Professional Documents
Culture Documents
377827-Text de L'article-545877-1-10-20201217
377827-Text de L'article-545877-1-10-20201217
Bernard R. Goldstein
Dietrich School of Arts and Sciences
University of Pittsburgh
brg@pitt.edu
Abstract: In this paper we examine several aspects of the Parisian Alfonsine Tables
[henceforth PAT] that appeared in Latin around 1320, first in Paris and then diffused
throughout Europe. Our focus is on the tables for precession/trepidation, the motion of the
planetary apogees, the radices for mean motions, and the mean motions. The goal has
been to identify sources for the parameters, and the result is that, for the most part, Anda-
lusian zijes provided the required information, indicating continuity from the table-mak-
ers in al-Andalus to Latin Europe. We derive the parameter for the Alfonsine motion in
precession from the length of the tropical year ascribed to Azarquiel by Abraham Ibn
Ezra. Although the sources for some parameters in PAT have not been identified, there is
no evidence that new observations played any role. Among the sources for PAT were the
Toledan Tables and al-Battānī’s zij. In the case of the planetary apogees it is shown that
the values in PAT were probably derived from those in al-Battānī’s zij, whereas those in
the Toledan Tables were almost certainly derived from that zij, despite the fact that the
Toledan Tables use sidereal coordinates and al-Battānī used tropical coordinates. In the
case of the radices, again values in the Toledan Tables were derived from those in al-
Battānī’s zij, but the values in PAT show no such affinity. The mean motions in PAT are
closely related to those in several Maghribi zijes, which supports the suggestion that a lost
work by Azarquiel may be the common source.
* The contribution of one of us (JC) was written in the framework of the European Research
Council project ALFA, Shaping a European Scientific Scene: Alfonsine Astronomy, under the EU
program Horizon 2020 (Grant Agreement 723085)
Chabás, José; Goldstein, Bernard R. (2020-2021). «New Approaches and Parameters in the Parisian Alfonsine
Tables». Suhayl 18, pp. 51-68. ISSN: 1576-9372. DOI: 10.1344/SUHAYL2020.18.3.
Keywords: Alfonsine Tables, trepidation, apogees, radices, mean motions, al-Battānī, To-
ledan Tables, Azarquiel, John of Murs
Introduction
The Parisian Alfonsine Tables were the most widely diffused set of astronomical
tables in Europe in the fourteenth and fifteenth centuries. They originated in Cas-
tile in about 1272, but only the canons (or instructions) in Castilian survive, that
is, the earliest version of the tables appeared, in Latin, in Paris around 1320. There
are several sets of canons to PAT in Latin, but none of them addresses the sources
for the parameters underlying the tables, and little information can be gleaned from
the Castilian canons either. In general, each manuscript of a set of medieval astro-
nomical tables tends to have a slightly different collection of tables. Fritz S. Ped-
ersen’s edition of the Toledan Tables (2002) provides the clearest example of the
variation in the manuscripts of the same set of tables. No comparable study has
been made of the many manuscripts of PAT, and so we have to rely in part on the
early printed editions (Ratdolt 1483 and Santritter 1492). The goal of this paper
is to investigate the sources of the underlying parameters in PAT, where we focus
on precession and trepidation (§ 1), planetary apogees (§ 2), radices (§ 3), and
mean motions (§ 4), based on the zijes of al-Khwārizmī (fl. 830), al-Battānī
(d. 929), Ibn al-Kammād (fl. 1100), Ibn Isḥāq al-Tūnisī (fl. 1193 - 1222), Ibn
al-Raqqām (d. 1315), and Ibn al-Bannāʼ (d. 1321), as well as on the extant Latin
version of the Toledan Tables.1 In previous studies we have addressed the plane-
tary equations and the velocities of the Sun, the Moon, and the planets, and the
results of this new study are consistent with them, that is, most of the parameters
in PAT originated in the Iberian peninsula.2 As will be shown below, the approach
to the general problem of the frame of reference for the coordinates of the celes-
1. For the zij of al-Khwārizmī, we consulted Suter 1914 and Neugebauer 1962a; for the zij of
al-Battānī, we consulted Nallino 1903 - 1907; for the Toledan Tables, we consulted Toomer 1968
and Pedersen 2002; for the zij of Ibn al-Kammād, we consulted Chabás and Goldstein 1994; for the
zij of Ibn Isḥāq we consulted Samsó and Millás 1998 and Samsó 2019; for that of Ibn al-Raqqām,
we consulted Samsó 1997 and Samsó and Millás 1998; and for that of Ibn al-Bannāʼ, we consulted
Samsó and Millás 1998.
2. See Goldstein and Chabás 2001; Chabás and Goldstein 2012, pp. 63 - 81, 95 - 101, and the
literature cited there.
52
tial objects relied on a compromise between previous approaches and it was based
on round numerical data of no astronomical significance. For the apogees, radi-
ces, and mean motions, early Alfonsine astronomers borrowed material from their
Arabic predecessors and adjusted them to their needs in much the same way as the
author(s) of the Toledan Tables treated al-Battānī’s material, in particular for the
planetary radices and apogees at the Hijra. Although new approaches and param-
eters are found in PAT, essentially there is continuity between the work done by
table-makers in al-Andalus and Latin Europe. Among the common sources for PAT
and zijes in al-Andalus and the Maghrib, it is likely that one of them was a lost
work by Azarquiel.
3. According to chapter 1 of the Castilian canons, the era of Alfonso began at noon, Sunday,
Dec. 31, 1251, and the first day of the era, Jan 1, 1252 (JDN 2178351), is considered to be the epoch
of these tables. However, the printed edition of PAT uses a different epoch, June 1, 1252, the date
of the beginning of Alfonso’s reign (Chabás and Goldstein 2003, 21 - 22, 144).
53
4. Other authors take this epoch to be a day earlier, May 16, 16 ad: see, e.g., Mercier 1977, p. 59.
54
2. Apogees
For the linear term in Alfonsine precession/trepidation, one has to refer to the
tables for the «radices of the apogees». The editio princeps lists separately the
radices for the apogees of the Sun and the planets for 10 eras, from the Flood to
Alfonso (Ratdolt 1483, c8r - d1r). In all cases we are told that the entries do not
include the motion of the 8th sphere, that is, they include the linear term for pre-
cession/trepidation but not the periodic term (trepidation). We also note that, fol-
lowing the standard tradition, the apogees of the Sun and Venus are considered to
be the same; they are given here to seconds, whereas all other apogees are dis-
played to thirds. However, this is a false precision because for all planets the
number of thirds is the same for each era, indicating that the entries were com-
puted following the same procedure. Moreover, the differences between eras are
the same for each planet. For example, between the Flood and the Incarnation,
Jan 1, 1 ad (JDN 1721424), the apogees of the four planets (Saturn, Jupiter, Mars,
and Mercury) have increased by 22;47,21,23º, whereas that of the Sun and Venus
has moved 22;47,21º. We may also note that even though the apogees for the
planets are given to thirds, for each of them the number of seconds and thirds at
the Flood and Alfonso’s is the same, indicating an unlikely coincidence or an ap-
55
proximation somewhere in the computation of these values. It follows that the un-
derlying rate for the motion of the apogees is about 0;0,0,4,20,41,17,19º/d, very
nearly the same as the value found in the editio princeps 0;0,0,4,20,41,17,12º/d,
for the mean motion of the eighth sphere, that is the linear term in Alfonsine pre-
cession. This unique value is equivalent to 1 revolution in about 49000 years and
differs from any analogous value in previous zijes. Indeed, the identification of
these two quantities is clearly stated in the heading of another table, entitled Ta-
bula medii motus augium et stellarum fixarum (Ratdolt 1483, d4v). The coinci-
dence of these two values can be considered a characteristic feature of the Alfon-
sine Tables that were reworked in Paris in the early fourteenth century.
By identifying one of the terms of precession/trepidation with the motion of the
apogees, the fact that these apogees are sidereally fixed is somewhat obscured. For
background, note that in the Almagest the planetary apogees are sidereally fixed
and the solar apogee is tropically fixed (Neugebauer 1975, pp. 58 and 150). But in
the above-mentioned treatise, On the solar year, it is argued that the solar apogee is
also sidereally fixed (Morelon 1987, p. 66; Neugebauer 1962b, p. 289, § 146).
In any case, the new parameter 0;0,0,4,20,41,17,12º/d (or about 0;0,26,27°/y)
for both the motion of the apogees and linear precession, corresponding to a pe-
riod of 49000 years, in turn depends very strongly on a calendar year of 365;15d,
as explained below.
The round number 49000 was not based on any observation; rather, it was
derived from an approximation of the length of the tropical year. The initial value
taken for the tropical year was 365 ¼ - 1/136 days, which is ascribed by Ibn Ezra
(d. 1167) to several astronomers including Azarquiel (d. 1100). The data given in
this text is that the tropical year is 365 days plus an excess of revolution of 87;21°,
where a day is taken to be 360°: Millás 1947, p. 83:14. Now, 87;21°/360° =
0;14,33,30d. The difference between a Julian year of 365;15d and a tropical year
of 365;14,33,30d is 0;0,26,30d, and 0;0,26,30 = 1/135;51 ≈ 1/136. Hence, the
implied length of the tropical year is 365 ¼ –1/136 days. In some manuscripts of
Ibn Ezra’s text this deficit is given as 1/106, but in one manuscript it is given cor-
rectly as 1/136: Millás 1947, p. 95:13 - 14; cf. Toomer 1969, pp. 318 - 319. At a rate
of 0;0,26,30º/y, the time needed to progress 360º is 360/0;0,26,30≈360 · 136 =
48960y, which was rounded to 49000y. The deficit from a Julian year of 365;15d
is obtained by dividing the time needed to progress 1º (360/49000) by the daily
velocity of the Sun (360 / T), where T is the length of the tropical year. Thus,
56
T = 365;15 – T/49000.
T = 365;15 · (49000/49001).
57
For astronomical purposes the epoch of the Hijra is noon on Wednesday, July
14, 622 (JDN 1948438), where the civil epoch of the Hijra is 6 hours later — at
sunset, considered the beginning of Thursday: Neugebauer 1962a, pp. 9 - 11, and
Sachau 1879, p. 34. These values may be compared with those for the Hijra found
in previous zijes and computed for different meridians. See Table 2, where we
have also included al-Battānī’s apogees for 880 ad.
Table 2: Apogees for the Hijra, as well as those in al-Battānī’s zij for 880 ad
a. Note that for al-Khwārizmī the positions of the apogees of the Sun and Venus differ (which is
not the case in the other zijes): Neugebauer 1962a, p. 99, and Pedersen 2002, p. 1228.
b. The values given for al-Battānī are tropical and correspond to year 1191 of the Seleucid Era
(879/880 ad): Nallino 1903 - 1907,1: 239 - 241, and 2: 108, 114, 120, 126, and 132.
* For discussion of these problematic values, see below.
In Table 2 for Mars the text of al-Battānī has 126;58º (Nallino 1903 - 1907, 2:
120). But the differences displayed in Table 3 suggest that the value for the apo-
gee of Mars attributed to al-Battānī by subsequent astronomers was 126;18º, for
they yield differences similar to those for the other planets.
Likewise, the value 164;30° for the apogee of Jupiter seems to be a «faulty»
value used by the author(s) of the Toledan Tables, rather than 160;30° which
would conform with the values in other zijes: see Table 3. Moreover, as shown
in Table 2, it is clear that the apogees for each planet — including Jupiter (as cor-
rected to 160;30º) — in the Toledan Tables, Ibn al-Kammād, and Ibn al-Bannāʼ
are quite close to one another and probably have a common source, suggesting
that they all reflect a tradition going back to Azarquiel: cf. Samsó and Millás
1998, p. 286.
The best results in comparing the values of the apogees in PAT are obtained with
those in al-Battānī’s zij and the Toledan Tables (see Table 3, where the two prob-
58
lematic entries for Jupiter in the Toledan Tables and for Mars in al-Battānī’s zij
are included).
Table 3: Comparison of the apogees for the Hijra in the Parisian Alfonsine Tables
and the Toledan Tables with those in al-Battānī’s zij for 880 ad
We can also offer an explanation for the difference of about 4;23° between al-
Battānī’s values for the apogees and those in the Toledan Tables. According to
many sources, the sidereal coordinates were equal to the tropical coordinates about
40 years before the Hijra, that is, about 582: Samsó 1997, p. 108; cf. Goldstein
2011, pp. 78-79. At a rate of 1° in 66 years for precession (al-Battānī’s value: see
Nallino 1903-1907, 1:128), in 297 years (= 879-582) the apogees would have
moved 4;30°, which is close to the differences in the apogees that we found. How-
ever, 4;23° in about 297 years corresponds to about 1° in 68 years (more precisely,
297/4;23 = 67;45,52), which is within the range of values for precession in the Mid-
dle Ages. It should be noted that this epoch, about 582AD, is unrelated to the epoch
of the periodic term in PAT, which has been shown to be 16 ad.
In sum, despite the variation in the readings of some entries, al-Battānī seems
to be the source of the apogees for the Hijra in the Parisian Alfonsine Tables as
well as in the Toledan Tables.
3. Radices
The editio princeps of the Parisian Alfonsine Tables also lists the radices for the
mean motions of the 8th sphere and the celestial bodies for the same 10 eras used
59
for the apogees (Ratdolt 1483, c8r - d1r). The entries are given to fourths for the
two luminaries and to thirds for the rest. Table 4 displays the radices for the mean
motions at the Hijra in the editio princeps of PAT and in previous zijes, as noted
in Table 2.
* The sidereal longitudes were obtained by adding the corresponding radix for the center of the
Sun and the planets to their respective apogees: see Chabás and Goldstein 1994, p. 33.
** Ibn-Isḥāq has exactly the same value for the lunar anomaly as in the Toledan Tables.
*** The entries in the tables are 360 – n, where n refers to the longitude of the ascending node.
A comparison of the radices in the various sets yield interesting results. As can
be seen from Table 5, the radices for the 5 planets and the lunar node in the To-
ledan Tables are clearly related to those in the zij of al-Battānī, for they differ by
a constant of about 0;23°. This amount is mainly due to the difference between
sidereal and tropical coordinates. As mentioned above, the date when tropical and
sidereal coordinates agreed with one another was taken to be about 40 years be-
fore the Hijra and, with a rate of precession of 1° in 100 years (Ptolemy’s value),
in 40 years precession would amount to 0;24°. However, this correction for pre-
5. In the zij of al-Battānī, the radices are for the Hijra, although the apogees are for 880 ad (see
Table 2).
60
cession should only apply to planetary longitudes, not to the anomalies of Venus
and Mercury (despite the textual evidence displayed in Table 5).
al-Battānī TT Batt. – TT
(Raqqa) (Toledo)
Tropical Sidereal
Sun 113;58, 4 113;41,11 +0;16,53
Moon, λ 119;43,16 120;58,18 –1;15, 2
Moon, α 106;30,40 108; 8,39 –1;37,59
Moon, node 233;45,18 234; 9,55 –0;24,37
Saturn 116;15 115;51,15 +0;24
Jupiter 332; 3 331;39,37 +0;23
Mars 211;48 211;24,59 +0;23
Venus, α 45; 7 45;28,37 +0;22
Mercury, α 73;23 73;46,18 +0;23
61
4. Mean motions
The editio princeps of the Parisian Alfonsine Tables displays tables for the daily
mean motions of (1) access and recess of the 8th sphere (d4r); (2) apogees and the
fixed stars (d4v); (3) Sun, Venus, and Mercury (d5r); (4) Moon in longitude, anom-
62
aly, and argument of latitude (d5v-d6v); (5) the lunar elongation from the Sun (d7r);
(6) lunar node (d7v); (7) Saturn, Jupiter, and Mars in longitude (d8r - e1r); and (8)
Venus and Mercury in anomaly (d1v - d2r).
Of special interest is the value of the mean motion for the Sun, Venus, and Mer-
cury, 0;59,8,19,37,19,13,56º/d, already reported by John of Murs in his Expositio
intentionis regis Alfonsii as belonging to the «Tables of Alfonso» (Poulle 1981, p.
253). This value results from the length of the tropical year, 365;14,33,9,57,4,26...
d which, in turn, derived from a period of revolution of 49000 years (see section
2, above).6
A potential source has been suggested for the Alfonsine parameters of the plan-
etary mean motions. Chabás and Goldstein (2012, pp. 57 - 59) listed values for the
mean motions used by different pre-Alfonsine authors and concluded: «Compari-
son of Ibn al-Raqqām’s values for Venus and Mercury and those given in the
Parisian Alfonsine Tables show very close agreement (the difference is about
0;0,0,0,0,0,30º/d), another clear indication of the Andalusian character of the Pa-
risian Alfonsine Tables that has not previously been recognized»: see Chabás
and Goldstein 2012, pp. 57 - 59. Based on a comparison between the mean mo-
tions in PAT and those in the zij of Ibn Isḥāq, Samsó (2019, p. 363) concluded:
«The fact that most Alfonsine mean motions derive clearly from Ibn Isḥāq’s pa-
rameters shows the Toledan origin of the Parisian Alfonsine mean motion tables»,
and he later added that «Ibn Isḥāq’s mean motion parameters seem to derive from
those of the Toledan Tables» (Samsó 2020, ch. 7, section 7.6.2.2.2.2). Thus, the
Arabic origin of the Alfonsine mean motions of the planets seems to be firmly
established.
In order to compare systematically the Alfonsine (tropical) values with those
used by Andalusian and Maghribi astronomers (sidereal), we start with those quan-
tities not affected by trepidation, namely, lunar anomaly and the anomalies of Ve-
nus and Mercury. The results are shown in Tables 7 and 8.7
6. The value for the daily solar mean motion confirms that of the tropical year, 365;14,33,9,
57,4,26... d, and shows that the value for the tropical year reported by John of Murs in the Expo-
sitio (Poulle 1981, p. 251), 365;14,33,9,59,20,... d, has to be understood as an error in computing by
John himself or an error that was already in the source on which he depended. Cf. North 1996, pp.
458 - 460.
7. The following values are taken from Chabás and Goldstein 2012, except those for Ibn Isḥāq
and Ibn al-Bannāʼ, which were taken from Samsó and Millás 1998, p. 262.
63
PAT–
PAT Ibn al-Raqqām Ibn Isḥāq PAT– Isḥāq
Raqqām
Moon 13;3,53,57,30,21,4,13 13;3,53,56,17,51,25,27 0;0,0,1,12,30 13;3,53,56,17,52,4 0;0,0,1,12,29
Venus 0;36,59,27,23,59,31 0;36,59,27,23,58,51 0;0,0,0,0,0,40 0;36,59,27,23,59,25 0;0,0,0,0,0,6
Mercury 3;6,24,7,42,40,52 3;6,24,7,42,40,23,56 0;0,0,0,0,0,28 3;6,24,7,42,40,49 0;0,0,0,0,0,3
In the three cases, the values in Ibn al-Bannāʼ and Ibn Isḥāq are the same,
with differences of 3, 7, and 16 sixths. The differences amount to about 30
sixths between Ibn al-Raqqām and Ibn Isḥāq. Since these are values not explic-
itly stated in the zijes but are derived from entries sufficiently far apart in time
in the corresponding tables, we consider these values to be essentially the same in
all three zijes, despite the minor differences among them displayed here. All
three zijes compare well with the values in PAT for Venus and Mercury. In
contrast, those in the Toledan Tables do not seem to have served as the basis
for PAT.
For the lunar anomaly, all three Arabic zijes and the Toledan Tables give ap-
proximately the same value, which is different from what we find for the anom-
alies of the two inferior planets; thus, all differ from PAT by the same amount.
It turns out that this is the same value used in al-Battānī’s zij as well as in
Ptolemy’s Almagest (13;3,53,56,17,51,59°). The reason for the small discrep-
ancy in PAT of about 0;0,0,1,12,24° from this «long-standing consensus» is not
clear to us.
We now turn to the mean motions in longitude. The results of the comparisons
with PAT are shown in Tables 9 and 10.
64
As was the case for the lunar anomaly, the zijes of Ibn al-Bannāʼ, Ibn Isḥāq,
and Ibn al-Raqqām have almost the same values for the six parameters considered
here; clearly, they all belong to the same tradition. The values given in the Toledan
Tables are generally higher in the fourths, except for the Sun, where the value dif-
fers from those in the three zijes by less than 1 sixth, and for Saturn and Mars,
where the values differ from them by about 1 third.
The comparisons in Tables 9 and 10 show that grosso modo the three zijes (those
of Ibn al-Bannāʼ, Ibn Isḥāq, and Ibn al-Raqqām) share the same differences with
PAT, concentrated around, say, 0;0,0,8,20º (for the Sun and the Moon) and
0;0,0,7,40º (for the superior planets). The entries in theToledan Tables also share
these differences, but for Saturn and Mars. We conclude that the values for the
8. All values for Ibn al-Raqqām are slightly different in Samsó and Millás 1998 (reproduced
here) from those in Samsó 1997.
65
Sun, the Moon and Jupiter in PAT agree with the three zijes and the Toledan Ta-
bles (adjusting for the differences in coordinates), whereas the values for Saturn
and Mars agree with the three zijes (adjusting for the differences in coordinates),
but not the Toledan Tables.
The differences between the entries in PAT and in the Arabic zijes is due to
precession. Their value is compatible with 0;0,0,8,9,22º/d, which is the rate of
precession equivalent to 1º in 72 ½ years ascribed to Alfonso by John of Murs in
his Expositio.
A few conclusions can be drawn for the mean motions. First, the parameters
in the zijes of Ibn al-Bannāʼ, Ibn Isḥāq, and Ibn al-Raqqām are the same, despite
their differences in the sixths, and produce the same results. It should be recalled
that 1 sixth is about 2 · 10-11, a tiny amount that, even when multiplied by a huge
number such as the number of days in, say, 1000 years, it does not affect the result
in the seconds, which is the precision usually found in astronomical tables. Sec-
ond, comparisons of the mean motions in anomaly show that PAT borrowed the
values used in these, or similar, zijes for the superior planets, indicating that An-
dalusian/Maghribi astronomy is at the origin of the corresponding Alfonsine pa-
rameters. Third, as regards the quantities subject to precession, the daily mean
motions used by Alfonsine astronomers were taken from then same source, to
which a fixed quantity was added to convert sidereal coordinates into tropical
coordinates.
The two authors of this paper, in agreement with Julio Samsó, have repeatedly
characterized the Castilian Alfonsine Tables, and thus the Parisian Alfonsine Ta-
bles, as a zij, as Arabic astronomers had been compiling for centuries. And this so
not only because of the structure and contents of this set of tables, but also for the
approaches and parameters that have been identified in them. The birthplace of
these tables is undoubtedly Toledo — not any other place — as we have been argu-
ing for a long time.9
References
66
67
Samsó, J. 1997. «Andalusian Astronomy in 14th Century Fez: al-Zīj al-Muwāfiq of Ibn
ʽAzzūz al-Qusanṭīnī», Zeitschrift für Geschichte der Arabisch-Islamischen Wissenschaf-
ten 11: 73 - 110. Reprinted in idem 2007. Astronomy and Astrology in al-Andalus and
the Maghrib. (Aldershot), Essay IX.
— 2019. «Ibn Isḥāq and the Alfonsine Tables», Journal for the History of Astronomy 50:
360 - 365.
—2020. On Both Sides of the Strait of Gibraltar: Studies in the History of Medieval Astro-
nomy in the Iberian Peninsula and the Maghrib (Leiden and Boston).
— & E. Millás 1998. «The Computation of Planetary Longitudes in the Zīj of Ibn al-
Bannāʼ», Arabic Sciences and Philosophy 8:259-286. Reprinted in J. Samsó 2007. As-
tronomy and Astrology in al-Andalus and the Maghrib. (Aldershot), Essay VIII.
Santritter, J. L. (ed.) 1492. Tabule astronomice Alfonsi Regis (Venice).
Suter, H. 1914. Die astronomischen Tafeln des Muḥammed ibn Mūsā al-Khwārizmī (Co-
penhagen).
Toomer, G. J. 1968. «A Survey of the Toledan Tables», Osiris 15: 5 - 174.
— 1969. «The Solar Theory of az-Zarqāl: A History of Errors», Centaurus 14: 306 - 336.
68