The Origin of the Two-Tower Faade in Romanesque Architecture

Author(s): Herwin Schaefer

Reviewed work(s):
Source: The Art Bulletin, Vol. 27, No. 2 (Jun., 1945), pp. 85-108
Published by: College Art Association
Stable URL: http://www.jstor.org/stable/3046994 .
Accessed: 31/07/2012 04:56

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

College Art Association is collaborating with JSTOR to digitize, preserve and extend access to The Art
Bulletin.

http://www.jstor.org
 THE ORIGIN OF THE
FACADE IN ROMANESQUE
ARCHITECT URE1
HERWIN
O NE of the problems least resolved in the architec-
tural history of the early Middle Ages is that of
the origin and the source of the two-tower fagade.
We know that this scheme came into use some time during
the eleventh century in France, Germany, and elsewhere,
but beyond that we know very little indeed. A number of
theories as to the origin and source of this motif have been
brought forward, usually in the form of suggestions within
the framework of a larger subject, but the problem itself
has never been specifically dealt with in detail or at any
length. A discussion of this problem must deal with two
distinct questions: where in Western Europe was the motif
of the two-tower facade first used during the Romanesque
period, and what is the ultimate source of this motif?
Before we proceed with our investigation, it is necessary
to define what is meant by the two-tower facade. To do
this, we must refer to the thirteenth-century cathedrals of
France, which in more than one sense represent the end
point of a development and which are the norm and ideal
toward which earlier formulations had striven. At no time
during the preceding centuries do we find the plans of
churches so closely approximating one standard ideal as do
the cathedrals of Paris,2 Chartres,3 and Reims.4 During the
Romanesque period one is still able to discern the distinct
contributions of diverse origins of the style as well as the
individual contributions of local inspiration. It is for this
reason that it is impossible to define precisely the two-tower
facade in the Romanesque period, particularly in its earlier
phase, when the conception of a two-tower fagade seems to
i. The author is especially indebted for assistance and criticism
to Professors Wilhelm Koehler and Kenneth J. Conant of Harvard
University.
2. Marcel Aubert, Notre-Dame de Paris, Paris, i920, frontis-
piece and plan at end of volume.
3. Rene Merlet, La cathedrale de Chartres, Paris, n.d., frontis-
piece and p. 33.
4. Louis Demaison, La cathedrale de Reims, Paris, n.d., frontis-
piece and p. 41.
TWO-TOWER
SCHAEFER
have developed within a variety of forms of expression. It is
only when church architecture is created in a uniform style
that one can attempt to find a definition of the two-tower
fagade.
It appears that the end point of the development is
reached, and therewith what might be termed the ideal
two-tower fagade, when the two towers form the natural
termination of the nave aisles and do not extend, or extend
but little, beyond them. The space under the towers opens
into the aisles as well as into the central portion between
them. This central portion between the towers is no longer
set off as an entrance hall or narthex, but is a part of the
nave itself which extends clear through and thereby finds
expression in the organization of the facade. There is no
longer a separate facade structure, but the motif is now
integrated into the whole, and at the same time articulates
the parts of this whole.5
With slight deviations this is true of the classical thir-
teenth-century cathedrals of France. In the plan of Notre-
Dame at Paris only the thickness of the outside walls and
buttresses and of the inner piers of the towers indicate their
existence. They are not separate entities and do not in any
way stand apart from the building but form a part of the
whole within strict limits. At Chartres, because of their his-
tory, the towers extend somewhat beyond the aisles.6 At
Reims this deviation is reduced to a minimum. In Germany
the cathedral at Cologne7 represents the ideal on a large
scale; and St. Elisabeth at Marburg,8 the cathedral at Lim-
5. Cf. Hans Kunze, Das Fassadenproblem der franzosischen
Friih- itnd Hochgotik (Diss. Strasbourg), Leipzig, 1912.
6. Significantly the towers were originally erected in connection
with a narthex and incorporated into the body of the church when
the new church was built after the fire of 1194. Cf. Merlet, op.
cit., pp. 19-26.
7. Paul Clemen, Der Dom zu Koln, Diisseldorf, 1937, figs. 26,
86.
8. Richard Hamann, Die Elisabethkirche zu Marburg und ihre
kiinstlerische Nachfolge, Verlag des kunstgeschichtlichen Seminars
der UniversitatMarburga.d. Lahn, 1924.-29, I, figs. 2) 32.
86 THE ART BULLETIN

burg a.d. Lahn,9 and the church at Andernach10represent
it on a smaller scale.
Even though the conception of the two-tower fagade
undoubtedlyreaches further back, we have in the two ab-
bey churches at Caen the earliest preserved examples in
which the towers are brought into a closer connection with
the structureof the church itself."1In the plan of La Tri-
nit (10o62-83)12 the fagadestructurestill receivesan em-
phasisof its own, chiefly becausethe towers extend beyond
the aisles, but at St. Etienne (io64-87)"1 the towers no
longer do this and appearas the termination of the aisles.
Aubert says of the facade of St. Etienne: "sa composition
grandiose, simple et puissante,a inspire les architectesdes
eglises anglaises apres la conquete de io66, comme les
grands maitres des cathedralesde l'Ile de France, dans la
deuxieme moitie du xIIe siecle."14In Normandy, then, we
have the earliest preservedstructuresthat foreshadow the
ideal reached in the thirteenthcentury. Whether the Nor-
man churchesare also the first to use this motif is a question
which is dealt with in this article.

9. Georg Dehio, Geschichte der deutschen Kunst, I, Berlin, I 9 I 9, I. THE EARLIEST ROMANESQUE TWO-TOWER FAgADES
figs. I52a, 201.
10. Ibid., figs. I58, 200. To determine the first use of the two-tower facade, in
ix. The church of Notre Dame at Jumieges, built c. 1037-67, which the towers are used as an organic part of the church,
has a facade of a distinct character, with its central portion pro-
we must considera group of early monumentsin the Upper
jecting about three meters. It is an important step forward that
the churches at Caen present a straight front. G. Lanfry, "Fouilles
et d6couvertes a Jumieges," Bulletin Monumental, LXXXVII,I928, 13. Ibid., p. I2, pl. 32.
pp. 129-131. Marcel Aubert, L'Art francais a l'ipoque romane, 14. Ibid., p. I2. Cf. Marcel Anfray, L'Architecture normande,
Paris, 1929-32, I, p. II, pI. 22. Louis-MarieMichon and Roger Paris, I939, pp. 231 if. In England notably Durhamand Southwell
Martin Du Gard, L'Abbaye de Jumieges, Paris, 1935, pp. 38-46. show this scheme; Francis Bond, Gothic Architecture in England,
12. Aubert, L'Art franqais, p. i , pl. 24. London, 1905, p. 28, fig. i on p. I49; p. 520, fig. 2 on p. 152.

FIc. 34. Distribution of Early Towered Facade Types

 THE ORIGIN OF THE TWO-TOWER FACADE 87

Rhine region (Fig. 34). Since the Miinsterbaumeister (Fig. 2)21 certainly allows for a two-tower reconstruction,
Knauth had reconstructeda two-tower facade for the early if indeed it does not favor it. Reinhardt himself reconstructs
cathedral at Strasbourg"1(Fig. I) this was consideredthe one tower over the central portion of the faqade and two
first.16This view, however, was challenged recently by staircase turrets built onto the eastern corners of the central
Reinhardt,'7 who had previously also accepted the early tower. He states that the facade of the cathedral was identi-
appearanceof the two-tower fagade in the Upper Rhine cal with the present state of St. Thomas at Strasbourg (Fig.
region,18but who now wished to prove its non-existence 3).22 Furthermore, he claims that the abbey church at
there in order to show that this type of fagade was created Lautenbach, which is now restored with a two-tower fa-
first in Northern France and representedthere by the ex- gade, was originally also built on this plan and was a true
amples in Normandy which are indeed the earliest pre- replica of Strasbourg. For this he cites an old print of the
served.'9 abbey church before its restoration in the nineteenth cen-
The cathedralat Strasbourg,founded by BishopWerner tury (Fig. 4).23
in I 01 5, is clearly of the greatest importance,not only be- Could the cathedral of Strasbourg have had a single
cause of its early date, but also becauseof its size - it had square tower over the center of its facade? The first doubt
the same dimensions as the present cathedral- and be- arises over the oblong shape of the space it would have had
cause of its position as cathedral church of the cultural to cover. The foundations along the entire front of the
center of southwesternGermany and the adjacent regions. facade go to a depth of 5.80 m.,24 which would indicate
Since Reinhardt has brought forth what appearsto be im- that they were destined to carry a considerable load. By
portant evidence for a different reconstruction,it is neces- completing Reinhardt's plan and continuing the outline of
sary to deal at some length with his thesis and to base our the foundations for the outside walls of the side chambers,
own conclusions on Strasbourgand other monuments in which are hidden under the later foundations, in the same
this region on a discussionof his views. width as they were found for their eastern walls and the
As Gall has pointedout,20Reinhardt'splan of Strasbourg walls facing the entrance hall, Kunze worked out the
following: the north and south outside walls respectively of
15. Joseph Knauth, "Das Strassburger Miinster," Strassburger the two side chambers extended beyond the body of the
Miinsterblitter,VI, 1912. I was unableto obtainthispublication, church, and on the front they extended the same distance
but it is reliablyreportedby HansReinhardt,"LaCathedrale de
l'EvequeWernher,"Bulletin de la Societe des amis de la cathedrale beyond the central portion of the facade, i.e., the triple
de Strasbourg,Ser. 2, II, 1932, pp. 41, 47-48. In the following arcade. For Kunze this disposition alone characterizes these
thisarticleis referredto as Bulletin, 1932. Cf. Gustavv. Bezold, side chambers as the bases of towers. He interprets the
"Zur Geschichteder romanischen Baukunstin der Erzdi6zese
foundations as running in a straight line all along the fa-
Mainz," Marburger Jahrbuch fiir Kunstwissenschaft,viii-ix,
1936, p. 31. cade; the walls of the side chambers placed statically cor-
i6. Cf. Dehio, Geschichteder deutschenKunst, I, p. 76; idem, rectly over their axis; the arcades in the center, however, as
Das StrassburgerMunster, Munich, 1922, pp. 7-8; in his earlier
being pushed back and not standing on the axis of the
publication,togetherwith G. v. Bezold,Die kirchlicheBaukunst
des Abendlandes,I, p. 574, Dehio had maintained that the motif foundations. This could be practicable only if their load was
didnot originatein Germany, butcametherefromCluny,the ab- small and did not consist of the weight of a central tower.25
beysof Limburga.d.H.andHirsaubeingthe intermediaries; Paul
However, even if the piers in the central portion had been
Frankl, Die Friihmittelalterliclhe und romanischeBaukunst,Wild-
centered on the foundations, this arrangement would argue
park-Potsdam,1926, pp. 72-73; Rudolf Kautzsch, Romanische
Kirchenim Elsass,Freiburgi. B., 1927, p. 71; Hans Kunze, "Der against Reinhardt, because it would make the foundation
StandunseresWissensumdieBaugeschichte desStrassburger Miin- wall here
considerably narrower than that of the two side
sters," Elsass-Lothringisches Jahrbuch, xviii, 1939, pp. 68-75, 8o-
82; Mettler,"ZumUrsprungderdoppeltiirmigen Westfassade der compartments, which would then logically have borne the
mittelalterlichenBasilika,"Zeitschriftfur Geschichte der Architek- major weight of towers. The plan itself, therefore, suggests
tur, VI, 1913, pp. 145-146. a two-tower reconstruction of the fagade, rather than a
I7. Hans Reinhardt, Bulletin, 1932, pp. 39-64; idem, "Das
ersteMiinsterzu Schaffhausen unddie FragederDoppelturmfas- central tower (Fig. I ).
sade am Oberrhein,"Anzeigerfiir SchweizerischeAltertumskunde, St. Thomas at Strasbourg has a square tower, which is,
xxxvII, 1935, pp. 241-257 (in the following this article is re- moreover, Gothic in its forms and represents a restoration
ferred to as Anzeiger, 1935).
of the nineteenth century.26 The tower in the old print of
18. Hans Reinhardt,Die spatromanischeBauperiodedes Basler
Miinsters, Dissertation,Basel, 1926, p. 74; idem, "Die Ausgra- the abbey church of Lautenbach (Fig. 4) also is Gothic,
bungender erstenAnlage des KlostersAllerheiligenin Schaff-
hausen,"Jahrbuchfiir Kunstvjissenschaft,I928, pp. 38, 41. 2I. Reinhardt, Bulletin, 1932, pl. II.
19. HansReinhardtand EtienneFels, "Etudesur les eglises- 22. Ibid., fig. I i.
porchescarolingienneset leur survivancedans l'art roman,"Bulle- 23. Ibid., fig. 9.
tin Monumental, XCII, I933, pp. 331-365; XCVI, 1937, pp. 425- 24. Ibid., p. 43.
469. 25. Kunze, Elsass-Lothringisches Jahrbuch, xvIII, I939, pl. II,
20o. Ernst Gall, Zeitschrift fiir Kunstwissenschaft, I, 1932, p. fig. i, pp. 71-75.
298. 26. Jos. Walter, "L'Eglise Saint-Etienne de Strasbourg d'apres
88 THE ART
and the broken off pilasters on each side and their absence
in the center would more easily serve as proof for a two-
tower reconstruction than for a central tower.27 The use,
therefore, of these monuments for a reconstruction of the
fagade of the cathedral at Strasbourg is not above criticism.
In further support of his reconstruction of the cathedral
with one central tower over the facade Reinhardt cites a
Strasbourg coin of 1249 (Fig. 5)28 which shows a central
tower over an oblong building which opens in a triple arcade
below and which has two smaller towers behind the oblong
structure. These two appear to be round, and it is of great
importance that their eave line is above that of the central
tower. This coin picture is compared by Reinhardt to the
abbey church at Maursmiinster (Fig. 6)29 which does show
a plan similar to this, but differs in one important respect.
At Maursmiinster the central tower, which is square, is set
back, and the side towers are in the plane of the fagade.
Both the building on the coin and the abbey church of
Maursmiinster are then compared by Reinhardt to Stras-
bourg. Having asserted, however, that the cathedral had
only one tower with attached staircase turrets, as at St.
Thomas, Reinhardt must, if he also wishes to use the coin
as evidence, call the detachment of the side towers in the
building depicted on it "une licence du medailleur."30 In
considering this particular aspect of the placing of the tliree
towers to each other, we may compare neither the building
on the coin nor Maursmiinster to the cathedral at Stras-
bourg, because from the plan it is obvious that if indeed
three towers existed, they were certainly all in one plane. In
order to reconcile this fact with his use of the coin and
Maursmiinster as evidence, Reinhardt assumes such di-
minutive attached staircase turrets for the cathedral as ac-
tually do exist on St. Thomas (in our photograph one of
them can barely be discerned on the southeast corner of the
square tower). For the rest he assumes relatively low square
side chambers at either side. The discrepancies among the
coin of 1249, the church of Maursmiinster, and his recon-
struction of the cathedral make his use of the coin and
Maursmiinster as evidence unacceptable.31
The use of the coin of 1249 may be questioned on other
des dessins et textes inedits," Archives Alsaciennes, VI, I927, pp.
i-I4; Dehio, Handbuch der deutschen Kunstdenkmdler, Iv, Suid-
westdeutschland, Berlin, 9 11, pp. 402-403.
27. This has also been pointed out by Gall, Zeitschrift fiur
Kunstwissenschaft, 1932, p. 298; cf. also Dehio, Handbuch, iv, pp.
402-403 Kautzsch, Romaniscihe Kirchen im Elsass, pl. 44.
28. Reinhardt, Bulletin, 1932, fig. 10.
29. Reinhardt and Fels, Bulletin Monumental, 1937, facing
p. 428.
30. Reinhardt, Bulletin, 1932, p. 50.
3i. Gall, Zeitschrift fur Kunstwissenschaft, 1932, p. 298, has
tried to argue that perhaps the coin cutter saw the cathedral from
the eastern end with the tower over the crossing in the foreground
and the fagade towers further back. Therefore, he must, of course,
call the appearance of the open arcade at the eastern end a "kiinst-
lerische Freiheit."
BULLETIN
grounds as well. Its face bears a bust of Bishop Henry of
Stahleck, who was Bishop of Strasbourg from I238 to
I260, and this identifies it satisfactorily as a Strasbourg
coin. The reverse displays the building in question and the
date 1249. Of the building Reinhardt says: "une eglise ro-
mane qui serait celle de la cathedrale."32 Elsewhere he says:
"dass . . . diese [the fagade of the Romanesque cathedral]
. . . gemeint war, darilber diirfte kaum ein Zweifel
bestehen."33 Nevertheless, it may be questioned that this is a
replica of the facade of the Romanesque cathedral at Stras-
bourg. If we examine the early coins of Strasbourg, they
present a variety of architectural types on the reverse during
the period under discussion. At the outset it might be asked,
why one particular type is chosen from this variety and
claimed as a replica of the cathedral?
Coin types were used over long periods of time and there-
fore it is well to begin with the earliest showing architec-
tural representations. This is a coin of Louis the Pious (8 I 4-
840), showing a gate structure with a rectangular opening,
and a small turret at each side. I do not believe this can be
called an ecclesiastical building.34 Next there is a type used
by Otto I (936-973), with a building which Engel calls a
temple.35 It shows two rows of three arches each, one above
the other, and covered by a pointed roof topped by a fleur
de lis. The whole is surrounded by the legend ARGEN-
TINA CIVITAS.
During the reign of the Ottos, chiefly Otto I, II, and III
(936-I002), a number of bishops of Strasbourg, who had
been granted mint privileges by Louis the German in 873,
used the Carolingian temple type. This presents a gabled
temple-front of four columns, with a cross between the two
center ones.36 The type was universally used throughout
the Carolingian epoch in all parts of the Empire and per-
sisted in many cases until much later.37 It was originally
32. Reinhardt, Bulletin, 1932, p. 48.
33. Anzeiger, 1935, p. 25 .
34. Arthur Engel and Ernest Lehr, Numismatique de l'4Alsace,
Paris, i887, pp. 152-153, nos. 15, i6, 17, pl. xxX, figs. II, I2.
35. Ibid., p. I56, nos. 49-55, pl. xxv, figs. i, 2, pl. xxx, fig. 29.
36. Ibid., pp. I63-I64, nos. 146-i53, pl. xxxI, figs. 22-24.
37. Hermann Dannenberg, Die deutschen Miinzen der sachs-
ischen und frankischen Kaiserzeit, Berlin, 1876. It is impossible to
give all the numerous references to the type in this work. A glance
at the volume of plates will indicate how universal its use was. Cf.
also Heinrich Philipp Cappe, Die Miinzen der deutschen Kaiser
und Kdnige des Mittelalters, Dresden, 1848; Hans Gebhart, Die
deutschen Miinzen des Mittelalters und der Neuzeit, Berlin, I929,
p. Io; Albert Escher, Schweizerische Miinz- und Geldgeschichte,
Bern, i88x, figs. 21, 22, 30, 34, 35, 36, 38; E. Caron, Monnaies
feodales francaises, Paris, I882-84, pl. I, figs. 14-18; Adrien
Blanchet et A. Dieudonne, Manuel de numismatique francaise,
Paris, 1912-36, I, p. 351, fig. 247; Iv, fig. 79; Heinrich Philipp
Cappe, Beschreibung der Mainzer Miinzen des Mittelalters, Dres-
den, i856, pl. I, fig. 8, pl. VIl, fig. IOI; Walter Haevernick, Die
Miinzen von Koln, vom Beginn der Priigung bis I304 (Die Miin-
zen und Medaillen von Kln, I), Cologne, I935, pl. I, fig. 5, pl. viI,
figs. 222a, b, g, i; pl. viIi, figs. 278a, b, e; 279, 283, 284.
FIG. i. Strasbourg, Cathedral: Reconstruction of West Fa- FIG. 2. Strasbourg, Cathedral: Detail of Plan, 1015 (Reinhardt)
cade by the Author with the Measurements Given by Kunze
~,~
/!i? . :: - /X:f :::a....
: -?
fS:f
'
~i:.'~
?!~
!i~i

_~~!!~!i~~~~i?

FIc. 3. Strasbourg, St. Thomas FIG. 4. Lautenbach, Abbey Church: View after an Old Print

FiG. 5. Strasbourg Coin, 1249 FIG. 6. Maursmiinster, Abbey Church

 a b c d

FIG. 7. Cologne Coins: (a) Archbishop Anno, 1057-75; (b) Archbishop Philipp, 1181-90; (c) Archbishop Sigwin, 1079-89;
(d) Soest (Archbishop Henry, 1225-38)

b b c

d e
FIG. 8. Strasbourg City Seal and Others: (a) Strasbourg City Seal; (b) Seal of the St. Peter Type of Unknown Provenance; (c) Wetzlar City
Seal; (d) Cologne, Oldest City Seal of the St. Peter Type; (e) Hagenau City Seal; (f) Zabern City Seal

f g

e h
FIG. 9. Imperial Seals: (a) Henry III, oz28-56; (b) Henry IV, 1053-1o06; (c) Frederick I, 1152-90; (d) Otto IV, II98-1218;
(e) Frederick II, I198-I250; (f) Frederick II, 1198-I250; (g) Henry VII, 1220-42; (h) Charles IV, 1346-78
 THE ORIGIN OF THE
characterizedas a type by the legend XPICTIANA RE-
LIGIO which surroundedit. Bishop Erkenbald of Stras-
bourg (965-991) used the Carolingian type in various
forms, some of them with arched portals instead of col-
umns.38Bishop Widerold (991-999) used a wide variety
of forms, all derived from the Carolingian temple type.39
In this they are by no means unique, becauseone finds simi-
lar variationsthroughout all of Germany and France and
the adjacent countriesat this time.
Under BishopAlwich (999-oo1001) a new type appears.
It shows two towers connected by a flat structure. Both
towers and the central section have arched portals,and the
whole appearsto be behind a wall.40This building is very
similarto one on the following coins. These were the first
to be struckduring the time the Romanesquecathedralwas
actually under construction, during the reigns of Henry
II (I002-24), and Conrad II (1024-39), and they are
therefore of specialsignificance.They bear the legend AR-
GEN-TINA in cross form in the center and three sepa-
rate buildings, which Dannenberg calls churches, and
Engel calls temples.41The first of these buildingsstands at
the base of the cross formed by ARGEN-TINA and is
composedof a low flat center portion with two arched en-
trances, and a massive high tower at each side. (It is this
building which resemblesthat on Bishop Alwich's coin.42)
The other two are in the uppercorners formed by the cross
of the legend and present what looks like side views of
similar buildings. In each case there is an identical tower
to which a rectangular structure with arched openings is
attached at one side (left and right respectively). Could it
be possiblethat these are meant to representfront and side
views of one and the same building ratherthan three differ-
ent buildings? It is more likely that it is a play with forms.
Dannenberg remarksthat this type is not at all rare.43
During the twelfth and thirteenth centuries a type ap-
pearsin a numberof variations,to which our particularcoin
under discussion belongs. It shows a building usually of
three towers, of which the center one is the largest and
widest but not always the highest. The side towers are al-
most always placed in back of the main structure.44This
type was also used throughout the Empire, by emperors,
38. Engel-Lehr, op. cit., p. 157, nos. 64-65, pl. xxv, figs. 8, 9;
Dannenberg, op. cit., p. 355, pl. XLI, figs. 930-933.
39. Dannenberg, op. cit., pp. 355-357, pl. XLI, figs. 934-942.
40. Ibid., p. 357, pl. XLI, fig. 943.
41. Ibid., pp. 351-352, pl. XL, figs. 920, 922; Engel-Lehr, op.
cit., pp. I6I-I63, nos. 128-133, 143), 44, pl. xxxi, figs. II, I2,
3>I39, 2I.
42. Cf. also Dannenberg, op. cit., pl. xi, fig. 263, a coin of
Tuin, a city in Lower Lothringia; pl. xxvIII, figs. 65i, 656, 658,
coins of Magdeburg; pl. xxxIV, fig. 794, pl. xxxv, fig. 807, coins
of Mainz.
43- Ibid., p. 35I.
44. Engel-Lehr, op. cit., p. i65, nos. 156-158, I6I, pl. XXVI,
figs. 1-3.
TWO-TOWER FACADE 89
feudal lords, bishops,cities, and monasteries.Its use is, how-
ever, not so widespreadas that of the Carolingian temple
type because the Empire had been divided and this type is
restrictedto the easternhalf and does not occur in France.45
Within the limitsof the easternempireits use was neverthe-
less so widespread and it appears in so many variations,
which are clearly derived from one type, that the only
analogy is the earlier Carolingian temple type. For the
same reasons one is forced to the conclusion that this too,
like the earliertype, representeda symbol and was not ever
meant to representa specificbuilding.
It is clear that this type owes its popularityto its use by
the city of Cologne, and indeed, it seems probablethat the
type originatedand was developedthere. Because of its re-
liabilityand the excellence of its alloy the mint of Cologne
was the largest in Germany during the High Middle Ages.
Its greatestimportanceand widest activityfall between 936
and I288.46 During this period almost all German cities
based their coinage on that of Cologne. Some adopted the
type without change. Others substitutedtheir local sov-
ereign on the face of the coin, or even simplyadded a small
symbol to the picture of "COLONIA URBS." For ex-
ample, the city of Soest added a grain of barleyin the space
under one of the three arches under the three towers.47
During the Saxon and Franconian dynastiesthe following
cities were among those using the Cologne type: Brussels,
Antwerp, Liege, Dinant, Miinster, Maastricht, Remagen,
Andernach, Trier, Halberstadt,Fritzlar, Bremen, Pader-
born, Corvei, Osnabriick, Soest, Speier, Basel, Strasbourg,
etc.48
By far the great majority of the Cologne coins of this
period and also those of other cities based on the Cologne
coinage, are of the three-tower type, which is of particular
interest here because the Strasbourgcoin of 1249 also be-
longs to the same type. The motif appearsin many varia-
tions, but the three towers are almost always retained: a
larger central one and two smaller ones at the sides. The
smallertowers are usually placed in back of the main struc-
ture, as has been noted for the Strasbourgcoin.49
The motif first appearsunder ArchbishopHermann II
(10o36-56), and is characterizedas a symbolic representa-
45. The type does penetrateas far west as Verdun:Blanchetet
Dieudonne, op. cit., pp. 280-282, fig. I58. The neighboring bish-
opric of Toul uses a variety of architecturaltypes, among them a
two-tower facade: ibid., pp. 279-280, figs. 154-I56.
46. Haevernick, op. cit., pp. I-2.
47. Ibid., pp. 212-213, 241, pl. 31, fig. 995b.
48. Ibid., pp. i62 ff. maps I, 2; pls. 22 ff. ArthurEngel, Traite
de numismatiquedu moyen dge, II, Paris, I894, p. 527, figs. 1019,
1040, 1050, 1053, 1065, 1127, 1128, 1141, 142) 1155 I 75,
1210o. Dannenberg, op. cit., pp. 38-39.
49. Haevernick,op. cit., pl. 12, figs. 43ia, 432, 455b; pl. I3,
figs. 458a, d, 46oa, 461 pl. i4, figs. 48Ib, 482, 483, 485a, 492,
496a, 497a, 499, 500o pl. i6, figs. 55I, 553, 554, 555c; pl. 3I,
figs. 995b, 996, 998, 999.
90 THE ART
tion of the city by the designation COLONIA URBS
which surroundsit. Thus the two side towers are placed in
the background,becausethe original idea of the motif was
that of the surrounding wall with towers at intervals or
within the enclosing wall, which is still quite plainly visible
in a great many coins.50It is for this reasonthat Haevernick
in describingthese coins speaksof "walls."51He observes:
"Wenn man die ganze Bilderreiheiiberblickt,kann man
nicht daran zweifeln, dass hier niemals auch nur im ent-
ferntesten der Versuch gemacht worden ist, ein Bauwerk
der Wirklichkeitnachzubilden.Die Reihe dieser Gebaude-
darstellungen ist vielmehr langsam mit immer denselben
Elementen (Tor, Mauer, Turm, Zinnen, Giebel, Fahnen)
in streng symmetrischerForm weiter entwickelt, fast wie
ein Ornament. Jeder Realismusliegt dem Stempelschnei-
der fern" (Fig. 7).52
The Strasbourgcoin of 1249 unquestionablyalso be-
longs to this type. Hanauer observed of this coin that "il
etait absolumentconforme au poidsde Cologne."53There-
fore, as one instanceof a very widespreadtype representing
a symbol and not an architectural portrait, it loses any
validity one might attach to it as a help in the reconstruc-
tion of a lost building. To place it entirely in its right per-
spectiveit must also be borne in mind that Strasbourgused
a number of types, and variationswithin these types, and,
as has been indicated above, it is impossibleto justify the
choice of one particularcoin pictureas the true replica of a
building in the city in which it was used.54
50. Ibid., pl. 9, figs. 306a, c, 337a, c, 338b, d, 3395 pl. Io, figs.
34oa, 34IC, 342, 343, 344, 345b, 350oa, 35I, 352a.
51. Ibid., pp. 5, 74. If we consider the two illustrations which
most likely represent the then extant cathedral at Cologne, namely
the picture on an enamel plaque of the Heribert reliquary in Deutz,
and the miniature in the Hillinus Codex (Clemen, Der Dom zu
Koln, pp. 43-49, figs. I8, 20), we find discrepancies between the
two, which cannot be resolved. The picture on the enamel plaque
shows no transepts whereas the miniature shows an eastern as well
as a western transept. The enamel plaque shows a central tower
over the nave, the miniature shows two, which, however, seem to
come from behind the nave and may well have been put there by
the artist simply to fill in space in his picture. It is generally as-
sumed that the cathedral was a double apse structure with transepts
and towers at each end. The question of a central tower is unre-
solved. Even though the final type of the Cologne coin may evoke
a structure like the then existing cathedral, in view of the observ-
able development of the coin type it is safe to say that a real picture
was not intended. It is possible to find confirmation of certain fea-
tures of the old Cologne cathedral through the eleventh-century
cathedral at Bremen, which was avowedly a copy of that at
Cologne, and had two apses and two towers flanking the western
apse. Arno Konig, Die mittelalterliche Baugeschichte des Bremer
Domes, Dissertation, Bremen, 1934; Otto Lehmann-Brockhaus,
Schriftquellen zur Kunstgeschichte des x . und s2. Jahrlunderts
fiir Deutschland, Lothringen und Italien, Berlin, 1938, I, no. 230.
52. Haevernick, op. cit., pl. 9, figs. 333a, 338b; pl. ii, figs.
378a, 390oa pl. i6, figs. 554, 573; pl. 31, figs. 996, 999.
53. A. Hanauer, Etudes economiques sur l'Alsace ancienne et
moderne, I, Paris, 1876, p. 8. Cf. Engel-Lehr, op. cit., p. xxi.
54. The same may be said, for example, of Trier, Magdeburg,
BULLETIN
A second coin cited by Reinhardt for his reconstruction
of the cathedral facade also belongs to this group of the
Cologne type. However, even though Reinhardt calls it a
"StrassburgerPfennig mit der romanischen Miinsterfas-
sade,"55it cannot even be assignedto Strasbourg,but only
to Alsace as a whole, along with a number of others which
are also of the three-tower type.56 The particular coin
quoted by Reinhardt comes from the so-called Find of
Bergbieten, which consistedof six to seven thousand silver
pennies, most of which are unidentified.57If a bishopwere
shown on the face of the coin, as is the case on some in the
find, one might have some basisfor attributingit to Stras-
bourg (though there were other bishopsin Alsace as well),
but in this case, without inscription,it is impossibleto de-
termine its provenance.Muller saysof the coins in this find:
"[es] entstehen . . . durch ihre Anonymitat kaum v6llig
losbareSchwierigkeitenbeziiglich ihrer Zutheilung an eine
bestimmte Priigstatte, geschweige denn an einzelne Bi-
schofe oder Abte."58 Among the coins of this find are
architectural representationsof all kinds, buildings with
one, two, and three towers, and buildingswith no towers
at all.59
The city seal of Strasbourg,which is also cited by Rein-
hardt as substantiatingevidence for his thesis,60is essen-
tially of the same "type" as the coins discussedabove, a
walled city with towers (Fig. 8).61 Here also the same ele-
ments are used as a symbol, and it is found again that the
seal's use as such was widespread,either as the symbol of an
individual city, or of the power over urbs et orbis of the
emperors.In characterizingthe Strasbourgseal as a repre-
sentativeof a type, Walter has shown that it cannot be used
as evidence for architecturalhistory.62
A number of Rhenish cities used this same type as a
symbol on their seals.63The purely symbolic character of
Mainz,Speier,and Regensburg,all of which useda varietyof
coin typesfrom whichit wouldbe impossibleto choosethe one
someparticularbuildingin the respective
whichtrulyrepresented
city. Cf. Dannenberg, op. cit., pls. 20, 28, 34, 35, 36, 46-49. For
Mainz see also Cappe,Beschreibungder Mainzer Miinzen des Mit-
telalters, pls. II, III, VII.
55. Reinhardt, Anzeiger, I935, fig. 4; idem, Bulletin, 1932,
p. 49, note i.
56. Engel-Lehr, op. cit., pl. XLIII, figs. 14, I5, I8, I9, 20; pl.
XLV, figs. I I, 19.
57. L. Muller, "Der Fund von Bergbieten," Bulletin de la
Societe pour la conservation des monuments historiques d'Alsace,
Ser. 2, xvIII, I897, pp. 315 ff.
58. Ibid., p. 315.
59. Ibid., pp. 319, 323, 324, 326.
60. Reinhardt, Bulletin, 1932, p. 48; idem, Anzeiger, 1935, pp.
250-25 I.
6 . Jos. Walter, "Zur kunstgeschichtlichen Bewertung des
romanischen Stadtsiegels von Strassburg," Anzeiger fur elsdssische
Altertumskunde, III, 1918, figs. 5 0-55.
62. Ibid., pp. 952-957.
63. Wilhelm Ewald, Rheinische Siegel, Bonn, 1906-33, III, pl.
37, figs. 8-9, pl. 41, figs. 1-2, pl. 82, fig. I, pl. 96, fig. 1; v, pl. 4,
fig. 3.
 THE ORIGIN OF THE
such a seal is strikingly brought out in the example of the
seal of Ettenheim in Baden, which until I803 was under
the temporalrule of Strasbourg.Almost all the seals of this
city, from the earliest times to the present day, have the
picture of a transverse structure with a large portal and
three towers, but these elements appear in all the styles
from the earliest Romanesque through Gothic, Renais-
sance, Baroque, to modern forms. If we accept the thesis
that such seals are portraitsof important buildings in the
respective cities, it would be necessary to assume that the
relevant building was entirely rebuilt each time the style
of architecturechanged.64
Imperialseals presenta similarpicture (Fig. 9).65 In the
time of the first dynastiesa number of different types were
used, some of them closely approximatingthe Cologne type.
Later the elements used are diminishedand stylized until a
purely symbolic picture of three towers and large portal
with the inscriptionAUREA ROMA crystallized,and this
was then used again and again.66It is significant that this
form is very similarto that on the COLONIA URBS coins
and those modeled after them.
From this investigationof coins and seals several facts of
importance for this discussion may be noted. Strasbourg
used coins with a wide variety of architecturalrepresenta-
tions. Some of these designswere used by many other cities
as well and did not have Strasbourgas their center of origin.
In the instance of those coins chosen by Reinhardt to sub-
stantiatehis thesis,it was possibleto show that they belonged
to a type particularlywidely spread, which, moreover, did
not have Strasbourg,but Cologne as the center of origin.
The same observationswere made concerning the city seal
of Strasbourg.67The use of this material as evidence for a
particularreconstructionof the Romanesque cathedral at
Strasbourgmust thereforebe discounted.
It has been necessaryto deal with Reinhardt'sarguments
64. BadischeHistorischeKommission(ed.), Siegel der badi-
schen Stddte, Heidelberg, I899-1909, III, pp. 53-54, pls. XV-
xviii. For similar examples,cf., Emil Schulthess,"Die Staidte-und
Landes-Siegel der Schweiz," Mittheilungen der Antiquarischen
Gesellschaft in Ziiurich,Ix, 853, pp. 86-90, pl. XIII; C. H. Baer,
Die Kunstdenkmiilerdes KantonsBasel-Stadt,I, Basel, 1932, p. 13,
pl. I, fig. i; H. Lloyd Parry, TheExeter Civic Seals, Exeter, 909,
p. 3, figs. i, 3, 4, 8 i G. BaldwinBrown, TheArtsin Early England,
II, London, 1925, pp. 269-270, fig. I12.
65. Otto Posse, Die Siegel der deutschenKaiser und Konige,
Dresden, 1909-13, I, pl. 14, fig. 4, pl. i6, fig. 6, pl. 22, fig. 4, pl.
25, fig. 6, pI. 28, fig. 5, pl. 30, fig. 5, pl. 31, fig. 6; II, pl. 3, fig. 7.
66. Ibid., I, pI. 28, fig. 3 II, pl. i6, fig. 2, pl. 8, fig. 3, pl. 24,
fig. 2, pl. 26, fig. 3. Cf. also ibid., I, pl. 20, fig. 6, pl. 23, fig. 6,
pl. 25, fig. 3, pl. 4I, fig. 7, pl. 47, fig. 4, pl. 50, fig. 8. Carl Heffner,
Die deutschenKaiser- und Konigs-Siegel,Wuirzburg,1875, pl. v,
figs. 35, 49, 62, pl. vi, fig. 50, pl. x, fig. 87, pl. xiv, fig. 103, pl.
xvII, fig. ii6. Wilhelm Ewald, Siegelkunde, Miinchen, 19I4, p.
200, pl. xxiv, figs. 8, 12, pl. xxv, fig. 3. Wilhelm Erben, Rombilder
auf kaiserlichen und pdapstlichen Siegeln des Mittelalters, Graz,
Vienna, Leipzig, 193 1.
67. Note, for example, the details of the Strasbourg seal in
Fig. 8 and those of various imperial seals in Fig. 9.
TWO-TOWER FACADE 91
in some detail, because at first glance the sum of his evi-
dence seemed convincing. However, in view of the posi-
tive evidence of the Strasbourgplan for a two-tower recon-
structionand becauseof the negative nature of Reinhardt's
substantiatingevidence, I believe that the old reconstruc-
tion of Strasbourgwith a two-tower facade must stand.
Reinhardt'smethod may also be questionedwhen he tries
to prove that the variousother churchesin the Upper Rhine
region, which had previously been assumed to have had
two-tower facades, did not have them because the motif
was unknown in this region.68This is preciselywhat must
be shown. The non-existenceof a motif is not proven by an
assertionthat it was unknown. That is merely a restatement
of the problem.
At Basel a cathedralwas consecratedin the presence of
the Emperor Henry II in ioi9.69 Of this church only the
lower stories of the northwest tower have been preserved;
the church itself was destroyedby fire in I I85.70 It is pos-
sible, however, to make certain deductions from the re-
maining tower, the so-called Georgsturm.71Over a high
ground story without any decoration,a second story shows
on its front side four delicate blind arcadesand above them
a rectangular recession with six small corbels. This same
decorationis repeated on the northern flank of the tower,
where, however, only the upper part is visible, the rest
being hiddenby the roof of a fourteenth-centurychapel. On
the back of the tower the rectangle does not appear, and
only three archesinsteadof four, and these three, moreover,
are moved to the side sufficientlyto allow for the abutment
of the wall of the nave. On its fourth side the tower is with-
out any decoration. A small door, which leads into space
today, evidently connected the tower at the second story
level with an entrancehall which stood between this tower
and its partneron the other side. It seems exceedingly diffi-
cult to deny such a reconstructionof the facade at Basel,72
68. Reinhardt,Anzeiger, 1935, p. 243.
69. Reinhardt,"Die Urkundenund Nachrichteniiber den bas-
ler Miinsterbaubis zum Jahre 1300," OberrheinischeKunst, iii,
I928, pp. i8-i 9. It is interestingto note that Poppo von Stablo
and Werner of Strasbourgwere presentat the consecration.The
influenceof suchpowerful personalities,and also their contactswith
the emperors,must be borne in mind, becausethey appearin con-
nection with a numberof crucial monuments.Cf. G. v. Bezold,
op. cit., pp. 70-76; Paul Ladewig, Poppo von Stablo und die
Klosterreform unter den ersten Saliern, Berlin, 1883, pp. 38-11 3;
W. Manchot, Kloster Limburga.d.H., Mannheim, 1892, pp. 35-
42; W. Meyer-Schwartau,Der Dom zu Speier und verwandte
Bauten,Berlin, 1893, pp. 3-14; E. C. Schemrer,"Bischof WernerI.
von Strassburg," Elsass-Lothringisches Jahrbuch, II, I923, pp. 26-
48.
70. Reinhardt,Die SpatromanischeBauperiodedes BaslerMiin-
sters, Dissertation,Basel, I926, p. 74.
7 1. Idem, Das Basler Miinster, Basel, 1939, pl. p. 2 0.
72. And yet Reinhardt says that after his researches at Stras-
bourg "il devient tres peu certain," Bulletin, 1932, p. 50, note 5;
however, in his latest publication on Basel he does not venture to
give any definite opinion on this question, but through his presenta-
92 THE ART
and so we may assumethat in I o1I 9 the cathedralat Basel, a
church of considerabledimensionsand important patron-
age, was completedwith a two-tower fagade.
Another buildingof this period, i.e., the first decades of
the eleventh century, is still very controversial,namely, the
cathedral of Mainz, built under ArchbishopWilligis and
consecratedin I0o9."7 Kunze believes that it had a two-
tower fagade (at its easternend),74 precededby an atrium.
Metz agrees with this reconstructionof the Willigis cathe-
dral,75and pointsout that in the rebuildingof the cathedral
after the fire of I009 (on the day of its consecration) the
squaretowers of this easternfacadewere not rebuiltand the
adjacent round towers enlarged with a high bell-tower in
the center; and that the lower stories of the square side
towers were finished off with pitched roofs,76 so that a simi-
lar transformationtook place as at Minden and Hildesheim.
In I919, before the work on the foundations, begun in
I910 and ended in I928, had progressed sufficiently to
allow for any judgment, Kautzsch and Neeb thought it
probablethat the Willigis structure had been substantially
like the present-daycathedralat its easternend.77L. Becker
and J. Sartoriuspublisheda history of the cathedral based
on the finds made duringthe years of restorationwork. It is
their opinionthat two towers existed at the east end of the
Willigis cathedral,with a high structureconnectingthem,78 notes various points of relationshipbetween Limburg and
as at Gandersheim (Fig. 3 ), in other words, a "Saxon"
fagade.However, this would be quiteunusualin this region.
Kautzsch in his review of the book states that either this
form or a rectangularcentral tower with lower side wings
is possible.79G. von Bezold has pointed to the similarity hang beider Bauten nicht gezweifelt werden kann."87
with Strasbourg:"Der StrassburgerDom, begonnen I 015,
steht Mainz nicht nur zeitlich, sondernauch in seinem Plan
und seinen Abmessungen sehr nahe, der Frontbau, Vor-
halle mit zwei quadratischenRaumen, hat die gleiche Tei-
lung wie Mainz."80 In that case the round towers at the
sides and the eastern apse must be assumedto be of a later
tion of the materialhe suggestsat leastthe probability of a two- is this possiblefrom other sources.89Manchot states that it
tower facade: Das Basler Miinster,Basel, 1939, p. 10.
73. Fora full andannotated bibliography seeEdgarLehmann,
Der friihe deutsche Kirchenbau, Berlin, 1938, pp. 126-129.
74. Hans Kunze, "Der Dom des Willigis in Mainz," Mainzer the springingpoint of the roof. He admitsthat the western
Zeitschrift,xx-xxI, 1925-26, pp. 39-40.
75. Peter Metz, Der Dom zu Mainz, Cologne, I927, p. iI.
76. Ibid., pp. I6-17.
77. Rudolf Kautzschand Ernst Neeb, Der Dom zu Mainz (Die
Kunstdenkmaler im Freistaat Hessen, pt. I), II, Darmstadt, 19 9,
pp. 27-34.
78. L. Beckerand J. Sartorius,Die Baugeschichteder Friihzeit
des Domeszu Mainz, Mainz, 1936. I was unableto obtainthis pub-
lication,but in his reviewof the book R. Kautzsch,"ZurBau-
geschichtedes MainzerDoms," Zeitschriftfur Kunstgeschichte,vi,
1937, pp. 200-217, gives a resumeand deals with the views of the
authors.Figure i on p. 200 showsthe reconstructionof the Willigis
cathedral by L. Becker and J. Sartorius, as taken from their book.
79. Kautzsch, Der Dom zu Mainz, pp. 209-210.
80. G. v. Bezold, Marburger Jahrbuch fiur Kunstwissenschaft,
1936, p. 29.
BULLETIN
date, which Kunze believes, since they are not connected
with the rectangular part of the front.81 At the present
state of our knowledge of Mainz it is impossibleto gain any
further clarity; both the written sources and the various
finds on the present structure give contradictoryand un-
clear reports.82
A monument where we are once more on firmer ground
is the abbeychurch of Limburg a.d.Hardt. It was founded
in 1025 and completed in I042.83 Being in an isolated
place it escaped destruction by fire and was preserved in
its original state until 1504, when it was in large part de-
stroyedin the course of a plunderingexpeditionby a hostile
neighboring count.84Since the funds were never raised to
rebuild the abbey church, the remnants which escapedde-
struction in I504 are still extant. The only later addition
that is still preservedis a little staircaseturret and part of
the southwest tower, which were built about I3o0.85 A
comparisonof its plan86with that of Strasbourgshows an
astonishing similarity, particularly of the western parts
(except that two round staircaseturrets are added at the
western corners at Limburg, as they are, for example, also
at Trier). At Limburg there is also a square chamber at
each side of an oblong entrance hall divided in the same
manner as at Strasbourgand later at Lautenbach.Manchot
Strasbourg,among them the masonrytechnique,a compari-
son which, he says, "ergiebt eine solch frappanteAhnlich-
keit nicht nur in der ausserenForm, sondern auch in den
absoluten Gr6ssenverhaltnissen,dass an einen Zusammen-
Kunze has shown that the dimensionsof Limburg are ex-
actly three-fourths of those of Strasbourg throughout.s8
Therefore, a two-tower facade in one place would throw
light on the other as well.
From the monographof Manchot, which is still the basic
work on Limburg, it is unfortunately impossibleto get an
altogetherclear idea of the presentstate of preservation,nor
was possibleto measure in situ the height of the transept,
nave, and choir to the upper edge of the nave walls, or to
two towers are reconstructedby him from the uppermost
point of the nave walls upward and that the little round
8 . Kunze, Mainzer Zeitschrift, I925-26, p. 40. Yet G. von
Bezold, op. cit., p. 28, has pointed out that farther up the round
towers do connect with the rectangular structure.
82. G. v. Bezold, op. cit., pp. 22-25.
83. Manchot, Kloster Limburg a.d.H., pp. 5-9.
84. Ibid., pp. 26-27.
85. Ibid.,fig. 23.
86. Ibid., pl. III.
87. Ibid., p. 49.
88. Kunze, Elsass-Lothringisches Jahrbuch, 1939, pp. 68-69.
89. G. v. Bezold, op. cit., pp. 57-6I; Meyer-Schwartau, op.
cit., pp. 5-6, 165-I 66.
 THE ORIGIN OF THE
staircase turrets are reconstructed from 1.20 m. above
ground upwards.90 However, the reconstruction of a two-
tower facade seems to have been a matter of course for
Manchot, for he admits of no doubt.
As was stated above, a Gothic turret remains with parts
of a Gothic square tower four stories high.9" Kunze has
shown that these were built onto the remnants of the old
square tower when the original round Romanesque turret
collapsed and tore part of the main square tower with it.92
The fissure is clearly visible and part of a round arch re-
mains in the second story masonry to prove the originality of
the tower. Kunze has argued that in the fourteenth-century
restoration of this tower the same measures were kept for
the various stories as had been used in the Romanesque
tower, and if these are multiplied by four-thirds the result
would be the measurements of the towers at Strasbourg
(Fig. i).93
G. von Bezold accepts Manchot's interpretation of a
two-tower facade without question,94 as does Meyer-
Schwartau, both independently and after the appearance of
Manchot's work.95 Meyer-Schwartau points out particu-
larly the oblong shape of the central section which would
speak against a tower over it.96 To judge from the available
material and the close relationship of the monument to
Strasbourg, it would seem safe to assume a two-tower fa-
gade for the abbey church of Limburg a.d.H. as well.97
At about the same time (1030-60), the cathedral at
Speyer was built as another of the great representative
monuments of the mediaeval emperors.98 Authorities agree
that the present day facade is very much the same as that
of the eleventh century (in form, not in substance).99 Even
though this has been called a two-tower fagade and been
compared to Strasbourg and Limburg,0?? it is essentially
different in that the front of the fagade presents a solid rec-
tangular mass, two stories high, with a central tower which
was perhaps also present in the eleventh-century building,
90. Manchot, op. cit., p. 65.
91. Ibid., fig. 23.
92. Kunze, Elsass-Lothringisches Jahrbuch, i939, pp. 69-71,
fig. 4.
93. Ibid., pp. 69-7 i, fig. 5.
94. G. v. Bezold, op. cit., pp. 6o-6i.
95. Meyer-Schwartau, op. cit., pp. 5-6, I65-i66.
96. Ibid., p. 166.
97. The coins of Limburg are also of the Cologne type and are
here of as little value for the interpretation of the abbey church as
at Strasbourg. Cf. Paul Joseph, "Die Miinzen des graflichen und
fiirstlichen Hauses Leiningen," Numismatische Zeitschrift, XVI,
1884, pp. 109-216, figs. a and b; pl. I, fig. 2.
98. Meyer-Schwartau, op. cit., pp. 36-40; Dehio, Handbuch,
IV, p. 373; Bernh. Hermann R6ttger, Stadt und Bezirksamt Speyer
(Die Kunstdenkmndler von Bayern, III), VI, Munich, 1934, pp.
62-70.
99. R6ttger, op. cit., pp. 53, 204-205, fig. 20j Meyer-Schwar-
tau, op. cit., pp. 136-148, pls. II-v, VII, Ix,XI, XIv, xxvIII, xxIx;
Dehio, Handbuch, iv, pp. 377-378.
IOO. Rottger, op. cit., pp. 204-205.
TWO-TOWER FAgADE 93
and, most important,the two towers are placed behind the
actual faqade,so that they do not characterizeit."?0
The Romanesque church of the abbey of Einsiedeln,
built between 1031 and IO039, appears to have been a
church with two western towers. It was entirely rebuilt in
the seventeenth and eighteenth centuries, and of the old
building we have only a plan of I633 and a number of
views, showing the entire monasterywith the old church.102
The picturesand the plan make clear that the Romanesque
church of Einsiedeln had two towers at its western end,
which separatedit from the pilgrimagechurch. Huggler has
shown that the variouscatastrophesthat befell the monas-
tery did not destroy the original church, and that it is the
original church which appearsin the pictorial evidence of
the sixteenth and seventeenth centuries.103
Another abbey church built in this critical time is Hers-
feld, which we have reason to believe had a two-tower
structureat its western end. At the present time the south-
west tower is intact, with its lower storiesdating from the
early eleventh century, and the upperstoriesat the earliest
from the end of the eleventh. Of the northwest tower only
the lower portionshave been preserved.Between these two
towers is a rectangular entrance hall on the ground floor
and above it a western apse.'04There cannot be much
doubt that the church of about the middle of the eleventh
century (begun IO38 after a fire; the nave was in use
107 I )105 had a two-tower structure at its western end,106
although changes did take place here during the twelfth
century,'?7and the original nature of the central portionis
not quite clear. Of greater importancethan this partly pre-
served building is the earlier Carolingian church of 831-
850 that stood in its place. Excavations have brought to
light that at its western end (slightly farther west than the
present facade) this Carolingianchurch stood on two mas-
sive foundationblockswith very strong connecting walls.108
Io . The third great imperial cathedral along the Rhine, that at
Worms, had a western apse from the time of its origin, so that it
does not represent a two-tower facade. Meyer-Schwartau, op. cit.,
pp. 26-28 von Bezold, op. cit., pp. 48-49.
102. Max Huggler, "Die romanische Kirche in Einsiedeln,"
Anzeiger fiir Sch/weizerischeAltertumskunde,new ser., xxxvi,
I934, pp. I8O-197. Cf. Rudolf Hengeller, "Einsiedeln im Bilde,"
Anzeiger fur SchweizerischeAltertumskunde,new ser., xxvii,
1926, pp. 237-250.
103. Huggler cites the monastery church of Muri as being mod-
eled after that at Einsiedeln, op. cit., pp. 187-188.
104. Lehmann, op. cit., pl. I8, fig. 39.
I o5. Meyer-Schwartau, Der Doms zu Speier und ver.wandte
Bauten, p. 7.
io6. Ibid., pp. 8-Io; v. Bezold, op. cit., pp. 40-42; Dehio,
Handbuch, I, pp. I84-185.
107. G. v. Bezold, op. cit., pp. 4I-42; Meyer-Schwartau, op.
cit., p. I 0.
io8. J. Vonderau, Die Ausgrabungen der Stiftskirchze zu Hers-
feld in den Jahren I921 und 1922, Fulda, 1925. This publication
was unavailable, but is reliably reported by v. Bezold, op. cit.,
pp. 17-21, figs. 7, 9.
94 THE ART
In spiteof contentionsthat this was a two-tower facade,?09
it must be conceded that all that can be said is that a tower
structure of some kind stood at the western end of this
Carolingianchurch."?0
Both in Wiirzburg and in Koblenz are importantrem-
nants of two-tower facadeswhich can be attributedto this
period. At Wiirzburg a two-tower facade precedes the
churchitself, which is much wider and of a somewhat later
date."' The narrower dispositionof the tower structure
indicates that it originally belonged to a narrower nave.
Also, the wall of the present nave does not connect with
the towers; in fact, partof the presentnave coverswindow-
openings on the eastern side of the towers."12The literary
sourcesare unclear, both about the present church and the
earlier one, and it is impossibleto come to definite conclu-
sions from the monument itself. However, it seems safe to
attributethe west front with its two towers (except the two
top stories which were added in the twelfth century)"13to
the early part of the eleventh century, because the main
body of the church is certainly somewhat later, after
IO42.114
St. Kastor at Koblenz has recently been put forward as
the first exampleof the two-tower fagadeof about I ooo."15
Here we are in a situationsimilarto Wiirzburg becausethe
western front of this churchwith its two towers is also nar-
rower and earlierthan the main body of the church"1 (Fig.
io).117 It is Michel's opinion that a Carolingian "West-
bau," a simple rectangular frontal structure without tow-
ers, was transformedshortly before the middle of the elev-
enth century into a two-tower facade,"8 and that the main
part of the church was rebuilt in the twelfth century while
the facade towers were heightened."9 Busley believes that
the present church correspondsin its measures and in its
plan to the Carolingian church built under Bishop Hetti
109. Vonderau, op. cit., as quotedby v. Bezold, op. cit., p. 19;
Kunze, Elsass-Lothringisches Jahrbuch, 1939, p. 8I; Gall, Karo-
lingische und ottonische Kirchen, Burg, 1930, pp. I4-16;
Kautzsch, Zeitschrift fur Kunstwissenschaft, 1932, p. 53.
nio. Cf. Fuchs, Die Karolingischen Westwerke, p. 14; Rein-
hardt, Anzeiger, 1935, pp. 249-250; v. Bezold, op. cit., pp. 20, 45.
iii. Felix Mader, Stadt Wiirzburg (Die Kunstdenkmiler des
Konigreichs Bayern, xii), III, Munich, 1915, pp. 24, 44-45, figs.
8, 19, pl. I; v. Bezold, op. cit., pp. 43-48, fig. 23.
I I 2. Mader, op. cit., p. 44.
113. Ibid., p. 37.
114. Ibid., pp. 24-25; v. Bezold, op. cit., pp. 47-48, attributes
the towers to the tenth century, i.e., 940, as does also Kunze,
Mainzer Zeitschrift, 1925-26, pp. 40, 43-44.
i x5. Frankl, Die frihmittelalterliche und romanische Baukunst,
p. 58.
ii 6. Fritz Michel, Die kirchlichen Denkmdler der Stadt Ko-
blenz (Die Kunstdenkmaiilerder Rheinprovinz), xx, i, Dusseldorf,
1937, figs. 55, 72; J. Busley, Der Hettische Griindungsbau von St.
Castor-Koblenz, Dissertation,Bonn, 1921, p. 32.
137. Michel, op. cit., fig. 72.
He believes that this was done on the
II8. Ibid., pp. 120-123.
inspiration of Strasbourg and Limburg a.d.H.
I19. Ibid., pp. 116-125.
BULLETIN
(consecratedin 836) but that actually it is an almost com-
pletely new structureof the eleventh century in which only
fragmentsof the older churchwere used, among them parts
of the facade wall.120That at least this is the case can be
seen from the Carolingian capitals which were either re-
used or are in their original place, as Michel believes.1'2
The west front of St. Kastor at Koblenz is assuredin the
eleventh century through the nearby church of St. Florian
in the same city, which is undoubtedly a reflection of St.
Kastorbuilt in the very first years of the twelfth century.122
Busley says "Die L6sung des Westwerkes von St. Florin
ist ohne das Vorbildvon St. Kastor undenkbar.""123
In the precedingwe have dealt with monuments which
are either unquestionablyof the first half of the eleventh
century, such as Basel, Strasbourg,Mainz, Limburg, and
Hersfeld, or can be attributedwith reasonablecertainty to
this period, as the cathedralat Wiirzburg and the church
of St. Kastor at Koblenz. All these belong to one region,
namely, that of the Upper Rhine, or more broadly speak-
ing, southwestern Germany and the adjacent regions of
France and Switzerland.
It remains to call attention to several monuments in
Italy of this period,which show connectionswith Germany
and therefore perhapscast an important reflection on the
monumentsalreadycited. The abbeychurchof S. Salvatore
near Monte Amiata, which was consecratedin 1036, is a
building of a single nave with a two-tower fagade.24 It
appearsthat the fagade is original. This dispositionis most
unusual in Italy, and it is therefore in a certain measure
justifiable to seek connections with the North. It is note-
worthy that at S. Salvatore the idea of a two-tower fagade
was here added, so-to-speakinorganically,to a single-nave
church. Furthermore, that it does not representan accept-
able solution in Italy is attested by the fact that it found
practically no following there. One other example does
exist in Italy, the ruin of the church of S. Martino near
Farfa."25This building was begun under Bishop Berardo
( o89-99)26 and was never completed.Today only rem-
nants of its walls, its towers, and its crypt are preserved.
Whether this building was influenced by the nearby abbey
of Farfa is as yet undecided, mainly becauseit is uncertain
whether the famous abbey church had a two-tower faqade
or a so-called westwork.127 Only one tower is preserved
120. Busley, op. cit., pp. 32, 50-55.
I21. Michel, op. cit., pp. 106-116, figs. 73, 74, 78, 79.
I22. Ibid., pp. 20-22, fig. 18.
123. Busley, op. cit., p. 2 ; cf. Michel, op. cit., pp. 3, 5 I-52.
124. Hans Thiimmler, "Die Baukunst des ii. Jahrhunderts in
Italien," Romisches Jahrbuch fur Kunstgeschichte, III, 1939, pp.
i96-203, figs. 193, I96.
I 25. Ibid., pp. 205-207, fig. 207.
126. Ibid., p. 206.
127. Ibid., p. 206; Fuchs, op. cit., pp. I3-14.; Giuseppe Cro-
quison, "L'Abbazia imperiale di Farfa e i suoi problemi ar-
cheologici," L'Illustrazione Vaticana, IX, I938, pp. 864-867;
 THE ORIGIN OF THE
there, of which the three lowest stories are Carolingian,
the fourth of the eleventh century, and the uppermostthree
of the twelfth and thirteenth centuries. Croquisonbelieves
the Carolingian structurewas very similarto Michel's re-
constructionof Hetti's St. Kastorat Koblenz, and that parts
of this early structure were re-used or built upon as was
done at Koblenz.28 It is known that Farfa was closely con-
nected with Germany, through its origin and its patron-
age.129Whether a Frankishwestwork was introducedhere
in central Italy during Carolingian times and how the
western portionsof the church were changed in later times
both remain uncertain. The significance of a few isolated
two-tower facadesin northern and central Italy lies in their
possible connections with Germany. The region most
closely connected with Italy was the Upper Rhine valley,
as the great trading and traveling route from North to
South.
Even though soon after the turn of the middle of the
eleventh century two-tower facades were also well estab-
lished elsewhere, particularlyin Normandy, if may be of
interest to call attention to a few examplesduring this later
periodin the Upper Rhine region, in orderto emphasizethe
continuedoccurrenceof the motif and its establishmentas a
traditionin this region. With the turn of the middle of the
eleventh century a church buildingwas begun at the Mon-
astery of All Saints at Schaffhausen ( 1050, consecratedin
Io64),130 the fagadeof which has been under dispute.The
entire building was torn down at the end of the eleventh
century to make room for a new and larger church of the
Hirsau-Cluny school which will be discussedlater. Obser-
vations are therefore limited to the plan and the excavated
foundationwalls, as at Strasbourg.After initially assuming
two towers here,13'Reinhardtlater insistedon a westwork
interpretation,with a dominating central tower over the
facade.13 Hecht stood for a two-tower reconstruction,133
as did also Kautzsch134 and Huggler.135
The plan shows a rectangle at the western end which
extends beyond the body of the church as the transeptdoes.
At each sideis a foundationmassof five sq. m., and these are
idem, "I Problemi Archeologici Farfense," Rivista di archeologia
cristiana, xv, 1938, pp. 36-7I; Paul Markthaler, "Sulle recenti
scoperte nell' Abbazia imperiale di Farfa," Rivista di archeologia
cristiana, v, I928, pp. 36-88.
I28. Croquison, Rivista, 1938, p. 69; Michel, op. cit., fig. 80.
129. Croquison, Rivista, 1938, pp. 69-7I; J. v. Schlosser, Die
AIbendldndische Klosteranlage des friiheren Mittelalters, Vienna,
I889, pp. 4I-66.
130. Hecht, Der romanische Kirchenbau des Bodenseegebietes, I,
pp. 278-279, pls. 8 I- 9 2.
13 . Reinhardt, "Die Ausgrabung der ersten Anlage des Klos-
ters Allerheiligen in Schaffhausen," Jahrbuch fiir Kunstwissen-
schaft, 1928, p. 38.
I32. Reinhardt, Anzeiger,1935, pp. 242-243, fig. 2.
I33. Hecht, op. cit., I, p. 283.
134. Kautzsch, Zeitschrift fiir Kunstgeschichte, I, 1932, p. 53.
135. Huggler, op. cit., p. i89, note 25.
TWO-TOWER FACADE 95
connected by two foundation bands of I m. and I.IO m.
thickness,leaving a spacewithout foundationsin the center,
measuring 2.97 X 7.60 m. Within the nave, 4.35 m. to
the east of this structure,fragmentsof masonrywere found,
which seemed to indicate another transverse foundation
band I m. wide. This is taken by Reinhardtto be the eastern
foundation wall of a square central tower, with entrance
hall and tribune, the dominating feature of a westwork,
very much like that at Maursmiinster. This structure of
entrance hall, tribune, and tower would have taken up
more than one-third of the nave, leaving at the most per-
haps two bays for the rest. A reconstructionwith such a
disproportionatedivisionof the church calls for some doubt.
It is furthermoreimportantto note that the masonryfrag-
ments on which Reinhardt bases his reconstructionwere
actually found projecting into the nave only at one side.
Their extension across the nave and aisles is purely con-
jectural. Hecht considers them merely reenforcements,
such as were also found in other parts of the foundations,
and he therefore ignores them in his reconstructionof the
elevation.
The foundations at the western end thus speak for a
two-tower reconstruction.The solid foundation masseson
each side were certainlynot without a purpose,and it would
seem highly improbablethat this purposeshould have been
to carry much less than a foundation wall only I m. wide.
The architectwould have been illogical and uneconomical
if he had built as Reinhardt suggests. Huggler points out
that the complicated and highly developed reconstruction
by Reinhardt cannot be reconciled with the small dimen-
sions and primitiveaspectsof the church."36
Another feature of the plan of Schaffhausen is highly
interesting: the church was preceded by an open atrium
surrounded by a roofed arcade and with a cistern in its
center. The west entrance to this atrium was made into an
imposinggate structureby flankingit with a chapelon each
side and by placing towers at the outermostcorners.137
Between 1059 and 1071 the church of St. Aureliuswas
built at Hirsau with western towers incorporatedinto the
church (Fig. 26).138 The western section of the church is
still standing to a height above the arcades of the nave. A
little later the Cluniac reforms were introducedhere and a
second church, dedicatedto Sts. Peter and Paul, was built
according to the preceptsof the great Burgundian monas-
tery (Fig. 25).139 For liturgical reasons these precepts
called for a large narthex flankedby two towers at its west-
ern end. This arrangement is characteristicof Cluniac
churches and will be discussedat greater length in a later
136. Ibid., p. i89, note 25.
1 37. Hecht, op. cit., pp. 283 ff.
I 38. Mettler, Zeitschrift fiur Geschichte der Architektur, III,
1909/I0, fig. 3. Cf. idemn, Kloster Hirsau, I928, figs. I, II.
I39. Idem, Zeitschrift fiir Geschichte der Architektur, III,
1909/10, fig. i.
96 THE ART
part of this article. The earlier St. Aurelius at Hirsau was
small and plain. With its western towers flanking an en-
trance hall, that is part of the church itself, it brings out
very clearly on this one site the difference between the
Cluniac interpretation and that of the Upper Rhenish
school.
At Constance the cathedral was built during the same
years, a preliminaryconsecrationtaking place in 1069, and
the final consecrationin 1089. Here alsothe western facade
consisted of two towers with an entrance hall between
them.140Only the north tower and parts of the entrance
hall are of the originalbuilding.The south tower collapsed
twice and was finally rebuilt in Gothic times. An old door
leading from the north tower to the central section above
the first story gives evidence for a chapel to St. Michael
above the entrance hall. Traces along the south wall of
the north tower show that the side walls of the central sec-
tion continuedthe arcadewalls of the basilicaand presented
at the front a gable which very likely conformed with that
of the basilica.The outsidewalls of the centralsection were
in line with those of the towers, so that an impressive,uni-
fied facade was built here. The towers of the cathedralat
Constance were part of the church building itself, and the
scheme is that of the Upper Rhenish group with its close
approximationof the ideal type.14'
Finally there is an entire group of twelfth-century
churches with two-tower facades in Alsace, all close to
Strasbourg:the church at Lautenbach, which has already
been mentioned,'42St. Leodegar at Gebweiler,143St. Fides
at Schlettstadt,144 and the churches at Andlau"45and Nie-
dermiinster,146the last-named now in ruins. These are
mentioned here becausetheir close proximityto Strasbourg
makesit very likely that they are reflectionsof the cathedral
there. They are therefore late membersof the same group
which has been discussedhere.
To sum up: in the Upper Rhine region we have a series
of monuments in the first half of the eleventh century of
which it can be said with some reason that they had two-
tower fagades. We do not have, scientifically speaking, en-
tirely conclusive evidence; it is only possible to bring to-
gether pieces of evidence, both direct material evidence and
conjectural evidence, which allow us to say that it is prob-
40o. Hecht, op. cit., pp. 205-207, pls. 119-131; v. Bezold,
cit., pp. 66-68.
141. Two small monuments of the eleventh century deserve brief
mention: the abbey church at Stein am Rhein, Hecht, op. cit., pp.
260-26i, pls. 172-176, and the small basilica dedicated to St. Mi-
chael on the Heiligenberg near Heidelberg, Meyer-Schwartau, op.
cit., p. I 1, pl. xxxI.
142. Supra,pp. 87-88.
143. Kautzsch, Romanischze Kirchen im Elsass, p. 69, fig. 4o,
pl. 36.
144. Ibid., fig- 39, pls. 147, I49a.
145. Ibid., p. 74, fig. 4I, pl. I5.
146. Ibid., p. 74, plI. 91, 93.
BULLETIN
able or very likely that these monuments possessedtwo-
tower facades. As long as positiveproof to the contrary is
not brought forward, this Upper Rhenish group must
therefore be reckonedwith as the first to have adoptedthis
motif.
II. SUGGESTED SOURCES OF THE Two-TOWER
FACADE
In the second part of this paper I proposeto investigate
the varioussourceswhich have been suggested for the two-
tower facade motif. I have found that the error in these
suggestionslies in their exclusiveclaim, and that ratherthan
attributethis motif to any one source, it would be more cor-
rect to say that, given the general idea of such a scheme as a
traditionin Western Europe, its adoptionand its form de-
pended on local tradition and local inspirationin the vari-
ous parts of Europe.
I. GRAVE MONUMENTS
As a possible source for the two-tower facade, grave
monuments have been mentioned,147both Roman monu-
ments and the so-called "Lanternes des Morts" (Fig.
I i ),148 which also occur in the Charente. Lethaby refers
to Roman tombs "like small towers crowned by a pyramid,
with a sculpturedfinial at the apex" as the source for Ro-
manesque towers. These tombs themselves he thinks are
derivedfrom the Mausoleumat Halicarnassus.The largest
tomb of this type in the territoryof ancient Gaul is at Igel,
near Trier, on the Moselle.149Others were found at Neu-
magen, but these are not nearly so well preservedas that
at Igel."50A somewhat different type of Roman grave
monument is that of the Julii at St. Remy."15It is a tetra-
pylon on a square base, topped by a colonnaded pavilion
with a conical roof.
These tombs, however, are strictly speaking not perti-
nent to the problem under discussion,because Lethaby is
concerned with the origin of the Romanesque tower as
such, whereas we are here concerned with the idea of the
two-tower fagadewhich certainlyrepresentsa development
147. W.Weingirtner, System des christlichen Thurmbaues,
W. R. Lethaby, "Byzantine
i86o, pp. 25-31, 52-53;
Gottingen,
and Romanesque Arts," The Cambridge Medieval History, iii,
op. New York, 1936, chapter xxi, pp. 552-55 7.
I48. R. de Lasteyrie, L'Architecture religieuse en France a
l'epoque romane, Paris, 1929, figs. 729-730.
I49. H. Dragendorff and E. Kriiger, Das Grabmal von Igel,
Trier, I924.
I50o. Wilhelm von Massow, Die Grabmaler von Neumagen,
Berlin, 1932, pp. 65-78; cf. Felix Hettner, Illlstrierter Fiihrer
durch das Provinzialmuseum in Trier, Trier, 1903, pp. 2-23.
However, compare with these relatively isolated tower-like tombs
the many of the stele type: Emile Esperandieu, Recueil gineral des
bas-reliefs, statues et bustes de la Germanie romaine, Paris, 193I.
151. M. Rostovtzeff, A History of the Ancient 'orld, ii, Ox-
ford, 1933, pl. LXVII, fig. i.
 THE ORIGIN OF THE
long after towers had been in use. The suggestion of tombs
and grave monuments has been introduced here primarily
for the "Lanternes des Morts." In some cases these consist
of a number of applied columns on a common base, and
above them an architrave which carries a tiny columned
pavilion crowned with a pyramidalscaled roof.152Precisely
this same device appearson the fagade of Notre Dame la
Grande at Poitiers (Fig. I2),153 and also at St. Jouin-Les-
Marnes."54At Santiago de Compostela it was used at the
beginning of the twelfth century in the transept facades
which are similar to the main facade of Notre Dame la
Grande at Poitiers."'55In the compositionof the fagadesof
Poitiersand St. Jouin-Les-Marnes it seems like a motiffrom
which towers might develop, or one might even call it a
rudimentarytower. One may also recall the round turrets
of the eleventh-centurychurch of Fromista in Spain,which
resemblea type of "Lanternes des Morts."'56As we have
no certain dates for the "Lanternes des Morts,"'57it is im-
possibleto say which served as model for the other, if indeed
they did at all. Fromista is dated in the second half of the
eleventh century and both Notre Dame la Grande at Poi-
tiers and St. Jouin-Les-Marnes are dated in the twelfth
century,158 so that it is out of the question that they should
have influenced the development of the two-tower facade,
which was at that time, as has been said, firmly established.
One can recognize certain points of contact, but the diffi-
culties of establishing a direct line of descent are too great,
and where similarities do exist, they are, because of their
chronology, without significance in this discussion.
152. Note the similarity with the monument of the Julii.
153. Aubert, L'Art franfais, II, pl. i 5.
154. Julius Baum, Romanesque Architecture in France, London,
1928, pl. 52.
I55. Kenneth J. Conant, The Early Architectural History of
the Cathedral of Santiago de Compostela, Cambridge, 1926, pl. IV.
At Ste Croix in Bordeaux the little turret which appears at the top
of the facade and which reminds one also of this motif is modern.
Cf. infra, note I67.
I56. Bernard Bevan, History of Spanish Architecture, London,
1938, p. 58, pls. XIX, xx; Walter Muir Whitehill, Spaxish Ro-
manesque Architecture, Oxford, 1941, pp. 194-198, pl. 78.
157. Cf. A. de Caumont, "Sur des colonnes creuses ou fanaux
que l'on rencontre dans quelques cimetieres," Bulletin Monumental,
II, i837, pp. 428-433; Tailhand, "Sur les fanaux ou lanternes
existant dans quelques cimetieres," Bulletin Monumental, v, 1839,
pp. 433-435, pl. I; A. de la Villegille, "Sur des colonnes creuses ou
lanternes existant au milieu d'anciens cimetieres," Bulletin Monu-
mental, vi, I 840, pp. 7-1 4 F. Deshoulieres, "Lanternes des Morts,"
Bulletin Monumental, xc, 1931, pp. 297-298; Jean George,
Charente, Paris, 1933, pp. 53, 199; Joan Evans, The Romanesque
Architecture of the Order of Cluny, Cambridge, Eng., 1938, p.
148, pl. 259.
I58. A. de la Bouraliere, "Guide arch6ologique du congres de
Poitiers," Congres Archdologique de France, LXX, 1903, p. 25;
Andre Rhein, "Notre Dame la Grande, Poitiers," Congres Ar-
cheologique de France, LXXIX, 1912, p. 286; Jos. Berthele,
"L'Eglise de Saint Jouin-Les-Marnes," Bulletin Monumental, ser.
6, I, 1885, p. 269.
TWO-TOWER FACADE 97
2. CITY GATES
Another interpretation of the sources has given rise to
the theory that the two-tower facade was derived from the
Roman city gates that were preserved in large numbers
during the Middle Ages throughout France. This theory
was mentioned by Weingairtner159 and fully expounded by
Schultze.'60
Here one is hampered by the scarcity of preserved monu-
ments. An examination of the existing monuments indeed
reveals several features which might well have been taken
from Roman city gates. In addition to the two towers,
Schultze suggested the frequent triple division of the facade
with its arched doorways and the rows of arcaded windows
or blind arcading above as being derived from Roman city
gates.161 It is true that some Romanesque fagades in France
seem to reflect the arcading of the Roman city gates, but
not many of the latter have a triple arcade. In fact, in the
case of Autun, which has been cited repeatedly as an ex-
ample of such influence,162 the two Roman gates (Fig.
I3)163 both have two large arches in the center and a
smaller one at each side, whereas the cathedral has the usual
triple arcade. As to the towers of the cathedral at Autun,
they were not built until the end of the twelfth century,164
when the two-tower facade had been a tradition in Western
Europe for at least one hundred and fifty years.
Other Romanesque facades which seem to reflect the
Roman city gates took only their central portion without
the adjoining towers. These facades are often rectangular,
without a gable, so that in this respect the similarity is
heightened. However, they are facades in the true sense of
the word, i.e., the features which may have been taken
from a Roman gate are applied flat against them. Such ex-
amples center in and around the Charente, for example the
twelfth-century church at Echillais (Fig. I4),165 and the
church of St. Nicolas at Civray,'66 also of the twelfth cen-
tury. At Bordeaux the center of the fagade of the twelfth-
century church of Ste Croix is formed by such a rectangle
with a triple entrance and arcading above, and while these
features may well be derived from a Roman city gate, the
I59.Weingartner, op. cit., pp. 68-69, 72.
I60.Rudolf Schultze, "Das r6mische Stadttor in der kirchlichen
Baukunst des Mittelalters," Bonner Jahrbiicher, cxxiv, 1917, pp.
17-52; cf. also idem, "Die romischen Stadttore," Bonner Jahr-
budczer, cxviiI, I909, pp. 280-352; E. Weigand, Byzantinische
Zeitschrift, xxvii, i927, pp. 149-158; Alfons Maria Schneider,
"Die Kirche von Et-Taijibe," Oriens Christianus, ser. 3, vi, 193I,
p. 18.
6 . Schultze, Bonner Jahrbiicher, 1917, p. 26.
I62. Ibid., pp. 35-36; Joseph Dechelette, "Autun," Congres
Archeologique de France, LXXIV, I907, p. 135.
I63. Schultze, Bonner Jahrbiicher, 5909, pls. XIV, XX, XXI;
Dechelette, op. cit., pl. facing p. 124.
I64. Dechelette, op. cit., p. 32.
I65. Baum, op. cit., pl. 57.
I66. Aubert, L'Art francais, II, pp. 0--I 1, pl. 17.
98 THE ART
towers are added as separate entities to the central rec-
tangle.'67
To judge from exampleswhere Roman influenceis most
evident, it would appear that when a facade was inspired
by a Roman gate, preservedin a given locality, the result
was a flat rectangular fagade with a harmonious division
into arched doorways and groups of windows, and a har-
monious relationshipbetween these elements in the several
stories.
Trier and Aachen have been mentioned as derivations
from this same source,168but the facade of the Carolingian
structureat Aachen, in spite of the Byzantinesqueoctagon
behind it, is essentiallyNorthern, mediaeval, and unclassi-
cal; and at Trier the western portion, although added to a
Roman building, is entirely mediaeval in character, with
its apse, its two squareand two round towers, and decora-
tive details.'69It would be easier to see a derivation from
an arched Roman gate in the so-called "Torhalle" at
Lorsch, a Carolingian building in which all the elements
mentioned as classicalabove are clearly evident: the divi-
sion into three arched doorways, the pointedblind arcading
above, the beautiful proportionsand decorativedetails. But
this is an entrance hall which stood alone. It was not a fa-
sade structure,nor is there the slightestevidence for towers
at the sides.170The double arched outer gate of the abbey
at Cluny III is even more classical in its conception and
detail, but it neither had towers, nor was it the facade of
any building.17'
3. SYRIA AND ITS INFLUENCE ON THE WEST
Ever since Vogueipublishedhis work on the Early Chris-
tian churches in Syria,172the thesis has been brought for-
ward again and again that these influenced Romanesque
architecturein Western Europe, and were the source of
the two-tower fagade motif. Such great churches as those
at Der Termanin (Fig. I5),173 at Kalb Lauzeh,174and
167. Baum, op. cit., pl. 95. The tower at the left and some other
alterations of the faGadeare the result of a 19th-century restoration.
Cf. Jean-Auguste Brutails, Les vieilles iglises de la Gironde, Bor-
deaux, 1912, pp. 8-13.
i68. Schultze, Bonner Jahrbiicher, 1917, pp. 46-49.
169. Wolfgang Graf von Rothkirch, Deutsche Kunst, Berlin,
1934, pl. 74. Effmann has suggested the possibility that merely the
recessed doorway, or the doorway in a niche, such as at Aachen,
we derived from the gateways to a Roman castellum: Die karo-
lingisch-ottonischen Bauten zu Werden, I, Strasbourg, 1899, pp.
310-312.
170. Friedrich Behn, Die karolingische Klosterkirche von
Lorsch an der Bergstrasse, Berlin, 1934, pp. 70-90, pls. 25-35;
Rothkirch, op. cit., pl. 64.
171. Francois-Louis Bruel, Cluni, Macon, 9 10, pl. Iv.
172. Charles Jean Melchior de Voguei, Syrie Centrale, architec-
ture civile et religieuse du Ier au VIIe siecle, Paris, 1865-77, 2 vols.
173. Ibid., pls. I32, 135.
174. Ibid., pls. 123, 124.
BULLETIN
at Ruweha (Figs. i6, I7),75' unquestionablyforeshadow
and anticipateRomanesquedevelopmentsin the West, and
one can easily see how they could inspirethe theory of an
influence from Syriato the West.
Vogiiuehimself, in scholarly reserved words, points to
the undeniablesimilarities,in plan as well as in details, and
suggests the possibilityof influence.176 Courajod cham-
pions Vogiiue'sfinds and the theory of oriental influence
on Western Romanesque architecture.177Butler, who
expanded our knowledge of Syrian architecture enor-
mously,178speaksof "the influence the Syrian towered fa-
cade may have had upon the Romanesque facade with its
two flanking towers."'79The most vigorousprotagonistof
oriental influence on the West, and also specificallyof the
Syrian two-tower fa?ade on Romanesque architecture, is
Strzygowski. In addition to the factors usually claimed
to have been responsiblefor the transmissionof influence
from Syria to the West, namely, trade, monasticism,and
pilgrimages, he believes that the decisive factor in the dis-
seminationof Eastern ideas was exercisedby the migration
of the Goths from the Black Sea to Western Europe.
Among the ideas broughtto Europe he includes that of the
two-tower fasade.180Oelmann, in a study of the Syrian
hilani, claims them as the source for the motif of the two-
tower fagade in the Early Christianchurches of Syria and
also in the Romanesque architecture of the Occident.18'
Another representativeof the thesis, Beyer, selects the two-
tower fagade as one of the outstanding features that were
transferred.182He is perhapsright when he says that an
175. Howard Crosby Butler, Ancient Architecture in Syria
(Publications of the Princeton University Archaeological Expedi-
tion to Syria in I904-1905 and I909, II, Sec. A), Leyden, 1907-
20, pls. xv, xvi.
176. Ibid., pp. 17-22.
177. Louis Courajod, Lefons professees a l'ecole du Louvre,
Paris, 1899-1903, I, pp. 117-119. It is apparent in the lectures
of Courajod that the discoveries of Marquis de Vogue and the theo-
ries concerning the origins of Romanesque art which resulted from
them were not at all popular in the scholarly world of France and
that Vogiiueand his work were held virtually in quarantine even at
Courajod's time. Courajod complains that one is regarded a bad
Frenchman "quand on osait laisser soupdonner qu'on accordait a
l'Orient une action directe sur notre art occidental. Les seules ori-
gines latines 6taient tolerees" (I, p. 309).
178. Butler, op. cit.; idem, Architecture and Other Arts (Publi-
cations of an American Archaeological Expedition to Syria in
1899-i900, ii), New York, 1903; idem, Early Churches in Syria,
edited and completed by Earl Baldwin Smith, Princeton, Published
for the Department of Art and Archaeology at Princeton Univer-
sity, I 929.
179. Butler, Early Churches, p. 264.
80o. J. Strzygowski, Kleinasien, ein Neuland der Kunstge-
schichte, Leipzig, 1903, pp. 213-216, 230; idem, Origin of Chris-
tian Church Art, Oxford, 1923, pp. 214-215; idem, L'Ancien art
chritien de Syrie, Paris, 1936, p. 64.
181. Franz Oelmann, "Hilani und Liwanhaus," Bonner Jahr-
biicher, cxxvii, 1922, p. 206.
182. Hermann Wolfgang Beyer, Der syrische Kirchenbau, Ber-
lin, 1925, pp. 174-175. This work is based on Vogue and Butler.
FIG. o. Koblenz, St. Kastor FIG. ii. Lanternes des Morts: Left, Chateau
Larcher (Vienne); Right, Fenioux (Charente-
Inf rieure)

l
%

i<a
j

K I'voao_
,. .

FIG. 13. Autun, Roman City Gate

FIG. 2. Poitiers, Notre Dame la Grande FIG. 14. lchillais, Church

 FIc. 6. Ruweha, Bizzos Church: Actual State FIG. 17. Ruweha, Bizzos Church: Restored by
Butler
FIG. 5. Der Termanin

FIG. 20o. Bamberg, Staatliche Biblio-

thek: Ms. Patr. 6i (HJ iv I5), fol. 29v.
The Monastery of Cassiodorus at Squil-
FIG. 18. Auxerre, St. Germain: FIG. 19. St.-Denis, Abbey Church, 775: Restored by Crosby lace
Plan

FIG. 22. London, Victoria and Albert Museum: 149 to i49b-- I866. Detail of
Werden Casket

FIG.2 I. Rome, S. Sabina:Detail of Door FIG. 23. Gannat, St. Croix: Detail of Ivory Gospel Cover
 THE ORIGIN OF THE
understanding of Romanesque architecture presupposesa
knowledge of the Syrianchurchesand that this fact has not
yet penetratedinto the "kunsthistorischeAllgemeinbewusst-
sein."183Yet, in spite of many who would still question
an oriental or specificallySyrian influence on Romanesque
architecture,or would even ignore the possibility,it is being
taken more and more for granted. Quite recently Conant
ventured to write without qualificationthat the Syrian fa-
Qadescheme came to the West and merged there with
indigenoustower schemes.184On a broaderbasisKrauthei-
mer has made it clear that oriental influences permeated
Western Europe in the early Middle Ages, particularly
from the fourth to the eighth centuries.185
In connection with the two-tower problem, however,
interest is usually centered on Syria, and quite rightly so,
becausehere a relatively large group of buildingsused this
motif in a conspicuousmanner, and, furthermore, because
of its own pilgrimage places and its proximityto the Holy
Places of Palestine, it representeda focal point of contact
between East and West. The period which is to be con-
sidered falls between the edict of Milan in 313 and the
conquest of Syria by the Arabs in 638. A large number of
monuments from this period have been preservedin com-
paratively good state, because these territories have been
almost entirely uninhabitedsince the Arab conquest, and
there has therefore been no successivebuilding on the sites.
The architectural monuments of Syria fall into three
distinct groups: Southern Syria, or the Hauran, North-
eastern Syria, and Northern Syria proper. Though both
Butler and Beyer give lists of possibletwo-towered fagades
which include examples from all three regions,186the evi-
dence we have for Southern and Northeastern Syria is
too scant to allow for any conclusion as to the importance
or even reliabilityof the occurrenceof the two-tower motif
in these areas. In most casesthe ground plan is the sole evi-
dence. In Northern Syria, however, which is the architec-
tural region par excellence, the largest and finest monu-
183. Ibid., p. 3.
184. Kenneth J. Conant, A Brief Commentary on Early Me-
diaeval Church Architecture, Baltimore, 1942, p. 23.
185. Richard Krautheimer, "The Carolingian Revival of Early
Christian Architecture," THE ART BULLETIN, XXiv, I942, pp. 1-7;
cf. also Paul Clemen, Die romanische Monumentalmalerei in den
Rheinlanden, Diisseldorf, 1916, pp. 670-687. Clemen brings the
whole discussion into a good perspective. He rebukes Strzygowski
for his vehemence and for his claim of priority, in view of the
French scholars who had pointed to the Orient decades earlier. Jean
Hubert, L'Art pri-roman, Paris, 1938, pp. 83-84, 169-170; Louis
Brehier, L'Art en France, des invasions barbares a l'Ypoque romane,
Paris, n.d., pp. 32 ff.; idem, "Les Origines de la basilique chre-
tienne," Bulletin Monumental, LXXXVI, 1927, p. 248; 0. M. Dal-
ton, East Christian Art, Oxford, 1925, pp. 57-65 Camille Enlart,
Manuel d'archiologie francaise, I, Paris, 1910, pp. 107 if.; Els-
marie Kn6ogel, "Schriftquellen zur Kunstgeschichte der Merowin-
gerzeit," Bonner Jahrbiicher, CXL/CXLI, 1936, pp. 2 if.
i86. Butler, Early Churches, p. 190, note 348; Beyer, op. cit.,
pp. 152-I53.
TWO-TOWER FA9ADE 99
ments of all Syria have been in large measure preserved,
namely, Der Termanin, Kalb Lauzeh, and the Bizzos
Church at Ruweha, and it is of great significance for our
discussionthat these great basilicasall had two towers with-
out any question.187
Although not one of these is definitely dated, it may be
assumedthat they all belong to the sixth century or, at the
earliest,the end of the fifth century,188and represent,with
the great church at Kal'at Sim'an, the highest development
that Syrian architecture reached before it was cut off by
the Moslem conquest.Butler says of this group: "they stand
in a category of their own, which representsall that is best
in the Syrianart of their day, archetypescentering in them-
selves the highest conceptions and the greatest skill of a
generation.""89
The church at Der Termanln (Fig. 15 )190was in large
part preservedwhen Vogiiuefound it in the middle of the
last century. Its nave is preceded by a narthex flanked by
two towers which terminate the aisles and are accessible
only from them. The storiesof the towers are separatedby
moldings which run straight across the entire facade. The
window openingsincreasein number and size in the upper
stories. The entrance to the narthex from the west is
formed by a single wide arch which is flankedby an arched
window at each side. Above this entrance was presumably
an open terrace.
At Kalb Lauzeh19.the fagade of the church is also pre-
ceded by a narthex between two towers. A single tremen-
dous arch opens the entire width of the narthex, and over it
runs a cornice on which stands the balustradeof a terrace.
187. The church of St. Simeon at Kal'at Sim'an is, of course,
excepted from this statement, although it also belongs to the so-
called "Cathedral" group and is certainly the climax of Syrian ar-
chitecture. It consists of three basilicas built in the form of a Greek
cross around an open court in which stood the pillar of St. Simeon.
The southern basilica served as the main approach and had a
narthex of one broad middle arch and two narrow ones, the three
being joined to the main fagade by four arches. Given only the plan
of this narthex one might have assumed towers here, particularly
since the compound piers on which the arches rest consist of three
pilaster piers carrying the arches and an exterior pier which acts
as a buttress, so that one might have assumed considerable weight.
Actually the superstructure is preserved and shows three gables
with pitched roofs, over each entrance. Very likely towers were not
built here for the simple reason that they would have destroyed the
balance of the cruciform structure. Cf. Vogii6, op. cit., pp. 141-
152, pls. 139-150; Butler, Am. Exp., II, pp. 184-190, fig. 73 and
illustration on p. 184; idem, Early Churches, pp. 97-105, figs. I00,
101.
8 8. Vogiiue gives his dates on the plates; Butler groups his
monuments and places these in a group of sixth-century churches.
Butler, Am. Exp., II, p. i8o. See below the references for the indi-
vidual monuments. Cf. Beyer, op. cit., pp. 69-76.
i89. Butler, Am. Exp., II, pp. 22-23.
190. Vogue, op. cit., pls. 132, 135; cf. also pp. 138-140, pls.
13 0-13 6.
191. Vogiue, op. cit., pp. 135-138, pls. 122-129; Butler, Am.
Exp., II, pp. 221-225, fig. 89; idem, Early Churches, pp. 71-72,
fig. 193 M.
I00 THE ART
The flanking towers are wider than the aisles, but com-
municate with them as well as with the narthex. Their
uppertwo storiesare divided by a prolongationof the cor-
nice of the side aisle walls.
The Bizzos Church at Ruweha is in its western section
the least well preservedof the three great churches of the
North, but enough remains to make a reconstructionpos-
sible (Figs. i6, 7 ).192 To the west of the nave facade of
the church stand a narthex and two towers. Two T-form
piers supportthe great central arch of the narthex and the
openings of the ground floor chambers of the towers. A
molding separatesthe storiesof the towers and, after step-
ping up three courses,is carriedabove the central arch.
One detail in the three facades should be observed,
namely the treatment of the moldings which separatethe
stories:at Kalb Lauzeh the molding steps up one course in
the width of the towers, giving the actual tower story a
stunted look; at Der Termanln the moldings are carried
straight across; and at Ruweha the molding is lowered for
the towers, thus giving them a lighter and more tower-like
aspect.This indicatesthat the architectsat work here were
already conscious of one of the greatest problems of the
towered fagade, namely, to harmonize horizontal and ver-
tical lines.193
Leaving asideall furtherdetailsof decoration,and view-
ing simply the dispositionand nature of the main elements
of the western section - the west wall, the towers, and the
narthex - we find that they differ in a number of respects
from anything known in Europe. The towers project from
the actual west wall of the church; the narthex between
them appearsto have consistedof an entrance hall with a
loggia or terrace above; the whole seems to have been
dominatedby the pediment of the west wall of the church
further back and above.
If we extend our investigationthroughout the Eastern
territories, it becomes evident that the use of two towers
was very widely spread. The towers were usually com-
bined with a narthex which stood in front of the church as
a distinct entity. Thus this scheme resembles the Syrian
use of two towers.
To the north in Asia Minor at Binbirkilissenumerous
churches with two towers were built over a period from
perhapsas early as the fifth or sixth century to the eleventh
century.'94Of Church I Crowfoot says that "the walled-
off narthex between two high towers (reminding one of
the old Hittite 'Hilani'), . . . make this church fairly
192. Vogiiue, op. cit., p. io2, pls. 68, 69; Butler, Am. Exp., II,
pp. 225-228, fig. go; idem, Princeton Exp., II, Sec. B, pp. 142-
148, figs. I65-I67, pls. xv-xvIIIj idem, Early Churches, pp. 145-
148, figs. 155-156.
i93. This observation was first made by Beyer, op. cit., p. 73.
i94. Strzygowski, Kleinasien, pp. 9-27, 158 ff.; Sir William
Mitchell Ramsay and Gertrude Lowthian Bell, The Thousand and
One Churches, London, I909.
BULLETIN
characteristicof the site."'95 A similar church, also with
two towers, is reported by Strzygowski in Phrygia.196In
Lycia the ruin of the eighth-centurychurch at Dere Ahsy
has its esonarthex flanked on each side by staircase tow-
ers.197In their book on Meriamlik and Korykos, Herzfeld
and Guyer point out the similarityof the plans of the fifth-
and sixth-century churches there, particularlythe disposi-
tion of their western portions,with those in Syria, although
they do not believe that the churches at Meriamlik and
Korykos had towers.198At Ephesos the Austrian excava-
tions have brought to light the pre-Justinianbasilicaof St.
John which had a double narthex preceded by a deep por-
tico which was perhapsflanked by two towers.'99To the
East, in Armenia, Strzygowski reports a lone example of
the two-tower fagadeat Ereruk, which he dates in the fifth
century and which he believes to be an importedmotif be-
cause of its uniquenessin this region.200This same type is
also found in a monumental church at Aboba in Bulgaria
of the ninth or tenth century. Filow reconstructsa narthex
flankedby two towers before which a large atrium extends
with a portico and two towers at its western end.201An
even earlier example in the Balkans is the fifth century
BasilicaA at "ChristianThebes," which is precededby an
atrium flanked by two towers.202The atrium opens in a
multiple arched arcade toward the west and presents an
outline which is strikingly like S. Apollinare in Classe at
Ravenna, if the fagadeis imaginedcompletedthere with the
southwest tower.203Finally, the eighth-century church of
Sta Sophia at Thessalonika proves also to have had two
tower-like projectionsin the two corners of its faqade.204
At Constantinoplethe Byzantine churches are more cen-
195. Strzygowski, Kleinasien, p. 14, fig. 5.
196. Ibid., fig. 43.
197. Hans Rott, Kleinasiatische Denkmiler aus Pisidien, Pam-
phylien, Kappadokien und Lykien, Leipzig, I908, pp. 300-3I4,
fig. iIi.
198. E. Herzfeld and S. Guyer, Meriamlik und Korykos
(Monumenta Asiae Minoris Antiqua, II), Manchester, Eng., 1930,
pp. 29-30, 13, 13I. Krautheimer, op. cit., p. 6, compares the
cathedral at Korykos with S. Apollinare in Classe "with its narthex
flanked by west towers."
I99. Josef Keil, "xvi. Vorliufiger Bericht uiber die Ausgra-
bungen in Ephesos," Jahreshefte des Osterreichischen Archdologi-
schen Instituts in Wien, Beiblatt, xxviI, 1931, fig. 47.
200. J. Strzygowski, Die Baukunst der Armenier und Europa,
Vienna, 1918, I, pp. 397-399, plan on p. I53.
20I. Bogdan D. Filow, Early Bulgarian Art, Berne, 1919, pp.
8-9, fig. 5.
202. G. A. Soteriou, "Al XPLOTiaviKcalOY[3aL TY (OE)Ca-
X[ac," 'ApxaioXOYiKh 'Emypcplk, 1929, p. 19, pls. [3, y.
203. Oskar Wulff, Altchristliche und byzantinische Kunst, Ber-
lin-Neubabelsberg, 1914-I8, II, p. 398, fig. 345. Cf. Krautheimer's
comparison of S. Apollinare in Classe with Korykos, note 205.
204. Marinos Kalliga, Die Hagia Sophia von Thessalonike,
Dissertation, Wiirzburg, 1935, pp. I5 if. Cf. also the report of
later investigations by Kalliga in the American Journal of Ar-
chaeology, XLIV, 1940, pp. 246, 540-541; idem, "AvaactaCal ?
T') IV OOCraaaOovLK vaC Ty; 'Ay. 'o[tcac," HIpaKTIK&a'Apxato-
XOylKh ETalpca, 938, pls. I, 7.
 THE ORIGIN OF THE
tralized, and where towers are used, they are capped by
cupolas and also flank a narthex. Such is the case in the
Agios Theodoros or the so-called Theotokos of the twelfth
century,205and in the Chora church.206These examplesin
the Eastern Mediterranean are all of the Eastern basilica
type with the two towers used to flank a narthex which is
usually of the width of the church.
It is importantto bear in mind the wide diffusion of the
use of two towers throughout the East, when we consider
the Orient as the possiblesource of the motif in the West.
It must, however, be noted again, that in the East the two
towers are invariablyused with a narthex which standsas a
separate one-story flat-roofed hall before the nave of the
church.207It is true that we have in Western Europe a
numberof churcheswith a narthex or entrancehall flanked
by two towers, but the narthex is always either in itself
built up into a tower, or the gable of the narthex corre-
spondswith that of the nave,208so that the towers are inte-
grated into the actual facade of the church. This is not the
case in the East, and particularlynot in Syria, where the
towers project from the actual facade and have between
them a flat-roofed narthex. The composition finds its
closest analogies in the ancient Syrian hilani which were
also open flat-roofedentrance porchesflankedby towers.209
The positionof the Syriantower scheme is therefore one
that, along with other features,links the Christianarchitec-
ture of that land to its earliest native architecture,but af-
fords for the architectureof Western Europe at the most
a suggestion, which, if taken up from there, was essentially
altered into a new conception. We can only say that the
idea of two towers at the western end of a church may have
been transferred. A two-tower facade, in the strict sense
of the word, did not exist in Syria, because the towers pro-
jected from the facade, whereas in Europe the towers are
integrated into the facade as soon as they appear. Yet the
205. Wulff, op. cit., II, pp. 483-484, fig. 412. A similar arrange-
ment, but with staircase towers instead of cupolas, existed in the
Theotokos of Lips: Conant, op. cit., pp. 1S5-6, pl. xxiii b.
206. Wulff, op. Cit., II, pp. 473-474, figs. 404, 405.
207. This type also appears in Byzantine painting: Henri
Omont, Fac-similis des miniatures des plus anciens manuscrits
grecs de la Bibliotheque Nationale du VIe au XIe siecle, Paris,
I902, pls. xxxvIII, XLVII.
208. Where this is not the case, as, for example, at Paray-le-
Monial and at Tournus, which will be discussed in a later part of
this paper, it is not the result of design, but of the peculiar circum-
stances of the growth of those churches. Crosby has recently recon-
structed the Carolingian church of St. Denis with a pent roof over
a narthex between towers, but he himself writes that this appear-
ance of the narthex, particularly in conjunction with an atrium, is
more closely allied to the tradition of the Early Christian basilicas
in Rome and Italy. Sumner McKnight Crosby, The Abbey of St.-
Denis, I, New Haven, I942, p. 152, figs. 37, 38, 74-78, 89, 90, 92.
209. Oelmann, op. cit., pp. 189-236; 0. Puchstein, "Die Saule
in der assyrischen Architektur," Jahrbutchdes deut. archdologischen
Instituts, VII, 1892, pp. 1-24; H. Schaefer, "Two Gandharan
Temples and Their Near Eastern Sources," Journal of the Amneri-
can Oriental Society, LXII, 1942, pp. 63-65.
TWO-TOWER FA(ADE IOI
transfer of the tower idea as such is possibleand may have
resulted in suggesting the use of pairedtowers. This possi-
bility deservesinvestigation.
Historicallyit is not difficultto find connectionsbetween
Syria and the West. It is known that the Syrianswere the
merchants and traders of the early mediaeval world and
that their number in Gaul was large.210In fact it can be
observedthat bankerswere called "Syrians"in this period
in much the same way that they were called "Lombards"
at the end of the Middle Ages. During the seventh century
the catastrophesof the Persianwars and the Mohammedan
invasion, and during the eighth and ninth centuries the
iconoclastic controversiessent an unprecedented flood of
refugees'from Syria to the West,211 particularly monks,
who founded new monasteriesin Sicily, Italy, and Gaul.212
The orientalizationwent so far that a majority of the Popes
in the century from 650 to 750 were Orientals, among
them five Syrians.213It is noteworthy that the rite pre-
dominant in Gaul until the time of Charlemagne was ori-
ental in origin.214
There is not only evidence for this migration from East
to West, but also for a steady flow of pilgrims from the
West to the Holy Places in Palestine and Syria.215We even
have the records of architecturalobservationsby one such
pilgrimwho drew the plansof the buildingshe saw in Jeru-
salem and describedthem.216Of the churchesin Jerusalem
2io0. Louis Br6hier, "Les Colonies d'Orientaux en Occident,"
Byzantinische Zeitschrift, XII, I903, pp. 1-39; Jean Ebersolt, Ori-
ent et Occident, Paris, 1928-29, I, pp. I6-i8; Wilhelm Heyd,
Geschichte des Levantehandels im Mittelalter, I, Stuttgart, i879,
pp. 23-29, 99-142; Paul Scheffer-Boichorst, "Zur Geschichte der
Syrer im Abendlande," Mittheilungen des Instituts fiir Oster-
reichische Geschichtsforschung, vI, 885, pp. 52I-5 5 0.
21 . J. B. Bury, A History of the Later Roman Empire, II, Lon-
don, I889, pp. 207-269, Charles Diehl, History of the
428-469;
Byzantine Empire, trans. George B. Ives, Princeton, 1925, pp. 40-
67; Alfred J. Butler, The Arab Conquest of Egypt and the Last
Thirty Years of the Roman Dominion, Oxford, I902, pp. 154-
I67; Norman H. Baynes, "The Successors of Justinian," The Cam-
bridge Medieval History, II, Cambridge, Eng., 19 3, chapter Ix,
pp. 288-301; C. H. Becker, "The Expansion of the Saracens-
The East," The Cambridge Medieval History, II, Cambridge,
Eng., 19 3, chapter XI, pp. 338 ff.; Charles Diehl, "Leo III and
the Isaurian Dynasty 7 I7-802," The Cambridge Medieval History,
IV, Cambridge, Eng., I923, chapter I, pp. 1-21.
212. Lynn White, Jr., "The Byzantinization of Sicily," The
American Historical Review, XLII, I936, pp. 1-21; Ferdinando
Antonelli, "I Primi monasteri di monaci orientali in Roma," Ri-
vista di archeologia cristiana, v, 1928, pp. I05-12 1.
2I3. Louis Duchesne, Le Liber pontificalis, I, Paris, I886, pp.
33 1-426.
214. Louis Duchesne, Christian Worship, Its Origin and Evolu-
tion, translated by M. L. McClure, London, Society for Promoting
Christian Knowledge, 1923, pp. 86-I05; J. Quasten, "Oriental
Influence in the Gallican Liturgy," Traditio, I, 1943, pp. 55-78.
2 1 5. Publications of the Palestine Pilgrim's Text
Society, Lon-
don, The Society, I885-97, 32 vols.; cf. Jean Ebersolt, op. cit., I,
pp. 84-88; Julius Schlosser, Die Kunstliteratur, Vienna, 1924, pp.
41-42, where he gives an evaluation of this literature.
2I6. Lucas d'Achery and Joannes Mabillon, Acta Sanctorum
I02 THE ART
it was that of the Holy Sepulchrewhich served most fre-
quently as a model for churchesin the West; for example,
the priory church of the Holy Sepulchre at Moutier-les-
Jaligny in Bourbonnaiswas founded in Io36 by a French
nobleman after his return from a pilgrimage to the Holy
Land.217From the eleventh century we also have a letter
by Geoffroi, abbot of the Holy Trinity at Vendome, to
Hildebert de Lavardin, Bishop of Mans, in which he asks
whether the architect Jean (who was the architect of the
nave of the cathedral at Mans) had returned from Jeru-
salem and begs to have him sent as soon as possible.In an-
other letter he complainsbitterlyover the numberof monks
who go to Jerusalem.218
Unfortunately there is no such reference to show that
the idea of a two-tower fagade was imported from the
Orient. There is no direct evidence which would prove be-
yond doubtthat the motif was brought from the East to the
West. Since we do know that a great many connections
existed- that people went back and forth in relatively
large numbers; and that at least some of them showed an
interest in architecture- it is perhapssafe to suggest that
the idea as such, which was certainly a striking one, may
have been carried to Europe, either by one of the many
refugees, particularlymonks, or brought back to Europe
by pilgrims. In any case the difficulty of a great span of
time must be reckoned with. The greatest and finest de-
velopmentof Syrianchurch architecturewith its use of two
towers was reached in the sixth century, whereas the earli-
est preservedfacades with two towers in Western Europe
date from the eleventh century. Of course, it is fair to say
that most of the monuments in the Occident, particularly
in Italy and France, that might be comparedto the Syrian
architecture during the sixth and seventh centuries have
perished.
Sufficientevidence survives,nevertheless,to show that a
Ordinis S. Benedicti, Ill, pt. 2, Venice, 1734, pp. 456-472; cf.
Julius Schlosser, "Beitrage zur Kunstgeschichte aus den Schrift-
quellen des friihen Mittelalters," Sitzungsberichte der Kaiserlichen
A4kademie der Wissenschaften in Wien. Philosophisch-Historische
Klasse, cxxI I, 1 89o, p. 5 o idem, Quellenbuch zur Kunstgeschichte
des abendldndischen Mittelalters, Vienna, 1896, pp. 50-59; it is
interesting that the Venerable Bede gives an account of this descrip-
tion of the Holy Places: The Complete Works of Venerable Bede,
ed. and trans. J. A. Giles, IV, Historical Tracts, London, 1843, Pp.
402-443.
2 7. Victor Mortet, Recueil de textes relatifs ia l'histoire de
l'architecture et a la condition des architectes en France au moyen
dge, XIe-XIIe siecles, Paris, 1911, pp. 107-10o8 cf. also pp. 123-
125; Richard Krautheimer, "An Introduction
to an 'Iconography
of Mediaeval Architecture,' " Journal of the Warburg and Cour-
tauld Institutes, v, 1942, pp. 2-20; Gustaf Dalman, Das Grab
Christi in Deutschland, Leipzig, 1922.
218. Mortet, op. cit., p. 292. To the latter complaint Mortet
remarks in note 4 on page 292: "Certains moines durent mettre a
profit leur voyage a la 'Jerusalem terrestre' pour satisfaire leur
curiosit6 en matiere d'architecture, et ils purent tirer a leur retour
des resultats pratiques pour l'exercice de cet art."
BULLETIN
tradition of two-tower facades existed in the West before
we definitely know of it in the eleventh century. It is likely
that thistraditionhad its originsin the East, and in that case
the spanof time would be bridged. Krautheimersays of the
architectureof this period: "the Near Eastern elements in
early medieval architecturewere not just accidental infil-
trations,but . . . they formed the basisof this entire archi-
tecture."219 As an example he cites S. Apollinare in Classe
"with its polygonal apse flanked by pastophories,and with
its narthex flanked by west towers."220 Hubert has recently
claimed two-tower facades for several early examples in
France. Of St. Martin at Autun,221 which he dates at the
end of the sixth century (and a restorationin the ninth),
he says that it was "surmontee de deux tours." The basis
for this statement is a plan made in I658 and a description
of the building by Abbe Germain who lived from I689 to
175 I. A second example is St. Germain at Auxerre, which
was consecrated in 865. Here the south tower was re-
placed in the twelfth century by a larger tower which still
exists. The north tower and the narthex, of which the
towers were a part, were taken down only in I820. Hu-
bert's identification of the parts demolished in I820 as
belonging to the building of the ninth century is based on
the fact that eighteenth-centuryhistoriansof the abbey at-
tributedthem to Queen Clotilde (Fig. I8).222
A further example of this early period in France is the
abbeychurch of St. Denis built by Fulrad and dedicatedin
775. The abbot Suger, who tore down this Carolingian
church in the twelfth century, says of its western end:
"Therefore because in the front part, by the north (ab
aquilone), at the principalentrance doors,the arched door-
way was narrowed on both sides by twin towers, neither
high nor very beautiful,but menacing ruin . . . we began
to work."221 On the basis of this report by Suger and his
own excavation results at St. Denis, Crosby has recon-
structed a two-tower fagade for the eighth-centurychurch
(Fig. I9).224 He states that the "presence of the two tow-
ers flanking an entrance porch at St. Denis is conclusively
proved by the texts, and that the masonry found in the
excavationsof 1939 providesconvincing evidence that the
fagade was situated at the western end of the church."225
THE ART BULLETIN, 1942, p. 2.
219. Krautheimer,
220. Ibid., p. 6.
221. Hubert, L'Art pre-roman, fig. I8, pp. 11-13, 83.
222. Ibid., fig. 114, pp. 27-30, 83.
223. Quoted by Crosby, op. cit., I, p. I I 8, note 1 14: "Quia igitur
in anteriori parte, ab aquilone, principali ingressu principalium
valvarum, porticus artus hinc gemellis, nec altis, nec aptis multum,
sed minantibus ruinam, turribus angebatur . . . laborare . . .
incoepimus."
224. Ibid., fig. 37b. Cf. also pp. 118-127, 150-158, figs. 37a,
38, 74-78) 89, 90? 92.
225. In his review of Crosby's book Krautheimer,
Ibid., p. 123.
in-
of Archaeology, XLVIII, 1944, PP. 220-221,
American Journal
terprets the finds at St. Denis as giving evidence for "a central
structure dominating and slightly projecting over the rectangular
 THE ORIGIN OF THE
At Frankfort on the Main the Salvator church was com-
pleted in 874 with two square towers at its western end
and two round staircaseturrets between them. Thus the
entire western face of the church was blocked and portals
found their place on the north and south sidesof the square
towers.226Later a better solution for the staircaseturrets
was found when they were placed on the outside of the
squaretowers, as at Limburg a.d.Hardt. At Jumieges, the
church of St. Pierre, dating probablyfrom the first half of
the tenth century, appearsto have had two towers at its
western end.227Parts of these are preserved, and between
them an entrance porch with a tribune above. The two
westernmost bays of the tenth-century nave are also stand-
ing. In the first years of the twelfth century a part of the
north tower was torn down to make way for a chapterhall;
in the following century the south parts were changed;
and finally in the fourteenth century the entire church was
rebuilt, except for the fragments mentioned above. The
remnants of the towers preservedstaircaseswhich are cut
off where the towers were demolished. Though towers
cannot well be deniedhere, their height and form, as well as
the form and height of the central portion, cannot be pre-
cisely determined. Moreover, the central portionprojected
considerably,as it did in the later church of Notre Dame
at Jumieges, which was very likely inspiredby the earlier
church.
Aside from this group of early two-tower fagades we
have a number of illustrationsin the minor arts before the
eleventh century. As Krautheimerhas pointedout, mediae-
val representationsof architecturestand in a peculiarrela-
tionshipto the originals. They show "the disintegrationof
the prototypeinto its single elements, the selective transfer
of these parts, and their reshufflingin the copy."228In the
following examples the towers were evidently important
elements which were represented,though not in all cases
in the original relationshipto the rest of the building. The
most important of these representationsis a miniature of
the monastery of Cassiodorus at Squillace (Fig. 20).229
main bay and flankedby equally high or possiblysomewhatlower
towers on either side." In other words, he sees a sort of "West-
werk" at St. Denis.
226. G. Schoenberger, "Beitriage zur Baugeschichte des Frank-
furter Domes," Schriften des historischen Museums, III, 1927, pp.
7-31 -
227. G. Lanfry, "L'Eglise carolingienne Saint-Pierre de l'Ab-
baye de Jumieges," Bulletin Monumental, xcviii, 1939, pp. 47-66.
Cf. Aubert, L'Art fran~ais, I, p. 7; Reinhardt and Fels, Bulletin
Monumental, I937, p. 466, note z. The latter believe that the
staircases were not built into towers, but were encased in a block-
like fagade structure.
228. Krautheimer, Journal of the Warburg and Courtauld In-
stitutes, 1942, pp. 14-15.
229. Pierre Courcelle, "Le site du monastere de Cassiodore,"
Ecole frangaise de Rome, Milanges d'archeologie et d'histoire, LV,
1938, fig. 2; Friedrich Leitschuh, Katalog der Handschriften der
koniglichen Bibliothzeken zu Bamberg, Bamberg, 1903, Ms. Patr.
6i (HJ, iv, 15), Fol. 29V.
TWO-TOWER FACADE I03
This shows the sixth-century monastery church of St.
Martin with two towers. Courcelle sees in this church the
preciousmissing link which proves the transmissionof the
two-tower facade motif from the Orient to the West.230
One argument used by Courcelle to establisha connection
with Syriafor this buildingis the fact that Cassiodoruscame
of a Syrian family. As a matter of fact, the great-grand-
father of Cassiodoruswas already prominent in Southern
Italy, as were his grandfatherand father, so that his family
connection with Syriais a rather remote explanationof the
two towers at Squillace.-3'
One of the panels of the famous wooden door of Sta
Sabinaat Rome also shows a church with two towers (Fig.
2 I ) .22 These reliefs date fromthe fifth century,and though
their provenance has been disputed,233they are another
instance of the visual expressionof the motif in Western
Europe. Another fifth-century representationis that on
the so-called Werden casket (Fig. 22), in the Victoria
and Albert Museum in London.234Smith believes it to be
of Provengal origin, but points to the Syrian character of
the presentationand accounts for it "by the large Oriental
populationin Provence and the intimate relation of the re-
gion with the East."235In this case the two towers, which
stand at each side of the western end of the church, appear
to be round.
A numberof representationsof the Holy Sepulchreshow
a buildingwith two towers. There are two almost identical
ivory Gospel covers with such representations,the one
preservedat Paris,236the other in the church of Ste Croix
at Gannat (Fig. 23).237 They are both dated ninth-tenth
century and are eithercopiedone from the other or are both
copiesof a lost original. In an illuminationof the Benedic-
tional of St. Aethelwold (963-984), round towers like
those on the Werden casket stand beside the building in
the scene of the three Maries at the Sepulchre.238The last
230. Courcelle, op. cit., p. 307.
23I. Ibid., p. 280. Cf. Monumenta Germaniae historica, XII,
Cassiodori Senatoris Variae, ed. Theodor Mommsen, Berlin, 1894,
variarum, I, 3,4.
232. J. J. Berthier, L'Eglise de Sainte-Sabine a Rome, Rome,
1910, fig. 31.
233. Alexander Coburn Soper, "The Italo-Gallic School of
Early Christian Art," THE ART BULLETIN, XX, I938, fig. 43 and
p. i68; E. Weigand, "Der Monogrammnimbus auf der Tiir von
S. Sabina in Rom," Byzantinische Zeitschrift, xxx, 1930, pp. 587-
595.
234. Margaret H. Longhurst, Catalogue of Carvings in Ivory,
Victoria and Albert Museum, London, Published under the Author-
ity of the Board of Education, 1927-29, I, pp. 31-32, pI. IX.
235. Earl Baldwin Smith, Early Christian Iconography and a
School of Ivory Carvers in Provence, Princeton, 1918, pp. 221-
224.
236. Adolph Goldschmidt, Die Elfenbeinskulpturen, Berlin,
1914-26, I, p. 48, pl. xxxvI, fig. 86.
237. Ibid., I, pp. 49-50, pl. xxxvIII.
238. Eric G. Millar, English Illuminated Manuscripts from the
Xth to the XIIIth Century, Paris, 1926, pl. 4.
1o4 THE
example is an ivory plaqueof the tenth century in the Bar-
gello with a representationof the same scene.239Here the
Anastasis or Ecclesia Rotunda is included and it would
appearthat beyondit we are looking at the western- origi-
nally liturgical eastern- facade of the Martyrion, which
became the main facade of the buildingafter re-orientation
in the restorationof Modestus (seventh century).240The
Bargello ivory shows this facade with a gable and a tower
at each side. (The left hand tower is partially hidden be-
hind the angel.)
We have the evidence, then, of a traditionthrough these
centuries. This, as suggested earlier, helps to bridge the
spanof time between the great churchesof the sixth century
in Syriaand the western eleventh-centurybuildings.In the
transmissionof the traditionthe monasterieswere undoubt-
edly importantagents, and among them Cluny played the
most significant role. It was very likely due to this great
Burgundian monastery that the idea was given renewed
impetus in the eleventh century, as Dehio had originally
supposed.241 It is thereforeworth while to investigatebriefly
the relation of Cluny and the two-tower fagade.
4. CLUNY
The building which is of importancein this connection
is the church begun by the abbot Mayeul at Cluny before
963 and dedicated in 98I, commonly called Cluny II
(Fig. 24).242 We owe almost our entire knowledge of this
structure to K. J. Conant who has succeeded in recon-
structing the church. Two towers stood over the western
corners of a galilee of three bays which preceded and was
wider than the church. Conant bases his reconstruction
on the Ordo Farfensis (written between I 030 and 1048),
a view of Cluny by Prevost of about I670, and his own
findings at Cluny.243The Ordo Farfensisis only the oldest
of severalCluniac consuetudines,which regulatedthe plans
for the building of monasteries.In regard to the western
portion of the church they declare: "Let two towers be
built in the front of the galilee, and below them an atrium
where the laypeoplestand, so that they will not impede the
239. Goldschmidt, op. cit., II, p. 49, pl. XLVI, fig. I62;
Graeven, Friihckristliche und mittelalterliche Elfenbeinwerke,
Rome, 1900, pl. 26; A. Venturi, Storia dell'arte italiana, ii, Milan,
1902, pp. 616-617, fig. 450.
240. Hugues Vincent and F.-M. Abel, Jerusalem, Paris, ed.
J. Gabalda, 1914-22, II, pp. 218-228; George Jeffery, A Brief
Description of the Holy Sepulchre, Cambridge, Eng., i919;
H. T. F. Duckworth, The Church of the Holy Sepulchre, London,
I922, pp. 153-187.
241. Dehio and v. Bezold, Die kirchliche Baukunst des Abend-
landes, I, p. 574.
242. As reconstructed by Kenneth J. Conant in 1939. From the
photograph collection at Robinson Hall, Harvard University.
243. Kenneth J. Conant, "Mediaeval Academy Excavations at
Cluny," Speculum, IV, 1929-vII, 1932, XVII, 1942; idrem, "The
Third Church at Cluny," Mediaeval Studies in Memory of A.
Kingsley Porter, Cambridge, Mass., 1939, pp. 327-357.
ART BULLETIN
procession."244 This corresponds to the finds of Conant at
Cluny II.
At Hirsau the church of SS. Peter and Paul, begun 1082
and consecrated I091, is the first church in Germany to
be built according to these prescriptions of Cluny245 (Fig.
25).246 Here we have, in the beginning, an open court be-
fore which towers were built, the court being later trans-
formed into a galilee incorporating the western towers.
About the same time a similar development took place at
Lorsch, where towers were erected at some distance to the
west of a Carolingian church, leaving an older open court
between them and the church. This open court was then,
as in Hirsau, built up into a galilee some time during the
twelfth century.247 Mr. Conant has shown me with the
help of a series of plans of Cluny II how this same process
of open court, towers, galilee took place there.
The Cluniac interpretation with towers over narthex
or galilee was also adopted at Paray-Le-Monial, a Cluniac
church from the end of the eleventh century.248 The same
occurs some time in the second half of the eleventh century
at Romainmotier, a Cluniac priory in Switzerland.249 In
the first decades of the twelfth century a narthex of three
bays with towers at its western corners was added to the
abbey church of Vezelay, thus making it conform to the
Cluniac pattern.250 Two other abbey churches received
their Cluniac pattern directly from Hirsau. In Switzerland,
at Schaffhausen, an earlier church was torn down after the
monastery had been reformed by the abbot from Hirsau
after 1079, and a new church was built which corresponds
244. Otto Lehmann-Brockhaus, op. cit., No. 2187: "duae turrae
sunt ipsius galileae in fronte constitutae, et subter ipsas atrium
est, ubi laici stant, ut non impediant processionem." Mortet, op. cit.,
I, pp. 135-136, gives the variant "duae turrae sint . . ."
245. Eduard Paulus, Der Schwarzwaldkreis (Inventar der Kunst-
und Altertumsdenkmale im Konigreich Wiirttemberg, II), Stutt-
gart, 1897, pp. 43-66; A. Mettler, "Die zweite Kirche in Cluni
und die Kirchen in Hirsau nach den 'Gewohnheiten' des XI. Jahr-
hunderts," Zeitschrift fur Geschichte der Architektur, III, i909/o0,
pp. 273-286; IV, I910/Ii, pp. I-I6; idem, "Die beiden roman-
ischen Minster in Hirsau," Wiirttembergische Vierteljahreshefte,
1915, pp. 89-108; idem, Kloster Hirsau, Augsburg, 1928; E.
Fiechter, "Das Westwerk an der Klosterkirche von St. Peter und
Hans Paul in Hirsau," Wiirttembergische Vergangenheit, 1932, pp. 35-
162.
246. Mettler, Zeitschrift fur Geschichte der Architektur, III,
I909/0o, fig. i.
247. Friedrich Behn, op. cit., pp. 20-59, 122 ff., plans 3, 5, 8;
Heinrich Walbe, "Das Kloster Lorsch," Deutsche Kunst und Denk-
malpflege, xxxvii, 1935, pp. 126-142.
248. Jean Virey, Paray-Le-Monial et les eglises du Brionnais,
Paris, 1926; E. Lefevre-Pontalis, "Paray-Le-Monial," Congres
Archeologique de France, LXXX, 1913, pp. 53-64; Baum, op. cit.,
pl. i68.
249. Albert Naef, "Les phases constructives de l'eglise de Ro-
mainmotier," Anzeiger fiir Schzweizerische Altertumskunde, new
ser., VII, 1905/06, pp. 210-230; idem, "Les dates de construction
de l'eglise de Romainmotier," Bulletin Monumental, LXX, 90o6,
pp. 425-452.
250. Charles Por6e, L'Abbaye de Vezelay, Paris, 1930.
 THE ORIGIN OF THE
in proportion and detail to SS. Peter and Paul at Hirsau.251
In Thuringia, the abbey church of Paulinzelle was built
between 1108 and I 132. The monastery was settled by
monks from Hirsau and its church is distinctly modeled
after SS. Peter and Paul.252
The influence of Cluny, which reached Hirsau ca.
1077-78, brought there the Cluniac interpretation of the
use of two towers. In view of the earlier church at Hirsau,
St. Aurelius ( 105 9-7 I), which had two towers within the
body of the church (Fig. 26)253 and which is therefore a
much closer approximation of the ideal two-tower facade,
we must regard the Cluniac interpretation with its use of
the towered narthex as a parallel development. Moreover,
St. Aurelius belongs to the group of earlier churches of the
Upper Rhine region, which all approximate the ideal type.
Whether this group received its inspiration from Cluny or
from a source common to both cannot be determined.
Close to Cluny in place and time is the narthex of St.
Philibert at Tournus As it stands today it
(Fig. 27).254
also represents a scheme like that of Cluny, a main church
preceded by a narthex with towers. Although the original
form of this church and the dates of the various phases of
its construction stand in question, it may be permissible to
group this monument with the Cluniac examples of nar-
thex and towers.255
In all these instances that represent the Cluniac pattern
of the use of two towers, it is essential to note that the tow-
ers are at the western end of a narthex or galilee and are not
251. Josef Hecht, Der romanische Kirchenbau im Bodenseege-
biet, Basel, 1928, I, pp. 293 ff., pls. 199-200; Konrad Escher, Die
Miinster von Schaffhausen, Chur und St. Gallen, Frauenfeld/Leip-
zig, 1932, pp. 25 ff-
P. Lehfeldt, Fiirstent/izm Schzvarzburg-Rudolstadt (Bau-
252.
und Kunst-Denkmaler Thuiiringens,xix), Jena, 1894, pp. 127-
148. Adolf Zeller, Friihromanische Kirchenbauten und Klosteran-
lagen, Berlin/Leipzig, 1928, pp. 45-47.
253. Mettler, Zeitschrift fiir Geschichte der Architektur, III,
I909/10, fig. 3. Cf. idem, Kloster Hirsau, I928, p. 8, figs.
254. From the Photograph Collection, Robinson Hall, Harvard
University.
255. Lasteyrie, L'Architecture religieuse, p. i56, attributes the
whole of the narthex and its facade to the tenth century, whereas
Virey, "Les dates de construction de Saint-Philibert de Tournus,"
Bulletin Monumental, LXVII, 1903, pp. 549, 552-553 5 idem, Saint-
Philibert de Tournus, Paris, 1932, pp. 314-339i F. Deshoulieres,
Au debut de l'art roman, Paris, n.d., pp. 32-34; Puig y Cadafalch,
Le premier art roman, Paris, 1928, pp. 107-108; Marcel et Chris-
tiane Dickson, Les eglises romanes de l'ancien diocese de Chalon,
Macon, n.d., pp. 314-339; and H. Masson, Saint-Phiilibert de
Tournuzs, Tournus, Amis des arts et des sciences, 1936, would all
assign the lower story of the narthex to the tenth century but its
upper story and the towers to the first half of the eleventh cen-
tury. Both Charles Oursel, L'Art roman de Bourgogne, Dijon,
I928, pp. 44-45, and Ernst Gall, "Die Abteikirche Saint-Philibert
in Tournus," Der Cicerone, IV, I912, p. 632, believe in the unity
of the construction of the narthex with its facade, but while Oursel
dates it just before the great fire of ioo008, being' of the opinion
that the solidity of the narthex preserved it in the fire, Gall tries to
demonstrate that all was destroyed in this fire and that the entire
narthex and fagade date from the years 1007 or ioo8 to 1019.
TWO-TOWER FACADE Io5
a part of the church itself. It is nevertheless possible that
Cluny gave the impetus for the use of two towers, and that
this suggestion was soon re-interpreted and developed into
the true two-tower facade elsewhere, notably at Stras-
bourg, while Cluniac monasteries continued to use their
own scheme for some time.
5. WESTWORKS
The last, and perhaps the most important, thesis con-
cerning the origin of the two-tower facade is that of its
derivation from the so-called westworks.256 Nordhoff was
the first to use the term "Westwerk" and he defines it as
"ein hohes Glocken- und Mittelhaus mit eigener Beda-
chung, begrenzt von den Flankenthiirmen."257 Fuchs calls
them "zentrale Vorkirchen, deren Mittelraum zweige-
schossig und deren SeitenrHume sogar dreigeschossig sind,"
and he adds, "man darf wohl das Merkmal der Dreitiir-
migkeit, mag es auch weniger wesentlich sein, hinzufiu-
gen."258 I believe they can be defined as complex structures
containing the church vestibule, a tribune above, and stair-
cases at the sides. The tribune may be augmented by gal-
leries; both tribune and stairways are usually built up into
towers, the center one having the greatest height.
It is generally thought that these westworks originated
in the idea of a chapel over the entrance at the western end
of the church. Usually these chapels were dedicated to St.
Michael as the guardian of the gate.959 Other ideas have
also been brought forward: that the upper stories were used
for baptisteries, and also served as parochial churches, and
that the balcony which faced the nave of the main church
was for the use of the sovereign when he came to visit the
cathedral or abbey church.260 Effmann believes they were
used for judicial courts which were held by the bishop, a
256. Effmann, Die karolingischl-ottonischen Bauten zu Werden,
I, II. I, pp. 432-435 j Alois Fuchs, Die karolingischen lI'estwerke, Pader-
born, 1929, pp. 47-48, 65; Reinhardt and Fels, Bulletin Monu-
mental, xcII, 933, pp. 331-365; xcvi, 1937, pp. 425-469; Kraut-
heimer, 4American Journal of Archaeology, XLVIII, 1944, pp. 220-
22 I.
257. J. B. Nordhoff, "Corvei und die westfalische Friuharchi-
tektur," Repertorium fiir Kunstwissenschaft, xII) 1889, p. 380.
258. Fuchs, op. cit., p. I6.
259. Ibid., pp. 5-6, note i on p. 6, p. 3 ; Friedrich Ostendorf,
Die deutsche Baukunst imnMittelalter, Berlin, 1922, I, pp. 44, 217.
Fuchs (p. 31) mentions that the old Germanic idea of having a
cult place on top of a hill influenced the placing of a chapel dedi-
cated to St. Michael in the upper stories of a church. He writes that
it is well known that chapels of St. Michael were often built on
hill tops which had earlier been heathen cult places. Cf. also Otto
Lehmann-Brockhaus, Schriftquellen zour Kunstgeschichte des I i.
und 12. Jahrh/underts fiir Deutschland, Lothringen und Italien,
Berlin, 1938, I, nos. 6, 123, 256, 317, 569, 894, 955, 1049.
260. Fuchs, op. cit., p. 43; Otto Gruber, "Das Westwerk: Sym-
bol und Baugestaltung- germanischen Christentums," Zeitschrift
des deutschen Vereins fur Kunstwvissenschaft, III, 1936, pp. 150-
15 ; Wilhelm Rave, "Sint Servaas zu Maastricht und die West-
werkfrage," 14estfalen, xxii, 1937, pp. 56-59.
 Io6 THE ART
duty imposedon him by the reform movement of Carolin-
gian times.261At any rate, it is clear that these tower-like
structures were created primarily for practical purposes
and grew out of practicalneeds.
The first grandiose example is that of the Carolingian
abbey of St. Riquier (Fig. 28).262 Other Carolingian ex-
amples are the Capella Palatina at Aachen (Fig. 29),263
and the abbey church of Corvey.264As has been stated
above, the thesishas been brought forward that this type of
structure represents the source of the two-tower facade
motif. As Fuchs says, these Carolingian structures repre-
sented a "Gestaltungsmaximum"265 and a number of later
types are derivations or reductions from this maximum.
The type persistedas such also, for example,in the first half
of the eleventh century in Essen266and in St. Pantaleon267
and St. Maria im Kapitolin Cologne,268and in the twelfth
century in the church at Freckenhorst,269 the abbeychurch
of Maria Laach (with the additionof a western apse in this
case),270and the abbeychurchof Maursmunster,which was
discussedat greater length above (Fig. 6). In the Nether-
lands the type is representedby St. Servaasat Maastricht.271
By bringing the staircasesinside of the main tower and
suppressingor eliminatingthe sidestaircaseturrets,the one-
tower type of fagade is created during the eleventh and
twelfth centuries,accordingto thistheory.272Suchwould be
26I. Effmann, op. cit., pp. I76-I83.
262. Effmann, Centula. St. Riquier; eine Untersuchung zur
Gesclhichte der kirchlichen Baukunst in der Karolingerzeit, Miin-
ster, 1912. Our figure is from the collection at Robinson Hall,
Harvard University.
263. G. Dehio and G. v. Bezold, Die kirchliche Baukunst des
Abendlandes, Stuttgart, 1887-1901, I, text, p. 570. Cf. Albrecht
Haupt, Die Pfalzkapelle Kaiser Karls des Grossen zu Aachen,
Leipzig, 1913.
264. Effmann, Die Kirche der Abtei Corvey, Paderborn, I929,
fig. 42.
265. Fuchs, op. cit., pp. 47-48.
266. Edgar Lehmann, Der friihe deutsche Kirchenbau, Berlin,
1938, pp. III-II2, pI. 12, fig. 27.
267. Ernst Gall, Karolingische und ottonische Kirchzen, Burg
bei Magdeburg, 1930, p. 8i. Lehmann, op. cit., p. 121, pl. 10,
fig. 22.
268. Hugo Rahtgens, Die kirchlichen Denkmaler der Stadt Koln,
Dusseldorf, 1911, pl. xvi, figs. 132, 137, 139, 145.
269. Dehio and v. Bezold, Die kirchliche Baukunst, I, Plates, pl.
169.
270. Lehmann, op. cit., pl. 32, fig. 66.
271. F. A. J. Vermeulen, Handboek tot de geschiedenis der
Nederlandsche bouwkunst, I, Plates, The Hague, I928, pls. 10, iI.
272. A number of early single western towers make this thesis
unlikely: in Germany a small basilica was erected at Fritzlar in
732 with a west tower, Becker, Denkmalpflege, xxi, 1919, pp. 85-
88, and the church on the Petersberg near Fulda received a west
tower in 836, Lehmann, Der frilhe deutsche Kirchenbau, p. 113,
fig. 98. (I was unable to obtain G. Weise, Untersuchungen zur
Geschichte der Architektur und Plastik des fri/ihen Mittelalters,
Leipzig, 1916, pp. 78-98, who has a more detailed account of this
monument.) In England a large number of Anglo Saxon exam-
ples still remain: Bond, Gothic Architecture, pp. 595 ff.; Brown,
The Arts in Early England, II, pp. 330 ff., 385 if.
BULLETIN
the Patroklus church at Soest (Fig. 30),273 St. Quirin at
Neuss,274the cathedralat Minden,275the abbey church of
Rolduc in Holland,276and in France, where it is called
"clocher-porche," the churches at Ebreuil,277 Saint Benoit-
sur-Loire,278Morienval,279and le Dorat.280
The so-called "Saxon" fagade is the result of the drop-
ping of the centraltower, but substitutingfor it a three-story
rectangular structureconnecting the two flanking towers,
which gain in importance. One might even term this a
"two-tower" facade, but it does not conform to the ideal
type, because the high rectangular central portion is not
a part of the nave nor a prolongation of it, but is still a sepa-
rate structure which obscures the nave behind it. This type
occurs in Lower Saxony and is most clearly represented by
the church at Gandersheim (Fig. 3 I ),21 the former cathe-
dral at Goslar,282 and the abbey church of Corvey in its
present state.283
Finally, it is held, the two-tower facade itself is another
derivation from the westworks. A fagade such as that of St.
Peter at Wimpfen im Tal (Fig. 32),284 which probably
dates from the end of the tenth century, is considered as a
last step in the development to the true two-tower facade.
Since this fagade was originally part of a central type build-
ing for which a basilica was substituted only in Gothic
times, and since the ideal type of two-tower facade pre-
supposes a basilica, St. Peter cannot itself be considered in
the line of those monuments approaching it. This facade is
mentioned here as an analogous solution for a different
architectural type of a period from which no basilica with
towers is preserved. There would follow then the group of
Upper Rhenish churches discussed above, which would
represent the first group of two-tower fagades properly so
called, namely, Basel, Strasbourg, Limburg a.d.H., Hers-
273. Lehmann, op. cit., pl. 32, fig. 68.
274. Ibid., pl. 31, fig. 64.
275. Ibid., pl. 31, fig. 65.
276. Vermeulen, op. cit., pl. i2.
277. Ren6 Colas, Le Style roman en France, Paris, 1927, pl. 96.
278. Baum, op. cit., pl. 158; Aubert, L'Art francais, I, p. 15.
279. Baum, op. cit., pl. de Lasteyrie, op. cit., fig. 406.
213;
280. De Lasteyrie, op. cit., fig. 405.
281. Lehmann, op. cit., pl. 27, fig. 57.
282. G. Dehio, Geschichte der deutschen Kunst, I, pl. 174i
Erwin Panofsky, "Der Westbau des Doms zu Minden," Reper-
torium fiir Kunstmwissenschaft,XLII, 1919, fig. 8.
283. Panofsky, op. cit., fig. 9.
284. From the photograph collection of the Germanic Museum,
Harvard University. Cf. Adolf Zeller, who connects this facade
with Aachen: Die Stiftskirche St. Peter zu Wimpfen im Tal, Wimp-
fen, 1903, p. 7. Against Zeller, Dehio, "Zwei romanische Zentral-
bauten," Zeitschrift fur Geschichte der Architektur, I, 1907/08,
pp. 45-46, points out the differences between Aachen and St. Peter
at Wimpfen, and similarities between the latter and a number of
Syrian central type monuments, and thus comes to the conclusion
that St. Peter at Wimpfen can only be explained by assuming di-
rect Syrian influence. Cf. also idem, Handbuch der deutschen
1908, p. 561; Leh-
Kunstdenkmader, IIi, Siiddeutschland, Berlin,
mann, op. cit., p. 144.
 FIG. 25. Hirsau, SS. Peter and Paul

FIG. 24. Cluny II, Abbey Church: Plan (Conant)

FIG. 29. Aachen, Cape

FIG. 30. Soest, St. Patroklus FIG. 31. Gandersheim, Minster

FIG. 32. Wimpfen im Tal, St. Peter

 THE ORIGIN OF THE
feld, Koblenz, and others. This group is far from the west-
work region, and only the twelfth-century abbeychurch at
Maursmiinster can be cited as an example of the type in
this region (Fig. 6).285 It would therefore seem hazardous
to consider this Upper Rhenish group the end point of a
development for which there is in this region no evidence
in any of its preceding stages.
A development from the original westwork as a "Ge-
staltungsmaximum" to the independent expression of its
variouscomponent partsseems logical and is probablyto be
assumed for the region in which the original type flour-
ished, i.e., Lower Saxony. However, even Fuchs admits
that other factors may have been involved; in fact, the
Upper Rhenish group of two-tower facades puzzles him,
and he explains them by assuming a Carolingian westwork
for Maursmiinster which would have served as a model
for the present church there and from which the two-tower
facades of this region were derived.286 But for this there is
no evidence.
Another difficulty for this theory, even on its own soil
in Lower Saxony, appears in the history of the cathedrals
at Minden and Hildesheim, both in the westwork region.
Panofsky has shown beyond doubt that the eleventh-cen-
tury structure (under Bishop Eilbert I05 5-80) of the
cathedral at Minden had a two-tower facade and that only
later, in the twelfth century, the present westwork-like
structure was built. He attributes the early appearance of a
two-tower facade in this region directly to influence from
the Upper Rhenish school in the eleventh century and be-
lieves that later the more indigenous form of the westwork
won out and superseded the two-tower facade. He bases
his conclusions on observations of the monument (Fig.
33).287 The large built-in arch over the entrance is set on
the corners of an original gable as a reenforcement for the
later superstructure. Originally the center ended in a gable
at the height of this arch, whereas the structure of the sides
is original up to the horizontal break above and includes the
narrow slits just under this line.
From the available material of Hildesheim - the west
facade was demolished in 1840-41 and only documentary
evidence remains - Panofsky assumes an analogous struc-
ture for the eleventh century under Bishop Hezilo (1054-
79), and also a change in design in the twelfth century
similar to Minden.288
The cathedrals of Minden and Hildesheim must there-
285. Reinhardt and Fels, Bulletin Monumental, xcvi, 1937,
facing p. 428; Aubert, L'Art frangais, I, p. Io, pl. 19.
286. Fuchs, op. cit., p. 65.
287. Panofsky, op. cit., pp. 51-77; cf. also Nikolaus Pevsner,
"Die Bautaitigkeit des Heiligen Godehard am Hildesheimer Dom,"
Die Denkmalpflege,xxxv, 1933, pp. 213-214; Wilhelm Ritter,
Der Eilbertdom zu Minden in 'Westfalen, Minden, 1926, pp. Io-
19, 27-42, 44-56. Our figure is from Lehmann, op. cit., pl. 31,
fig. 65.
288. Panofsky, op. cit., pp. 67-74. Cf. Ritter, op. cit., pp. 33-
TWO-TOWER FAgADE I07
fore occupy a specialplace, if one assumesa reductionfrom
the westwork to the two-tower fagade, because here the
two-tower fagadeis anteriorto the westwork-likestructure
which still standsin Minden and existed in Hildesheimuntil
i 840-4 I. It is perhapspermissibleto join Panofsky in sug-
gesting an influence from the Upper Rhenish region.289
It must be emphasizedthat the two-tower facade differs
in two ways from the westworks. As has been stated, the
westwork grew out of and satisfied primarily practical
needs, liturgical and otherwise, whereas the two-tower
facade is a solution which satisfies primarily aesthetic
needs.290With their own separatefunctions the westworks
were added and combinedas separateentitieswith the main
body of the church, and often with their transversestruc-
tures they even negate the existence of the church beyond
them. The two-tower facade, on the other hand, one
might say, represents or expresses the structure of the
church with its nave and aisles, it is a part of a homogene-
ous whole; it has become united with the body of the
church. Aesthetically it conceals the abrupt drop from the
gabled roof of the nave to the sloping roofs of the aisles and
gives the entire facade a nobler aspect.
If this is taken into account, and also the possibility of a
penetration of South German influence, we may at least
assume the correctness to this extent of the thesis of Eff-
mann, Fuchs, and Reinhardt, that the westworks are the
source of the two-tower facade in Lower Saxony, but by
no means in all of Western Europe. Nevertheless, in this
theory we come closest to an observable development from
a known starting point, through intermediary stages, to
clearly identifiable two-tower facades.
To sum up: a knowledge of two-tower facades may be
35, 51-52 Pevsner, op. cit., pp. 210-214. Lehmann, op. cit., pp.
16, 130-131, assumes a two-tower facade for both Minden and
Hildesheim in the eleventh century, but he believes they were the
first examples of "Saxon" facades, i.e., the central portion between
the towers built up with a transverse pitch roof. However, the gable
which was found in the center would make such a reconstruction
impossible. Effmann, Zur Baugeschiichte des Hildesheimer Domes
vom 9. bis zum I2. Jahrhundert, Hildesheim/Leipzig, 1933, tries
to show that Godehard's structure at Hildesheim was a real west-
work. The basis of his interpretation is a literary source, Wolfher's
Vita of St. Godehard (Monumenta Germaniae historica, ed. G. H.
Pertz, Scriptorum, XI, Hannover, 1854, ca. 37, p. 195), which is,
however, differently interpreted by Pevsner, op. cit., pp. 2II-2 4,
who points out that it allows for a two-tower facade if the bell-
structure is assumed to be over an entrance hall, i.e., the paradisum.
289. Pevsner, op. cit., p. 213, points out that Godehard was ab-
bot of Hersfeld in 1oo005and was consecrated as abbot by Arch-
bishop Willigis of Mainz, so that he was undoubtedly familiar with
the works of the Upper Rhenish school when he was sent to Saxony.
Cf. also Godehard's connections with Henry II: Ernst Landers, Die
deutschen Kloster vom Ausgang Karls des Grossen bis zum Worm-
ser Konkordat und ihr Verhzaltniszu den Reformen, Berlin, I938,
pp. 36-37.
290. It is therefore not correct when Gall says: "Westwerke und
Doppelturmfassaden waren andere Losungen ein und desselben
kuiinstlerischenProblems," op. cit., p. 20.
io8 THE ART
assumed for Western Europe at the very beginning of the
mediaeval period. Though the ultimate origin of the idea
must remain obscure, it is likely that it was brought to
Europe from Syria and the Holy Land and that in subse-
quent centuriesit was carriedfar and wide through West-
ern Europe. It must, however, be borne in mind that in the
East the two towers invariablyflank a narthex that projects
as a one-story,flat-roofedloggia from the actual facade and
stands as a separate entity before the nave of the church.
In Western Europe, especially in the Cluniac group of
churches, the towers are also sometimes used with a nar-
thex, but here the narthex is either itself built up into a
tower or the gable of the narthex correspondswith that of
the nave, so that the towers are integrated into the actual
fagadeof the church. It could only be as a general sugges-
BULLETIN
tion that the use of two towers was introducedto Europe
from the East. In the form of a traditionthe idea was main-
tained over centuries, very likely by the monasteries.It is
where this tradition then merged with local factors and
resultedin the practicalexpressionof the idea that the more
immediate sources of the Romanesque two-tower fagade
must be sought. The tradition was reinterpreted and
brought to life by local talent in invention and adaptation,
which was in turn inspiredby the various locally existing
Roman and Germanic monuments. It was not one particu-
lar factor that was responsiblefor the creationof the motif,
but rather a combinationof factors which differed in the
variousparts of Europe.
[HARVARD UNIVERSITY]