 497 /H4B /: cod. 497 JKap Y, ˜
˜ ˜
saepius cod. 498 4M? /0D4 ˜ (sic) !-. !3(

JK, 4 cod. 498 +>4 ˜ /0D$4 cod. 507 +>4 ˜
˜
/0D$4 cod. 508 /0D$4 +>4 cod. 510 ˜ cod. 510 Gs p 1
˜ G(W
I cod. 510 (6M> ˜ (JKM. cod.
4
U% ', / Z4AB 9W: '% U U

2 ,P ?
/I !` , 4 /# bM, b,4 &, q>%
' /= !I GM? p I ' p W:
Z2 c Z, /2 !I bM?" @k8 5 9
G ,P
j '
&B ,4 /, &. !`9, !-. &. GY4,/, P &N Z, /2
&B
[ U '
515 &. V?)n8 9N /: s1
? @31/,
, / &N &I (A.II.7)
U ', / +2 '
!I

;?, !-,4 ?
JKL?, T59 !I 6MP JK2 /I
/=
(" M2 !I /I &. /0m +2 ?U ', /2 @31/, %, 9E, 4 b @31

,4 JO !` &B9 +2 ',4 /
&. !` &(?, '@k8 &' , (I GM?
4
p
I
'
' '
' % '
520 6M?)n8 !3: Z, / T5 !I JK,/HI Z, / b !I ,\ ,4 / &.
4 /H4E Z', / T5 !I + 'H4? , JKU a5 6'MP
4, /H4E Z', / !I 9E,
[ ' ' ' ' 4
/: G,) 7 8 +2
JO &? ,4 W2 G,B VP +2 JO &. M , JK a5
4'
?', / bM, !I 9E, 4 2 /: [ !` + H4 , +2 BU
&.
U
JK4I @31/HI

' ' U
,4 / &. 9E I 6)
7 8 VP +2 ,4 WY2 V()7 8 9E , 9S,
' ' ' !I & %Y :
525 E2 4: / &. 4?2
9^ !I v#. Z, / VB U 4

9; &N
!-. ?,
,9: Z4AB JO
p

/I

/#
!3LI
+2 &B9 & B %, !36I JO
, '4 4" 2' I (A.II.8)
&? M '@k8 JKI &. /H" 9N ?
&? b !-,4 & ', / bM,
4
U% ˜ U ˜
,P ,4 P cod. 522 , 4 W2 W2 cod. 524 , 4 WY2 &U % Y 4: GW: cod.
˜ ˜
511 ,4 W2 cod. 525
˜
U ˜
4 4 6 cod. 527 !36I 96I cod.
526 ZAB JO ZAB +2
YI JK2 T59 !` !3JKa , !3: ' U % '@k8 /H"
&? G 4 q > b &"
,
+2 4M,4 +, 4 &M + HI/# G Y, ? ', / +2 T5 GY 4 !I T5
4 p 530
{12r } + HI/#
+, 4 &M (JK4I G Y, &I G " +, 4 &M + HY !I /I
4 4 4
b &" '@k8 j !I j Z', /2 !I !3 ?W Y &. +2 #5 4M,4

4I bMA. P j !I j Z', /2 !-. !R %, !3: '
&. !I, !I V? ,

' u u ' c c ' .
Z, /2 !I !I Z, /2 !I !I Z, /2 !I bMA P b &"
!I !3 ?a ', / ZaAB !-. !R %, !3: '
b V? , q>% t !I t
U
4 '
535
M B&
4?
9;
(JK4 !3JKa , +, 4 &M
+2 + HI/#
4
& Sa I b GM? p I I (A.II.9)
' ' T5 GY 4 !I (JK2 V4M GY ,I ',4 / &.
]; +2
' 7 8 +2 ' '
@k8 M, ,4 2 !3( 4 6) ]; !I ,%
, /0D4 6) 7 8 +2

7 8 +2 ' ' !I VM b 9M, & VM &? GY ,
6) ]; 4 4 4 540
+ H4I/# &SaI M, , 4 2 4M? 6) 7 8 !I ,%

,
T5 GY 4 !I !-W
G,B !` !-W 6) 7 8 !I (JK2 VM ,I (JK !3JKa , +, &M
4 GY 4 4
GY , '@k8 M, , 4 & /# 4M? 6) 7 8 !I ,%
,
/0D$ U
4 T5 GY 4
GY 4 G,B !` /0D4 6) 7 8 !I /# b 9M, & VM /#
4 4
V4M &? GY , '@k8 M, , 4 2 !3( 4 6) 7 8 !I ,%
,
9 I +2 T5
'
545
7 8 +2 4 !I &? b 9M,4 &

6) ' !R ]; +2 T5 GY
'
GY , @k8 M, , 4 2 !3( 4 6) 7 8 +2
! ];
R !I ,%
, /0D4
' VM
!R ]; +2 T5 GY 4 !I /# b 9M,4 & 4
&?
VM '
b 9M,4 & 4
/# @k8 M, , 4 2 ,%
, !-W 6) 7 8 +2

' ' 7 8 !I &?
,%
, ]; +2 T5 GY 4 G,B !` 4M? 6)
' ' '
550

b 9M,4 & V4M &? @k8 M, ,4 2 V5 +2 ]; !I

˜ (pr.) ˜ +, & cod. 531 (JKI ˜ JKI cod. 532 !I !3
JK2 JK4I cod. 531 +, 4 &M ?W &. ˜
Y
529 4˜ 4 ˜4
<N !3 ?WY !I &? cod. 533 I JKI cod. 537 b b&" cod. 546 9M, 9M, saepius
˜
4 4 4
cod. (« alif otiosum »)
' '
]; !I )n8 T5 GY 4 G,B !` !3( 4 6) 7 8 +2
/#
p ' !I b GM? p I I M, , 4 2 /0D$ U
&I G " + H4?, / +, 4 &N 4 ,%
,
p _ &B9 !` G p
G" I &,4 /, 2 !-. b GY , I /H4B /# 4JKa ,
'
555 bMA. !3JKa , & I t Z, /2 !I
' ' p
!3JKa , !3: b V2 VA. I fj ff Z, /2 !I G " (A.II.10)
M &. /: [ 9M, & VM ' !`
M 4 4 4 &?' @31/, ' @k8 Z^9 &N
GY 4 !I !-W 7 8 +2 p
+2 G 6) ]; !I JKa Y, !3
!I b 9M,4 & VM &? G Y, '@k8 M, , 2 ,%
, U
/0D$
4 4 ' 4 T5

'
9I +2 T5 +2 G GY
4 !I /0D4 6)7 8 +2 ];
, '@k8 M, , 2 {12v } !3(
560
9M,4 & VM /# G Y 4 !I ,%

,
4 4 '
& T5 GY 4 G,B !` !-W 6) 7 8 +2 !R ]; !I b

!I b 9M,4 & VM
4 /# G Y, '@k8 4M? 6) 7 8 +2
,%
, M, , 4
T5 GY '
M, , 4 & 4 G,B !` !R ]; !I /0D4 6) 7 8
' 7 %
!I 4 !3JKa, & V4M /# GY, @k8 !3(4 6) 8 +2 ,
,
Y, '@k8 M, , 4 2 ,%
, '
565
/# G /0D$4 G ]; +2 T5 GY 4

p G ' 4 !I b 9M,4 & VM
,%
, V5 ]; +2 T5 GY 4
!I b 9M,4 & VM /# G , '@k8 M, , 2 M? 6)
Y 7 8 !I
4 4 4 '
6)7 8 +2 %,
,
7
' T5 GY4 G,B !` 4M? 6)' 8 +2 !R ];
!R ]; !I b 9M, & VM Y, @k8 M, , 4 2 !-W
570 4 4 &? G
7 8 +2 % T5 GY 4 G,B !` !3( 4 6) 7 8 +2
2 /0D4 6) ' ,
, ' '
4M? 6) 7 8 +2 ]; !I b 9M, & VM G Y, @k8 M, , 4
4 4
G Y, '@k8 M, , 4 2 !-W 6) 7 8 +2
' ,
,
% T5 +2
' G GY 4 !I
G GY 4 !I !3( 4 6) 7 8 +2 ]; !I b 9M, & VM
4 4
575 GY , & I b GM? p I I /0D4 6) 7 8 !I M, , 2 ,%
,
4 T5 +2
564
&
& pr. scr. et corr.
˜
565 4
˜
add. in marg. cod. 575
˜
T5 T59 cod.
[ ' !I ', /

T59 !I N/: !` + H4?, / +, 4 &N 4
&.
' p . fu fc Z', /2 !I bMA. P
!3: b V2 GMA I (A.II.11)
9M,4 & VM G , '@k8 fj !I GMA
Y
p .
(" G

'
]; !` !3JKa ,
4 ' '
2 T5 +2 Z, / GY 4 !I !-W 6) 7 8 +2
]; !I b
7 8 +2 ' ' ' % M, ,
/0D4 6) ' ]; !I b 9M,4 & V4M @k8 ,
, 4 580

9M,4 & @k8 {13 } M, ,4 2 ,
, T5 +2
r % Z, / GY4 !I
4 !I !-W 6) 7 8 +2 !R ]; '
T5 +2 G GY !I b
' '@k8 M, , 2 ,%
,
!I /0D4 6) 7 8 +2
! ];
R ! I b 9M, &
4 4
G '
!I b 9M,4 & Y, @k8 M, , 4 2 ,%
, T5 +2 G GY 4
7 8 +2 '
2 ,%
, T5 GY 4 G,B !` !-W 6)
G
];
' '@k8 M, ,
585

G,B !` /0D4 6) 8 +2 G ]; !I b 9M,4 &
7
4
' ' %
!I Z, / ]; !I b 9M,4 & @k8 M, ,4 2 ,
, T5 GY4
!I b 9M,4 & '@k8 M, , 2 ,%
, /0D4 6) 7 8 +2 T5 GY 4
' 4 '
@k8 M, , 4 2 ,%
, !-W 6) 7 8 +2
Z, / ']; +2 T5 GY 4
4 G,B
GY !` 4M? 6) 7 8 +2 G ]; !I b 9M,4 &
'
'@k8 M, , 2 ,%
,
590
7 8 +2 ]; !I b 9M,4 & T5

6) G ' 4
!I b 9M,4 & @k8 M, , 2 ,%
, T5 GY 4 G,B !` !3( 4
' 4 '
@k8 M, ,4 2 ,%
, T5 +2 G !I 4M? 6) 7 8 +2
!R ];
!I !3( 7 8 +2 !R ]; '
T5 +2 G 4 6) !I b 9M,4 &
' ' '@k8 M, , 2 ,%
,
!I 4M? 6) 7 8 +2
]; !I b 9M,4 & 4
' ' '
595
]; !I b 9M,4 & @k8 M, , 2 ,%

, T5 +2 Z, /
p 4
b GM? I I M, , 4 2 ,%
, T5 +2 Z, / !I !3( 4 6) 7 8 +2

2
p I + H4?', / +, 4 &N !I
p 2 !-. /I !-. GY p 4, % & I
576
˜
!` +. cod. 579T5 T59 cod.
˜
580
˜ post b præb.
V4 M cod. 585

!- W
˜
/0D4 cod. 588-9 4 !I ˜ postea

T5 GY et iter. !I
hab. cod. !I 2 (. . . )
7 8
/0D4 6)
T5 GY 4
wn8 4
& !I , +2 9S &N 9) p n8 (6I9 I
' U ' ', / &)n8 (A.II.12)
600 !I JK2 @3DI VP )n48 2WI @3D , &U %Y 4: ?
[ B VA. !` , / !I ': / &. GY,/,
/: VA. {13v } r 4 4 4

&B ?W h h &. +2 JK4M.
2 fu fj t 9EI
h h &. +2 JK4M.
2 /0D. ?W Y /0D. W( ) 7 8 /0D.
' < ,4 ', I 6.9, +2 P 2 W( ) 7 8 @31
2' I (A.II.13)
]; +2
'
VB& Z', / , TS '
605 JO V;a, 4, 4> ,4 / 4, ?, ' 4 '

, / < , 6MP .
I Y, G,B
?, 4 ' 4', !UR +2 4 ' 4' U VP +2
VP &, b &" 6MP .
I JO V;a, Y, /0
!-. &,4 /, 4 ]^ 4 D.
fi ,4 / 4, &?, MI &
&N +2 /H4. !I 2 !-. 4I
9) n84 XY4 VA. T I c Z', /2 !I
' ' ', / /,5 % 2' I (A.II.14)
610 ]; I @3DI !I "
W $ !3 ?
' , t Z', /2 +2
&,4 /, 4 GY4,/H &N !-. 4I ,4 / &. GY !-.
' '
+ >4 ]; +, 4 /0D. V5 F , 4 +, 4 /0D?, G: 9 : !-.
' '
!R &,4 /, 4 fl . /0D$4 F , 4 fl : !-. &, /, 2
44
/0D$4 F , 4 jh + >4 !R &,4 /, 4 6M2 V5 F , 4 ji !-.
' '
615 &,4 /, 4 6M2 V5 F , 4 dl !-. &,4 /, 4 fj Z, /2 !I 6M2
'
F , 4 cd !-. !R &,4 /, 4 6M2 /0D$4 F , 4 di + >4
&,4 /, 4 6M2 /0D$4 F , 4 ce + >4 !R &,4 /, 4 6M2 V5
G
/0D$4 F , 4 uh + >4 G &,4 /, 4 6M2 V5 F , 4 ui !-.
599
I ˜ 6I9
(6I9 I cod. 599 , ˜ 4, cod. 602 &. ˜ /I cod. 603 W( ) 7 8 ˜ X2) 7 8
U ˜ +
cod. 607 Y4, /0
˜ ˜
D. , H4, cod. 608 &?, &?, cod. 608 fi dh cod. 609 +2 ˜ !I
˜ ˜
cod. 610 "
W $ corr. ex
W$ cod. 610 @3D I !I supra post h h in fin. A.II.12 in cod.
' ˜ ˜
D. cod. 613 &,4 /, 4 2 ˜ 2 &,/, ( &,4 / , pro &,4 /, 4
4 /0D?, +,/0
610 ]; I ]; !I cod. 612 +,
˜ ˜
sæpissime) cod. 614 ji fi cod. 614 /0D$4 V5 cod.
N/,5% b &" 6M2

&,4 /, 4 /0D. M fi MI &
!-. !R &,4 /, 4 ?W Y 9E , 620
' ' ' l ' '

&,4 /, 4 fu dh MI & !-. ]^ VP @k8 ?WY !-.
' ' ' l ' '
jd MI & !-. ]^ VP @k8 /0D. M +2 !-.
. ]^ ' ' U '
MI & !- VP &, b &" @k8
' : '
GY 4M G)n84 & !-. {14r } ,,4
!3?2 (A.II.15)
4, +2 GY ' '
4, VP I #5 9 &. V2 &? MA B +2
T59
' ' p B fj ' n8 n8 c '
625

/I +2 ]^ VP !I Z,4 2 GMA I ) G) 4 Z, /2
#59 +. '
@3D /0LI JK4I &. ZI !3 4, &. !-. (W p '
I @k8
U ' '
!I , /0L2 /H4; b
& > 6: +2 VI 4, !I ZI & I
&. GY4,/, 9E, 4 S4I b GW: !-?I
4,

U ' ' '
<, 4 Y4,/, .
4, +2 ]^ VP !I <, (4I b &?, 4: /
!I b + 'H4Y5 p
630
?^92 G ,P ,4

& 2 F , 4 P ; ,
/2 % + H4a4Y, &.

' < 4, ', I 6.9, +2 P 2 ? W @31
2' (A.II.16)
]; +2
!R ]; '
.
I ,4 / 4, h VB& ' '
+2 Z, / , T4S
]; ' VP ' ' U
G h G,B
+2 +2 JO V;a, Y4, fi ,4 / 4,
' ' U ' U
635
VP &, b &" .

I JO V;a, Y, jd , / 4,
!-. &,4 /, 4 ]^ 4 4
fi ,4 / 4, &?, MI &
9 &N !I ? ', / +2 !-. ' '
]; 4 (A.II.17)
9E,
' ' '
d : !-. + >4 &,4 /, 4 6M2 V5 F , 4 h : !-. &,4 /, 4
, ˜ 9E '
, cod. 622 ˜ ˜ ' ˜
620 9E (alt.) !R pr. scr. et del. 622 jd jj cod. 625 I
p B ˜ B ˜ ˜
corr. ex /
cod. 637
cod. 626 GMA GMA cod. 627 &. (post.) b cod. 634 ,4 /
˜
&?, &?, cod.
W +2 ' 4 (6M2 /0D$4 F , 4

640 !-. &,4 /, 4 ? !-. !R ]; 9E,
' ' '
: !-. ff Z, /2 !I !-. !R &,4 /, 4 fl + >4 !R de :
2 F , 4 ce + >4 G cd !-. G di + >4 !R dl
!-. !-. !R &,4 /, 4 /0D. W( ) 7 8 , / V2 b ZAB V, ,
4 4 4

% b &" jd !R !-. G jd '
l l &,4 /, 4 N/,5
' U % ] 4 %, 9E,
645 !` 6 &?,4 4 , JK4I
4 S4I (A.II.18)

%
4 )n48 2
& > &,4 / , !3 &. GM# /2 ] 4" @-?, b

;
&2 . F , , / +2
4 {14v }
' +2 /H4B /# + H , / , b I/?2 :
@k8
4'
'

(N +, 4 &M +H4I/# !` !3JKa, !3: b &" va, MI 9M,4 2 ?,
650

&. bM, #5
M p p '
MI & M
4
&. /HB &B b GM?
I b,I (A.II.19)
r4M JKI !3 6) 7 8 /0 '4

% /'H4^ 6: !I /'H^
4 LI (6M, ,4 2 (JO
, '4
'
h 6: +.
(JO9M,4 &M &? /H4^ h /H4B
' ji ' dl ' de de ' fd fd '
@k8 @k8 @k8 /0L,4 (JO9M,4 &M /0L,4
U

!` &, b &" ie @k8 cd @k8 hh ' '
&)n8 G , + H4 , ?,
'
655
!I F, &,4 /, > JK4I

GY4,/, !` )n8 !3
9 ;

' ' ' ' ' . JO6:
2 dl 2 de 2 fd 2 h 2 GY4,/H &N !- +2 + H46:

, / GA p I
U ji
2 &JO 4 I &N !- . &, b &"
p &? ' / M, , 4 2 JK2 &. VP ' p 1
G I H4E +2 g
, Gs
;
(6M2 6M2 cod. 641 fl dl cod. 641-2 di +>4
˜ ˜
640 ( . . . ) &,4 /, 4 ˜ add. in marg. 641 ff ˜
u cod. 642 F , 4 ˜ Z4AB add. et del. 643 Z4AB ˜ add. in marg. 646 /2 % ˜ ˜
!R cod. 647
cod. 647
& 2
& >
˜ cod. 648 / , Q I9 , cod. 649 M I 9M ,4 6M 2 6M, , 4 cod. 651
˜ ˜
p ˜ % ˜
cod. 652 (6M, , 4 ˜ 6M, , 4 cod. 652 r
JO
, JKI ˜ 6M, ,
4M
&B /HB cod. 652 (JO
, 4
+ H4MM cod. 657 JO6:
(corr. ex 6:) (JO6:
˜
cod. 658 ,4 / corr. ex / cod.
˜
'
< ,4 > (6 &?I (6M,4 &?, !` )n8 6: G 4, + H46: +2 9 ; 660

+2
'
JK2 !R F , 4 M M +2 &. ?,
p & B % I (A.II.20)
' _ 4 U _
( , b +2
VI ?AB Z, / G F , 4 2 &.
?AB
4 p ' p I
+ H.9()* + H,; , (JO&B
b &" 4 4 ];$ ; 9I Z, / p G?()

b V2 G !R G?AB
' ' B_ / '
665
(6: +. /0LI ?A M4M +2 +,4 H4B +,4 &? b 2

' G?A p B I /0D. !3 ,, /0'L, 4 (6MI &M ?,
,
/H4B
'
/0D. !3, ];$ V2 9N 4(, /0LI MI & G G?AB
U '
q>% M(?W$ b, I/?2 !` )n8 2 &N + H.9()* ?,
4

' !I {15r } VM ' p s1
(i ) (A.II.21)
6: 4 &? G
p s1
'@k8 M Gp Y% 5 &.
670
6M, , !3 /: 6) 7 8 !I 9M, & G M +2

' 4 4 ' p 1
6) 7 8 /0 M, , 4 2 (JK2 &B VP
!3 LI /H4E +2 ' '
g
, Gs

I (ii ) + H4 ,, /:
+ H4 ,, 6: +. V4M JK4I G p s1
6M, , !3 /: 6) 7 8 !I G

p s1
6: !I M

M
4
+2 &. U
Gp s1
' JKI !3 6) 7 8 /0 'LI &? b +2 G V4M &?
V4M 4
U p ('MP b &" '
675
&, d ;

&?, (JK 4, &., (iii ) h 6: +.
U
&?, (JO &., (MP b &" '
&, d ;
!3 /: 6) 7 8 !I GY
, _ 6: !I h GY
,_ b', b 2
h JK4I !3 7 8 G; , M, , 2 (JK2 &B VP ' i
+. 6) 4 g
, 6M, ,
V2 GM? p I l i \2 G p s1
I d i JK4I !3 6) 7 8

d 6: '
p ('MP b &"
680
h l 6:
!-?I (JK 4, &.,

h h 6: G;
, b

4

&I &N
JO,
' ˜ p 1
(6 &?I 6&?I cod. 664 + H4.9()* corr. ex (6.9()* cod. 672 (JK2 JK2 cod.
˜ ˜
660 674 Gs
˜ ˜ ˜
(pr.) G,
cod. 674 M4M V4 M cod. 675 JK4I 6M I pr. scr. et del. 676 d h
˜
cod. 677
U add. in marg. 677 d i cod.
˜ ˜
&, ( . . . ) b &"
$ !-. JK2 d VP ' _

QW ,4 / !I &. ?,
GY
, (A.II.22)
7 8 !I (JKa,/,
&I !-. + HMM p U
4M? 6) 4 4 4 G Y, /#I (NI/= (6M, 4I
,P ' 'L I
685 (6MI +, 4 &M h +, 4 /H4B ,P 6: +. 4M? 6) 7 8 /0

7 8 9E
4 6) h d ' l l U fd
!3( , !` !3J Ka , ! 3: ,4 /, &, b &"
, JKI ,4
&,U d d .
; F , 4 6: '
!-. &,4 / ,
4
9 &N !I
% ' ' '
]4 , !I 4 )n8 (2 ,I MP &N v6II
2 Z4AB !I A. 2
,/"
' Y, 96I c !I c Z', /2 !I V(? 2' (A.II.23)
690 ,4 / &. G
dj j dc f JK2 < 4I :/0 1 2 !-. #5 !I M, Z', /2 !I
7 8 !I ,%
,
6) dc V5 T5 GY 4 !I &B9 {15v } G YI

' 7 dj
@k8 M, , !3 !3(4 6) 8 !I /0D$4 T5 GY 4 !I j 4M?
/0D , / !I /# U % 4 4M? ,4 / !I d G Y,
!-W 4 4 I (JO
, !3(
695 2 /0D4 4M? ,4 / !I h G Y, dh 9N / +2 6M, , 4 2
Y, dd / +2 !3( 4 ,4 / !I /#U %
G 9N !-W I JO
, 6M, , 4
fe G Y, de 9N /0D4 !I 6M, , 4 2 JO%
, 4 6)
!3( 7 8 !I 'W
' 7 fd 7 l
2 JK2 &B VP g
, GY, !3(4 6) 8 !I !-W 6) 8 !I
M, , 4
+2
&N P 2 fu fj !` t !` W( ) 7 8 p
)
n8 I (A.II.24)
/= /: [ GY
700
4M? ,4 / !I

+2 M
M 4 ,I (i ) 9
p VB&
' 9N ' '
,9W: & Z, / !-,4 & ]; /0D4
+2
683 ˜
add. post fin lin. in marg. 684 G
,4 / p Y , ˜ GY
, I cod. 686 d ˜ i cod. 688 v6 I I ˜
˜
add. in marg. 688 !I (post.) +2

cod. 690 !I spat. rel. cod. 694 (JO%
,
˜ ˜ JO4
, cod.
˜ ˜
697-8 fe dl (in marg.) cod. 698 &B U &. cod.
' ' U %
M, , 4 2 ]; !-W
+2 !3( 4 ,4 / !I /# I ,
, /I,

' 4M? !3(
4 ,4 / !I MI 4M,4 & V4M /# GY ,
]; +2
/# U % '
I M, , 4 & /0D , / !I /#

!-W 4 4 I ,
, 705
' fi ' p p I
fi + >4 ]; ,4
!-. ]; ,4
&B b GM?
' '
!I d GY ,I (ii ) + >4 ]; ,4
V2 !-. ]; ,4

;I
' ' ' '
&?, &. @k8 6;$ &SaI !-. ]; ,4
@" !` /# $ !-. ];
&? GY ,I G p 4JKa , G 4)I M M
4

+2 ]; b V2 + H ,
4
' '

/# (JK2 &B VP g
, GY , V5 ]; !I 4 !3JKa , &
' ' 7 h
710
h !3. b &?, &. @k8 !3(4 6) 8 !I GY, (iii ) M, , 4 &
GY ,I G p 4JKa , G 4)I {16r} M M
4

+2 ,4 / ]; $ V2
' '
(JK2 &B VP g
, GY, /0D$4 ]; !I 4 !3JKa , & &?
M, ,4 & /#
!I 6) ' ' p I I
&N 7 8 Z4AB +2 ]; . & I b GM? 715
? ', / +2 9

+. (JO; , 9E, 4 +, 4 &M + HMM
4 4
+ HI/#
4
GY , '@k8 (i ) (A.II.25)
' ' 4 + H4,4 / !I fd ?AB_ (6: '
(6MI +E 4 !R ]; + H4,% (
+2
U
/#I (6M, ,4 2 H4,/0D4 H4,4 / !I (JO
, GY,I 4M? ,4 / !I
+ % + %
+2 V' I %, (6: ' ' B_ +
+. /0LI ?A +,4 &M H4M4M +,4 &? /# $ @k8 (ii )
'
720
7 8 +2 ' '
!3( 4 6) !R ]; !I (JK ,I fh + >4 !R ]; ,4
'
+ H4M4M +, 4 &? GM= @k8 (iii ) (6M, ,4 2 /0D4 !I (JO%
,
' ' % ' ' B_
(JK ,I l !-. ]; ,4
+2 VI , (6: +. /0LI ?A +,4 &M
703
˜ ˜ /#
2 2 cod. 704 MI 4M,4 M, , cod. 705 /#
˜ !I cod. 706 fi !-. !-.
˜
'
h ( !-. corr. ex ˜ I ˜
cod., !- . et h hab. in marg.) 706 fi (post.) d cod. 707
;

^ ];
cod. 709 ];
˜ cod. 710 (JK2 JK2 cod. 713 !I +2
˜ ˜ ˜
cod. 713 /0D$4 V5
˜
cod. 714 M, ,4 & deinde iter. verba M, ,4 & /# ( . . . ) h GY
, (cum erroribus
˜ % ˜

sed sine !3. b ) cod. 715 . G. cod. 719 (JO
, JO4
, cod. 721 +2 ˜ !I cod.
'
(6M, , 4 2 V5 !I (JO%
, ]; b !I
(JO; , 9E, 4 +, 4 &M + HMM +, &? GY '
725 +. 44 4 @k8 (i ) _(A.II.26)
,
' ]; '
+2 + H4,% ( 4 + H4,4 / !I de ?AB (6: '
(6MI +E4 G
U ' U %
2 /0D$4 G ]; + H4,4 / !I /#
+2 I (JO
, GY, 4M? !I
' ' B_ + '
(6: +. /0LI ?A +,4 &M H4M4M +,4 &? GM= @k8 (ii ) (6M, , 4
' !I (JK ,I dd + >4 G ' ' %
(JO%
, ]; &N ]; ,4
+2 VI ,
' '
730 + H4M4M +, 4 &? GM= @k8 (iii ) (6M, ,4 2 /0D$4 G ]; !I
l !-. G ]; ' ' % ' ' B_
,4
+2 VI , (6: +. /0LI ?A +,4 &M
' ' !I (JK YI
(6M, , 4 2 V5 G ]; !I (JO%
, ]; &N
(JO; , 9E, 4 +, 4 &M + HMM +, &? GY '
, @k8 (i ) (A.II.27)
+. 4 4 4
' + H4,% ( 4 + H4,4 / !I dl ?AB_ (6: '
{16v } Z, / ]; +2
735
'
@k8 (ii ) (6M, , 4 2 /0D4 !I /#I (JO%
,
U 4M? !I (6'MI +E 4
,
+2 ' % ' ' B_ +
4 ' VI , (6: +. /0LI ?A +,4 &M H4M4M +,4 &? GM=
' '

Z, / ]; !I (JO
, % ]; !I (JK ,I je + >4 Z, / ];
&N
' _ + HMM '
/0LI ?AB +, 4 &M
4 4 +, 4 &? GM= @k8 (iii ) (6M,

' %,4 2 /0D$4
' ' '
]; b !I (JK ,I l !-. Z, / ]; ,4
+2 VI , (6: +.
740

(6M, ,4 2 V5 !I (JO
, %
+ >4 ]; ' U U
4: !-. !I !I l l ,4
&, V?I I b &"
U ,
;$ 4 l l ;
; +2 ' ' p I I (A.II.28)
9 ]^ VP ,4

^ & I b GM?
< ,4 d : ' !-. !3( 4 +2 ]^ ' ' i '
+2 VP ,4
: !-. 4M?
' ' h ]^ ' '
745 ]^ VP !I 4, 4, 9; +2 VP !I .
94
'%˜
725 4 ˜ ,9E,
9E, cod. 731 VI , corr. ex VI cod. 734 + H4,% (
4 ˜ + H4,(4 cod. 737 je ˜ d
h ˜l
cod. 745 cod.
U U
, ' ' l
4
&, &EN fd ?, / +2 ]^ VP !I 4, +2
h h
h VP ' % < ,4 2 !` & %Y 4: &(?I (A.II.29)
+2 &B I &. +2
' ; ' '
]; !I G 4M? 9 +2 ]^ !I JK2 GYI M4M
g
, G
Y, /0D4 !I Z, / !3(4 !I !R !-W
+2 M, , 4 & 750
' ' ' +2 '
+2 ]^ VP !I @31/, !3: M, , 4 2 ?,
&N
&B VP
b V2 9 ;

' p # $ b GM? p I I (A.II.30)
GY ,I M4M +2 +,4 &. !` /
6) 7 8 !I !R 9 ; +2 '
!-W ' ]^ !I 4M? 6) 7 8 !I (JK2

[ '
!I (JK2 GY,I +,4 /: +,4 &. !` &A @k8 (6M, ,4 2 (JO%
,
.
'
755
(6M, , 4 2 (JO%

, !3( 4 !I !R ]; b +2 /0D4 6) 7 8

9; Z4() b V2 V?I n8
U
?2 ]^ ' ' p I I
&,4 /, 4 M, , 4 &M VP
^ & I b GM?
U
Y% 45 F?, !-. JKL?,
4, ?,

< , >' I
I g!31 4, +2 <, I (A.II.31)
4 760
' '
JK2 h VP &?, {17r} fd V, , fd l V, , l h V, ,
' . I W' Y2 &. ?,
!I :/0 1 & !- 4 +2
,

!R, + H ' ' + ' '
4 GYI H4WY2 JK2 +,4 &. VP 9E,4 &. ?,
,
g
,
6M, , !3 6) 7 8 !I +, /: [ ' + H '
! R , 6: ! +, /: [
I
4 4 4 '

M, ,4 & &? JK2 &B VP 765
/0D. ?W Y /0D. W( ) 7 8 /0D. &B ? W !I V(? (A.II.32)

, / !I
/ Y, 9N b 5 2
4 ' +2 M4M /' = 'H4B G
& VB& Z, / , T4S ]; 4I
+2 /0D M?
4 4
' ˜ !R
!I ];
fd ˜ fi cod. 749 ]^ ˜ cod. ˜
746 cod.
+2 750 & 754 4M? /0D4 cod.
761 l l
V, , h ˜
fl
V, , fh cod.
l ˜
767 /H4B /: cod.
' !-W !3( 4 ,4 / !I /# U % ,9 p

2 ]; &N +2 I ,
, /I, W:
+2
&N 4M? !3( 4 ,4 / !I MI & VM , M, , 4
770 4 /# U %
GY
'
M, , 4 2 !-W /0D , / !I /#
4 4 I ,
, ];
+2
&N ? ', / Z4AB !I ' '
]; G p &. ' p
&I b GM?I I
9
; +2 ]^ ' ' '
4M? 9 VP ,4
@" b &?, /#$ @k8 (A.II.33)
'
; , ?AB_ M M +2
!I (JK ,I 6M2 (6: +. 4 +, 4 &. GM=
' '
775
2 !-W !I (JO%
, !3( 4 !I (6MI +E 4 ]; b !3,4

, / ZAB !-. !R %, !3:
4I
' U
/#I (6M, , 4
' ' 4 4
2 VP P 6MP !-W 4M? 9 ;
G p &. ' p
&I b GM?I I
' '
h VP &?I h h ,4
b &" U h V, , h .

JO 4, +2 < ,4
'
780 & , (4I :/0
1 2 !-. &. ?,
, JK2

dh 2' fi 2' 6: '
!-. &,4 / , !3( 4 9 ; < , (A.II.34)
4
4, +2

<,4 l l +2 /)n48 2 hl 2 he 2' jd 2'
'
+, 4 &. GM#I h h ,4 /, b &" l V, , l h V, , h .

' _
G 4^ ,4 /, (6: !-. {17v }
?AB /H4E &. +2
'
785 M
M
+, &. ;
4 +2 4 ,4 W2 G ,P + >4 ]; ,
!`
4 _

!I GY, 6)7 8 &N !I (JK4a:^ +H4M4M +, 4 &? GY,I ?AB
'
(6M, , 4 2 g
, +, 4 &. VP &. ?,
V, , 2 /0D
p4 4

<,4 2 +H46)7 8 +H4,N +2 ' p

6: VP G &. &I b GM?I I'
+2
' '
JK2 h VP &?I h h
2 h V, , h .
b &?,
4,
M
˜
, ˜ ?,
cod.
775 M4 ˜ + H4M4M cod. 776 !3( 4 ˜ + H4(4 cod. 777 ,4 /
˜ ,4
cod. 780 ?,

782 he h cod. 783 V, , (pr.) ?, cod. 783 cod. 783 V, , ?, cod. 784

˜ ˜ ˜ ˜

cod. 785 M
M
M? cod. 786 (JK a:^ JK :^ (corr. ex )xm ) post + H ,N +2
˜ ˜

4 4 ˜ 4 4
+H46)7 8 præb. cod. 789 .

.
I cod.
'
& , (4I ^
(" N&?, &. ?,

,
790
,
&)n8 b', u Z', /2 !I !3( 4 6) 7 8 !I V(? 2 (A.II.35)
4
+, &? Gp Y, fl + >4 ,

; , 4 +, 4 &M '
4 4 de !-. ];
(JO%
, 4 !I (6'MI p /'H4^ + H4,
!3( % M? + H, / !I de (6:
4' p 4 4
'
+.
!` /# I fi !3( 4 6) 7 8 !I ,
&B (6M, , 4 2 !-W !I
4
p ' fd /
!` fd G ^ b, (JO,4
9E, +,4 &M +,4 H4E +,4 &? 795
+2 +, 4 &. ; ?2

dl ,4 /
^ 4 ,4
<N !3
!3( fi
&N !I G Y, !3( 4 6) 7 8 !I +, &? +, &N
4 4
G YI ?AB_ M
M
4

Y, dl (6: ' ' B_ + U 7
G +. /0LI ?A +,4 &M H4M4M +,4 &? ;$ 4 6) 8
'
M, ,4 2 JK2 &B VP g
, /0D4 !I

% !I b V2 V(?
N/,5
Y2 N&. ' #$ '
800
+2 I9W /0D &N .

? , / G s1
&I
4 !-
6MB !I N/I ' !-. &B &B , /, N/HB !` &B9
' 4 4 '

M ,9?>
/" +2
./I &?, s1/, @ 2 !-. JO & WYI
!`?,
&. @-. !I !R

805
' ', /2 !I V(? 2' I (A.II.36)

WI M, !-. JO I

?
/I

/I

!3. &N I +E
4 >4 , 9N
' ' '
?,
6I
U _
I &. 4I V?)n84
&B V?SI ?,
Z', /2 !` &(? &?M I (A.II.37)
810
˜dl cod. 796 fi ˜ corr. ex fd cod. 796 dl ˜ di ' ˜ U

792 fl cod. 798 /0L I /0LI cod. 798
dl ˜ fl fl (sic) cod. 801-2 &B9 +2 ˜ &B9 !-. cod. 805
˜
(JK2 add. cod. 808

/
˜ U

cod. 809 I Q I cod.
˜
?,

!I V4 /0D2 V2 ,

!I {19r } /:
4 ' #5 !I 4#5
9 !-, & &? @k8 /#I . 2 <N wn8 /#W
#5 4 !-
9 !-, 2 JK >4 +. !` /?;
/HE #5 4 4 /HE #59 G,B
' 9 G,B U GY 6M, 2
&? @k8 /?; #5
!` ;$
4 4 V 4 , ! I
815 / /
HE !-,4 2 HE #59 G) /?; #59 !-,4 & n
8
,4 / !I /?; #5 9 !-, & &? '@k8 V GY , ! I 6M, 2
4 4 4
4
!-,4 & &? 2 V4 GY4, !I M, ,4 & !-W !3(
JK2 V4 GY 4, !I M, , 4 2 4M? !3( 4 ,4 / !I /?; #5 9
&? G,B !` & GY 4 !I /?; 9 !-, &
#5 &?
4
820 V4 GY 4, !I fh 9N V, j 9N 4 G p 4JKa , & /H"

fh G)
n8 f 2 V4 GY
4, !I fc fj G)
n8 & GY
4 !I d 2
', / &N
? G q' > 2 V , !I fi
4 GY 4 4
' ' ' '
@31/, @k8 Z, / @31/, 96I /
Z, /2 !I V(? 2I (A.II.38)
'
' p
Z, / !I &. !3#5
Z, / !3#5 GM?: G4: #5 !I &
' % '
Z, / !-. !R , !3: (N/,% # $ 9M,4 2 P 4: (N9M,4 2 MI &

825 G
&.
) 7 8 Z', / b < I /I,
4I V? , Gp " (" VB&
4
+ HI/a# p % p I I (A.II.39)
(JK4 !3JKa , +, 4 &M 4
&a : b GaM?
N/H4B !` &B9 +2 QW Y P &,U
9N V4M G YI

M, , 4 & , / GY
4 4, G) n8 /0D4 6) 7 8 !I &B &B , /,
!I /AB , \ JKM. !3 V\WM n8 /S &N
) {19r , 18 → 184v }
830
% ' 4 4

+, 4 &? &B , @k8 ,
, %
M, , !3 !3( 4 6) 7 8 !I /0D S ;
4
/4HE !-W 6) 7 8 !I /I 9N (JK2 VM YI +, &N M, +, &M
4 G 4 44 4
+2
V?)n8 &,U b &" '@k8 ,%
, M, , !3 4M? 6) 7 8 !I

813
9 (post.) ˜ T59 cod. 815 &
#5 ˜ &M cod. 818 M, , 4 6M, , 4 cod.
˜
824 &.
˜
&? cod. 824 #5 #5 cod. 828 N/H4B N/: cod.
˜ ˜
832 /H4E /H4E cod.
˜
833 6)7 8 !I
M, , !3
,%
, 4M? post /I 9N (v. supra) hab. cod.
˜
% ' ' U ' U ' 4 + H46SM M M &.

& &6 /2 &6 /2 /0D4 !3( 4
U/2' W V?)n8 /0D +2
&6 !- 4M? + H46SM M M
4 +2
/I
4 835
/ &? !` !3JKa , !3: ' +2 %

& ' &6 ' U/2 '
V4M !-W
7 8 !I /I &?, Z', / ZM^ @31 Q I 9 / &? VI &
6)

ZM; ]; $ > 9N & GY 4 !I &,U Q ',4 b !3( 4
7 8 &N
6)
!I M, &
4 4 ' V4M / &? GY ,I N ap Y, I
&? GY , @k8 M, , 4 2 /0D4 !I ,%
, !3( 4 !3. ;$ U
V4M 4 840

&?I i ZM; @31 P !3. ZM;

@31 QI94, 2 9N 9M,4 &
U
!I /#I ,%
, 4M? /0D4 ,4 / !I fe &?I fe P i
&? &N 9M, &
4
VM
4 / &? , '@k8 M, , 4 2 !3(
GY 4 ,4 /
/0D , / !I /# U % 4 4M? ,4 / !I
&? !-W 4 4 I ,
, !3(
M, , 4 &
p b GM? p I I
845
!I M M
4
&. +2 < ,4 2 GM?:
ZM , !` /0D4, p %&, &6 U/2 ' U '
&6 /2 /0D4 !3( 4 + H46)7 8
&. GM?: p (W !3( 4 4, %, gI b &, V;: & I ?,

&6 {18r } U/2 ' U '
&6 /2 !-W 4M? +H46)7 8 !I M4M /I
' !I %&
!3JKa , !` M, , 4 2 &. VP g
, G
p Y , G , !- W
p
850
% 7
6M, ,4 2 JO
, !-W 6) 8 !I M GY, &N GM?I I M !`
&B9 (6M, , 4 2 !-W !I (JO%
, 4M? 6) 7 8 !I + H , M? !I
4 4
p %
% !-W 6)7 8 !I

!-. GY4, &I 9EI 4M? 6)7 8 !I M, ,4 2 ,
,
&,4 /, 2
' '
b, 96I / /

' /
!I V(? 2I (A.II.40)
855
{18r , 7 → 19r , 18} Za, / !I (N9M,4 !3

&. + H4#59 G p s1
(JO%
,
˜ JO4
, (6M, , 4 6M, , 4 cod. (N9M,4 N9M,4 cod.
cod. ˜ ˜
852 852 856
/ !I ^ (" M. !R %, !3: ' '

?, / &N VB !I 9E, 4 &
4

/
JK4 !3JKa , !3 &. p & B % l Z', /2 !I bMA. P I (A.II.41)
GY 4, &?, 4M? 6) 7 8 !I QW $ !-. M M + H4 , GY ,I
860 4 {19v } +2
M +2 ' % G
M 4 h
GY @k8 (6M, ,4 2 !-W 6)7 8 !I (JO
,
, Y, ,4 /
6M, , 4 2 4M? !I JO%
, !-W 6) 7 8 !I QW

$ !-.
!I 4, +2 b V;)n48 I 4 > +2 /H" bMA. P (A.II.42)
YU , fl 2' YU , fh 2' ' '
b &" 4 ' ' 4 ' 4, fe ,4 / 4, &?, ]^ VP
865

&. ?,
VP @-. l Z, /2 !I GM?: 2 !-. h h ,4 /, &,U
p
' GY , !36I &. ?,
V, , QW Y !-.
M 4M ?,
+2
GY , 6) 7 8 &N !I + H#5 /HE ?,
+2 !I /: [ JK2
6:
' 4 4 p
+ H46)7 8 &B G p &?
I &. +2 6M, ,4 2 /: 6) 8 !I JO%
, 7
h &? ' fe < ,4 & P I /:,
, 4, ,4 / 4, &?, +2
!3 &. +2 &. ?,
, V5 +2 JO4
9, ?,
, 9 I +2 JK2
870 '
fh <, b &" 6M, , 2 VB& Z, / 9W: &?, JK4 !3JKa ,
4
i 9 I +2 4, i b < ,4 !3: ' U > l G p &?'
+2 &, b &" 4 I

V5
+, 4 &. GY
,I b GM? p
I I

JK4 !3JKa, !3 M4M &. +2
875 h GY , '@k8 (6M, , 4 2 !-W !I (JO%
, 4M? 6) 7 8 !I QW

Y !-.
!I JO%
, !-W 6) 7 8 !I QW

Y !-. JK !3JKa , !3
4
M M
4 +2
6M, , 4 2 4M?
9N 4 !3JKa , & VM
4 &? GY ,I b GM? p I I (A.II.43)
861 (JO%
, cod. 861 (6M, , 4 ˜ 6M, , 4 cod. 863 I ˜ post hoc & del. (atramento rubro)
˜ JO
,
˜ ˜ ˜
867 6: 6)7 8 cod. 867 /H4 E + H4, add. 871 6M, ,4 M, ,4 cod. 872 G
' ˜
p &? I G&. cod.
˜
872 i i cod. 875 (JO ˜ JO4
, cod. 875 (6M, , 4 ˜ 6M,, 4 cod. 876 !-W
%
, ˜ corr. ex V5

cod.
/0D4 ,4 / GY 4, !I ZM; @31 Q I 92 QW Y &N !I &,U

'
4 ,4 / !I /#I ,%
, U
9M,4 & V4M GY , @k8 M, ,4 2 !3( 4M? 880
U
/0D4 ,4 / !I /#I ,%
, 4M? {20r} !3( 4 ,4 / !I /I 9N
M, , 4 2 !3( 7 8 !I &? &N
4 6)
9M, &
4
VM
4
GY , '@k8 M, , 4 2
/0D4 6) 7 8 !I &? &N &? /0D 6) 7 8 !I
p ' !-,4 & V4M 4
+ H46)7 8 + H4,N G &. & I &N GM? p I I !3( 4 6) 7 8 !I M, , 2
4 U
;$ 4 885
fd l V, , l h V, , 4, h JK2 b < ,4 2 9E, 4

?,
, JK2 4, h VP ' ' h h fd V, ,
!3 &?I ,4 /, b &"
6: !I /HB M M +2 '
GY, 9N ^ (" 6M, ,

4 4 !I /HE +2
/: 6)7 8 !I 6M, , 4 2 6) 7 8 &N
4 + H4#5
' % ' %
6MP JK4M. !R , !3: JO
, 890
+2 9E, 4 & /H" Z', / @31/, 96I

@3DI 2' (A.II.44)

' ' ', /2 ?,
, MB @3D ,_ '@k8 M2 !I
&. /0m
Z, /2 VP ZM^ ? 4
' U ' ' _ / '

VP ZM^ ?, /2 /0D. W$ @3D,4 H4E Z, / ZM^ ];$ V2 JK2
' ', /2 ?W Y, @3D ,_ /HE Z', / ZM^ Z,
_ V2 JK2 Z', /2
VP ZM^ ? 4 4
4 ? ' '
, /2 /: Z, /2 !I @3D ,4 /H" ZM^ GM _ _
, V2 JK2 &B
9E, 895
+2 JKL?, i !I i /0m +2 JKL?, l !I l /0m +2 JKL?,

+2 /H4E Z', / 4, Z4AB !I &. G',/, '@k8 h !I h /0m
' VAB !32 Z', / 9E
G , ,6: Z4AB +2 4I JKMA B ,4 JO !` &B9
'
, &B ' ' B
!-. MB !I !3 ?, / bM, +2 Z, /2 VA !32 V()7 8
!I /
&. !I %'JK, &N 6) 7 8 Z AB +2 MA B G ',

4 4 4 900
cod. 886-7 fd V, , fd l fd V, , fd fl cod., quorum fd V, , fd
˜ ˜
886 (pr.)
˜ ' ˜ _˜
in marg. 892 ?,
, Z,
, cod. 892 Z , /2 Z, /2 &. cod. 894 Z,
]; $ pr. scr. et del.
˜ ' ˜ B ˜ MAB cod. 900 &. ˜ &.
894 ?W Y, ZW$ cod. 894-5 &B VP &B9 cod. 898 JKMA
cod.
/ /
&.

U ' _ %'
&. 2' I (A.II.45)
@31/, .,
Z, / @3D ,4 4JK,4 rI /
U
'
#$4 /0D &N !` 9 '
/
&? MAB 4I &. JK4I
/0D" {20v } /H4. +2 o4)xm Z,
_ 9E, 4 %'4JK,4 4I Z, 4 &
b &M I /0D" ?
', / &N +2 g!31 !I G Y,4_ #$ /0D +2
4 XY4
'
905
U'
&?
^B ] 4;Y !I /
!I &N
,4
/H4. g/: +2 g/: _
, &
GM /
&. +2 P 2 2I '

_ %' '
g/) 7 8 !-. @3D ,4 4JK,4 , I b 5 2 je fl V2 ];
U 4 !3 g/: /H. I &B 9, &
B
N&. 9E, +2 ]; 4 4 4
' ' ' ' +
&. G,/, @k8 2 &I (" (JK4I M : + H ,
4 4 H4,
S
'
910
,/" 2 !` 9 4I 4I

i /0m +2
JK2 &B VP ? ' '
, /2 ?WY, @3D,4 fl Z, /2 I ' '
', / &N +2 ' U ' /
? &B VP !I 2 H4E !I V()7 8 9E I i !I
' ' t je ' U U '
!I fe /0m +2 &B VP ?, /2 @3D,4 Z, /2 ;$ 4 2
i !I i /0m +2 ' U ' n8 U
915 &B VP ?, /2 +,4 /0D. W() ;$4 @3D , 4 fe
' ' ;$ U
JK2 &B VP 9E, 4
? , /2 ?,
,
4 ,
Z,
! I 4 @3D, 4
Z', /2 VP ' ' t _
!I i /0m +2 ?, /2 V; ,4 fd !I fd /0m +2
7 8 Z4AB +2
6) JK2 /H4E !I V()7 8 Q 'aI JK4I &. G',/, '@k8 i
U
?; '
;$ 4 b &" +2 &B VP !I
/
&. +2 (N/H4. !I +, &? +, &N >4 (A.II.46)
920
4 4 !I +E
' ˜ (pr.) ˜ !I cod. 905 b &M I ˜ b&EI cod. 907 &. ˜ &. cod.
903 94I 9,4 cod. 905 +2
910 + H4 , (pr.) post hoc ,%
,
˜
M, , !3 4 6)
!3( 78 !I add. et del. 910 (JKI ˜ JKI cod. 912
4 4
˜ Y, ZWY, cod. 913 V()7 8 ˜ corr. ex M()7 8 cod. 914 U '2 ˜ Q2
@3D , 4 @3D, 4 cod. 912 ?W
˜ cod.

U' ˜ ˜ ˜ ˜ ˜
915 ?, /2 Z, /2 cod. 916 ,4
,4
cod. 916 ?,
, Z,
, cod. 917 t c cod. 920 /

/ pr. scr. et del.
' '
V()7 8 , !` 9I &. JK4I @31/, M ) * g/: !-. @3D, 4
/0D !-.
#$
' ' 4 '

VP ?, /2 ?,
N,4
Z,
!I 9E,4 +E >4 &I /0D? I

JK2 &B VP ' B h h
7 8 Z4AB +2
6) VA !I /0m +2 JK2 &B
' Z', / {21r } VP ' B '
MAB G , ,6: Z4AB +2 VA MAB G , 925
' ' h U
&B VP ? , /2 Z,
,4
!I 9E, 4 fh !I ;$ 4 +E>4
' ' '
G , ,6: Z4AB +2 JK2 Z, /2 VP VAB i !I i /0m +2 JK2
Z', /2 Z', / &N T5 !I 9E, U %'
!I i /0m +2 4 ;$ 4 4JK,4 MAB
' ' ' l
VP VAB h !I h /0m +2 JK2 &B VP ?, /2 , T4)n48 i
U
4 ;$ 4 fl Z, /2 !I +E ' '
>4 MAB G , , &B '
9E, !-. JK2 &B 930
' U' Z', /2 Z', / T5 !I

VP ?, /2 fd , T4)n48 fe !I fe /0m +2
/,5 % !I b &" ;$ U '
&N 4 MAB Q a, h !I h /0m +2 JK2 &B
&.
' _ ' ' ' ' '
Z, /2 JK2 @3D ,4
&? ,4 ,I ,4 / ?, / 2I (A.II.47)
' ' ' U ' ', />
#$4 /0D &N !` 9,4 @k8 /I JK2 &B VP &. 9E, 4 ? 935
' _ Z', /
Z, / !I V?: !3A I &B #5 !` )n8 /I &. &
' 4
'
' '
6MAB Qa, !` : /I JK2 &B VP &. ?, /2 &. /H4E
' ' p '
&? /H" Z, / #5 JK2 &B VP #5 4 !` :
9E,
U '
g!31 !I XY4 &. +2
<,4 >
!32 &. &. &N +2
?
'
, / Z4AB !I #5 p
9 G Y, I &B Z^92

!I ' G 'M2 940
TW $ !3: ' U p ' p 5

\2 QW$ !-. &. G Y, 6MP JO\2 G#

˜ Z,
cod. 927 ˜ I cod. 931 Z', /2 ˜ ?,U /2 cod. 932 MAB ˜ JKMA
'
923 ?, ˜ B cod. 935 VP
&B &B9 cod. 936 &. ,&. cod. 937 : o4: cod. 937 6MAB JKMA
˜ ˜ ˜ B cod. 938
'
˜ ˜ cod. 940 G Y , G , cod. 941 Gp Y ,
: o4 : cod. 938 &? /H " & ? /H4 B
p ˜

U ˜ U
&. &. GY, cod.
+E, '
Q I9
#$ /01 !-. &N @ + H4M, 2 +, 4 &. + H4#5 VP

4 4
7 8 V#, !` ' ''

/E /HE M() ,4 / &. !I Y4, 2 !` 9?,4
F?, 5 JKL?,
945
', / +2
? Z', /2 {21v } !I &. @31/,
p I (A.II.48)
' ' ', /> JK(W p I &I !3
& Sa I , Z, /> < ,
9#5 JK 4, V;, 4 @ ?
', / +2
? Z', /2 !I (JK ,I Z', / b !I Z , !3 &. !3#5
' ', /2 #5 G ,P p
!-. b &" @k8 h !I h ? G 4: JK4M. (W I !3
' ' [ '
+, 4 &M +H4I/# &B @k8 Z, / b /: !` !3JKa, !3: @31/ b

950 bM, +2 /: [ Z', /2 !I (JK , + H4#59 2 (6AI I (JK !3JKa ,
4
% ' U U ' p ', /
2 !-. !R , !3: &, b &" GM?I 2 V2 , V?I ?
&,4 /,
< ,
9#5 JK 4, V;, 4 ? ', /> ?U ', /2 JK2 G(W p I (A.II.49)
/#M
P
W2 /H4. G, 9)7 8 !I b +2 V;, 4 2 P ' , Z', />
' ' h i
/= l h JK)n8 4, h VP !I GY, b,I !I V; b

&^ !I ' Y', (" :^ /#5 VP '
955
V? , &N 4 JO &?I M, 2

?,
!I GMA p . l i '
W !-. !R , !3: % '
b
!I V; P
'
b V> W !I ?,

' ' U U ' _ %'
Y4, /I JK4, V; 9E,4 2WI Z, / @3D,4 4JK,4 XY4
960 i !I d h !I d l !I i 2 h !I i 2 V;
' ' 9E, 4 >'
G, 9)7 8 !I P h !I d V; P &,U 9S &N . l !I d
!-
U ' ' /H4.
2 &2 ,/0DI (" M,4 &?, !I V(?I /#M
W2
@31/, @3D , cod. 946 V;, 4 @ V;I @k8 cod. 947 (JK ,I JK ,I cod. 947 +2
˜ ˜ ˜
945 ˜ in marg.
˜ U '
I I cod. 953 ? , /2 Z, /2 cod. 953 JK 4, (JK,
˜ ˜ cod. 954
W2 ˜ W2
cod. 950 (6AI I (JKA
U
cod. 955 4, ˜ /#5 cod. 959 JK 4, ˜ JK4I cod. 960 (post l ) ˜ corr. ex 2' cod.
'
T5 !I /H4;4I /#
W$ 4 !3: T59 !I b P (A.II.50)
fe !I fe !I Z, 4 2 V2 d !I d /0m +2 9E, 4 Z', /2 Z', /
' ' _
V2 h !I h /0m +2 JK2 &B VP ?, /2 ?,
N,4
!I V?:
' ' _ '
965
VP ? , /2 ?,
{22r } ,4
Z,
!I V?: fh Z, /2 !I Z, 4 2
i !I i /0m +2 JK2 &B
/# U + + + 4H#59 @31/, b', I
+2 I 4HM, 2 4H,4
!I &. Z4AB +2
' Z', /
&?, /H, /H4E Z, / T5 !I d !I d /0m +2 9E, 4 &

+ HI + H, / !I (JK YI (N&?, +, 4 &M &B , +, 4 &. + H4#59 %
4 4 4 4 /#W & %Y : &SI + HM,
970
(N&B F,4 r4M &? 4I &

U 4 4'
4I & /#W j ,4
94, & &,4 /, 4 ,4
94, & +. j : +.
d ,4
94, & F I d &, /, +, /HE /H" + HMM
4 44 44 4 4 &B
i !I d 4:9, Z,
+2 G4M; !I < 4, & P I (A.II.51)
' ' ' + ' ' p :
GM; /H4;$4 !3: 6: VP +2 H4Y4, Z, / &?, !` GS 975
!I d 6) 7 8 Z,
+2 G4M; ( < ,4 U ?2 d !I i T59
'DI > 6 &?
,/0
' h
I
!3 p '

G Y, b, b !I V(? (A.II.52)
^ &. ?,
+ HY !I GY , & /#W !` / p
+,4 &M 4p 4
, I j F 4' #$ T59 !I
, +, 4 &M +, &? VI + H6)7 8 &B +2 '
&?, (JKa"/ 4 4 , r; ,4
p
980

(JKaY, +,4 &M +, 4 /H4E +, 4 &? &?, & /H4E &? +H4#59
&? /: 6) 7 8 +2
, r; ',4 +, 4 &M + HY !I GY
4 4 , T59 !I
G p 4JKa , & VM
4
&? 9M, +, &M
4 4
+ HMM
4 4
+, &? +2
4 V' I
&? b 9M,4 +, 4 &M
+, /HE +, &? +2 /H" &? 4M.
44 4
963 /H4;4I ˜ /H4;$ cod. 965 ?,
˜ Z,
cod. 966 ?,

˜ Z,
cod. 968 + H#59 ˜ + H#59 cod.
4
˜
968 &. &? cod. 968 +2 (alt.) cod. 973 /H" ˜ &B cod. 976 6)
˜ 7 8 ˜ 6:
cod.
p ',4 ˜ r;,4
981 + H4 #59 + H4 #59 cod. 981 &?, &;pI cod. 982 T59 ,4 /
˜ ˜ ˜
cod. 982 r;
' ˜
9M, +, &M post seq. /H4E &? hab. cod.
˜
cod. 983 VI V4M cod. 984 b 4 4 +, 4 &? +2
' p 4JKa , & /HE

I M, , 4 2 JK2 &B VP g
, GY' , 4 pG ' p4
985

4, W &N G &. &I b GM?I
!I + H46)7 8 &B !I GY ,I d F , 4 & /#W !` /# , '@k8
' p 4JKa , !3 /= +2 ',4 +, 4 &M + HY
6MI JK4 {22v } G , r; 4 4
GY, /: 6) 7 8 !I b &" (JK !3JKa , +, &M +, /HE /H" '@k8
' p ' 4 p 4 4 4 '
4, W &N G &. &I &N GM?I I M, ,4 2 &. VP g
,
' d h '
990

> 6: VP !I !I <N 6)7 8 +2
<,4 2 &UY4: &?I %
b /0 _ 'DI
h !I d 6: VP ' '
+2 T5 Z, / &?, < ,4 2 P I (A.II.53)
&B 6: +2 h j F , & /#W , &?, ' p

,
4 ! I 4 4 2 GM?:
!` )n8 @ d F , 4 & /#W 6: !I b &"

6: +2
4
995

7 8 Z,
+2 U
4a4E, b V?)n8

6)
+ H6)7 8 !I < , ' p ' p I I
> + H44I 4 4, h VP G &. b GM?
b /0 _ 'DI
' ' I (A.II.54)
&?, b, I fe !I d l !I d !I, !I V(?, b &E
' l
1000 <, 4, h 4, h T I + H46)7 8 !I G &. <, T5
'
h !I d 9N < ,4 2 &?, + H4Y4, + H4Y4, 6) 7 8 Z,
+2 G p &.' fe
U' _ 'DI 2 !-.
2 & 2 b /0
p ', / !` / p # $ b', bMI
Z^ 92 !-. G I N Y, !3 ?
' _ 'DI 2 !-. b &, GM& p 5 JKa4,/, &.
C ,9? > b /0
'
1005 !`?, C g5 G;$ H.I 4I9
/ , +W:

U /I !I ,/" 2 M 2 &N
?4AB

985 JK2 (JK2 cod. 988 +, 4 &M
˜
4 ',4 ˜ V;, cod. 994
˜ + HM cod. 988 r; 4, cod. 994
4, ˜ ,
˜
MAB cod. 995 )n8 o)n8 cod. 1000 cod. 1001 d h cod. 1003 bMI b ( b
˜ ˜ ˜ ˜
6:
p ˜
corr.) cod. 1003 GI G?I cod.
{23r }
!-. I9W
Y2 M, !I M,
' '
fc ]^ VP !I &B9
h t d
j c u
l f i
N/I 6M2 !I c !-. 'I9W

Y2 c !I c
'
6#5 !I ic ]^ VP !I &B9
h fl dc fi d i l dj dh h
de ff t ft i u fd fu fe ft
j df fj c dj c ff fj fc df
fh u fu fc fd dc fi t fh f
dh l f fe dd dd fl j d de
!-. 'I9W
Y2 u !I u
'
;$ 4 u !I u
fuc ]^ VP !I &B9
fi l hh hc hi d fh l fd fe hc hi hl i
fl dh jt ju j dd jd ff fl de jc ji fi jt
de t dj df jf hf je t ft dh dt dd jf hf
u fc jj dc fu jc hj u fu dj dc du jj hj
he ht ft dt du f fe hu ju dl df di fj j
jl dl ff fj hu di fd ht jh je fc fh jd f
ji hd i c h hl jh hh jl he c h d hd
{23v }
t !I t jit ld
fe fi fh fd uc ui ul le l
fc dh dl di if id ih dd iu
fj du jh ji cf cd jd cc it
ff dc jc he hc jl hu cu uf
t dj jj jt hf hj ht ct uj
uu ij cj hh ju hd dt ft c
ut ic ce hi jf je hl fu j
lf ie ch ci df de fl cl f
uh ii il ue u i h d ud
t !I t

92 95 &,4 G. /. w" 9" &

G, &5 . T5 : w7
o. /"
5 )x1 )" G
&. T" T / /2 )n8 S.

,4 " T2 2 )7 / , /01
.
G,4 I 5 2 )* /, 4
G" T$ , P T$4 , G5
wn8 &2 wn84 ,4 )n84 T. G2 &"
w* 9,4 wx1 9. &, 9, 9
{24r } ff !I ff
G,4 wn84 9,4 &,4 S4I &4I 94I o4I bI
wxm w"
T$4 9 & G ^ 9^ )I
/, 4 &2 w* 92 I GI &I G2 / I I
,4 )7 &,
/2 9, . G. G, . TI / I

)n84 2 , 5 wn8 /.
/01 ^ TI
,4 T" )* )n8 T$ 5 )x1 T5 T. )xm 4I

4I /0m
)I S. &5 / , G5 T2 T "

/ H4I T^ I 95 , wx1 / )"
T4I I &. 9. 2 w7 o. P :

\I &^ 9I wI /" 9" &" G" G^

G4I GI &I 9I wI !I
{24v } ff !I ff iuf

9 9,4 wI !I S4I 94I o4I &
w* &" / I /0m

GI wn84 92 G.

G, 9, )I ^ TI )" : wn8 95

9^ &, /, 4 /2 , I )I )* I wx1 9"
&^ )n84 T" T2 T$ / , /01 S. )xm TI w"

,4 /" 2 T5 5 )n8 / I
/. ^ 4I
G^ / 4HI I I , 5 )x1 / P
G \I T. . 2 T . G2
w7 o. &5 T$4 )7 T^ T4I G5 &2 &I
GI 9. wxm G" ,4 " 4I &I &.

wI 9I &,4 G,4 G4I &4I bI 9I
ff !I ff (24r ) ˜ w" : )" . wxm : )xm . / I : VI . ^ : 2 . wn8 : w7 . /0m

: T" (v. loculum supra) .

I : I . wI : wI . ff !I ff (24v ) GI : &I . TI : : 5 . / , :
TI . I : I . )* : )x1 . , : 2 . /01
˜
/I . 4I : I . 2 : , . / : / I . )x1 : . I : T. . / H4I

: TI . 2 : I . T. : . . &I : &I . )7 : )n84 .
" : &" .
h !I h h !I h
{25r }
G,4 )n84 ,4
,4 : ,4
&,4
9,4 )n84 G4,

&,4
,4 9,4 :
fe !I fe

wxm &^ wI ,4 &I
wn84 9" &. S.
G. . G " )I
I /.
w7 wx1 & 95 / &" 9,4
&,4 )" G2 /2 G, / , 92 T$ o. / I
TI T. 5 )n8 )* 9, G" G,4
P G5 &2 , &, T2 T ,4
9 wn8 2 w* , 5 I )xm
^ GI &5 )7
/01 2 )x1 T$4
9. /" w" T"
T5 . /0m
G^ T^ : 9^
)n84 9I /, 4 ^
l !I l
{25v }

i !I i wn8 &5 : 5 T$
9 & w" 92 wn84 &2 )* &" 9,
,4 &,4"
/" ,4 9" T" ,
T" P wn84
,4 &" ,4 T2 /" / G & 9,4 &,
G4, )" G" /, 4
G,4 /, 4 9 " T w* )n8
9" )n84 9,4 T$4 )7 G, )" )7 w7 w" G2 )n84
G : , 2 /2 P G" 2T$4 &,4
/ , )x1 G5

fe !I fe ˜
TI : TI .
{26r }
fd !I fd
G,4 GMI rI VI TEI /, 4 wn84 T$4 EI &EI G" ,4
MI G 9I / !I 94I /"
bI &" GEI
w* 9^ ^ &^ )xm )n8 &, /2 I 9MI

/M I / I T^ GI wI 9, T$ 92 w7

9 2 &5 T5 &. T. wx1 I TI oMI

4I I )I . G. 5 o. G5 &2 & TMI
@I G4I )* &I 95 9. .
/. 5 GI )7

AI G2 wn8
/01 S. / I )I 4I

: / 4HI &I 9I )x1 / , TI , I 2 w" GAI
9" wxm T2 , G, G^ ^ /0m T4I SAI

&AI &4I T wI T" o4I " \I )" S4I
&MI )n84 &,4 ,4 9,4 oEI /E I 9EI P SEI SMI
fd !I fd (26r ) ˜ &EI : /EI . /M I : /EI . oMI

: oI . : . I : 2 . &I : &I . )I : )I . SEI :
I : /MI . oEI : oMI .
SMI . /E
{26v }
fd !I fd

GI wI 9, T$ &" o4I &EI 9EI )"
&5 T5 &. T. wx1 I w" )7 !I 4I G / H4I
o. G5 T4I 4I
)I . G. 5 9 T" &4I 9"
&I 95 9.
/. . 5 bI S4I G4I "
wn8 /01 S. / I G" 94I & TI SEI
9I )x1 / , TI
, I GEI /" P EI T$4 \I

9MI GAI &AI G2 9I w7 wI 2
,4 oEI SMI &,4 MI 92 ,
G^ /0m T^
/M I TEI wn84 ,4 GMI I )xm &, /2 9^ &2
oMI G4, rI VI /, 4
GI w* ^ &^ )n8 )*
&MI )n84 9,4 I AI wxm
/E T2 G, ^ I
@I
: SAI TMI &I 2 T / I / )I
fd !I fd (26v ) ˜ T5 : TI . : corr. ex . ,4 : , .

: quasi I : /MI .
. /E
{27r }
fd !I fd
T5 &. T. wx1 T" &4I T4I w" 5 GI / I
. G. 5 o. 4I G " o4I )I &5 / , 9I
/ 4HI 94I
95 /. 9. . 9" wn8 I &I )x1

/01 S. bI /" S4I wI T$ G5 I
I )n8
P GEI /E ^ G, )n84 VI MI G4,
SEI &" /, 4 9EI ^ 9, T2 &^ rI 9,4 &MI
wn84 EI &EI )" )xm G^ , SMI GMI ,4
oEI T$4 G" \I 9^ , &, TI 9MI ,4 &,4 TEI
oMI SAI
2 wxm )I &2 / 9I 4I 9

T^ w* 2 GI TMI GAI / I )7 !I

G2 I /2 AI @I
& TI wI T
&I )* 92 /0m &AI : /M I G4I w7 I
fd !I fd (27r ) ˜ GI : G, . bI : bI . : corr. ex (et iter. in marg.) . : /0m

. TI : TI .

&, : I . &2 : &I . G4I : GAI . &AI : VAI . )* : oI .
/(; &? /S 5 !I G

9 {27v }
+2 96S /S 5, 4,

!I 9 &N ? 2 ' (A.III.1)
VY5 !-. , " ''
VB& 4 4 4I > 2 Y4, &I /H4WY4, GWY 4 &.
@3D /(p ; &. /S 5 !I (N&B + H(WI M?: _
1010
4 &I g
/(; gs1 /S 5 !I !R
'
@-. (N&B + H4 + H4(WI @3D , 4 /(; &. /S 5 (A.III.2)
'
@3D , : 9 : 9 @-. 9N ' " @-. /: [ /(; &? '(" '
4 4 4
'
!I 2 &I 2 .9,, !3. /?, 4 2 /: [ .9, (N&B + H4(WI

/
<N !3 +, 4 &N 9, / / +2 &N , "
1015
'
' '
T59 G G"/ / / / /
/
/?, 4 2 G'"/2 /H4. ' /H4. &. G'"/2 W , !I 9N +, 4 &N + H,
' ' 7 % ' 4
@3DS T4WY <#) 8 &,/ FI aEI
'
MP /2 2' + H(WI @3D , '(E '
I ;2 2 ?, 4 4 (A.III.3)
' '
1020 2 b 9I 2 M, M, +H4, +H4,P ?2 2 +H4(WI @3D ,

' ' ' + U
"/2 2 # 4 W$ 2 H4(WI @3D, ;$4 <N M, + H4 ,P M)*

' U % ' ' +
@3D , 4 G"/ g/: &,' / 2 ].; ' 2 H4(WI @3D, ' #4WY
&B 9, +2
& 2I + H4.9, +2 2 &B 9, +2 2 + H4(WI
U % ' + ' gU/: I
&,% /\
g/: &,/ G"/ \I H4.9, +2 & 2 g/:
1025 g/:
)7 8 M(" U 2 Z4AB @-?I (A.III.4)
@-? M 4 ' 9: 92 ' 9E, 4 ,/"

9, +2 9, !I (M. 4(" (M. ,I {28r } /(; '4(E
' '
'
' ˜
/ S 5, / S 5 cod. 1013 4 4 " 4 (" pr. scr. et del. 1013 9N <N cod. 1016
˜ ˜
1008
' ˜ ˜ U ˜ U ˜
G "/ !R cod. 1017 9N 9N & 9N cod. 1022 g/: g/: cod. 1024 g/: g/: cod.
U ˜ ˜ add. cod. 1026 @-? M )7 8 ˜ @-?, M()7 8 cod. 1027
1024 g/: (alt.) g/: cod. 1026 @- ? I 4
/(; ˜ / cod.
' &. '4(" '

9, !I b &, (M. /(;2 (M. I 9N &?

9M ' 9) 7 8 ZAB 9M?2 ', I 9M?2 &. VP ' '

Q: 4 <N ;
'' )7 8 GM# &; , <? Y, U I 9
JO &? 4(E 9: 9 M

4 4 1030

/(; &? Y5 &,4 /
> &? b @-. Z4AB &B %
+2
' ' ' U
(4I 4?2 2 &2 b VI & M I ?2 (N ,%/) n8 4.; $
9 &N !I &, /,
4
'
&. !` W$ MI &. +2 ' '
&. VP 9 I (A.III.5)
P
' ' ' ' '
] 4?; $ P &. , ].9 ^ &. VP b G,P W$
_ F 4?, , F ?, @3'D)* T4W$ +2
' ' _
1035

b V2 &B9, V?I P P
'
&? b !` F4 G4"/H &?, &? +2 P 2 W$ I
4?, V? b V2 , V?I_ &B9 +2 9E, 4 2 WY" 2 + ,\
%

% &B9 b !`
2 + ,\
: /S 5 4, &,4 /, 2 + 4Ha4Y, !I & B %M I &N 2 &' I &I I (A.III.6)
%
1040
_ +2
&,4 / /2 ,4 6: /S 5 !I : 9 9 I /(; &. &N

(2 ' ' ' p
,/"
+2 Y4, 2 &?, V?, 4 /(p; &? b Y5
M?I +2 V;: AI b V2 &B9 !I 9N V(?,4 ,%/S ] 4?;
'
/(p; /29% 4I b V2 /(; +2 &B V\ V;: & I &B9, b
?2 V;: & &? V2 /(; &? +2 ?2 V;: (2 '
+2 <M,4
' U B
1045

b ?4A (N ?^ ,/) n8 {28v } +2 % 4?, &B9, M?I

% ' _ /2' @" &? '
& B I & , 4 !3: /(p; V^)7 8 +2 b < , , V?I
&? +2 /sm _ P 2 96I b +2 Z(: AI &B U ' '
/2 V\
'
9 ) 7 8 melius rescr. in marg. 1030 9: 9 &B 9 cod. 1031 &B 9: cod.
˜ ˜ ˜
1029
U
post &? b (v. supra) praeb. cod. 1032 4.; $ ˜ corr.

1031-2 ?2 (N (. . . ) 4.; $ +2
˜
' ' ˜
ex ]4 . ; $ cod. 1034 &. add. in marg. 1034 P ˜ G ,P
cod. 1035 &. VP
˜
&B9
˜ ˜ ' ˜ '
cod. 1039 &B9 &? cod. 1043 V(?,4 V(?, cod. 1044 V\ b cod. 1045 V;: (2
?2 V()7 8 cod. 1047 b in marg., post +2
˜ ˜
ponendum indic. cod.
'
?;$ 4 /(p; ,/2 % &I j /(; &? / , 2 (A.III.7)
' % d '
1050 d ?2 (2 <M,4 ,/2 I 9E4I &B9 +) n8 ] ?; $ i \ I
p
GEW2 &I 9 I j g /2' G: /: ?2 2 Z4AB & , 4 !3: ' d
j
' '
^ V?I , b !I M? '
(JK4I JKa ,%/) n8 / ?
'
, M2 &B9, G , GM? p I 2 rU?I /(; &? !I !3. ?U 4AB

&B9, bM?I +2 Z(: 2 WY" /(; p &? !` &B9 W$ 9E,
1055 4
_
2 !-. Z(: 2 @3DI I /(; &?, 9N M?I +2 Z(: 2 !`

4?, /sm 2 /: &B9, 4?, bM?I +2 Z(:
< S 4 ,%/S ? ^ !I &,4 /, 4 <? Y, 4 /S 5 ' '
' %
U% '
!I &,
P >
g45 G"
I /(p; !-. /S5 !I V(?
' '%
1060 /S 5 !I V(?, ] 4" @-? (JK4M. & )n8 + H42 b V2 5 JO% :
'
C ,9? > b
b,\
' p % '
I GY5 &. !I ?2 2 /0L$ 4 /(; /2 %, (A.III.8)
' ' p 4 /2 %,
V?, M, !I , /0L$4 Z(: 2 ]?;$4 @k8 ?,
!I , /0L$

V?
?, &N &B9, bM?I MA B 9E 4I b V2 &B9, G ,

1065 &?, Z4()7 8 <?M, 4 /2 %, '@k8 5&5_ Z4()7 8 +2 <M, /2 %I +, /0D.
4 4
U
4I &B (JKM, ' %
_ %
2 9E 4 4 +, 4 /0D. V\ &B , de de &W b, %
%
b,;: % MAB /sm
&?, <, 2 ]; $ <M, /2 %, {29r } Gp Y% 5 (A.III.9)
4 ' U ' % _
g95 , I &B /0D. V\ & B , &W
_ /2 %, Gp Y% 5 (A.III.10)
1070 <, 2 XAB ]; &?, <M, 4
1049 / , ˜ add. cod. 1053-4 !I !3.
( . . . ) ? ^ ˜ in marg. cod. 1054 ', ˜ , cod.
˜
1063 V? , melius rescr. in marg. 1064-5 +,

˜ /0D.
4 /0D. ?, ?,
cod. 1065 <?M, 4˜

˜
sæpissime <M,4 cod. 1066 (JK4M,4 JK4M, cod.
U ' %
&B 4 > V\ G , & B ,
/H4;I 6 ?' ^ '@k8 fi /H4;I h !I JO /0L 4I h bW2 ',\ I (A.III.8a)
_ I ti
;I j !I Z(: 2 /0m '@k8 jd
I le < ,4 JK1&5 JK2 Q
sm _ /: de de <, 2 <?
/ & 9N h g
p %
& B 9?,

<, ?2 <, 2 ];$ b?2 <, 2 !-.
X() 7 8 g % 1080

/(; 9N
2 !-.
9 , I ,% : !I &,4 / , G Y% 5 (A.III.12)
U ' % _ _
&B fe V\ & B , ?W$ Z(: 2 !-. > b?2 Z(:
% _
& B , /0D. b?2 Z(: 2 !-.
9 , G Y% 5 (A.III.13)
U '
/(; 96I /: AI &B ff V\ 1085

g5 2 2 <M, g5 2 M. &, /, /2 %, Gp Y% 5 (A.III.14)
p 4 4 44 '
+2 %
GY5 @" 4M. &,4 /, 4 Z4()7 8 ] ?;$ 4 c !I <, 2 /0L$
4 '@k8
%
/2 %, '@k8 ,
/0D? <M, 4 / bM, G , T ) n8 &.
4
,2
I9 % Gp # :
(2 '
/0D. G, <M, 4I G,P 9 1090
I ˜ <I cod. 1075 <

Q ˜ <
cod. 1078 JK? cod. 1088 2 ˜ (2 cod.
˜ N
1073
' 4 1089
% ˜ !3, cod. 1090 G ,P
˜ P % ˜ ,2
cod. 1090 (2 ˜ 2 cod. 1090 ,2 cod.
[
g95 ?2 2 B V2 b?2 < 4I
i < 4I j JK2 <M I t /H;I c JKM. &, /, h bW>4 b 2
4 4 4 '4 4
% '
&,4 /, 4 /2 , @k8 {29v } ie /H4;I 6 ?; 4I je /H4;I c !I JO /0L
4I
' /H" Gp Y% 5 9 rU2 c '@k8 c '@k8 c '@k8 c '@k8 c JK4M.
2 VI b +2
> G , V? , /0D? % <M, /2 %, '@k8 ]M ) * Q '2
1095 4
p & B % ?2 <, 2 V2 <N c b?2 < 4I b V2 Gp # :
U U
;$ 4 g!31 ?2 Q ,4 @-I g!31 b?2 Q ,4 @ B
W( ) 7 8

' '
] ;, 4 @k8 j !I /sm 2 /0L$
4 /(; /2 %, 9N (A.III.15)
' ' p j 4 b
1100 4I P I /#,4 @k8 Z(: 2 ];,4 !I <, 2 /0L$
/2' V\ ' % t t 4 '@k8 h !I b /0m_ /0D"
G, &B , Z(: 2 <?M,
' ' ' %
V\ b W)n8 t @3,4 @ (2 <, & B , h ?W

Y +2 JO%
s _
m U
I _ + %
/ 2 96I Z(: A &B Z,
H4,

]; $ P ; $ ?2 2 !-. &, /, /(; /2 %, (A.III.16)
1105 44 p '

&? !I b /0L$ '
4 @k8 g/: +2 P g/: ?,
_ M,_
' 7 ' ' '
@k8 2 g/) 8 b g/S M <(W &? !-. &,4 / , @k8 /0DE b & <(W
+2 Z(S &? <M, /2 %, '@k8 <(W '
b !I Z(: 2 /0L $
4
b !` &B9 W$ V2 9E , JO' I g /2' T ) n8 <(W
'
'
1110 /(; &? /S 4I g/S M <(W &?, 6. ; I /(; &?
j !I JO /0L 4I h 9E I JKM ,_ JKM. &,4 /, 4 '@k8 j /(;$ 4 2
4
ie /H4 ; I ˜ marg. (in fin. ult. lin.) U
V2 <N
˜
1093 1096 ?2 (. . . ) post ;$ 4 (v. infra) in
cod. 1101 <?M, 4 ˜ corr. ex ZM, cod.
, /0L I h 9E I M,_ j 9N <(W '

&? !-. G , &,4 / , fd 9E I
9N f /S 4I fd fd ?2 Z(: 2 <?M, 4 /2 %I fd 9E I j !I
9 I j 9E 4I g/) 7 8 &?, . ; I /(; &? GM ,_
p
j GEW2 1115

2 ]; $ <M, '@k8 h !I /(; &? /0L$
4 4 /2 (A.III.17)
<, 2 /0L$ 4 Z(: 2 ]; $ <M, h !I <, 2 /0L$
4
4 Z(:

' 4 '@k8 Z(: 2 ];
/) n48 , I l l ?2 <, 2 <?M, $ <M, h !I
4

4?, &? {30r } 1120
2 <?M, 4 ?,
_ <, 2 !-. &,4 /, 4 /2 %, Gp Y% 5 (A.III.18)
'
4?, &? /) n48 , I fe fe Z(:
_ _ % p
Z4()7 8 XAB Z, / &?, 4M. &,4 /, 4 /2 , G Y% 5 (A.III.19)
fd fd Z(S <?M, 4
_ p
X() 7 8 ,4
&?, 9 I ,% : !I &,4 / , G Y% 5 (A.III.20)
' %
1125
U _ '
&B t V\ G , & B , ?,
Z(: (2 Q
% _ ' % p %
G , & B , ?W$ <, (2 <M, 4 /2 , GY5 (A.III.21)
U '
&B l V\
2
&, <, 2 <?M, 4 '@k8 Gp Y% 5 2 &N +2 <M, /AI M()
4 '
7 8,

4?, /(; &? /)n48 ,I &? +2 <, 1130

' ' U ' %
?; $ @k8 &B G , bW> ?2 2 ] ;, 4 /(p; /2 , (A.III.22)
1128 l ˜ u cod.
U ' p %
4, , ;, 4 ,/2 4I /0D" P I /0D" 4I VN MW % $
/0D" P I /0D" 4I VN MW % $ U , , b?2 2 G , ] ?' ; $
U% U U '
' 4 U ' p %
!` q> &EN , /2 &? G, ]?;$ , ;,4 ,/2 4I
I 9N ] ;' , 4 /2' VP ' '
1135
P !I ]?;$ G, g!31 ?2 < ,4

'
b?2 2 V2 9N /
&. bW2 &I , I JK2 g!31 !I /0D"
p % ' ' ' /
, !*/,4 ,/2 /0D" &,4 !I / +2 /2 !I HB 4?,

].; &? +2 b?2 2 Z4AB V2 /: &4 !I G , bW>
p B /(; & , I /0D" ?2 ' /H) n48 /2' VP '
1140 G?A !I V? , b &"
2 V2 9E, 4 ', I Z4()7 8 ]; $ p & B % b, &, !I +, &M
4 4 4 +, 4 &?
_
4?, /sm
V(? !I, M, !I &B9 G,
p /0m ?2 2 GM ,_ <M, ,/2
4
p %

p _ p % '
G, /0m ?,
<M,4 ,/2 b &" g95 ]4;Y !I &, 2 !-.
&B9 /0L $ b &" g95 ] ;Y V2 V(? !I, ?,
!I &B9
4
's1 !I
1145
&B V(? !I, 4

$ TW$ {30v } j !I /sm 2 /0L$ %
; 4 4 /( p % , (A.III.23)
; /2
p %
,/2 /0D" 4I P I /0D" ]; !I VN 5 b V?I I
1150 j !I ?2 Z(: 2 /0L$ 4 /2 %, '@k8 &B U %
G , & B , /H)n84
/0D" ];
P !I VN %5 b V?I I ; $ TW$
' ' p d p % / n8 p % 4
@k8 &B9 !` (JK^ G, &B H) 4 ,/2 /0D" 4I

' % t t %
I I h JK4M, t V\
4' p G , & B , ?2 Z(
: 2 <
?M, 4
/2 ,
'
/2 G ^ , &?, @ ?W $ <, & '
2 !` g
U ' ˜
4, , (. . . ) G , ] ?; $ marg. cod.
˜ ˜
1134 1138 /0D" /0D" cod. 1140 /0D" /0D" cod. 1145
˜
&B9 &. VP cod.
4 @
9W" ] 4;Y !I P
+E, JOp & B % !3
B[ +2 b?2 1155
p % p %
2 96I 2 Z(: AI ?W Y g% &. JO &B !3 ?,
& B
!I +E,
4 @ /0D" !R ] 4;Y !I P 4M. XI v6I I ?, & B %
4%
'
/2 !-. (JO p

p
d !RM &B '
g

' %
4M. &,4 /, 4 ?;$ 4 d bW2 2 !-. &,4 /, 4 /(p; /2 , (A.III.24) 1160
' p % ' fe !I '@k8 c !I , /0L$

V\ & B <, (AI hce 2 <M, 4 U 4 c
bW2 2 96I P AI &B fee
fd 9E I 6 ?' ;$ 4 i /H4;I d JKM. &,4 /, 4 h bW>4 2
' c 4 fu / c
<M4I lce 9EI fe !I @k8 !I JO /0L4I H4;I JK4M. &,4 /, 4
U fee ' % hee hce
4I &B
bW2 2 9N h 9E V\ &B I +, 4 &? +2
Z(S &? +2
VI !I /?^ /0L$4 @k8 /H"
' ' p %
!I /?^ M()
+2 7 8 !I /H" +2
+ H
4 % + HS#W
,\ 4 Z() n8
@k8 GY' 5
4 1170
' p p %
<, AI (JKMA B Z, /2 +2
{31r } 4 P AI G Y5 > (JKMA B +2
VI
p
!I , /0m & JK2 V' I &? M() 7 8 + H, 2 V; !-. (W p
I
4
9E
4I M()7 8 +2
b <MI /?^
&? 96I /: AI /?^
&?
/H" &? <,4 2
4 fj ?U 4AB P
9E, (JKMA p B I u i rU2 /(; /(;$ 4 2
p 1175
' ˜
2 cod. 1161 (AI AI cod. 1165 hce e add. supra 1170 !I (alt.) +2
˜ ˜ ˜
1156 cod. 1171
˜ ˜ p
˜
(JKMAB JKMAB cod. 1172 JK2 2 cod. 1175 (JKMAB (JKMAB cod.
' tf fj 7 8 !I u 9N /H" /0L $ fit 6?', /2

@k8 /S4I <N M()
' ' ' p %
2 VI 9N (2 G Y5 (4I i 9N /?^ /0L $

M()7 8 +2 VI !3.
p fcf tf p I ie \ I fe !I ,p /0m b',\ I
JK4 \I Z2 ie G?()
fe fj + H4, V; <N !3 j !-. b @3D I fl < I fit +2
u <, fj <N M() 7 8 +2 p
1180 JK4 I /?^ &? 9N i /S 4I
/H" &? 9N
% p %
!I (JKMA B /0L$
4 +, 4 &? Z()n84 /2 , G 'Y5 (A.III.25 )

fj <N +, 4 /(; +, 4 &? MA B !3. , /0L$ 4 @k8 Z4()7 8, /H) n48 h

<, AI /: +2 + H4M()7 8 &B <M , '@k8 ;$ U
4 Z4()7 8, /H)n48 i !I

d !-. Gp # : 2 @3D I d < I 'W +2
4 ?,
G
p 4
'@k8 p# :
'
1185
< 4I M()

7 8 +2
4M, 2 &I (" +, 4 &? &B /S4I M()7 8 /SI

/H" &?
p ' p B Gp Y% 5 (A.III.26)
&B !I M() 7 8 G,
' /0m [ @k8 +, 4 &? G?A
7 8 +2
M()
VI !I % /: &? G, p /0m '@k8 M() 7 8 Gp # :

+, &?
4
p ' p
1190 /I
&B &B +, 4 &? MA B !` + HS#W G?A
4 B &B9,
+2 <M , +, 4 &? 9()* V2 &,U 9E, 4 ', I Z()n84 2 !` &.
' < ,4 +, 4 &? MA B Z', /2 Z()7 8
@k8 +,4 &? &B V2 Z4()7 8 +2 4
[

&? +2 !I 9E,4 /: &? < 4I M() 8 +2 !I &? <M ,
7
7 8 !I /0L
M() &? V2

(JKMA B Z', /,4 +, 4 &? Z()n84 /2 %, Gp Y% 5 (A.III.27)
1195
U ' ' '
{31v } ;$ 4 , /H) n48 (6?()n84 (JK2 &B VP Z, /,4 @k8 , /H) n48
' ˜ ˜ ˜ '
1176 6? , /2 (6?,/2 cod. 1179 fe fe !- . cod. 1183 <N 9N cod. 1188 @k8 +,4 &?
p /0m ˜ add. in marg. 1189 &? ˜ add. in marg. 1193 /: [ ˜ /H" cod.
G,
96I P AI ; $ p & B % <, AI M() 7 8 Z', /2 +2 + H4?', / MA B <M, '@k8
' ' [ ' 4

+, 4 &? o#W2 /0m 9, @k8 /: !I +, 4 &? &B o#W2 V2

/H) n48 ? ^ +2 Gp Y% 5 2 I? ^ M, /: [ ]? ^ !I
p %
&B !I JO /0L$ 4 ,/2 !3 ?^ !-. M() 7 8 @3D I M()
'
7 8,
1200

/0L &? /S4I +, 4 &? o#W2 !I +, 4 &?
&B ]; $ !I o#W '

/0L$ 4 /2 %, Gp Y% 5 (A.III.27)
AI P ' _ _
@3D I , /HB P g/: ?,
M, +, 4 &?
' '
&? /S 4I 4(W$ . ; I &? g/: /S 4I o#W !-. M() 7 8

/0L 1205
' U U % p
@k8 ,4 W2 &. bW>4 /(p; /2 , G Y% 5 (A.III.28)
' ' '

+2 +2 <, 2 !-. &,4 /, 4 @k8 g5 2 +2 !R !-. &,4 /, 4
'

+2 Z(: (4I !R !I <, 2 /0L$ 4 4I 2 V2 !R
@3D ' I !R !I P 2 !-. b @3D, 4 2 <, 2 V2 4M.
<, 2 @3D +2 /) n48 2 !-. &,4 /, 4 '@k8 ' U
!I 2 V2 &, /) n48 1210
/(; &? ]? ^ &,U 9E,

'
, I , / H n8 '@k8 !R +2
)
4 4
6 ,4U W2 4U, , &. U % c
JK4M.
I c P &B I /(;$4 2
'
d !I <4I j JK4M.
,\ '
I g5 2 !` W( ) 7 8 +2

^
4I 2 V2 d 4I '
P !-. l 4I !R +2
I l 4,
2 !-. 6(WI I dh
^ h !I i /0m I i !R h '
P 1215

i 9N !R !-. N
I h /: i 9N + H4 , +2
!R !I
/(; G ,P !3 W( ) 7 8 ]? ^ 9N fe Z(:

2
W$4 !I g5 2 >4 !I bW>4 /2 %, Gp Y% 5 (A.III.29)
4
7 8 Z', /2 ˜ Z,/ MAB cod.
M()
' ˜ ˜ /?^
o#W2 &B cod. 1205 /0L
1197 1198 cod. 1208
˜ ˜ P
cod. ˜
(4I AI cod. 1217 G ,P 1218 !I corr. ex cod.
' ' '

Z(: 2 !-. &,4 /, 4 {32r } ?;$ 4 @k8 d 4>4 !I 2 !-. &,4 /, 4 @k8 g5
1220 !I Z(: 2 /0L$ 4 '@k8
W$4 !I 2 4M. &,4 /, 4 '@k8 c !I , /0L$ 4 '@k8 c
AI &B U ' %
P fee V\ & B I !I , /H) n48 hce 2 <M, fe
4
2 96I P ' '
AI f <, (2 fe V\ &B , @k8 4>4 !I P % ' 2 96I
P

W$4 !I
' % U U % U
Z, /,4 /2 , /I B
bW>4 /2 , ;$ 4 2 (A.III.30)
U ] ?' ;$ (6?()n8 (JK2 &B VP '
1225 !I , /0L$ 4 &B 4M. &,4 /, 4 / 4 4
9E,4
&) 7 8 ' I 4 &. /I
&B & B %, Z()7 8 Z()n8 /
4 4 4
' 7 ' ' U U
<,4 ,I M() 8 +2 Z, / b &?, <M,4 ,I &, +,4 &?M ,4 W2
< I
&) 7 8 +2 U
/ 4 4M4I &, /
' B
@k8 (JKMA !I +,4 &? &B /0L$ 4 /2 %, Gp Y% 5 (A.III.31)
' Z', / <M, '@k8 Z()7 8, /H) n8 &? b Z', /,
1230 < 4I o#W +2 4 4 [ 4 4
' '
/S4I Z, / &?
&B !-. (WI /: !I +, 4 &? &B o#W2
[
/: &?
(JKMA B Z', /> /H) n48 +, 4 &? Z()n84 /2 %, Gp Y% 5 (A.III.32)
[ 4 /2 %, '@k8 &,U ?,
_ G , & B %I
/: !I +,4 &? &B /0L$
+, 4 &? MA B Z', /2 Z,
_ +2
p I '

# 5 P A o#W M() n8 /H) n8
1235
4
I '
A +,4 &? Z, /2
&B ];$ +2 p

#5 P AI !I
&B & B %,
'

96I <, A v#. Z, /
&B +2 #WYI /?^ &? 96I <,
I
/H" &?

1240 !R !-. &,4 /, 4 '@k8 ,4 WY2 &. M, 4I bW>4
+2 (A.III.33)
!I (ult.)
˜ ˜ ˜
1220 e corr. 1221 fee ult. e add. vid. 1231 &B add. in marg. 1231 +, 4 &? ˜
'
;$ 4 add. cod. B Z, /> (JK?,/2 M()n8 cod.
˜ ˜
4 &? (6?,/, add. cod. 1233 (JKMA
1233 +,
' ' U U U U '

+2 !-. &,4 /, 4 @k8 29M?2 Y% 45 G +2 29M?2 Y% 45
4M.
2 V2 T5 !I < ,4 ', I 2 <, 2 V2 T5
+2
T5 +2 TW$ ?U AB '
+2 4M.
2 ]? ^ V2 G

4 4
4I <, 2 ] ;' , 4 G
4M.
2 9E
+2 4M.
2 U;$ 4
+2
'
{32v } 1245

!" #

' 2' 9E,
M, 2 + H4I/: 4 +2 @31 9M) n48 + (A.III.34)
U' M, +2 /H" 2' /:
/S 5I /: /H" P I b +2
' + ' ' '
@k8 H4I/: M, @k8 M, + H4a4I
+ HI/)

4 7 8 @k8 + H4 + H4I/)7 8
M, 1250
' 7 8 '("
'
vS? /: !`9, !-. 4, /? /) 4 /" &N (A.III.35)

&6

fi fc fh fj fd ff fe t l u i c h j d f

dt dl du di dc dh dj dd df de ft fl fu

' 7 8 +2 '
/, (" 4, /? /) JK4M. /,% &S a, 9N (A.III.36)
/: '
!` ] +2 % &?,4 /2 %, '@k8
9; &N
' !I
/: +2 1255
' p % U U ' ' %
@k8 GY5 @" B
&. 4M. &,4 /, 4 b ] ?;$ 4 @k8 /,& !I @31
' ' % +2 ' ' %
& , 4 !3: g5 JO4 /,& /: /:
+2 &?, 4 ;, 4 /2 ,
U
!I +2 ut vid. 1243 V2 ˜ V2 cod. 1243-4 ;$
˜
1242 4 T5 ˜ post 4M.
2 scr. cod.
+2
' ˜ ' ˜ x1
1244-5 ( . . . ) <, 2 in marg. post ult. lin. 1254 &Sa, &) cod.
' &?
/: !` /H) n48 ] 4;Y +2 ?2 <, {33r } P &
7 8 !` 2 &?, 9W /) 7 8 +2 '
/) % g& , & &? I &? !3JKa ,
% 7 AI 4 !3JKa , &
1260 !` ] +2 /) 8 +2 P 2 ]; $ 96I P
I &, /, ,/2 p % $ U ?' ;2 7 8 b
2 Q 44 & / &? ]; /)
% &? < I / / &? ];
!` ] /: +2 P & 4
$
p 4JKa , G 4)I &? b V2 ] % ' ' /) 7 8
4 G +2 &?I 4 &. &
4I
@31 /: Z AB @-?, '@k8 /)
)7 8 &JO
M 7 8 b 96I
4 4

&?, ' p % ' * /: ' /? ,
p 2
1265 +2 4 ,/2 I &() @31 +2
U p
'@k8 hl
; I JKp? ^ I dh ,&B p %
r2 d JK4M. 9I @34 !` ]
+, 4 /0D. W(
+2 ) 7 8 &N &? p I dc
; I JKp ;' $ '@k8 ce
; I
9N f b +2 G p 4
I 9 /: !` &? !3JKa , I rU2 g /:
%
] /: p I dh < 4I ,p
&
JO&?
+2 / &? ]; $

@34 9N 2 /: ' ' G(M? p I @3 /: !` &? b, !3JKa , I
1270
4
'
N/"& 5 &N +2 ] : <N /: V4Y5 @-?, 9= 4I &N
I @31 /: % '
P
+2 &. @" /H) n48 /2 , 6I (A.III.37)
% ˜ 7 8 ˜
/H) n48 , add. cod. 1259 9W
% ˜ 9W cod. ]
˜
1258 1260 cod. 1263 /)
p ˜
/: cod. 1269 JO&?I JO&?, cod.
n8 /H) n8 '@k8 G p %

b M() 4 !R + H4I/) 7 8 Z()n84 ,/2 /: M,
' ' n8 / n8 ' ' ' ' '
@k8 !R 4(W$ M() H) 4 @k8 G Z()n84 @k8 4(W$
' ' n8 / n8 '
M V()7 8 Z()n8 @k8 G 4(W$ M() H) 4 !R Z()n84 1275
' '
@k8 /) 7 8 < I !` M() 7 8 2 <M I Z()7 8 ]; $ & B %,
4 ' 4
M()
7 8 2 <M , @k8 !R /) 7 8 < I , M() 7 8 2 <M ,
4 4
G /)7 8 < 4I

!R /: M G?Ap B 4U.,
{33v } @31 P (A.III.37)
/H /) 7 8 @31, U' Z(: 2 G p ' 1 ' p

4s G"/, Z, / G 1280
U ' ' '@k8

4, , Z(: 2 <(W$ !R /H, Z, / G Z()n8 b &"
' '@k8 ;$ U
/H, Z, / !R Z()n8 b &"
4 /H /)7 8 @31,
' ' U '
Z()n8 @k8 /H /) 7 8 @31,
, Z( : 2 < (W $ G
_ % ' U
, & B , @k8 /H @31, ?,
Z(: 2 <(W$ G ' !R
Z4AB GM
/) 7 8 % b &" M() 7 8 XAB_ !3.
WA B
_ p % U
&B 45&5 @31 P
'
b !I M? &B9, I/: B +2
MA V' I &? !` <N !3 WY
V' I 9(S +2 /2' VP ' n8 '
1290
G: 9I &B9, /) 7 8 &. +2 !I Z() b,

'
2 <M , @k8 6?4AB +2 9 : &? b !` &B W$ & B %,
%

JK4?, " H /: < I V()7 8
/
+ H4I/) cod. 1280 Z, / ˜ Z, / cod. 1283 Z()n8 ˜ Z()n8 cod. 1285
7 8 ˜ /: < 4 I (pr.)
˜
1273
<I cod. 1289 M() 7 8 ˜ V()7 8 cod. 1290 V' I 7 8 ˜ corr. ex
˜ cod. 1291 /) 7 8 cod.
/)


p I ' ' %
1295 MAB G # : P A !R Z, /,4 /(p; /2 , (A.III.38)
U ' ' ' '
, /0L$ 4 &B 4M. &,4 /, 4 !R ] ?;$ 4 @k8 !R Z()n84 @k8
/I
&B & B %, (JK?', /2 9()* . &, /, ' 7 8 !I
/)
4 !- 44
' ' U ' 4 ', I 4 &.
Z, /2 F , @k8 &, + H4 + H4I/) 7 8 9()* V2 9E,
7 8 < I
&) 7 8 +2 I &,U
' 7 8 < I M() b
/) 4 ; /) 4 7 8 +2
1300 !R

j !I M() 7 8 /0L$ '
4 !R + 4HI/)7 8 Z()n84 9N (A.III.39)
' ' 4 '@k8 ZM,
2 T ) n8 &,U
@k8 {34r } j !I !R h !I /0L$
7 8 +2
7 8 96I M() <, 2
1305
$%
!R G?()

G p I &,U 4
/S W$
p b',P '
(A.III.37a)
' p '
!36I ff \ I G G?A B @k8 !` M() 7 8 !36I fh \
I
' 7 ' p B ' 7
@k8 M() 8 !36I fu \ I !R G?A @k8 4, M() 8
p I df \ I b ]; p %
$ & B I hd G ,\ I V(_)7 8 G?A p B
JK2 G 4
p ' '
1310
< I 4, M()

7 8 G 7 8 <N u < I !` M() 7 8
4 @k8 /) p ' 7
7 h
/) 8 9N <I M() 8 G4 @k8 !R /) 8 <N 7 fe

G
' ˜ ' ˜
b Z, /2 Z, / b cod. 1302 corr. ex + H4 cod. 1303 T ) n8 T ) n8 @k8 cod.
˜
1298-9
p ˜ p
cod. 1307

˜ GM I pr. scr. et del. 1308 !36I (post.) ˜ corr.
1304 #: & JK#: !3

ex <N cod. 1309 4, corr. ex cod.
˜
' + 7 ' / U
+ H4 H4I/) 8 Z, HI ;$4 &,4
!I !R : 9 2 (A.III.38a)
' '
4I !R
9E Z()n8 @k8 + H4?,
?W $ ,2 % (JKMA B 9E

4I 1315
df /H4;4I &B U 4I !R ] ?' ; $ /0D. ? 5

4M. &,4 / , +, 4 /0D. 9E
fht 9N (JK4?', /2 9()* !-. &,4 / HI fhu 9E 4I u !I , /0L $ '@k8
'
<M , @k8 fu b
&S &. /I
&B & B %I dti /H;I
' 4'

/)7 8 <N u <4I dti 9N 9(S +2 dlt 9N fu Z, /2
!R /) 7 8 9N fe < I + HI/) 7 8 MA
B +2 JK , '@k8
M
4 4 4 1320
' U
+ H4 + H4I/)7 8 Z()n8 ;$ 4 &,4
!I G : 9 2 (A.III.39a)
' '
u 9N /0L $ @k8 6# S I cf 9E 4I j !I , /0L I fu 9E 4I
U 4I Z4()7 8 {34v } Z()n8 j !I !R h !I
&B 2 <M I cl 9E
fe < 4I M() 7 8 +2 b <M , '@k8
' 7 8 <N u 

&' () *+
,
-+

/(; &. 2 /S W$ /H2 V\5 '
+2 6I (A.III.40)
. p

9; &N !- G Y% 5 2 !` &B9 1330
fdl ih jd fi l h d f
lc df df c c f f
fh fh fh fh fh d
cd cd de h h
fde dh dh l
le fi fi
ti jd
ih
1317 fhu ˜ hoc u add. vid. 1320

7 8 cod.
+ H4I/) 7 8 ˜ corr. ex /) (in tabula) le ˜ l cod.
' U V(?
p I
&. bW >4 /(p; /AI &)7 8 &JO
' '
&? !-. &,4 /, 4 @k8 2 /H" bW>4 /#W +2 P &.
' p %
/0DE ] ?;$ 4 ,/2 v?, P I /0D" 4I VN MW % $ ; $
' ' '
@k8 /I +2 &? b
9#5 !I &? G , bW>
U %
4I P /0D" 4I VN MW % $ ;$
4 ;$ 4M. &,4 /, 4 /2 ,
% ' ' p %
1335
&,4 /, 4 /2 , @k8 !R &? G , bW> /0DE ] ?;$ 4 ,/2 /0D"
' p %
/0DE ] ?;$ 4 ,/2 /0D" 4I P I /0D" 4I VN MW % $ ; $ M.
4
[ U
/: !` !3JKa, !` &, &EN /, G &? G, bW>

b,4 &,4 !I Z(: {35r } (2 ' '
W2 +2 <M, @k8 /#W b !I &.
' '
1340 &? < ,4 , I /
/#5 !I r. !I GY &?
'
/ +2 '
/2 !I +E,
4 @ I /sm &
bW> rI /0D" /H)n48 !3
&? ' n8 &?
4 @ 9E,
+E, 4 &
) ,/ bM, !I &
' ' '
/#W b !I &. bW2 &I , @-. I /0D" JK2 /2 !I
$ JKM.
I fj /0D. 'W /#5 +2
6; /(p; /(;$ 4 2
4 U ft 7
1345 G , bW(I /0D" 4I 9E, 4 ; $ Z4() 8
^ ]; $ '5 9N
f 9E 4I /0DE /H)n84 /2 %, c 9N /I !I '
&? /0DEM
% '
je /H4;I 6; $ +, /0D? !-. &, /, /2 , @k8 de ?2 /H;$
4 44 4 4
% '
/H4;I 6;$ je !-. &,4 /, 4 /2 , @k8 !R &? &?aI JK4I /0D" '
% '
V2 hc !-. &,4 /, 4 /2 I G &? &?aI JK4I /0D" hc
l 9N /H4B &? G , bW2I /0D" I ; U iu /
4 $ H4;I 6;$
'
1350

bW2 &I , &U%Y 4: @-?I &. /#W &N !I < ,4 fj b?2 9E 4I
fj
' '
G;$ /H" VI g95 /" & , 2 !-. JK4I, !I VA. I &N !-.
'
!`?, C g5
'˜
, corr. ex cod.
˜
1340
˜
/ cod.
&? &?, cod. 1340 / 1343 1345
U ˜ U ˜
; $ ];$ cod. 1350 ; $ ];$ cod.
/0D. > '

{35v } 4 5 ?,
, &. ?5 WAB M, 45B

: : : : : : : :

:

:
:
:
: : ,4 ,4
G,4 G,4
: )n84 : )n84
&,4 &,4

: : ,4 ,4
9,4 9,4

/, 4 : /, 4
/(; &. [ {35r, 16} (A.III.41)

2 /S W$ /: &B &N 1355
/(; &? <?M, 4 /(; /2 %, , V(?

/I !I
' p
/#W !I !3 V()7 8, BU
Gp Y% 5 U /I Gp Y% 5
/2' VP' ' ' !I 2'
!I !I9, /H) n48 u u @k8 c c @k8 j j /I
/HB !3 ,4 V2 JKI Q ',4 !3 9 4 !` / p # $ JO /HB I
4 ' ' '
ZI9 5) 7 8 !I M. ZI92 /(; &? I G, JO, JO
4 4 1360
_ %
/sm 2 V2 &? /#5 5
!-.

' ˜
& cod. ,4
˜ ˜ ˜
1356 post fin. lin. in marg.

1357 !3 1358 2 2 cod. 1359
cod.
/2 I
' h h ' d d I U ' %
@k8 @k8 4?M,4 /A B
4?M,4 ,4 /2 P
' ' ' ' '
!` /# , @k8 < ,4 2 <?M, 4 /2 VP !I /H) n48 fe fe @k8 l l @k8 i i
' ,4 V2 JKI Q ',4 !3
9N /(; &? I JO /HB !3 4
9 4
1365

{36r } &)7 8
9^ &N &? b /#5 !I 45)
7 8 !I GY

' j j U %
@k8 g!31 <,4 rI /I 4?M,4 /2 I fc /(;$ 4 2
f < 4I u u 4?M, 4 /2 %, '@k8 ;$ U
4 g!31 < ,4 rI c c 4?M, 4 /2 %,
'
c !I g!31 j !I < ,4 JK4I Q ,4 !3 94 !I /# I
fc / p sm &I 9 I fc 45) 7 8 !I I &SI &B u !I < ,4
!I VA. I &N

1370 !-.
p bE
!I &,4 / , G Y% 5 2 &)7 8 &. !I &,4 / , V(? &JO >4 &I
' ' p 4, % 2 &?, M /'
!`?, C ,9? > M 2 2 !-. GY
'

GM(" , " @3S M I + &N +2 Z2 4I 2 ,/" &I

/,5
JO % !I JO. /)* !-. ? #
4
' '
1375 S W2 MN 9N (" ( M?2 .& &()7 8
V; v?, +W: 9N
˜
1371 &. add. in marg.
Arabic text of B 335
!RB
9
g !R "
I9

?', / !I Q I9
&? GY,/, !I
4
(Extracts)
Y, V^ ,4 WY2 '
!I G M ) * &. 9N Q I9
&? @-. (B.1)
U U U U ' ', / 9
&. /#I ^/. 9= &. 2 ] 4" N&. 9E, 4 ? 4,
U
,4 WY2
&? JK4I GY , _ 9 $ !-. V(W$
4, ?W 4 ', I M Z', /2 V2
,U WY2 &. U U U U ' '
Z,
&N 4 /#I ^/. 9= JK2 ]^ VP P QI9
U U
4?4= (#$ (# 49 &. JK4I (Fig. B.1a) !`
9^
U B /# U U U ' ' ,& %
&B /0D. WA I ^/. 9= JK2 ]^ VP &B9 +2
%
!` NJKa, W()7 8 +2 %
,& /I &. JK4I (Fig. B.1b) 4,
+2 ,& / &. JK4I (Fig. B.1c) +, 4 /0D. &B
%
&. JK4I (Fig. B.1d) ?, / +, /0D. ?,
4

!` NJ
% Ka , 4,(

+ H4?,
M, !` NJ B V^ , M +2
B WA
% Ka , WA ,& %

l hj fl fe dh fh u df ff d t h
jj dj fj de fi fd fu fj t u c j
dl j jl fl l dd fc c ft i f l
Fig. B.1d Fig. B.1c Fig. B.1b Fig. B.1a
(Edition, lines 2–12)
&' 2
W$4 2
JO ? !I V;I (B.2)
+ H4(WI @3D , 4 / ', /2 !I Q I9
/ ?

&? , ' @-. (i )
' &. , /H4;$4 U Y4,/, Z', / !I &? G',/, (N&B
VP !I !3
U U ', / +2
2 I &. F? , JKL?, # S / /
4
? &B
T4S WA B Z', /> T)n8 & ? 5 Z', /> T)n8 ', I ?W $ Z', /2 b
44 4 4
!-. 4 /# $ ? _ '
, / &N +2 &B VP P ' I M, Z', />

/,
' U U U U '
b V4I I /#I ^/. 9= &. JK4I !3 &. G ,P
' &. , /H4;$4 U Y4,/, &. Y ',
v# 2 GY4,/H
Z, / !I !3
QI /H. /HE Z', / VB !I & /H?; ' U

4 4 4 Z, / &. I /H4E
v# /H4. GY4,/H b & V4I
U U ', / !I &. G',/, ,
(ii )
4M. VJK1 (# 2 Y4,/, ?
' ' ]^ ' '
+2 ]^ VP !I Z,4 & QI9 &? I/. ]^ !I 6M,4 &?,
I9 ^
' &? /? , ,
/U I P Z', / '
VP !I Z, 4 & I
&,4 /, & Z', / ZM^ !I , /0m
T5 &? , & B % 2 ]^ '
' ' ,
+2 ]^ VP !I Z,4 & &? 96I V;: AI Q I9 &. /?
Q I9 &? /? , ,
?W Y Z', /2 V2 Z', / b 9 ^

&B 9N T5 &. , /0m MB !I W( ) 7 8 Z', /2 !I 9E, 4 &

!I Z, 4 & &? ', 9 I WA B + H,2 % I WA B !I 9?,

4 \
? ', / /,5 . W( ' ' '
/I % &N !- ) 7 8 Z, /2 9^ +2 ]^ VP
' $ !I &? &. , /0m BU
Z', / P
Z, / b ZM^ ];
& &? /? , ,
/0D? Z', /2 V2 9M# 96I V;: AI
' '
/0D? Z', /2 VB !I Z, 'W Z', /2 9 ^
+2 ]^ VP !I Z,
4 4
M, 9N 5 ];
' $ !I , /0m &B ,2 % 9N &? &. , &B %
' ' ' '
]^ VP !I W Z, /2 !I Z, 4 & &? 9N M, ,( % M, \ I
JK2
(Edition, ll . 205–228)
(Construction of ordinary magic squares)

(Squares of odd orders)
'
M Z, /2 !I Q I9 &? GY,/, !I V;I (B.3)
4
' &. ' ' 9 I (i )
M Z, /2 !I JKM#$ g&Y, <? Y, 4 !3
4I /HW$ 4 9N & ,9, +2 4 !I Z, 4 & &? 9N
4 T5 GY
4 !I Z, 4 2 &B _ I ?, _
<, &I 9E, 4 T5 GY
9
;
9# /#5 !I 9 4, ?,
+, /#
4
!I ?,
9
4, 4 > 4M.
p ' '
9: &B ]^ !I Z, !3 &. GM= 4M. < ,4 /?
T5 GY 4
'
b WA B ' T5 GY 4 !I 9E, 4 +E >4 ', 9 I
$ NJ
?W % Ka , &B N& % , Z', / &N 9 4, !I Z , !3 &.
&B9 +2 ( # &. Z', / &N +2 49 9 4 !I Y, I

) 7 8 G?
W( I ,
&B +2 N& % , M?:
<?4# v# !-.
' 4
+2 &B VP !I Z,4 & &?
;I T5 GY 4 !I Z#,
' ' &? 9N &B U B
]^ VP !I 9E,4 <? Y, 4 & /0D. WA +,4 /#
' 9 ^ +2
Z, / &N
&B !I (N&B G Y, T I +, 4 /: [ +, 4 &. GM#$ <? Y, 4 >' (ii )
_ ' ^
+2 4, !I !R ,4 /
&B !32 , /? 9# 9 T5 GY
^
+2 &B ]^ ' 9E, 4 M &. &N
9# 9 !I 6.9I
' U ' 4 !I G Y,4 2 &B
, rJK1 ]; ( G GY P /?
B !3. ' '
/0D. WA ]^ VP !I Z,4 & &? > !` !I

' '
I Z, / +2 4 !I GM;: '
!I T5 GY W( ) 7 8 9 I
Z,
+ H4M ) * + H4(W, @3Da , <N /0D. 9E, 4 /0D. W(a ) 7 8 a> !`
' $ (N +, /, JK2 + H,/2' /2'
JK2 + H4,/2 M, ? 5 &B ?W
4 4
, <? Y 4I / WI 2' I '5 ?,

9E
,
4 > , (N + H4B /,

5 9 4 !I / WI I 9 4 +2 ,4 / 9 4, !I
4
!I /I &B 9E, 4 +E >4 @ I b /H4. 9E, 4 +E >4
( &B V, GY 4 !I 9E 4I JKI ?W Y +E MI ,4 / &B
U ' 4
2 &. P GY4, .
94 +2 GY 4, !I Z; $ /0D. WA B
'
2 b +2 V;S4I 9 4 !I, v(' , /U I P BU
Q ', ] 4"
' '
Y
?W !-. &,%
&. JKI &. .
I 9
4 4
/E2' &.
, 2
!I /I
, / !I
/M 4I )* 6MP &N
4 9E, b +2
5 9 4
$ M, 2 !I 9E U
&B 4 &B !I Z; I (iii )
( ?W 4 I 5 9
GY
4I & 4 !-, !3 ,4 / &B !I Z; $ G)n84 /0D. WA B

+E >4 XY4 r /H" &. ?,
4MI
<N BU
&. U
&B9
4 '
?W$ +2 /H" /0D. WAB ( !I I +H4M4M &B )n8 Z; $

v(' , '5 > JKI G
4 !I,
9 4 4 Y I
(Edition, ll . 38–69)
'
!I /#
&? &B 4 4" (B.3) ,/"
> V;: & I (B.4)
G
&. &N Y, +E
>4 &I ?W
Y
&B !I G
Y,4_ &
Q I9

M Z', /2 !I
V /0L$
d t h j d f
u c j i c h
i f l t l u
!I <?4# v# !-. &B9 +2 49 &. Y, !32 ', JK2 (i )
Fig. B.4b Fig. B.4a
' % '
&B VP
+2 !I 2 , & B (Fig. B.4a) !`9 !-. Z, / &N 9 4,
'
5/I GY 4, !` ,4
VP !I 2 M , + H4#5 +, 4 /#W +, 4 /#
,4 / !` GY
!I ) n8 !3 4 b !I P 2 M , &B 6: +2

&N !I Q I9 &? 9E, 4 V;: AI 6) 7 8 &' ^ 2 9 ]; '
'
(. . . ) (Fig. B.4b) Z, /
&B9 +2 <?4# v# !-. 49 &. Y, !32 ', JK2 (ii )
&B U B +, /# +2 ' ' !I
P /0D. WA 4 &B VP !I 2 P Z, / &N
B /? 9# +2 + H4#5 + H4; ' '
/0D. WA &B VP
+2 !I 2
' '

&B VP b !I GYW (Fig. B.4c) &B U
4I 9; &N +2
' _ ' U ' U
@-. &I ,4 WY2 6M4^ , !3. ,4 &. W$ 5 2 &. M,
4M. <, &I T5 ]? ^ I/#
9E, 4 5 &N '^' B +2
' 4
VP !I 2 M , I T5 GY 4, # S 9 ;
V, &?, b &?,
' 4 4'
/H4;$4 !3:
4M,4 & GY4 !` T5 GY4, T4S ]; +2 GY4,
'W Z^92 !I M M Z^92 !I , + H4 , Z^92 !I &B9
4M. VJKDI (Fig. B.4d)
9; &N !I (" V;: N/,5 % b &"
' / '
!I + H4M + H4,4 / +2 &B VP !I 2 H4;$4 , b JK4I 2 V,4 &?,
B V, ]; ' 4 !I 2 Z2 &B ]^ '
/0D. WA +2 T5 GY
I N/H4. b &" /0D. WA B 9N M, > ?,
' /, ' &B U
' 4 '
GY 4 !` T4S ]; 5 9
+2 4 +2 &B VP !I 2 M ,
(Fig. B.4e) 9 ;
G& . V,
d t h d f h j d f
u c j j c u i c h
i f l i t l t l u
Fig. B.4e Fig. B.4d Fig. B.4c
(Edition, ll . 71–78, 88–101)
% !I Q I9
/,5 &. &B 9 V^% 9N V /H4. /0m JK2 (B.5)

T5 GY
+2 4 !I T5 &? G Y, 9N /I ? ', /
GY S 9
4, #4
;
&B T5 !I &B9 G
+2 Y, '@k8 Z', /
9N ,
/ I GY
4, !I Z , M 5/I GY Y, T5
4, !I + H4 , G
' !I 9 % ' % B +

4, M, , r2
Z, / &N !32 , /2 MA H4, /I GY 4,
5 &B G ,4 / &B !R T5 GY 4 N&B 9E, 4
% ' v(' , Q ', 2 ] " &B ]^ '
, 2 9 4 !I, 4 +2 , 9E,4

Y, /0D. WAB +2 &B ]^ +2 '
+H4Y4, !I V;: &I 2 TW
$
G
' +
&B ]^
+2 +2 H4I/# &B !I 2 [TW $ G GY 4 !I !I

/: /# !I !I GY, /0D.
) 7 8 Z', /2 !I Q I9
&? GY,/, !I [
W( 4 (v . B.15) /: : (B.6)
V ,

V +2 /0L$ Z', / &N
!I Q I9 &? G',/, +E >4 &I (i )

v# !-. &B9 +2 ,& % 9 &. +2
4 JK4M. V(?, 9^ %,
U I Z', / &N !I <?#
&N !-. I +, 4 /#
!I !3 &. 9E 4
(Fig. B.6a)
9;
U ' ' "/ , '@k8 (ii )
M , JO9
4, +2 Y% 45 JK2 /H4? , 6)n8 6MP &. &N
' !I 2 95 M Z', /2 !I Z', / T5 !I !3 &.
VP +, 4 /#
!` &B 6: !` MI GY
' 4 4 4, !I 2 M4I GY 4, !` JK2 &B
. /H;$ !3: [ !` &B9 JK2 + H , VP ' + * +
&N !- 4 4 /: 6: 4 H4M) H46:
' 9
4I '
!` T4S ]; +2 ' 4 !'I 2 M , @k8 (Fig. B.6b)
9;

9; &N &, !`9 !-. 6 M, 9 4
!-. /H4;$4 !3:
(Fig. B.6c)
c fd dh dj f c fd j d f c h j d f
ff t de u fl fe t de u fl fe t l u i
fi dd fj h fe fc dd fj h ff fc fh fj fd ff
l ft i fu fc l ft i fu fi de ft fl fu fi
dc j d fh df dc dh dj fh df dc dh dj dd df
Fig. B.6c Fig. B.6b Fig. B.6a
'
T5 Z, / !I !3 &. V ,
9) n84 V !I /: [ : (iii )

4, !I /0D. ?,
/H4;$4 !3: '
+ H4a,4
92 +, 4 /#I !-. (Fig. B.6d)
GY
+, 4 /0D? GY 4, !I 4( /H4;$4 /0D. ?,
GY
4, !I , + H4 ,
' '
!I W W GY 4 , !I /0D. 4( /H4;$4 4( GY 4, !I /0
D?

+, /0D. ?,
?,
GY 4, !I /0D. , /H4;$4 /0D. 4,( GY 4,
U 4
,4
92 wn84 &" G,4 /#I !I 2 /H4;4I /0D. !3 , GY 4, !I /0
D.
4I 9 '
+2 4 !I 2 M , @k8 (Fig. B.6e) &,4 /#I !I
% ; '
4I /2 P b M , I 6 M,
9

9 4, !` T4S ];
(Fig. B.6f )
9; &N !I 2 .
!-
c dj dd fh f c h j fh f :
ff t de u fl fe t de u fl

fi d fj dh fe fc d fj dh ff ,4 )n84 G,4 ,4
&,4
l ft i fu fc l ft i fu fi T$4 wn84 /, 4 9,4
dc fd h j df dc fd dj dd df " &" )" G" P
Fig. B.6f Fig. B.6e Fig. B.6d
(Edition, ll . 316–334)
(Squares of even orders)

' ' '
Q I9
&? JKI V(?, +E
4 >4 !3

'
? ,
/2 2 (B.7)
' '
, / b /0D. 'WI

+E >4 XY4 ?,
9N V?, ?
,& ' '
&B9 +2 % &.
I +,4 /# !I QI &. JK4I 9E, 4
' ?,
JK, . G ,P <?4# v# !-. JKI G Y, !32
VP !I 2 P 4 4
' ' B
]^ VP !I 9E,4 <?Y,4 & &? 9N WA +,4 /# +2 &B
2 /? 9#
9#5 !I Z, 4 +E >4 @ Z', / &N 9 ^
+2

' U
/#5 /HW2 ,4 / GY 4, +,4 /# !I ,4 W2 .9()* 9E, 4
' ' _ I /# /?
GY 4, ]^ VP +2 !I P JK2 T5 9#
'
/: [ GY
)* &N 4 !I ,4U W2 4I 2 9E, 4 G)n84 &B

M ) * Q I &. JK4I 9E,

4 !3 ', / !I &. 9E
? , /
(v . B.1)

Z', /2 !I Q I9
?, &? GY,/, !I V;I (B.8)
4
' %
, / !I QI9 &? GY4,/, !I V^ ,r Z', /2 ' + 'H4, &I (i )
?

Z', /2 ;$ U ' V^ % ' + ' n8
&?, ?, 4 / ? ,
/ ! I H4, +) /
'
4I &. @31
+ H4,
' <?4# v# !-. 49 &. ' 9 I (ii )
Z, /2 !I N? ^
' '
&B VP !I 2 ,, P (Fig. B.8a) &B9 JO&, M?:
% ?,

' U +
VP !I 9E,4 <?Y,4 & &? 9N &B H4M, ?,
+, 4 /# +2
9 4, ?,
, # '

V, &?, < , JK2 ]^ '
!3 4S /#5 U
?,
U 4 4
^/. 9= <#59 /#5 ?,
T59 !I

' ' '

V;: &I 2 Z2 VJK1 6M,4 &?, (iii )

VP b /H4)7 8 , +2
_ ; Z', / &N ,
+2
Z2 (JK4I 2 ZAB 9 &B +2
4 + H4,4

U V, ]; ' + H4#5 + H4Y4 !I 2
;$ 4 P &. +2 +2
' ' +

!` ]^ VP +2 H4#5 + H4Y4 !I 2 M , +S I + H4M, ?,
' ' + +
Z, / , #4S /#5 ?,
P V, ]; +2 H4#5 H4Y4

9; &N
N&.
^ JKI 2 &. &I T5
!-. + H4M, ?, 4
(Fig. B.8b)
h fh fc f h fc fh f h j d f
t u i fd c u i l l u i c
c ff fe l t ff fe fd fd ff fe t
fi d j fj fi j d fj fi fc fh fj
^
Fig. B.8c
+2
Fig. B.8b

&? ' Fig. B.8a
9# 9 <#59 /#5 ?,% , 4M. <, &I (iv )

4 !` !R GY
4 !I 2 V , U
GY , M,4 &?, VJK1 ;$ 4 &N
/?
Z, / !` /0D. X2) 7 8 GY 4 !I 2 4 G GY 4 !I 2 G
!I 2 /0D. !R !` +2 GY 4 !I 2 4 /0D. Z, / !I 2 /0D.
& I 4 Z5 !I 2 Z5 !` X2) 7 8 !I 2 M , 4 /0D. !R
U U U ' /#5 ZAB V, &?, V;:
&N !-. /#I ^/. 9= Z, / &N 4 4
(Fig. B.8c)
9;
Q I9 &? +2 , 9M#2 P 2 95 Z', / &N !I V;: &I (v )
' ' ' '
T4S ]; +2 + H4#5 + H4Y4, VP b +W: &. ,,
U _ '
& ;$ 4 P JK4I 2 ZAB (6 +, 4 /H4# + H4Y4 Z2 T5 Z, / ,
B /0D. ?,
WA
V2 WA B ?W $ Z2 ?AB_ > /0D. !3 , V2
4
+ H4Y4 !I 2 ?A '
B !32 , ;$ 4 V;: + H4 , M, Z2 ?AB_ /0D.
U
'
+.
+, 4 &M +H4Y4 !I 2 !` ,4 / +2 &B VP !3 : +. +, 4 &M
' U
/0D. !3 , ,
!32 , b & ;$ 4 P 6 M, , / !3 :
4
4, > !-. ?W $ /0D. ?,
,
!32 WA B + H , !-. /0D. WA B
4

VP ' B ' U
^r
2 9 4, ?, !I 2 ?A !32 , ;$4 V;: & P M,
'5 /0D. !3 , /0D. WA
P B &B V2 & P ,4
&B &.
&
'

Z, /2 !I Q I9 &? GY,/, !I V;I (vi )
b ,
I V , ?, 4
'
6?^ 92 !I T4S ]; +2 5 9 4 !I !3 &. "/ ,
4 !I 2 V , 6 M, '
, / !` &B VP , / !I 2 V ,
9
U 4' '

4

!` !3. M, 2 !` ;$4 &B VP Z, / T5 !I !3 <#59 ?,

9; !I V;: 2 ZAB Q I9 &? +2
4I V;S4I ,
/I GY 4,
4
(Fig. B.8d) JO
9^ !`
&N
fj j d fi
l fe ff c
fd i u t
f fc fh h
Fig. B.8d
' I ', +2

Z, /2 !I Q I9 &. !I GM;: 9(S
!3 (vii )
(iii & iv ) ,&B ', oM;$4 /U H4" rU^% M?)n8 ,
2 ?,

%
Z2 I\2 &B9 +2 %
N&, P Z,
,4 / !I !3 &.
/#a5 +2 6 M, ?a,
5 9a 4 !I !3 &.
U

9
?, 4, Z2 2 ;$ <#59 9
%I\ 4, # S /a?
4 4
'%
NM 2 , <#59 ?,
9
4, # S 9# /#5 +2
4 5

!I !3
?,
9 4 ,&B (. . . ) &B9 +2 49 &. !I

Z2 %I\
?, 2 9# /#5 +2 5 9 4, ?,
Z2 T59

U
(. . . ) %I\2 ;$ 4 /? /#5 +2 5 9 4,
Q I9
&. Z^ !I 4M. &(?, rU^% M?)n8 ,
', I (viii )
4I 9: 9 &. 9E , v# $ !` , /' I b +E2 ' !I
Z, / &N
4 Z,
,4 / +, 4 /#
9 /? 9# 9 ^ !I U I
4, ?,
9
T59 !I Z', / , TS ]; '
!I M, 5
4

&N /H4. QI &. JK4I &B 9,4 XY4 T59
(Edition, ll . 446–485, 497–502, 505–513)

Z', /2 !I V2\ Q I9
?, &? ( !I !-'P : (B.9)

'
>4 &I + H42 &(M N,&B (i )
&B 9,4 +E !3 <N
9; &N

+W: GY4,/, # $ 6 V /H4. /0L$ Z', / &N !I Q I9
&?
', / /,5
? % !I Q I9 &? &B !I rU^% b V?)n8 +E >4
g&Y, bE >4 G Y% 5 GY '
4, +2 (B.26ii )

&. Z^9,
p
JK4I &. b?^ 4 JKI
&. !`9, !-. &B9 +2 g&, 9E,
4
U
' ' % _
JK4,/, P I JK2 4, ?,
VP 9E,4 6MP 94 r> !`
'
(. . . ) V2\ Z2)7 8 QI9 <(W$4 & 9N 9W2 # $ !-.
' U
n 8
g&, V?) <?Y4I Z, / &N !I &. GY, ,
I (ii )
!I M + H4 , Z; $ ,4 / &B !I Y, &B9 +2 6?^
!I ?,
Z; $ 6 M, , / + HY
4M,4 +, 4 &M
!-,4 & GY 4 4 4 4

9; &N
!-. <#59 94 ?,
+2 /# !-. !` ,4 /

+2 &B VP ' &N !I VB ', @-._ &I (Fig. B.9a)
WI +2 Z^9
4, +2
9 GY 4, 9N &B GY 4, Z', / &N !I 9 4 JO (W I !3
GY 4, ?,
I <#59 ?,
4

9
4, +2 GY 4, &B9 4I ,4 /
9 , 4I /? 9 ^ +2 GY 4, M, 4I 9# 9 ^ +2
I b
B
,
M Z^92 !I , + H4 , Z^92 !I M P
!-. Y, 4 ?M, 2 !`9, !-. /: 9 4, ?,
!I &? G
Y,

&N !-. /H4;4I N/,% # $ !I 2 &'
9; !I GY ,_ 2 XE?
(Fig. B.9b)
9;
JK4I P 9 4 !I, !I &.
, v(' , ,
!32 ', I (iii )
' &? &? >
I GY 4, VP '
Z, / &6 !I Z;$ (N&B 6:
@-?, ,
',\ I M4I GY 4, !3. /# !-. 2 G GY 4 !I (2 '
/0D. !R GY 4 9N M4I GY 4, !` ,/ # $ !R GY 4 !I Z, 4 2
/0D. ? 5 9N Z', / &6a &? &? +2 N#a 5 I M, ,&B 9I

GY
G 4 !I Y, b &" !R GY 4 !I N Y,I /0D. ?,
< I

/
(Fig. B.9c)
9; &N !-. H4;4I 94 /,5 &N !-. /0D. &B %
(. . . )
l ff fh f l f f
fe c h fc c h h
j fi t i j i j
fj d u fd d u d
Fig. B.9c Fig. B.9b Fig. B.9a
(Edition, ll . 514–521, 533–552)

' ' [
W Z, /2 !I Q I9
&? GY,/, !I
4 (v . B.18) /: : (B.10)
V ,

6 V +2 /0L$ '5 Z', /2 !I Q I9 &? G',/, +E >4 &I (i )

_ ' p
V(? 4M. ] ) n48 &,4 /,4 + q +W: GY4,/, # $
<?4# v# !-. &B9 +2 49 &. G Y, ', b (ii )
9N Z#, Q I &. (JK4I V;: &I +, 4 /# &S I (Fig. B.10a)

' 4 <? Y, 4 W2
2 !-. (N&. /H I Z, / &N 9^ !I 9E,
?,
9 ' ' U
4 Z, / &N T5 !I & ?,
Z, /2 +2 V;S4I <N
'
/H4? a, N&. ,4 / 9 4, ?,
<#59
6B +.
?,
Z', /2 !I !3 9 '
4 !I &. GY4,/, v(, ,
I (iii )

/"
/" GY 4, !` GY 4, !` M , V , Z', / &N T5 !I
+2 ?,
4I
9 4, V? , b &" GY 4, !` w" w" GY 4, !`

4, !` &,4 &,4 GY 4, !` V , b T4S ]; '
!` )" )" GY
' &. '@k8 &I 9E 4I GY
T5 Z, / !I !3 4, !` /, 4 /, 4 GY 4,
' '
5 Z, /2 +2
: G & )7
G,4 ,4
" ,4 w"
wn84 /, 4 9,4 ,4 &,4 )n84 )n84 )" 9,4 ,4 &"
&" )" G" P T$4 T$4 &,4 G" P wn84
/, 4
T" w" /" 9" "
G4, T" /" 9"

9 & )7 G 9 :
' <#59 9

Fig. B.10a
4 !I !3 &. M , '@k8 (iv )

T4S ]; +2

&N !-. (6 + H4M, + H4Y4 !` (6M , + H4#5 + H4Y4, g&Y,
4, !` GY
)7 )7 GY 4, !` & & GY 4, !` : V , b : 9
4, !` )n84 GY4,/H &N '
GY 4, !` wn84 wn84 GY !-. V , @k8 : GY 4, !`
5 9 4 M , &I 9E I )n84 GY
4, !` &" &" GY 4, !` T$4 T$4
' 4

!` G G GY4, !` GY4, !` V, @k8 T4S ]; +2
4, !` " " GY

G,4 G,4 GY 4, !`
V , '@k8 GY 4, !` GY 4,
/
&. GY4,/, +2 9E,
4 b &.
GY 4, !` GY 4, !`

(Fig. B.10b) JO
9^ V ,
&N Z', / &N
!I
i jd jh jj c f i c h j d f
dc ff t dl l je fd ff fe t l u
fj dj fi fc de dh fl fu fi fc fh fj
ft fh dd df fu fl dh dj dd df de ft
fd dt du fe di u je dt dl du di dc
ji d j h jc jf ji jc jh jj jd jf
9: 9, 'W Z', /2 !I &. G',/, +E

/H" V +2
Fig. B.10b
(v )
>4 &I
4 '
N/H. . &
&N W$4 "
g& !-. VJK1 2 JK2 Y, &I
4 !-

JO
9^
(Edition, ll . 653–675)
'
V ,
4( Z, /2 !I Q I9
&? GY,/, !I V;I
4 (B.11)

: l u i c h j d f
9,4 ,4 &,4 )n84 G,4 ,4 fi fc fh fj fd ff fe t
&" )" G" P T$4 wn84 /, 4 dh dj dd df de ft fl fu
G T" w" /" 9" " jd jf je dt dl du di dc
T w7 / 9 & )7 he jt jl ju ji jc jh jj
w* /2 92 2 &2 )* G2 2 hl hu hi hc hh hj hd hf
9, , &, )n8 G, , T2 ci cc ch cj cd cf ce ht

&5 )x1 G5 5 T$ wn8 / , ih ij id if ie ct cl cu
', / Z', / &N &. G',/, +E

!I Q I9 (i )
Fig. B.11a
?
!3 !I >4 &I
' ' '
XY4 W Z, /2 !I 2 &I 2 4I !-. V +2
9: 9, &?,
' ' !I JKa,/, /"& , g
#$
V(? @-? !-. / 4 Za^9 &N
4
'
Z, / !I &B9 +2 49 &. Y, b ,
I JKI,
4 !I
(Fig. B.11a)
' ' ' . '@k8 (ii )
2 V2 4( Z, /2 T5 !I & W Z, /2 !I 6M, !I MA
' '
(Fig. B.11b) W Z, /2 !I &. V , !I (B.10) MA .
P GY 4, !` V , ?,
Z', /2 !I !3 &., 6M , g&Y,

U '
!` ;$ 4 V , @k8 GY 4, !` 2 2 GY 4, !` &2 &2 GY 4, !` P
GY 4, !` GY 4, !` w7 w7 GY 4, !` /" /" GY 4,
4, !` )n8 )n8 GY U '
G, GY 4, !` )n84 )n84 GY 4, !` G,4 ;$ 4 V, @k8
4, !` GY U
4, !` 9" ;$ 4 M , G,4 GY 4, !` G,
!` & & GY
, , GY 4, !` &, &, GY 4, !` ,4 M , '@k8 9" GY 4, !` T T GY 4,
)" GY 4, !` G2 G2 GY 4, !` wn84 M , ,4 GY 4, !` &,4 &,4 GY 4, !`
!I & 'W Z', /2 &. /H;I wn8 GY
4 4 4, !` /2 /2 GY 4, !` )"
; ' 2 U I 6'MP 4( Z', /2 T5
,2
% 9 +2 &B VP !I 2 ZM
(Fig. B.11c) JO
9^ 9?W
&N $ WA B
fc cf cj cd fh fe fc fh fj fd ff fe
hd dd de hc ft hu dj dd df de ft fl
di jl dt dl jc jt jf je dt dl du di
jh du ju ji je jf jt jl ju ji jc jh
dj hi hh df hj fl hu hi hc hh hj hd
cc ff fd fj ch ce cc ch cj cd cf ce
!I 5 "/ 4, !` M , '@k8 (iii )

Fig. B.11c Fig. B.11b
(6?^92 , GY 4, !` GY

!` wn8 wn8 GY 4, !`

GY 4, !` )x1 )x1 GY 4, !` M , '@k8
&N '
GY 4, !` G5 G5 GY 4, !` : 4 !-. M , @k8 GY 4,
^ ZAB /H;I : GY , !`
5 9# 9 4 4 4 4, !` T$ T$ GY 4
'

% 9N JK2 ]^ VP !I 9E, ' 4 <? Y,4 QI92 N&. U I

,2
;
+2 ]; ' ' ' U '5
/? 9 +2 ]; ;$ 4 /H4;$4 9
'5 ,2
9 % (JKI 2 &. U I
' 4 '
@k8 (JK4, (4I G "/, " GY4, !` ) ) GY4, !` " M , @k8
7 7
^ ' '
2 4I !-. 9# 9 +2 +2 ]; !I 9 4 M ,
' ^ ' '
M , , b /?
9 +2 +2 ]; !I 9 4 M ,
' 4, !` 9, 9, GY 4, !` GY 4, !` 9,4 9,4 GY 4, !` T2
@k8 T2 GY
4, !` /, 4 /, 4 GY 4, !` &" &" GY 4, !` 2 ;$ U
!` w* w* GY 4 M,
'
Z, / &N +2 ;
ZAB &a. &?aI 2 GY
a,
9^ &N 9 4 4,
(Fig. B.11d)

)x1 G5 T$ wn8 l ij id h c ct cl f
T2 ,4 , )n8 G, &,4 9,4 ht fc cf cj cd fh fe fi
2 G2 G" 2 T$4 /2 &" hf hd dd de hc ft hu dh
G 9" w7 T" w" T )7 jd di jl dt dl jc jt jj
& /" / 9 " he jh du ju ji je jf dc
/, 4 )" 92 &2 P )* wn84 w* fu dj hi hh df hj fl hl
, ,4 G,4 )n84 &, 9, t cc ff fd fj ch ce ci
&5 : 5
/ , ih d j if ie i u cu
', /2 /,5 &? G',/, +E

% !I Q I9
&N
Fig. B.11d
? >4 4 !-. (iv )

U 2 U%
2 ?,
ZM, P ,P /
(Edition, ll . 719–749)
(Construction of bordered magic squares)
(Squares of odd orders)

/I ', /2 ' _
VJK1 Q I &. ?W
$ 9N & ? &B (B.12)
&? &B _ b &"
!I Q I9 /I
?
' %
, /2 /,5 !I b &B

, /2 '
?
'
!I &. GM;: ) 7 8 Z', /2 !I VB

I W( M Z', /2 '
'
: /S 5 VJKD4I &B ]^ , 9: Z4AB +2 !I P
M Z', /2

GM;: W Z', /2 !I VB
? W( ) 7 8 Z', /2 b &" 'Y, 2 .
!-
' ' ' 4

W Z, /2 b &" &B ]^ !I P
? VB& Z, / ' !I &.

(. . . ) ?WY Z, /2 !I VB

' /0D. 'W !I 2 ;$ U _
Z, /2 !I 2 VJK1 Q I9 &? +2
4 @-. I
' 'W
4( Z, /2 !I 2 &"
(Edition, ll . 18–24, 36–37)
', /2 !I Q I9
?

4 &' 2
&? GY,/H 2 !I V;I (B.13)
/I
I
I9 &?, ' M Z', /> T)n8 W( ' ' '
4 4 )7 8 Z, /2 (B.12) &, &I (i )
' '
Z, /2 !I ,, 4M. VJKD$4 M Z, /2 !I Q I9 &? , !-.
' T4)n8 !3 ', / /,5 W(
!32 , I F? , JKL?, ? % !I b &" ) 7 8
' /U I P BU
2 Z', /2 !I Q I9
Z, / !I JK4,/, P &. M?:
'
Z, / , T4S ];
'
!I Z , !3 &. V,4 &?, 9N >' 4M,4 &
/I ? ', / + H4, V4^ + 'H4, <? Y, 4 b VB +A I VB&
JK4I Z , !3 &.

4 W2 (6M^ , ' + ' ' '
+2 9E, I H4?, /2 +,4 &. VP 9 I (ii )
?', /2 &. U P b &" P (6M^ , !I (JK?M^ 9()* /0m

9E,4 <? Y, 4 + H4a492 JK2 + H4 , VP ' ' 4
V^, I
/I

/ ? ', /2 !I r^ V^ , ' + '
H4,/2 (JK4?M^ 9(S ,4 W2
U
+ H4a492 + H4?', /2 + H4, V; /? , ,

!32 I , 9N /

P V;: AI 9(S ? ^
(JK?M^ ?A B / / ? '
, /2 +2
4 '
JKI Z , !3
P &. &. + H4?, / + H4, !3 9 4 &. b
4
!I &. V,4 &?, 4M. !I P Q I9 &? I G',
&I ' '
Z, /
4
+H4?', / V^ , <N !3 9 4 !I !3. T4S ]; '
&? I G',/, ,

Q I9 Z', /2 ?2 P b &" P (iii )
4
&. 96I P AI MI 4M,4 & Z', / ZM^ !` Z', / b ZM^ ^
'
]; +2 94 ];$ !I Z, !3 &B9 +2 ,& 49 &.
%
&? 2 > <N !3 6 , 4 / &. Z , /: [ ]; !I TS
4
+, &? ' 9N +, /
+H4, 4 / &? 9
I !3?2 '
Z, / b & &?
4 4
&? 9N &? &? &? &?M W2 (6.9()* &M &? (N
' U
+ H4 , V2 ? , / !I Z^9 , !3 49 &. !I/= +2 Z(S
? ', /2 !I T5 &? ]? ^ V2 9N ?W Y Z', /2 !I + H,
4 >
/ ? ', /2 !I + H4#5 +, 4 &? V2 /
_ ' ' ',
I (iv )
] 4^ 4Y, 2 !-. T4S ]; 4, !I &. bM, GY
9
&. b P V;: AI MI 4M,4 & Z', / ZM^ !` ?M^ ;$ U
' ' 4
b &" G Z, / , T4S ]; ]; $ 9 4, !I Z , !3 &.
&. &. b P Z, / Z', / ZM^ !` G Z', / ZM^ ^
' ' $ 9 4, !I Z , !3
!3JKa , !` VB& Z, / , T4S ]; ];
/I ? ', / !I &B9 !` 9 4 !I
^ +2 ' , ,

9 ]^ !I Z , !3 &. /? 2 (v )
/H4?; Z', / , # S 9
4' 4 &. ZM 2 9 , > &B 9N ?W $ Z', /2

'
MI 4M,4 & Z, / ZM^ !` + H4, > &B ZM^ ^ /H4E Z, / +2
' '
&. JO M I /0D. 5 \ I 9?,
?W $ 9N /I &. +2
' ' '
!I Z , !3 &. ? 5 Z, /> TS ];
4
+2 9 4 ]; $
' '
/0D. 5 !` &B9 +2 ,& % 9 &. <N ];
'4
&N
<N !3 6 , 4 /
&. JKI Z , ]; &N 9 4, +2 !R ];
4
49 &. !3. 9 > , 9N ?W
, Y Z', /2 &. &. 2 >

> &B !` + H4'W 'W +2

+ H4, ,& %

B
ZM^ 9N WA !` +H4?,
?W$ ZM^ 9N ?W ^ b &"
' $ JO' M I /0D. !3 , P B
!R ]; 4, &. ];
9 +, 4 /0D. WA
!I Z , !3 &. ' +, 4 /0D? W( ) 7 8, #
4S ?5 Z, /2 +2
'
U
4 > !` /0D. ? 5 +2
' 492 &. /0D. , <N 9 4 &N ]; $
&. JK4I Z , ]; &N 9 4, +2 !I ]; ' +, /0D.
4
+H4'5 WA B !` + HWAB ?,
4

+2 %
,& 6 , 4 /

9N 4 > \ I ?W $ ZM^ !` +, /0D. WA
4'
B ZM^ ^ b &"
'
4 > JK4I Z , !3 &. M, Z, /> T4S ]; 4, &. ];
9 $
!I Z , !3 &. + H4M, '5 !` +, 4 /0D. ?W $ +2 492 &.
' ' U '
5 +2 6 9E,4 &. 4 > ;$ 4 <N ]; +2
&N 4I 9
4

+ H4WAB M, !` + H4?,

$ 9N ?,
P
]; &B9 ZM^ !-. ?WY ZM^ ,
b &"
NJ % Ka , 9 M, ? 5 &. g&, &B9, # S 9 4 &.
B NJ 4

WA
9?, % Ka , 9?,
, , 4 / &. g&, 9?,

9N &B (N&B +, 4 &M + HI/#
4
+2 ?U 4AB &? !3JKa , & I
<, 4I G Y,4 2 /H" 9N 9 , > &B /: [ Z', / !I G Y,4 2 V' I
' 4 !I Y, & T5 &?
Z, / &N +2 T5 GY
' 4, V4;)n8 !I /2 % ' 2' I (vi )
&. T4S ]; 9 I

, >' /I
6I
? ', / !I ,
2 2 !-. 9N JK4I Z , !3
Z , XY4 ?,
, # S /0D. !3 , ' I ?,
Z', /2 !` gJKa , +2
' 4 '
]; !I &. G,/, V^ JK4 !3JKa , 4 !3 &. JKI
4
' '
!3 ?, / +2 4I 9; !I G,/, 2 GW: ?,
, T4S

Z', /2 !I Q I9
?, &? GY,/, ' I /I
+2 2 /H"
' 4
(B.8–B.9) 4Y, 2 !-. ? ', / !I, !I Y,/, ]
4
) n8
4
(Edition, ll . 106–168)
'
]a; 4, !I Zaa , !3
9 !I /a: [ a:
&a. a I/?2 (B.14)
T4S

!I Z, 4 & &? 9N T5 &? M?: b @-?, ,

!3 : , /H2
+.
, & B % '@k8 /
&. ?, Z', / +2
T5 GY 4
'
Z, /2 !I Z , !3
&. <N T5 &? Z2 P AI T5 &?
?
&N W Z', /2 Q I !I <N M

&" )" G" P
9" /" w" T"
+, &? !3:
+, 4 &M
+.
, B % '@k8 /0
& D. WA B T5 &?
4
<N , /H2
&N &. > (JK JKa ,
4 4 4
T$4 wn84 /, 4 9,4 ,4 &,4 )n84
G )7 & 9 /
) 7 8 Z', /2 +2
W( M Z', /> TS ];

'
!I Z , !3 &. <N
4
U
<N &. /0D. !3 , (JK4 4JKa , +, 4 &M +, &? !3:
4 +. , & B % '@k8

&N
G,4 ,4
:
w7 T 2 G2 )* &2 2 92 /2 w* T2
' 7 ' ' 4, !I Z , !3 &. <N
Z, /2 +2 W() 8 Z, /> T4S ]; 9
!I G Y,4 &. V' I 9N &B9 !` + H46)7 8 &B !I 4JKa , ? W
/
+ 7 '
Z,4 &. H" 9N H4?,
?W$ !` /: 6) 8 !I Z, / &N
4I
4 &a? !3a:
+a, +. +a, 4 &a. &a : %, ', I

a? ', /a 2' I
'
& (N 9 M, M, 9M, , 4,( Z, /2 !I (N + H4#5
% &. '5 & !3: , & B % '@k8

N, & B !3 &. !-, , /H2 +.
U'
&N <N
T" w" /" 9" "
9 / w7 T

Z', /2 !I N2 &' I !3
?, &. ?,
Z2 Z , !3 &. <N

%' ' YU 4, /0D. 'W <N
&. /0D. , & B @k8 4( Z, /2 T5 !I !3
<N (JK JKa , +, &M
+. , /H2

&N 4 4 4 +, 4 &? !3:
&" )" G" P T$4 wn84 /, 4 9,4 ,4
2 G2 )* &2 2 92 /2 w* T2
% ' ' ' '
, & B @k8 /0D. W Z, /> T4S ]; !I Z , !3 &. <N
+, &? !3: U

&. <N (JK4 4JKa , +, 4 &M 4 '
+. &. /0D. ?,
&N <N + HM, W, TS ];'
4 4 !I Z, !3
&,4 )n84 G,4 ,4
:
G, )n8 &, , 9, / , wn8 T$ 5 G5 )x1 &5
,
7 8 !I

!` /: 6) ' &B9 !` + H46)7 8 &B !I &.' !I 4JKa ,
Z, / 9
&,4 /, & 4, &? W &? 9N + H45 ?,

QI9 &? 4I G Y,
(Edition, ll . 169–204)
,&
+2 % /I
', / !I Q I9
? &? @31
!I V;I (B.15)
) 7 8
W(
&B9 +2 ,&
% <?# v# !-.
4 49 &. G ,P
(i )
U ', /2 !I %

2
&, ?, /I
#5 ? T5 &? , & B /H4. +2
9^9 M Z', /2 !I , +2 , g&Y, & &? 96I <, AI

,& &. !`9, !-. &. @31/, W()7 8 Z, /2 T5 !I
% '
+2
];
' 5 9
+2 4 &B !I Y I M Z', /2 !I !I
&?

' 1
!I &?, 2 GY, I (B.3) M Z, /2 !I s
2 !-. V(?, T4S
&N !I (" V;SI M Z', /2 +2 9
4I 4 9W) n8 5/I GY 4,
4
^ W( ) 7 8 Z', /2 T5 !I !3 &.

G
'
&? I (Fig. B.15a)
9;

$ JK2 ]^ ' '
M, !I /0D. M, /0m +2 Z ,/,4 2 9N + H4M, ?W VP
'
(v . B.2ii ) Z, / ZM^ <N !3
fe fu fd
fc fj ff
fh t fi
B +2 , + 'H4, G)n84 (ii )

Fig. B.15a
MA 4'I '&., T4S ]; V,4 &? '

7
Z, /2 +2 ]^ VP !I 2 9E4 W() 8 Z, /2 !I Z, !3 &.
> 4M. VJK1 &N
,&B Q I9
&? I/= !I G Y,4 2 Z2 W( ) 7 8
' 4
!32 , b (Fig. B.15a)
9; &N !I N Y, &I !3 &. +2
'
+ H4M, 2 JK2 + H4Y4, VP !I !3
+2 &. ,&B
9; &N M 2 %,
'
' '
V2 /0D. ' W /0D? V2 &? &?M ,4 W2 T4S ];
'
JK2 + H4 , VP I /0D. !3 , /0D. ?,
V2 /0D. ?

W ?W

Y
' '
!I !-. Z, / &N !I ,&B 2 ' & I /0 D. '5 + H4M, 2
' 4 <? Y, 4 T4S ]; ' 4I 9
JK2 + H4 , VP !I 2 9E, +2 4
' I &? &?M ,4U W2 ;$
P U +
+, 4 /#5 !I Z; $ , I b &" 4 H4M, 2
' B +2
<N Z, / &N !I GY, <?Y,4 !3 &. MA 4I &.
&N

:
wn84 T$4 P G" )" &" "
' 4 <? Y, 4 !3

JO T4S ]; Z4AB V,4 &?, 9E, &. <N
+, 4 /# V, &?, ' I +, /#, U' Z, G)n8 V, &?I b &" P
4 4 4 4 4
' ' '
4M. ] ) n48 @ (N/H4. /#5 &. ZAB T4S ]; V,4 &? /H5
g!31 4I /#5 !I 2 V, &?, +2
4
!I !3 &. +2 +, 4 &. + H44a2 ,4
+2 + H4,4
!I Z; I (iii )
+ H4,4 / !I G Y, '5 4, !I ?,
!` !I +E
4 +, 4 /#W
?I &? &? > (JK2 &B VP !I (6 + HM, '
> /# 4
' 4 B '
<, &I 95 WA (JK2 &B VP !I V;: ,4 / !I Y,

' '
+2 /#5 VP !I/= !I Y, !3A I /? 9# /#5 &?, 4M.
!I N Y, !3 &. +2 + H4, /H2 +, &. /?
4 9#
9#5
' B U
+ H45 WA ;$4 Y, 2 Z2 b
^ +, 4 /#W
% ' ', b 2 (iv )
, 9E, 4 6M,4 &?, 9# /#5 &?, ,

4, M, <N ' .

9 4 !I G Y,
&. M, 9 /#W +2
4, !` ,4 / !I +, 4 &M +, &? Z2 JK2 V;)n8 2 MA B 9E ,
4 4
B WA B /#W &N )n8 & &? +E ' + ' B
9WA 4 4 H45 WA
!I N Y, !3 4I , /H &. MA

B +2
GM#$ <? Y, 4 I
B B U
+ H4WAB WA JKMA 9E, &. M, &. /0D. , <N +, 4 /#W
> +, 4 /0D. M, +, 4 /0D. ?,
4,
b ,&B
+, 4 /#W M 2' %, I

9#5 +2 ' .
9 4 !I N Y, I

^ /? /#W +2
6 , 4 / &. V, ]; ' ' B
P !I Y, + H45 WA 4I 2
GY
4 !I M, +, 4 /0D? G
!I , +, 4 /0D? !R GY 4, !I
U ^ '
;$ 4 /? 9 +2 X2) 7 8 ];
^ /0D. 4 > +, 4 /0D? Z, /
!I !R %, 2 . + H'5 WA
(Fig. B.15b)
9; &N !- 4 B
, &?,4 & 9N 9?,
&B 6.9()* &. M, M= (v )
'
Y, 2 .
9 4 !I N Y, 9#
9#5 +2 /#W
' [
/: /# !I &B VP 6 M, 9 ' 4 !I JO% ,/ I

&I " /#5 +2
'% '
!I M 2 , b M?
I /?
9# /#5 +2 <, 2 Z4AB &.
U U %
I &B +, 4 /0D. &B /0D. ?W
$ M &. ,&B
+, 4 /#W

9a; &N
!-. /H4;$4 9aM# V;: a .

9
4 !I N Y,
(Fig. B.15c)
i l dj dh h i l dj dh h
u fe fu fd ft fe fu fd
c fc fj ff df fc fj ff
dc fh t fi f fh t fi
dd fl j d de dd fl j d de
+ H4,4 / !I (N? ^ +, &? ' +' # $ <? Y, XY (vi )

+, 4 &M 4 ' 4 4
&. +2 U '
/H4" I ,, (JK4I ?I (JO JK2 + H449 '
!I !3
@ + H4,4 / + H4,N !I (N? ^ JK2 + H4 , , & B % + H4, /H +, 4 /#W
+ H4,4 / + H4,N ' b +A I Z', / &N !I Q I9 &? V, &?, (JO '@3,
4
' p , _
, &B ?,
&B W ?,
' &, (JK4I GY
&. + H4, /H +, 4 /#W !I &B 9,4 @ ? 5 5 WA B ?,

_
'
, I + H45 WAB + H4,4 / !34, !I 2 Z2 JKMA B 9E ,
(JK4I GY
B '5 ?,
,
WA 4 > , ?5 M, ' M, &B
JKMA B 9E, 4 &. M, (6 &B 9,4 +E >4 P 4, > 5 ? 5

' ' _

ZM^ VI & &? GY, +H4?MI +H45 WAB (JK4I GY, 2 Z2
' &? !` , / !I Z', /
,4 / !I &?, Z, / ZM^ !-,4 & 4 '

!I, !I 4I &. &B 9,4 +E>4 (JO &?,4 /#W I 4,

9 ;
&N !I G Y, GSa4I ,/" 2 !-. /a2 %
Z^9 P I (vii )

Y, !3 &. Z2 N/" 2 &I !3 ' &.
9# /#5 !I G
b &, @-?4 W( ) 7 8 Z', /2 !I Q I9 &. G',/, &, /, + qp /?
' 4 4 4
&)7 8
9^ & , &I 6.^ r : 6.9, /H"
' ' +2 ' ' _

9, 5 !-. 9 Y,4 9: 9 &N : VP @-. &I (viii )
' &. M ' b ?' ;2
!I G2 & , !3
9; !I N? ^ !3
&" GY ,_ (Fig. B.15d) )" &" <N /?
9#5 +2 '
/#W
_
)" !R GY 4 !I V?)n84
9) &I
n84 P /#W +2 !R GY 4 !I
!R GY 4 !I V?)n84_
9) G
n84 P GY 4 !I GY , _ 2 b &"

9#5 +2
'
/#W !I B
'5 2 G'"/, &EN Z, /
'
/#W !I B ' '

5 G"/, b &" /?
9#
9#5 +2
!I G, /0m_ 6:
U
U (!) /0D. , &6I !` ,4 / + H4,
6: . . . b +2 &)7 8 !I !3
P 9: 9
)" &"

/, 4 G,4 T$4
,4 )n84 ,4 P
" &,4 9,4
G" wn84 :
) 7 8 Z', /2 !I G &? ' @-._ & I (ix )

Fig. B.15d
2 !-. W( Y,4 +E >4 Q I9

&? I G
!-. Q I9 ) 7 8 Z', /2 P
&N 4 Y,4 +E >4 W( ,/"
/H" 6.r^ !3 ', / !I /2 %
+2 ? 9E,4 /, ] 4EI 9: 9
' ' '
!-. @-? JO ? MI 9: !-. JKMA
W$4 M4 B +2
/0L, , ' WAB
4I /S 5 Q, 4 /= !-. ] , 4 @-?,4 2
(Edition, ll . 229–314)
?
/I ', / /,5
% !I Q I9 &? GY,/, !I V;I (B.16)
4
&. &)n8 <? Y,4 ]4" (B.13–B.14) &' , (4I ' Y'4, &I (i )
!3
'' ', / +2 ' ' '
] 4" (B.15) Y4,

/I ? ]^ ]^ !I G,/,
' U ) 7 8 Z', /2 !I Q I9
Z, / 2 JK4 ] 4; $ <? Y, 4 W(
&? @31/,
' U
@-? !-. VJK1 /H4;4 JK2 ?M^ /H"
[ '
9N T5 &?, g&Y, B ?W $ ZM^ +2 Z, / V?S I (ii )
' 4 !I ?; $ &B U ' '
@k8 T5 GY 9> &B Z, / 9?,
&B
'
4 &B !I 6MI G',/, , /H2 &. ?,
Y: B %,
5 9 4 ' +. &

!I JK49, !-. &. !I, GY, T5 GY4, T4S ]; +2
+2 M Z', /2 !I s1
2 GW: ?W Y 9 4 +2 4I 9
4
' ' U ' 9 ^
,2
% JK2 ]^ VP !I I Z, / &N &. /H;I 9: 9
4 4
/0D. M,
(N (JK4 ?M, &I +, 4 &M +, &? !3:
4 +2 &. 4 > & B %, '@k8 (iii )

9; &N
!-. +, 4 /#5 !I JK Y, 2 /
9 & )7 G T"
92 /2 w* T2 , G, )n8
' !` ,4 / !I G 9N JK2 Z, / G Y,
W T4S ]; +2
' 7 ' 1
G,/, W( ) 8 Z, /2 !I s
2 GW: 2 4, ,4 / !I & 9N
B +2
!3 , MA
' ' &. M, X(M, (6 +H4M, +H4,4 / !I (JO4 /H4#$
+2 /#5 VP b + H4M, ?W $ ,2 % JKMA B 9E,

4 , /H /0D.
<? Y, 4 ?W Y Z', /2 !I 9^9 ) 7 8 Z', /2 !3.
W( Z
'
, / &N
9#5

' ' U '
(JK2 &B VP !I V;: &I +, 4 /# +E I 6MP 4I &. 9E, 4
^ +2 ' B ,2
/? 9# 9 &B VP !I 9E,4 <? Y 4I WA %
B + H,2 % F , & G I & ]; ' B ,2 % ;$ U
WA 4 +. 4 4 WA 4
+,4 /#W !I 4I &. MA B +2 %
% W( I +H4M, ?W $ ,2
9
4 !I NY, +4HM, ?W$ ,2 JKMAB 9E, &. M, +4H, /H
'
^ 6 , 4 / &. M, 9

4 !I & G ];
+2 .

% JK2 ]^ ' ' ' U ?2 9# 9 ^ ZAB
,2 VP !I QI 6MP N&. & 4
' % ' ' B
,2
% 6?M 2 &. M, (JK4I ,&B +, 4 /#W M 2 , q E WA
]; '
& G 4I & !I N Y, I 9 G, , <N 9 M, ?W $
' ' &. Z', / &N +2
]^ VP
^ 6 , 4 / 6 M, 9
4 !I
&. M, GM#$ b &?, <? Y, 4 WA B + H,2 % 9# 9 ^ +2
4
!` + H4,4 / !I GY , _ 2 Z2 JKMA B 9E,

4 4I &. +2
/:
M, , /H &. &N MA B +2 I WA
W( B + H,2 %
4
, 4 /
T" /2 T2 N,&B +, 4 /0D. WA B ,2 % JKMA B 9E , &.

6 M, 9
4 !I JO% ,/ I 2 .
9 4 !I N Y,I )n8 )7 6
B !I WA
WA B 9N Z', / &N 9 ^ G& . & I

' ' &. &N &?, &? '
]; &N !-,4 & ]; , <?Y,4 (iv )
+2 U
!`9 !-. &. /0D. !3 , & B I W( % ) 7 8 Z', /> T)n8 &
!3,B 44
4M. VJKD4 +, 4 /#5 !I JK Y, )n8 T" (N (JK4 4JKa , +, 4 &M +, &?
4
U , /H2
&. U
/0D. ?,
<N &N <N T4S ]; !I JK4,/,
'
w" /" 9" " &" )" G" P T$4 wn84 /, 4
&, , 9, / , wn8 T$ 5 G5 )x1 &5 5
&? 4, ,4 / !I JK2 W &? !` , / !I G
4 Y,
,4 / !I &" , , / !I
4 ' 4
G" !` , / !I 9E
4 4I +2
+, 4 /# +2 &B VP !I V;: & I ?, / , / !I wn8
4
' ' 9 > ?
GM# I JK2 ]^ VP !I 9E,4 <? Y, 4 b &" , 5 ,2

%
U % % JKMA , , /H &. &N MA B +2 B
&B + H4,2 B 9E
&. WA
> ? 5 + H,2 % + H, / !I Y, 2 Z2 9E ' +
+ H4, 4 44
, ! 3: H4?,

&N JKMA B &. WA
B &" G" ]^ ' !I G Y, +E >4 ,&B 9I
'
VP 6 M, 9 4 !I JO% ,/ I Y, w" 9" 5 G5 &5 <N &?

B wn8 G" ]^ ' U ' [
WA !I ;$4 ,&B ^ +2 /: /# !I &B
9 4 !I JO% ,/ I Y, , / , T$ T$4 /, 4 <N ? 5 ,2

% JKMA B &.

'
T5 !I 9^9 ?W Z, /2 !I !3 &. GM;: &I 6 M,
> ?
, 5 ,2 ' ' U '
% JK2 ]^
9 VP I ?WY Z, /2
fc dh dl di if id ih dd iu
fj du jh ji cf cd jd cc it
ff dc jc jl hc he hu cu uf
t dj jj hj hf jt ht ct uj
uu ij cj hd ju hh dt ft c
ut ic ce hi jf je hl fu j
lf ie ch ci df de fl cl f
' &. GM#$ [ G)n84 (v )

Fig. B.16a
T4S ];
+2 !I G Y, !3
U
&EN &. '
?WY Z', /2
/0D. 5 <N 4I &. +2
9,4 ,4 &,4 )n84 G,4 ,4
:
95 /01 wx1 T5 . G. S. &. . 9. /. o. T. I
/01? 4, !I +2 &? &' , 2 !-. !` ,4 / !I G Y I
'
VP !I /H4;4I (6 + H4, 4 / +, &? (6 + HM, + H, / !I G
4' 4 44 Y,
<? Y, 4 & &? 9N 9 5 ?W $ ,( +2
% M, +, /# &B
[ ' 4
' '
5 GM#$ <? Y I Z, / &N
? 9 ^
+2 ]^ VP !I 9E, 4
4
I 9,4 &,4 G,4 . 9. o. <N 9WA B &B% ,( % M, JKMA B &.

,94, !I Y, + H4'5 ?W
$ ,( 7 8
^
% M, M()
+
H4, N Y,
+2 X(M, b &" jit &. G. &. Z2
^ JO% ,/ I 6 M,

jit G. &. Z2 JKMA B 9E , &. ? 5 , /H I
&.
' 4

G. ]^ !I b Y, : S. . T5 /01 b ,&B 9I
U U ^ !I 2 ZAB
^ 6 M, 9
4 !I JO% ,/ I
^/. 9= I9 4

(Fig. B.16a) JO
9^
&N + H4'5 ?W
$ ,( U
% M, /# I
(Edition, ll . 335–404)
/I ', / Z4AB !I Q I9

? &? GY,/, !I V;I (B.17)
4
&N !I /"& , <? Y, ?W Z', /2 !I Q I9
Y &? Y ',
&I
4
p ' U

!I W, G?,4
)n48 + q @-? !-. G,4 /M ,4 /= Z^9
' '
Z, / !I G,/, !3 &. GM=

:
' &? I G',/, &, /, & Z', / +E
Fig. B.17a
: Z, /2 Q I9 4 4 4

+2 /? 9# 9 ^ +2 5 9 4 +E (Fig. B.17a)
T5 GY 4 !I G',/,
<^ '
!I !3 9 4 Z', / &N

'
,4
+E G ,P ,4
,4
+2 ,4
!` &; , T5 &?
'
+ H4; 9 4, !I 4Y: +. &B9 +2 49 &. G',/,
' ' + H; ' U
!R ]; !I 4M,4 & 4 &B !I &B , + H4#4S
' $ g !->4 !` + H; ' +2 + H4#5 + H4Y4
+ H4; 9 4, ]; 4 +, 4 &N
' ' '
@k8 !R ]; !I
+ H4; &B !I /I &. V;SI
: ]^ ' 4 !I 4 4JKa , & &? G Y,
& ' +2 T5 GY
: ]^ +2 T5 GY 4 !I 4M,4 & , , / !I Y, M,
' 4 4 44
+H4; 94, !I 4M, !3 &. G Y, ?, / , / !I M, &
U 4 4 4
6I9 ^ /= !I !3 9 4 !I ;$ ?, / ,4 /, + H4#4S
U ' 4 U '
/I ;$ 4 V;S4I + H4; 9
+, 4 &N 4, ]; $ ;$ 4 g !->4 !3: &.
' ];M ' '

]; !I
I &
/I 4 V, 9N + H4; &B !I
' 4, ]; $ !I &. Y ',
&I 9E I 4M,4 &
T4S ]; 9
_ ' ' '
G Y, !3 &. +2 &B V\ +,4 / &? Y, !32 , I Z, / ,
% !I ,/" 2 !-. ^ ' ' !I
/,5 +2 V, GY 4 !I ]; &N
% ' ', /
&., r2 &I T4S ]; (v . B.15–B.16) ?
Z4AB P
'
]; !I <?4# v# !-. JK4M, !3 &. G Y, )n48
2 @31
!-. 4I &? GY4,/, +2 ];
./ I & '
!` ,/" !-,4 &
/2 +2
, /
b &. 9E 4I T5 GY 4 !` ZM ,
(Edition, ll . 405–429)
(Squares of even orders)

&?, / ?
?, ', /2 !I Q I9 &? GY,/, !I V;I (B.18)
4
'
, /2 /,5 % !I Q I9 &? GY,/, (B.12) 2 &' I &I (i )
'
/ ? 4
% T5 !I Z^9, 4 , ' '
?,
Z, /2 !I Y4,/, +2
/
&?, VJK1
/,5
%/,5 T5 !I .9 U ' ' ? ', /
^92 P M Z, /2 ("

/ ? ', /2 !I Q I9 &? GY,/, M. VJK1 /I
4 4
? ', /
!I Q, /#, Y5

/I 4 4 !I Q,4 /#
[ ' # 5 b ,
I (ii )
AI B ? 5 &? b Z, /2' ];$ +2
?M^ & T59 !I & Z, / !I Z , !3
&. ' 96I <,
!-. ?2 & !I &? 6' &. U
/0
D. '5 I G',/, '@k8 ?,

' ' ' ' 4

& ]; !I &. G,/, @k8 Z, / 9W) !` &. !`9, n 8
9 ZAB !-. !R %, !` Z', / b &, T)n8
;
4 44
' ' ' '
5 Z, /2 JK4I &. G Y, &,4 /, !3 ? , / +E 4MI (iii )
' ' 4 U Y4,/, JKI G',/, &,4 /,
]^ VP !I V;)n8 !3 &. Z4AB 9E,
' 4
[ ' Y I U I JK2
< 4I B ? 5 /0D. > 9N W Z, /2 ];
4
$ +2 TW
T5 !I & ?,

Z', /2 !I &. +2
&?, 2 ',/, /0D. &B
U ' ' '
/0D. &B N&% , &. /0D. 5 9
4 Z4AB g !->4 !3: W
'
(Fig. B.18a)
9; &N !I 2 !-. /0 D. 5 NJ% Ka ,
fh dh dc ff
ft fu fi dd
fc df de fl
di fd fj dj
@31/ , /: 2 !-. + H4, /H2 +, /#5 !I &. !I, G',/, '@k8 (iv )
Fig. B.18a
4

JO
9^
&N

:
w" T" G )7 & 9
/"
(JK4, 4 /I G Y, + H44a2 + H4,4
!I +, 4 &. +, 4 /#W &B +2 G',/,
+2 &B !I G Y, _ &. ?,
X(M, (6 + HM, + H, / !I
4 ' 44
% % /#5
&B ,2 9N QI9 &? 4I/= !I 2 Z2 P T4S ];
' U U
V;)n8 b Y, /Ha" &a. ,&a: b Wa( /aW.
9;
'
oM;$4 +, 4 &. VP + H4I + H4,4 / !I G Y, 2 V2' %, G)n84 (v )
U Y, &I ', I (JKI G',/,
4, !I + H , &B
4U
9; &B ! I
/0D. . . . &B ' B 4
6&, Y, 9 5 WA !I /0D. ?W$
, /H &. +2 Y, 9: ]; ' '
G !3, JO @3,4 @ oM;$4 @
' Z', / ZM; W &?
,4 / !I 4I &. G,/, &,4 /, & '
'
X(M, @k8 4, ,4 / !I MI & &? Z, / b +2
!`
' '
GY4,/, !I +) n8 & ]; +2 5 9 4 !I G,/, !3 &.
' ' ' !I b V2 I &?
&? 5 !I 5 /0m +2 9N Z, / &N 4
' ' '
!` ,4 / !I Y, I 5 &?I 5 9N Z, / ZM; W
Y, !3
4 !I G
9 &. W( B Y,
4, ,4 / !I WA
' 1

(Fig. 18b–18c) Z, / &N !I s
2 !-. ,
9^
^ 5
c du t dt jc i c ji jh dl d i
ji fh dh dc ff f jj fh dh dc ff h
j ft fu fi dd jh du ft fu fi dd fe
h fc df de fl jj l fc df de fl dt
je di fd fj dj u u di fd fj dj je
jf fe dl l d jd jf f j t jc jd
(Edition, ll . 618–652)
4,( Z', /2 !I Q I9

&? GY,/, !I V;I (B.19)
4

9: 9 V> 4,(
!3 Z, /2 !I QI9 &? G',/, +E
' >4 &I (i )
'
(B.12–B.18) / / , / !I &' , (4I N' Y'4,
?
D. WA
/0 B < I ?
4
5 ,(
4 Z', /2 ];
$ +2 TW $ ', b (ii )

Z', /2 !I + H?,
!` 9 &. +2
?, '/
& 4' 4 N&?, 2 JK, HI
'
Z, /2 T5 !I
!I &. GY4,/, !I 2 &I !3 9: 9 F? , 4,(
!I 2 . ?, '

&.
^ & I (Fig. B.19a)
9; &N !-
Z, /2
U '
I Z, / &N !I !3
dl jl jt dc
jj jf je ji
dt jc jh jd
he di du ju
' '
Fig. B.19a
'
GW)n8 Z, / &JO T4)n48 & ]; !I &.' G,/, <? Y,4' (iii )
% , , /H2
,4 N& &. /0D. Z; I ? % !I 2&I 2
, / /,5
&" NJ
&EN % Ka ,
&" )" G" P T$4 wn84 /, 4 9,4 ,4

2 G2 )* &2 2 92 /2 w* T2
T4S ];
' G ,P
+2 ,4
'
! I 9N J W &? G',/,
K2 '
, / !I T$ 9N MI U & &? !` , / +E
?,
Z, />
4 4 4
+ H4M, + H4,4 / !I (JK4, 4 /I G Y, 4, ,4 / !I +E 4 JK4M, !3
' &' , !3
!I GM#$ @k8 (Fig. B.18b) G2
9; !I s1
&I (" (6
& /#W JO v(' , &. ?,
+ H, /H +, /#W +2 4I &.

U ftc 4 4
<? Y, 4 & &? 9N &B (6?2 JKMAB 9E, T$4 4I

' T5 !I & 'W Z', /2 9 ^ +2 ]^ ' '
Z, / &N VP !I 9E,4
6 M, 9
4 !I ,4 /, 4 )" T2 <N JO% ,/ I G Y, w* G2 9,4 <N
&
4I/= /#W +2 .

9 4 !I JK',/, &. ?,
X(M, '@k8

U

JO% ,/ I G
Y, &2 )* &" wn84 <N ftc ;$ 4 (6?2 JKMA B +E 92
'
s1
&I 2 !-. Z, / /H4;4I 6 M, 9
4 !I P G" 2 /2 <N
(Fig. B.19b)
9; &N !I
T$4 w* G2 9,4 ft ce hl hd fi de

/2 w" w7 T " wn84 hu dl jl jt dc fl
2 )7 9 &" hf jj jf je ji dh
G" T" & G )* dd dt jc jh jd hj
P 9" /" / &2 df he di du ju hh
2 ,4 /, 4 )" T2 92 hc fc fu dj ht hi
Fig. B.19b
T4S ];
' '
!I Z, / &. +2 &. G',/, <? Y, (iv )
4I' ' 4 '
'
Z4AB G
Y, b MI !3 ?
, / !I 2 &I (" 5 Z, />
+, /#5 !I U , /H2
!-. + H4, /H2
U
&.
<N G
/0D. ?, Y, !3
&.
4

9; &N !I 2 2
&,4 )n84 G,4 ,4

:
, G, )n8 &, , 9, / , wn8 T$ 5 G5 )x1 &5
4 Y% 5 ,4
' Y,
,4 / !I +E !I 4,> 9N JK2 +2 &? G
B !I X(M, , , / !I ?
MA 5 9N Z, W &? !`
4 4

&. Z2 JKMA B 9E , &. '5 +, 4 /#W !I !3 &.
' '
&N +2 ]^ VP !I 9E,4 <? Y, 4 & &? 9N + H'5 + H,2 %
4 4'
+2 .

9 4 !I N Y, I G5 5 T$ <N Z, /
^ ' '
&5 )x1 : <N JO% ,/ I Y, /? 9 +2 ];
' 9 ^

^ '
Z, / &N , 4 / V, GY 4 !I JK2 &B VP
U ' ' ^
+2
&B + H45 + H4,2 % JK2 &B VP 9# 9 5
' &. !I W(
!I 6s1/, &. 5 + H4, /H +, 4 /#W !I 4I
B 9E , 9# 9 ^ +2 ' ' .

9
4
2 Z2 JKMA ]; +2
U + ' + %
;I 6 M, 9

4 ! JO% ,/ I Y, &B
I H45 H4,2 4I/= !I
(Fig. B.19c)
9; &N 2 .
!-

G5 5 T$ u id if ie ct d f l
9, T$4 *w G2 9,4 ci ft ce hl hd fi de t
/2 w" w7 T " wn84 , fe hu dl jl jt dc fl cc
,4 2 )7 &" &,
9 ff hf jj jf je ji dh ch
)n8 G" T" & G )* G,4 cj dd dt jc jh jd hj fd
G, P 9" /" / &2 )n84 cd df he di du ju hh fj
&,4 2 ,4 /, 4 )" T2 92 , fh hc fc fu dj ht hi cf
/ , cu j h c i ij ih cl
: )x1 &5 wn8
9: 9 /H" @-?, Z^9 !I Y, &I > (v )

Fig. B.19c
+E
>4 !3 &N
? ' '
, / !I Z, / &N !I QI9 &? JO &B 9,4
&?, !3
(Edition, ll . 676–718)
(Particular cases of odd-order squares)

Curiosities presented by the magic arrangements in odd squares
?
/I ', /2 !I /# Q I9
+2 &? /?, 4 (4I V;I (B.20)
4^,
9
W)7 8 +2 4 / M(?W$4 Q I9
&? +2 ' @-. (i )
9 &N
% ' U
JK4 Z: /, JK4M. &(?, 9^ !-. !3 2 ;$ 4 9N
(Squares with separation according to parity)

Z', / T5 !I 6'MP /I
!I
&. V?)n8 ', b +A I (ii )

' _ ' %
$ . ,
Z, /2 #5 !I VA. &I , P Z', / /H;I ,

; !- 4 4 4 4
9E ' '
, 6MP /I Z, / b !I 9E 4I .r^
Z,
!I

!I /" Y, (" ,

(Fig. B.20a)
9; &N 4

/I

Fig. B.20a
(Edition, ll . 751–756)
, Z', / T5 !I V;)n8 !3

!-. 9E ', /
? ' U
I ;$4 (B.21)
+ H46:
+2 ]; Z^92 !-. '"/2 ,4
& Z', / 9E,
4 (N&B (i )
, <? Y, 4 ,94, !I
9E /U I 9E,
4 ,94, &. /H4E Z', / r^
' , & B % /I
"/2 6I /= !3
r^r ,4
9 6I9 ^ &.
&.r ,4 W2 6.r^ ,&B /H4E Z', / +2 r^ ; $ .
!-
' ' '
% <?# v# !-. 9
Z, / ZM^ /, + H4 , +2 ,& 4 ' 4
' '
Z, /2 !I Z,4 & Z, / ZM^ , 9N M Z, /2 !I Z,4 &
ZM^ ?,
9N ? W Z', /2 !I Z, & Z', / ZM^ M, 9N W( ) 7 8
4 '
&. !-. b &" WA B 9N ?W Y Z, /2 !I Z, & Z', /
' ' 4

ZM^ !-. ,

9 Z, / &N ZM^ /?, ,
!3A I 49
' '

9 Z, / ZM^ 96I P AI Z(: 2 ]; $ , & B % &B U '
Z, /
]^
'
ZM;M ,4 W2 ,94, 9E, 4 ]^ '
JKA I M ) * 9E , >' 6I9 ^
GY 4, 2 FI ,
'' ' !R : 9 2' (ii )
!I Y4, 2 @31
!-. #5 !I Z, / 9E, 4 I
' ' ' ? ', /
v#$ !-. JK2 &B VP Z,4 Z, / &N +E / /
W Z', /2 W( ' ' I + H?', /2 !I 9 /I &. !`9,
? ' )7 8 Z, /2 4 4
/0D. &B Z, /2 ?W Y Z', /2 M Z', /2 9N &B Z', /2 (JKI Z,
' ' ' ' 4 4

Z, /2 /0D. M, Z, /2 b &" W()7 8 Z, /2 9N &B Z, /2 (JK4I Z, 4

W Z', /2 (6#5 !I Z, /0D. WA
? B
4
2 T5 GY 4 !I Y, &. JK4I G',/, ,
', (iii )
' A I <, AI , > 2 # '@k8 T5 &?
5
Z, / &B9 2 4
Y
?W Z', /2 !I &. !`9, !-. 2 /I &. Y ',
?W Y
'
+2 TW $ @k8 (B.3–B.5) &' , (I N' Y', !3 /#, ,94, g !-> !3:
'
4 4
&B9 2 <, 2 A I /0D. '5 &B9 2 A I '" & &?

A I & &? !` /I +2 &?, 2 <, 2 G',/, W( ) 7 8 Z', /

M Z', /> T)n8 & ]; '
4, !I , 4 /I &B VP &B9 2 '
4 4 9
' ' ' ' ' ' '
@k8 MP ]; g !->4 !3: (B.15) W()7 8 Z, /2 !I Y4, 2 G W:

2 A I +, 4 /0D. ?,
&B9 2 U, , A I & &? +2 TW $
4
!3 &. +2 TW $ / , '@k8 ? W Z', / &B9 2 ;$ U
' 4 <,
!I Z, 4 & Z, / g >4 !` > > , /, &. ?2 < ,
!- 4 4 4
+ H4Y4, + H4, (4I
9 ' '
9 4, +2 Z, / ZM^ !` ?M^ !3JKa , 4 T59
' '

9 Z, / +2 ZM;
<, 2

&. Z4AB ? ^ Z^9 &N !` JKa , (iv )
4

4 4JKa, & &? !` &B9 +2 %
,& /I &. +2
?2
' '
@k8 @31/ , /: 2 GW: , 4 /I Z2 &B VP +, 4 /#5 !I , !I
!I + 'H, 2 . 9 '
&N 4 !- 4 !I, &?,
(Edition, ll . 765–802)
' ' ' '
Z, / T5 !I G? I &I /I
I M Z, /2 2 (i ) (B.22)
,4
VP'
(B.3ii )
&. ,4
+2 !I P Z#,
GY 4 !I )n84 9N T5 &? Y, ', I W( ) 7 8 Z', /2 2' (ii )

'

N&?, 2 NY,
I WAB !I P 4> 2 #5 T5
+2
9N T59 !I & M Z', /2 !I JO% ,/ I Z2 T5 &? !` /I

V\W &N !I ,
9^ '
2 !-. 4I &. GY4,/, +2 ' Y'4, 2 !-. ?W
$
&. !I, GY,/, M. VJK1 (Fig. B.22a)
4 4
u df ff
fu fj t
fc c ft

!` + H , +2
+, 4 /0D. ?,

Fig. B.22a
' '
!3 4' !3 &. G,/, , [b
!I Z , !3
!-. + H4, /H2
+, /#5 !I Z, / &N
4 / &. /: <N

&N . /I B +
MA +2 H4a4I +,4 / +,4 &? Z2 @31/ , /: 2
!-

: G,4
)" " &,4 9,4 wn84 G" &"
' 5 9 4 +2 Y% 5 GY 4, '
: T4S ]; +2 !I GY,
'
Y, @k8 (6 +H4M, +H4Y4 !I (JK4,4 /I 4M,4 (4I

!` +H4,4 / !I G
!I (JK4, 4 /I G Y, G,4 2' 4,

' +, 4 &. GM#$ + H4M, + H4,4 /
(JO v(,4 + H4, /H +, 4 /#W +2
]; ' ' ]; '
& (JO v(,4 +, 4 &. 4, !` ,4 / + H4, &
&., M, 9 9W) n8 , / !` , / + H,
4 !I,
4 4 4
+2 JO E # JO 9 /H4"
9^ V;)I 6 , 4 /
, ,

(Fig. B.22b)
9; &N JKMA B

h fl dc fi d
de u df ff i
j fu fj t dj
fh fc c ft fd
dh l f fe dd
4 !I " 9N T5 &? G Y, ', I ?

W Z', /2 2' (iii )
B.22b
GY
!` /I +2
&?, 2 Y, /,4 <4I 4> 2 TW $ T5
Y, T59 !I M Z', /2 !I +, 4 /0D. WA
JO% ,/ I G
+2 B !` ZM ,
. ?W 4 g !-> !3:'
(Fig. B.22c)
9; &N !- Y 9
/I
ft jj dj
dt dc df
du fu jf
@31/ , /: 2 !-. + H4, /H2

+, /#5 !I /I
&. !I, Z; $
Fig. B.22c
4

9; &N !I 2 2 .
!-
,4 )n84 ,4
:
/ T 2 )* 2 /2 T2
<N + H, /H2
+, /#5 !I

&N 4 4
&" G" wn84 9,4 &,4 G,4

9" w" G & 9 w7 G2 &2 92 w*
' ' Y, !3
<N M Z, /> T4S ]; !I G &. X(M, G)n84

' U '
JO ,4 / !I Z, !3 <N
&. ?,
JK2 &. /0D. 5

+ H4,4
!I G Y I /U I &.
U
/0D. , ,4 / !I !3
'M TWI +2
+ H4,4 / !I (JK4, 4 /I G Y, G" &" ,9E 4 + H4B
+, 4 &. + H44a2
4 !I JK Y, &. M, /I
9 &. !I X(M, '@k8 (6 + HM,
WAB ,2 % G" &" Z2 JKMA B 9E , (JK 4, G" &" ]^ ' 4
' ' ' !'I !3 M
+2 VP
!I WA B +2 9N & Z, / &N 9^ ]^ +, 4 /0D.
!I ,U W2 I !3 B
+, 4 /# 4 4 &. 9()* 9E, 4 G)n84 WA

U
&S I w" &" + H4, !3 M 9 4 !I JK Y, &. M, ;$ 4 X(M,
!3 &. &)n8 : T / <N G" &" /#5 !I G Y, !3 &.

, 4 /I GY 4, V\ V, GY 4 !I G Y, ,4 T2 w" &" /#5 !I G Y,
' 1
(Fig. B.22d) Z, / &N !I s
2 !-. /H4;4I /I &. +2
dd j jt ju dh
hf ft jj dj t
f dt dc df ht
jc du fu jf fc
di hu ff fj dl

+2 )* 2
&. ?,
/I &. +2
Fig. B.22d
< ,4

[ ) 7 8 Z', /> TS ]; ' Y,_ <? Y, 4 /0
/: 9N W( 4 !I G D.
+ H4Y4, !I /
&. ?,
+2
+ ' Y I 9 ;

H
4 +,4 &? G
' '@34 (6 + H4M, + H4Y4 !I (6 + H4, 4 /

T59 !I
9 Z, / Z^
,2
% JKMA B 9E , &. '5 X(M, '@k8 /I &., 95/

&N 9: I & ]; ' ' '
9N ]; !I JKY, + H4?5
4
/0D. (6.9()* 9E, 4 /2' GM , +, 4 &. , & B % !32 ', 4M. VJK1
];
' 4, g !-> !3:
9
'
!3 : +. N Y, B[ 4I 9E, 4
(JK Y, i h ,&B b W( , 9M#2 !` M^ ' '
+. "
+ H4,4 / !I (JK Y, ji jh '@k8 h i &?, (JK Y, hd hu '@k8 !3 :
'
!I 6 , 4 / &. Y, I fuc ]^
!I 2 Z4AB 9E 4I
&. +2 < ,4 T4S ]; +2 /#5 r% 2 &I P M, 9
4
<N &. >

&N 4
wn84 G,4
G w7

!I JK Y, + H4 > ?,
JKMA B 9E
JKMA B +2 X(M,
, &. '
?,
'

4 )n48 2 ( 9# 9^ +2 ]; !I 9 4, ?,
B %
NY, I G G,4 &SI 9?5 WA ,2 9N QI9 &?
9 4 !I N/,% # $ Y, ' '
2 !-. JO
9^ 6 M,
P ]; !I
' 9 ^
+2 ]^ ' ' ' !I
Z, / &N VP !I !3 &.
^ Z, / &N
' fuc 9N U I /# U U U
Z, / !I T59 !I /I GM;: I ^/. 9=

9; &N !I 2 . , / !I !3
!- 4 'M !I

9 '

(Fig. B.22e)
ji hd i c h hl jh
he dd j jt ju dh fe
jl hf ft jj dj t fd
u f dt dc df ht hj
de jc du fu jf fc je
fl di hu ff fj dl jd
fi l hh hc hi d fh
Fig. B.22e
p ' / Y,4 +E

>4 &N
+ 2 ? , / +2 N H4. !I &. G !-. (iv )
'
+ &JO ,
!R
(Edition, ll . 803–868)
B !-. /I
MA
? ', / !I Q I9 &? GY,/, !I V;I (B.23)
4

/#W !3? &N !I N/,5
% V()n8

'
/, +E
Q, 4 /= +2 Q I9 &? GY,/, +2
4 ' !3? &N
4 >4 &I (i )
(B.16–B.17) E &N +2 9 !I ,/" & V /H.
' ' 4
!3 ,4 /# / Z, / 94, !I &. GY4,/, !I bMW$ , b
/I &. G',/, &?, &B9 +2 49 &., N NEM5
' ' '
!I VI Y ,
2 GW: v#. Z, / T5 !I Z, 4 & Z, / !I
' '
94 /,
295/
94 +2
/H4> &?, 2 &I & V

/I &. 4 !3JKa , /, 95/2 GY
4, !` ?M, I /, 295/

&. 4 !3JKa, /, 95/2 GY
4, !` 4JKa, 4I Y,
&? / /, 4, !` 4JKa , 4I Y,
& 95/2 GY
g!31 4I G Y,4 @ ?U 4AB GY 4 &? ,
) n8 WY: /H. +2
4 4 4JKa ,
!I GY , _ &I /I &. !` 4JKa , /, 95/2 GY
4, !` 4JKa ,
Z^92 [
/: &? b & I 4I b G' Y,4 @ T59 !I & Z', /
+2
ZM, !` b V? , / , ?U 4AB GY
n8 94
' 4' &? ,
) '
T59 !I & Z, / !I GY
,
& !3
I &. !`

' %
4I
9
4 !I

/I &. +2 <, 2 &? , &UY4:
4 Z4AB g !->4 !`
9
&. 9E , Q I9 &? ?W $ Z', /2 !I @31/, ,
2 (ii )

,4 / !I !3 'M !I
T59 !I
9 ' '
Z, / !I /I
' '
&. Y ,
@k8 T5 GY 4 !I 2 9N T5 &? Y,I , %&,
' ' B Z', /2 !I Z , !3

Z4AB g !->4 !3: ?W$ Z, /2 !I WA /I
!-. ,4 / 9 4, &B %, , %&, '@k8 (B.15) /" &' , 2 !-. 9 4
Y, @-I
&? GY 4 b P
&B U GY 4 !I G Y,
G ) n8 &
@ + H4 , !R ];
' /#I !-. &
+2 GY 4 !I Y, YU% 45 4I
',
P YU% 45 ' '
];
+2
'
GY4 !-,4 & GY 4 !I G Y,
!I G Y, @ ?,
!R ];

+2 G GY 4 !I Y,
&?
GY 4 !I Y,
&? P ', YU% 45 ' '
]; +2 Z, / GY 4
' 4 !I Y, '5 !R ]; '
!R ]; X2) 7 8 GY
+2 Z, /
+2
, > !` , / !I Z', / +2 ', ? 5
!I 4 4 T5' /#W +2

/ , /0D. 4, ,4 / !I ?W$ Z5 ]; +2 T5 GY4
' ' ' 4 9W) n8
GY4,/, !I 2 &
9; !I Y4, (" V(? b !-. 9
'
]; +2 / , !` (Fig. B.16a) / ? ', / !I Q I9 &?
' ' ' ' '
v(, @k8 ,/" 2 @31
!-. 4I V(?, !R ]; !` 9?, @k8
. /H;I &? &?, M,
(Fig. B.23a) @31/ &N !- 4 4 9 4
dh dl ih dd
du jf if ij dj
dt jc ht jt cj
t dc hc hf ju cu uj
ic hj jj hu fu
ct cf df ft cc
ie ch fl cl
&" G" /#5 !I JK Y, &. M, / &. !I X(M, '@k8
Fig. B.23a
sm !32 !3: ' +

!I !3
&. ( $ ,2
H4?,
?W % JKMA
B 9E

,
<? Y, 4 & &? 9N + H, 5 + H,2 % JKMA B P /#W &N
4 > ? 4
' '
I +, 4 /# !I Q I 9 9N Z', / &N +2 ]^ VP !I 9E,4

!I G U U
Y, oM;$ /H" &. 6 ,&B &. &N W(
&N
B I ' Y,I 9
WA 4 &. !I X(M, @k8 T5 T. JKMA B +2
4
/I M, B
&.
JK2 9E, 4 ? 5 + H,2
4 JKMA
% B 9E

, &.
'
&. +2 4I 2 Z2 M() 7 8 G ,P
/#5 !I N Y, !32 !3:

,4 /. /I &. 9E 4I 9M# &? 9N + H4'5 ?W $ ,( % M,
9
6 M, 4 !I N/,% # $ Y, N Y, I G5 G2 B / ,4
&. +2
4I X(M, '@k8 U I &. U '
+2 !R ]; !I 2
^
' ' U U
/#5 +2 /#W !I JK Y, &. 5 B
&. /0D. , <N
&. JK ^ !32 !3: ' ' + % B
!3 4 /0D. 5 H4,2 JKMA 9E, 9#
,&B 9I QI9 &?M I 92 M()
7 8 P
9"
+2 /#W &N !I
9 4 !I JO% ,/ I Y, G. /#5 !I N Y, w* &2 9 G
' 9 ^ +2 &B VP '
Z, / &N !I !3 &. Z4AB
^ 6 M,
'
5 ?W$ ,(M, 9N U I
%
(Fig. B.23b) 9
9,4
&,4 G,4 . 9. o. fe fi fh fd uc ui ul le l
9, &" w" T5 T. &5 G" 9" ci dh dl f it ut ih dd di
G, /" 5 )x1 )" G2 cd he du jf if ij dj hd je
T" T2 ce c dt jc ht jt cj uu jd
T )n8 /. G

/01 " 2 2 / / , ,4 S. t iu dc hc hf ju cu fc uj
92 . 5 )* )7 /2 /, 4 ,4 9 hi uf ic hj jj hu fu ff ji
w7 T$ , P T$4 , G5 &2 jl de ct cf df ft cc id hh

& &,
I )n84 : wn84 wn8 w* jh ie ch lf fj j fl cl hl
&. 95 wx1
G. uh ii il ue u i h d ud
Fig. B.23b
(Edition, ll . 869–919)
(Composite squares)
(Odd order squares)
[
/: 9, (B.24)
[ $ Z', /2 !I &. G',/, +E
/ # W2 /: 9, !-. ?W ' ' >4 &I (i )
9 ^
!I !3 &. 9E , ? Y, Z, / &N
, /2 ?W @3D , 9N

' ' ' % U '
&. JK4I G,/, ? , /2 ?WY, Z, / @3D , , I JK2 &B VP
' ' ' '
!3: M Z, /2 !I G,/, " & GY4,/H !-. ?W $ !` &B +2
B 2 ]^ ' ' /
(Fig. B.24a) /0D. WA VP H4;$4
. / 0
1 2 3
4 5 6
Y
?W $ 9N &B9 I ZI
4, ?W
Fig. B.24a
& Z', / !I G',/, '@k8
4'
, ?W , I ZI
Y & Z, / !I ?W $ !` &B +2 !`
4 4
+2 <N ?W Y G Z', / !I
/0D. > !` /0D. +2
4 <N
% +, /0D. ?
&. Z4()n8 !R , !` V? , b &" 4
5 !` /0D. ?W

$
' ' 4I ?W Z', /2 !I Z , !3
Y
JK2 ]^ VP !I Z, !3 &. 9E
(Fig. B.24b) JO
9^ U I
&N
ff fl fj uh lf ui dt ji jf
fi fh fd ut uu uc jh jd je
fc fe fu ul uj le jj dl jc
ci ij cl jl hc he de du dd
if ct cu hj hf jt dc dj df
ie cc id hd ju hh dh ft di
hu ch ht d t h ic ud iu
cd ce hl u c j ue il ii
cf hi cj i f l it ih uf
Fig. B.24b
V2 M +2 G'"/2 /I Z', /2 VP '

!I b V(?, +E >4 &I (ii )
B Z', /2
@3D , 4 /0D. WA
' ' '
I +, 4 /0D. &B Z, /2 /0D. WAB Z, /2
', /2 ?W $ M, !I M, +2 JK2 &B VP ' U ' n8
? ?, /2 +,4 /0D. W()
?WY, @3D , +, /0D. &B Z', /2 WA B !I WA
B +2 '
4 4 &B VP
'
, /2 ?WY, @3D,4 M, !I M, +2
JK2 &B VP ' U ' +
? ?, /2 H4?,

' ' &B VP '
!-. 6MP JK4I QI9 &? G,/, +E>4 ?5 !I ?5 +2
' ''
(B.3–B.6, B.15–B.17) & , (4I Y4, 2
(Edition, ll . 920–937)
(Particular cases of even-order squares)

(Composite squares with different sums in the subsquares)

/# ', /2 !I Q I9
/ ?
+2 &? /?,
4 (4I V;I (B.25)
oM

&?, ' '
9N ?, 4 (N&B' + H4.9, !` @3D , 4 / &. @-. (i )
/
9N ?,
&?, /: [ / /
/

[ ' ' 4 ' ' '
B &?, Z, /2 &. &?,4 ?, /2 I ?,
&?,4 & &?
' ' ' ' ' '
@3D , 4 ?, /2 I ?,
&?,4 &. VP I /0D. 5 9N ?,
Z, /2

U' '
Z', /2 &?, (N&B + H,/2' ? '
!R ?, /2 /0D. 5 9N ?, 4 , />
U ' ' ' ' '
&? Z, /2 &?,
/2 /0D. G5 Z, / b &?,4 &
' I ' '
9N ?, /A /2 GM, ?,
&?,4 &. 9N /0D. , b 2
$ JK2 &B VP ' U ' '
4, ?W
9 ?, /2 /0D. W$ @3D , 4 9?,
?,
,2

%
YU , /0D. '5 JK2 &B VP ' ' U
b &" 4 ?, /2 ?WY, @3D,4 ;$4 9N
' '
/" XAB &? ' '
@3D , 4 ,2
% Z,
9N ?, /2
, ?,
I /0 D?
) n8 @3D , /0 ' U' '
+, 4 /0D. W( 4 D. WAB JK2 &B' VP ?, /2 /0'D. U W$ '

/0D. 5 JK2 &B VP ?, /2
' ', I ,/"

Y, ,
%
Z, /2 !I &. G 2 !-. /2 P I (ii )
' '
&. Y ,
?
% '
, / !.9, &B , Z, / (W I ?,
+2 G'"/H,4 ?M^
' ', / bM, !I <?4# v# !-. &B9 +2
2 &I 2 GW: ? ,& %

U ' ' ' ' '
&. JK2 Z, / !I G,/, @k8 / / ? , / !I Y4,/,
' ' [ % , ,94, &? ,U 4 W2
!I G,/, @k8 9 4 &. N/: &B9 N&
4 4JKa , & &? N& % , ,94, &? ,U 4 W2 &. U '
!R Z, /
' ' '
9W) n8 !` ,/2 &?, ,/2 &. G,/, b &" Z, / !I
'
Z, / !I ? ', / Z4AB !I Q I &. V;S4I Z', / 9 4, Z4AB
U
Q I &. ;$ 4 /H"
' ' '
@3D , 4 ?, /AI /2 GM, ?,
&?, /0D. !3 , !I b 2 (iii )
' I YU 4, /0D. '5 JK2 &B VP ' '
4I G,/, ,
' ?, /2 ?WY,
' ' ' '
/0D. 5 Z, /2 VP ? , /2 ?W Y, 9W2

, !-. MA . Q I9
&.
' ' &. Y ',
YU 4,
!-. /H4;$ !3: ?, / bM, !I ?W $ !` &B +2
(Fig. B.25a)
9; &N
' /0D. '5 !` &B +2 ' ' ' '
Z, / !I &. Z, / !I G,/, @k8
+ H4M, M, +2 + HM, + H , !` /0D. ? 5 +2
G Z, / !I 4 4 !R
U ' ' ' '
!-. &. /0D. 5 9M,4 Z, /2 VP !I G,/, b &" + H4?,
4 > !`
/H4;4I /0D. !3 , Z', /2 !I Z , !3 &. Z4()n8 !R %, !` g9

U U U
U '
I /#I ^/. 9= Z, / &N !I !3 &. Z4AB
6 55 50 5 . / 0
57 2 0 52
3 54 / 4
1 2 3
53 . 1 5. 4 5 6
, _ ?U ', /2 /0D. 'W$ Z', / &N
@3DI_
Fig. B.25b Fig. B.25a
49 &. JK4I GY (iv )


Za', /2 !I aa,/, ' Y', '" 2 V2 <?aaa# v# !-a.
a?, 4 4 4
' ' ,_ '@k8 (Fig. B.25b)
JK4I V;: 2 P
g9 !-. &. Z, /2 VP !I GY
U
I &. +2

(Edition, ll . 938–970)
(Composite squares with same sums in the subsquares)
6.r^ !3 ? ', / !I Q I9 &? G Y, +E >4 &I (i ) (B.26)
U ' '' _
&?, 29W2 Z, / 9E,4 9N Y4, & +2 +W: : 9, Z,
F? , JKL?, JKU a5 U I 6'MP 6I9 ^ !I !3 &. 9E , ? ', /2
4
96I (W +2 /: AI /0D. '5 !-. Z', / (W I b ,
I (ii )
' ' ' ' _ ? ', / &.
5 4I JK2 Z, /2 VP 9E,
4 Z, / b JO @3D,4 !3
' ' $ !I Y, I YU , /0D.
49 &. Q , 2 ] 4" JK2 Z, /2 ]; 4
' '
Z, /2 !I ,, Y,
& GY4,/H !-. <?4# v# !-. &B9 +2
> !` (B.9) !R 9 !I N' Y'4, &I !3
4, 9: 9 &B %, ?,

Q ', 2 ] " /: [ Z', /2 !` M , Z', / b & &? &?, N( q' >
4 ' '

&? &?, N( q> /0D. 5 !` ?W$ +2
;$ !I &? Y,
!3 &. & %Y 4: /H4;4I 4I ? ', / Z4()n8 !R %, !` V? , b &"
U U U '

9^ Z4A !I !3
B &. ;$ 4 9E,4 I Z, / 9 ^ ZAB !I
4
U ' '

I JK2 Z, /2 VP
/0D. '5 !-. @3DI_ + H4?,
?,
P
,2

% !I b 2 (iii )
' ' '
JK2 Z, /2 VP P ?, /2 ?WY, (WI I ?W$ (W +2
) 7 8
'
!I 4 > !` &B9 +2 &. Y ',
, %&, I YU 4, /0D. '5
!-. V4 GY 4, !I 4I 9
! 6 , /
I &. Y ',
JK2 Z', /2
4 4
U '
' ' '
'
&. /0D. 5 JK2 Z, /2 VP !I Y,
(B.9) !R 9 !I 2 &I 2
U ' U /
I Z, / P , H2
' _ '5 ?, Z', /2 !I 2 (iv )

/0D. 5 !-. @3DI 9

,94, 4,(
' ' ' '
%, (W
/0D. 5 JK2 Z, /2 VP P ?, /2 ?, I I ?,
/SaI
@31/ !-. &. JK4I G Y, YU 4,
(Edition, ll . 971–990)
SM
!I/ Q I9
&? !I FSM ,
4
(Extract)
'
Q MS
/ Q I9 !I (C.1)
GY4,/, ./2 Z2
+. /I I
/ I _ Q I9 +2 9, &N (i )
4
/I ? ', / !I ' !'R %,4 9 &N T59 !I Q I9
U ' U' /U I ?U ', /2 ? ^
,
92 ?, /2 #5 !I ## : b ,
I (ii )
Z', / (W ' ' )n8
/ I @k8 / Z, / r^ ' ;$ !-. ,4
Z, G 4

!I T5 &? 9N ;$ Y, ;$ &? &? I/. 2WI,
'

rB !-. &. 4,> T5 &? +2 :/= @k8 T5 GY4
' ' A I &? !3JKa , AI !`9
?2 &?, M Z, / &B9 2 4
Q, /#, M Z', /2 !I JK',/, T5 &? Z2 !`9 !-. /I ?W $
4 '

AI /0D. 5 &B9 2 A I & &? +2 TW $ '@k8 9M?
?2 &? ' 7 ' A I &? !3JKa ,
!` /I , W() 8 Z, / &B9 2 4
!-. G,)7 8 +2 ' U '
6M2 &?, &B9 2 AI & &?
+2
' ' ' U' &?
@k8 M Z, /> #4S
9#W !I JK,/, 4 4JKa , &

&B9 2 A I & &? +2 $
4 !3JKa , AI +, 4 /0D. ?, TW

&EN W Z', / /: [ G,) 7 8 +2
? J O /I
% ,/ I ?2 , & B % &?
' '
/I +2
<, 2 Z;$ @k8 JO? , /2 !I JK',/, 4 > 4 > ,4 /, V? ,
' ' '

,4
!I
&. G,/,
9 Z, / ,4
!I + H4I/#

+.
'
/H4;$ Z, / g !->4 !` +E2 2 ] 4" ,4 / !-,4 2 v#. '
Z, /
U
I &.
GY 4 !I T5 &? G Y, W( ) 7 8 Z, /2 ,/" '
V'( (iii )
' B $ '@k8 )n8 9N T5
6?2 &?, WA !` !3JKa I 4 > )n84 +2 TW 4
' ' &. ?W $
!I M Z, /2 !I JK,/, T59 Z2 !`9 !-. /I +2
384 Opuscule on the magic square by al-Kharaqı̄
'
Z, / /H4;4I /0 D. &B NJ B N&
% Ka , WA % , W( ) 7 8 Z', /2 T5

!` + H , +2 % '
& B , @k8
9; &N
<N +, 4 /0D. ?, 4
!-.
+2 '
<, 2 JK4 @3L $ + H4, /H2 +, /#5 !I 6?; $ Z, /
'
/I 4
' !-. + H4I/#
/I Z; $ @k8
9; &N !-.
!3 :
+.
+ H4,4
!I : (N JK2 + H4 ' ' '
Z; I
9 Z, / ,4
!I 4I
'
&N !-. Z, / /H4;4I )" " (N (JKM, !I (JK, /I
2 + Ha2
' 4'4 '
4 4
/?
9#5 !3,4
!I G,/, @k8
/? '

9#5 !3,4
!I (N Y,I 2 !I + 'H4?M I G,4
' '
+2 +, 4 /#W !I (2
+,4 &. GM#$ @k8 (JKaM, 2 !I (JK4,4 /I GY,
' ' '
!I !3 M &. Z2 , @3,4 (2 a ,/a a
&.
' ' &? 9N 9 '5 WA B /?
VP !I G,/, <? Y, 4 &
9#5
!-. (N Y,I 9,4 wn84 ,&B
Y, " !3 : 9I 2 !I Q I9 '@3 /#5
' [ 4

9#5 !I (6?; +, 4 /: +, 4 &. M= (JKM, 2 !I (JO4 /H4# $

)" !3a !-. (N Y,I G,4 ,&B
a: 9I 9 '5 WA B (JO '@3 9#
' 4
%
(Fig. C.1a) 46 &N !-. Z, /
;aI (JKM, 2 !I (JO4 /H4# $ Y,

wn84 " 9,4

P ,4
: /, 4 )n84 )"

&,4 ,4 T$4 G,4

&" G"
@-' ? v6I !-. (

U
Fig. C.1a
&N
.
& !I ,/0L
I (iv )
&I !-. @3D &N
Q I9
&? !I /: 5
!I . ,
,

&N !- 4
(Edition, ll . 364–401)
Bibliography
Abū Kāmil: The Book of Algebra, Kitāb al-Jabr wa l-muqābala [= Publi-

cations of the Institute for the History of Arabic-Islamic Science, C.24].
Frankfurt am Main 1986.
See also: Chalhoub, Sesiano, Weinberg.
Anaritius (al-Nayrı̄zı̄): Anaritii in decem libros priores Elementorum Eu-
clidis commentarii [= Euclidis Opera omnia, Supplementum], ed. M.
Curtze. Leipzig 1899.
A. Anbouba: L’algèbre al-Badı̄‘ d’al-Karagı̄, ed. with an introduction.
Beirut 1964.
J. Campanus: Preclarissimus liber elementorum Euclidis perspicacissimi
in artem Geometrie (or: Opus elementorum Euclidis Megarensis in
geometriam artem, in id quoque Campani perspicacissimi Commenta-
tiones). Venice 1482.
S. Chalhoub: Die Algebra, Kitab al-Gabr wal-muqabala des Abu Kamil
Soga ibn Aslam, ed. with comm. Aleppo 2004.
Euclid: Euclidis Opera omnia, edd. J. Heiberg & H. Menge (9 vol.).
Leipzig 1883–1916.
: Elementa [= Euclidis Opera omnia, vols. 1–4].
See also: Heath.
P. Fermat: Varia opera mathematica. Toulouse 1679.
: Oeuvres (4 vol.), edd. P. Tannery and Ch. Henry. Paris 1891–
1912.
B. Frénicle de Bessy: “Table generale des quarrez de quatre”, in Divers
ouvrages de mathematique et de physique (Paris 1693), pp. 484–507.
T. Heath: The Thirteen Books of Euclid’s Elements (3 vol.). Cambridge
1926.
Iamblichos: In Nicomachi Arithmeticam, ed., transl. with notes N. Vinel.
Pisa & Roma 2014.
Ibn al-Nadı̄m: Kitāb al-Fihrist, edd. with notes by G. Flügel, J. Roediger
and A. Müller (2 vol.). Leipzig 1871–1872.
of Mathematics and Physical Sciences, DOI 10.1007/978-3-319-52114-5
386 Bibliography
al-Kharaqı̄: see Sesiano (Herstellungsverfahren).

al-Khwārizmı̄: The Algebra of Mohammed ben Musa, ed. and transl. F.
Rosen. London 1831.
al-Kindı̄: Risāla fı̄ istikhrāj al-a‘dād al-mud.mara. Manuscript Istanbul,
Aya Sofya 4830, fol. 81r –86r (copied 1228/9 [627h]).
Liber mahameleth (3 vol.), ed. J. Sesiano. Cham 2014.
M. Moschopoulos: see Sesiano, Tannery.
al-Nayrı̄zı̄: see Anaritius.
Nicomachos: Nicomachi Geraseni Pythagorei introductionis arithmeticæ
libri II , ed. R. Hoche. Leipzig 1866.
: Tābit b. Qurra’s arabische Übersetzung der ᾿Αριθμητικὴ Εἰσαγωγὴ
des Nikomachos von Gerasa, ed. W. Kutsch (S. J.). Beirut 1959.
J. Sesiano: “La version latine médiévale de l’Algèbre d’Abū Kāmil”, in
Vestigia mathematica (Amsterdam 1993), pp. 315–452.
: “Herstellungsverfahren magischer Quadrate aus islamischer Zeit,
III”, Sudhoffs Archiv 79 (1995), pp. 193–226.
: “Les carrés magiques de Manuel Moschopoulos”, Archive for the
history of exact sciences 53 (1998), pp. 377–397.
: “Le Traité d’Abū’l-Wafā’ sur les carrés magiques”, Zeitschrift
für Geschichte der arabisch-islamischen Wissenschaften 12 (1998), pp.
121–244.
: Une Introduction à l’histoire de l’algèbre. Lausanne 1999. English
edition: An Introduction to the History of Algebra, Providence 2009.
: “Une compilation arabe du XIIe siècle sur quelques propriétés des
nombres naturels”, SCIAMVS 4 (2003), pp. 137–189.
: “Quadratus mirabilis”, in The Enterprise of Science in Islam,
New Perspectives (Cambridge Mass. 2003), pp. 199–233.
: “Magic squares for daily life”, in Studies in the history of exact
sciences in honour of David Pingree (Leiden 2004), pp. 715–734.
: Les carrés magiques dans les pays islamiques. Lausanne 2004.
Russian edition: Magicheskie kvadraty na srednevekovom vostoke, Saint
Petersburg 2014.
: Récréations mathématiques au moyen âge. Lausanne 2014.
F. Sezgin: Geschichte des arabischen Schrifttums. Leiden 1967- .
H. Suter: Die Mathematiker und Astronomen der Araber und ihre Werke
[= Abhandlungen zur Geschichte der mathematischen Wissenschaften
mit Einschluß ihrer Anwendungen, X]. Leipzig 1900.
Bibliography 387
P. Tannery: “Le Traité de Manuel Moschopoulos sur les carrés magiques,

texte grec et traduction”, Annuaire de l’Association pour l’encourage-
ment des études grecques en France (1886), pp. 88–113. Reprinted in
Tannery’s Mémoires scientifiques, IV, pp. 27–60.
: Mémoires scientifiques (17 vol.). Paris & Toulouse 1912-1950.
J. Tropfke (et al.): Geschichte der Elementarmathematik. Berlin 1980.
(Ps.-) T.ūsı̄: Kitāb tah.rı̄r us.ūl li-Uqlı̄dis min ta’lı̄f khōgah (ut fertur) Nas.ı̄r
al-Dı̄n al-T . ūsı̄. Rome 1594.
N. Vinel: see Iamblichos.
B. Violle: Traité complet des carrés magiques (3 vol.). Dijon 1837-1838.
J. Weinberg: Die Algebra des Abū Kāmil Šoǵā´ ben A
. slam (transl.). Mu-
nich 1935.
F. Woepcke: “Mémoire sur la propagation des chiffres indiens”, Journal
asiatique, s. 6, vol. 1 (1863), pp. 27–79, 234–290, 442–529. Reprinted
in his Etudes, II, pp. 264–461.
: Etudes sur les mathématiques arabo-islamiques (2 vol.). Frankfurt
am Main 1986.
Index
A
Abū Kāmil : 14, 137n-140n.
Abū’l-Wafā’ : see Būzjānı̄.
abundant number : 12, 180, 181n.
amicable numbers : 8.
Anaritius : see Nayrı̄zı̄.
al-Ant.akı̄ : 9-11, 13-15, 119, 120 ♦ (missing part of his Commentary)
10-11, 15, 132n, 135n, 136n.
See also: Text A.
Apuleius : 11n.
Archimedes : 11.
Asclepios : 11.
associated number : (def.) 224n.
B
Barlaam : 140n.
Bede : 15, 185n, 186n.
Boëtius : 11.
bordered magic square : see magic squares.
broken diagonals : 4, 5, 23-24, 26, 28, 29, 36, 37, 39.
al-Būzjānı̄ : 16, 207.
See also: Text B.
C
J. Campanus : 121n, 134nn, 140n.
central square, method of the : 113-115; 172-173, 175-176.
S. Chalhoub : 137n.
chess : 6, 135n, 162, 165 ♦ (moves) 6, 21, 22, 29, 30, 107, 165, 211-212,
213n, 216, 218, 228, 252.
complements : (def.) 4, 143n, 224n.
completing the sum (bordered squares) : 44, 47, 52, 60, 70-71; 229-234,
236-240, 243-246, 248, 254.
of Mathematics and Physical Sciences, DOI 10.1007/978-3-319-52114-5
390 Index
composite magic square : see magic squares.

construction methods : (general) 5, 6, 8, 9, 17, 23-26, 36-43, 47-51, 53-59,
59-66, 72-103, 105-106, 107-115; 143-148, 148-163, 166-170, 170-179,
234-235, 249-252 ♦ (empirical) 43-47, 69-72; 222-234, 235-240, 242-
248 (leading to general method) 43-47, 222-234 ♦ (not general) see
completing the sum, displacement.
See also: central square, cross.
continued proportion : see progression.
cross, method of the : 108-113; 172, 176-179.
D
deficient number : 12, 180, 181n.
diagonal, broken : see broken.
Diophantos : 11.
displacement (in the natural square): 20-23, 26-27, 30-36; 143, 211-214,
215-217, 218n, 219-222.
E
Elements : see Euclid.
equalize : (meaning) 209n.
‘equalizing number’ : (def.) 209n, 224n.
equations (systems of) : 195n, 199n-200n.
Eratosthenes : 12.
Euclid : 10-13, 15, 120n-136n, 140n-142n, 180n, 195nn ♦ (Greek com-
mentaries) 13, 133n, 134nn.
L. Euler : 136n.
even (-order squares) : (evenly even) 6, 30 (evenly odd) 6 ♦ (construc-
tion) 26-42, 51-66, 105, 107, 108-115; 164-179, 208n, 214-222, 235-240,
250-252.
evenly even number : 12, 121, 122, 164, 180, 188, 202, 250.
evenly evenly odd number : 122, 164, 180, 250.
evenly odd number : 12, 121, 122, 164, 180, 250.
F
P. Fermat : 8.
L. Fibonacci : 200n.
Index 391
Fihrist : 10.
B. Frénicle de Bessy : 6n.
general method : see construction methods.

Greek numerical system : see numerical system.
T. Heath : 120n, 133nn.

J. Heiberg : 120n, 126n, 128n.
Heron of Alexandria : 133nn.
hidden numbers : 15, 120 ♦ (determination of) 180-196, 201-205 (in
antiquity and early Middle Ages) 15, 185n, 186n ♦ (hidden names,
determination of) 196-201.
R. Hoche : 11n, 15n.
Iamblichos : 11, 16.

Ibn al-Haytham : 24, 26.
Ibn al-Nadı̄m : 10.
Ibn al-Qift.ı̄ : 10.
identities, algebraic : 140, 195n.
K
al-Karajı̄ : 15n, 140n.
al-Kharaqı̄ : 18, 102n, 253-255.
al-Khwārizmı̄ : 14, 136n, 137n.
al-Kindı̄ (K): 15, 180nn-183n, 186nn, 187nn, 190n, 191nn, 193n, 194n,
196nn-198n.
W. Kutsch : 11n.
L
‘large’ number : (def.) 43, 143n.
Liber mahameleth : 140n.
392 Index
M
magic squares : (def.) 3; 143, 208 (as talismans) 7-8 (origin of the
name) 8 ♦ (ordinary) 3, 6, 7, 17-18, 19-42; 143, 209, 219n, 210-222
(bordered) 3, 4, 6, 14, 17-18, 19, 30, 31, 36, 43-66; 143-148, 166-170,
208-209, 212n, 222-240 (see also, here below, separation by parity)
(pandiagonal) 4-5, 7, 28-30 (composite) 5, 6, 17, 18, 30, 104-115;
170-179, 208n, 248-252 (separation by parity) 6, 7, 14, 17, 18, 67-
103; 148-164, 208n, 240-248, 253-255 (symmetrical) 25 (with non
consecutive numbers) 7, 19, 147n, 208, 241n ♦ (no magic square of
order 2) 3-4, 26, 51; 165, 214, 225n (square of order 3) 6, 20-21, 26,
67-68; 143-144, 210-212, 215, 231n, 235-236, 242 (square of order
4) 6, 26-30, 38-39, 51; 165, 214-219, 225n, 227n, 235 (square of
order 5) 21-22, 43-46, 74, 102-103; 144-145, 153, 158, 213-214, 227-
231, 241-243, 253-254 (square of order 6) 30-31, 59-60, 66, 112; 167,
219-220, 236-237 ♦ (in Greece?) 6, 9, 15 (see also, here below, early
history) (in China) 6 (in India) 6 (in late mediaeval Europe) 8
(early history) 8, 9, 11, 17, 28, 43, 67, 74, 79, 104, 143, 240 (tenth
century) 6, 8, 14, 16-18, 19-22, 26-36, 38, 41, 43-47, 51-54, 58-61, 67-
103, 104-115, 219n, 238n (see also text A, text B) (eleventh and
twelfth century) 7, 19, 23-26, 36-42, 49n, 56, 59, 61-62 ♦ (number of
configurations) 6, 18, 43-46, 56-58, 60, 62-64, 102-103; 230, 235n, 237,
240, 243, 254n.
See also: completing the sum, construction methods, displacement,
even (-order squares), magic sum, odd (-order squares), sum due.
magic sum : 3, 4, 19, 20, 23, 27, 29, 31, 36, 37, 49, 50, 75, 84, 109, 208,
209, 210, 214, 216-218, 228n, 229, 245, 250n, 251.
manuscript Ankara Saip I, 5311 : see text A.
manuscript Istanbul, Aya Sofya 4843 : see text B.
‘median’ number : (def.) 43, 143n, 174n, 224n.
M. Moschopoulos : 6.
N
al-Nasawı̄ : 10.
natural square : 17, 19-25, 26, 30-31, 34, 36-42; 143n, 210n, 212n, 213-
216, 228n.
al-Nayrı̄zı̄ : 133n, 134nn, 140n.
‘neutral’ placements : 54-55, 58, 59n, 62, 65, 79-80, 84, 94, 96, 101, 110;
156n, 162n, 168n-169, 176n-179.
Index 393
Nicomachos : 10-13, 15-16, 119, 120, 121nn, 127n, 129n-132n, 134n-136n,

180nn.
numerical system, alphabetical : 7-8, 196n, 198n, 257.
O
odd (-order squares) : (def.) 6 (construction) 9, 20-26, 43-51, 67-103,
104-105; 143-164, 208n, 210-214, 227-235, 240-248, 249-250, 253-255.
order : (def.) 3 (categories of) 5-6.
ordinary magic square : see magic squares.
P
pandiagonal magic square : see magic squares.
parity (kinds of) : 12, 164, 180.
perfect number : 8, 12, 15, 135n, 136, 180, 181n.
Philoponos : 11.
plane number : (def.) 180n; 130, 132, 133, 181.
planets : 7.
‘predecessors’ of B : 28, 217, 240 (& n. 618).
prime number : 12, 108, 120, 125-128, 136, 180, 187n ♦ (infinity of pri-
mes) 12, 127-128 ♦ (relatively prime) 12, 120nn, 122n, 125-126, 127n,
128-130.
Proclos : 11.
progression, geometric : 12, 126-131, 135-136.
Pythagoreans : 8.
S
s.aff : 145n, 209n.
I. S. Sencer : 9.
F. Sezgin : 10n, 16n.
‘small’ number : (def.) 43, 143n.
solid number : (def.) 180n; 132-134, 145, 181.
sum due : 4, 54, 56, 65, 71, 74-75, 78-81, 83, 84, 87, 89, 94, 95, 97, 100,
104, 109, 110, 113, 153-164, 170n, 176.
H. Suter : 10n, 16n.
systems : (linear) 199n-200n (quadratic) 195n.
394 Index
T
P. Tannery : 6n.
text A : (manuscript) 8-11, 15, 28, 147n, 157n, 159n, 165n, 205n, 257 ♦
(content) 6-7, 9, 13-15, 17, 19, 21, 22, 26, 27-28, 43, 48, 51-53, 60, 61,
67, 69, 72-102, 104-115; 119-205, 257 (missing part) 10-11, 132n,
135n, 136, 205n.
text B : (manuscript) 8-9, 16, 70n, 208nn, 230nn, 257 ♦ (content) 6,
8-9, 16-18, 19, 20-22, 26-36, 38-39, 41, 43-48, 51-53, 60, 67-72, 102,
104-106; 145n, 207-252, 257.
text C : see Kharaqı̄.
Thābit ibn Qurra : 11.
J. Tropfke : 15n.
al-T.ūsı̄ (pseudo-) : 134nn.
V
N. Vinel : 16n.
B. Violle : 45.
W
wafq : 8, 208n.
J. Weinberg : 137n.
F. Woepcke : 10n.

Magic Squares in The Tenth Century: Jacques Sesiano

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Magic Squares in The Tenth Century: Jacques Sesiano

Uploaded by

Copyright:

Available Formats

Sources and Studies in the History of Mathematics

and Physical Sciences

More information about this series at http://www.springer.com/series/4142

Magic Squares in the Tenth

ISSN 2196-8810 ISSN 2196-8829 (electronic)

Library of Congress Control Number: 2017935465

© Springer International Publishing AG 2017

Printed on acid-free paper

This Springer imprint is published by Springer Nature

Chapter I. Ordinary magic squares . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Chapter II. Bordered magic squares . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

Chapter III. Separation by parity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

Chapter IV. Composite magic squares . . . . . . . . . . . . . . . . . . . . . . . . . 104

Part II: Translation

Part III: Arabic texts

Chapter III. Determining the hidden number . . . . . . . . . . . . . . . . . 316

Tenth-century construction methods

I. A square displaying this magic sum in the 2n + 2 aforesaid rows is an

But there are magic squares which display further properties.

composite squares, as an attempt to reduce the construction of larger,

its eﬃcacy), if possible at some astrologically predetermined time (which

§2. The two texts from the tenth century

a theorem and a test serves as a demonstration. A typical example is the

Also found, as A.I.124–A.I.133, are the ﬁrst ten propositions of Book

fact correspond to algebraic identities.16

they are anyway placed opposite (see above, p. 4).

Text B is divided into four parts. In the Introduction we ﬁnd various

The way B treats his subject enables us to reconstitute the beginning

opposite rows in order to obtain in each the required sum. Moreover,

The construction of ordinary magic squares, which was to attain a

§1. Construction of odd-order squares

Fig. 9 Fig. 10 Fig. 11

Fig. 14 Fig. 15 Fig. 16

1 2 3 4 5 0,1 0,2 0,3 0,4 0,5

6 7 8 9 10 1,1 1,2 1,3 1,4 1,5

11 12 13 14 15 2,1 2,2 2,3 2,4 2,5

16 17 18 19 20 3,1 3,2 3,3 3,4 3,5

21 22 23 24 25 4,1 4,2 4,3 4,4 4,5

Fig. 21 Fig. 22 Fig. 23

It is clear that neither A nor B knew of such a method. The ﬁrst

§2. Construction of even-order squares

Fig. 25 Fig. 26 Fig. 27

each place individually (A.II.37), even though to construct this square

Fig. 32 Fig. 33 Fig. 34

Fig. 35 Fig. 36 Fig. 37

Fig. 38 Fig. 39 Fig. 40

Fig. 43 Fig. 44 Fig. 45

magic sum in the square of order 6 is 3 · 37 = 111, the rows of the

1. Evenly even orders

1 143 142 141 5 6 7 8 136 135 134 12

132 121 +108

120 109 +84

133 11 10 9 137 138 139 140 4 3 2 144

1 134 135 136 8 7 6 5 141 142 143 12

133 11 10 9 137 138 139 140 4 3 2 144

Applied to order 8 (Fig. 50–52), these transformations give successively

2. Evenly odd orders

1 2 3 193 192 191 190 189 188 187 186 12 13 14

183 184 185 11 10 9 8 7 6 5 4 194 195 196

1 2 3 11 192 191 190 189 188 187 186 12 13 14

183 184 185 193 10 9 8 7 6 5 4 194 195 196 Fig. 55

1 13 12 11 187 188 189 190 191 192 186 3 2 14

154 141 +98