The Technology of Transition: Sexpartite to Quadripartite Vaulting in

unique occurrence, developed alnriostcertainly in response
to a series of unconventional design requirements. As a
resolution of the various problems faced by the designer in
his attempt to integrate a spacious and functional gallery
into a compressed elevation, the gabled gallery was an individualized solution that was never repeated in its
totality. Even so, it reflects the architectural milieux with
which the Durham nave designer was familiar and
provides an insight into the essential processes of his
thinking. He was clearly not a "structuralist" in the Early
Gothic sense of the term, not a designer with the same
concerns or the same ambitions as, for example, the man
who built St. Lucien at Beauvais using some of the same
technical apparatus. Instead, the second master at Durham
conceived of structure and of structural mechanics in
terms of strong, isolated units of support that were not
necessarily joined together in an integrated, systematic
whole. In this, his work continues the techniques of earlier
wood-roofed architecture and, by taking some of those
techniques to their logical extension, represents a culmination of that older Romanesque tradition. Even though he
might use rib vaults, he chose to hold them up by means of
thickened walls and not by lateral buttressing elements.
And even though he might use quadrant arches - and use
them in an inventive way - he still perpetuated their
original roof-supporting functions. The nave designer at
Durham was, without question, one of the great innovators of his period, but his personal solutions ultimately did not generate a consistent following. Many of
his elements may superficially suggest the dawn of Gothic
architecture, but, in the end, the conclusion must be that
his view of architecture was still dependent on ideas that
would play only a small role in the formulation of that
later style.
Columbia University
New York, NY 10027
Bibliography
Billings, Robert, Architectural Illustrations and Description of the
Cathedral Church at Durham, London, 1843.
Bilson, John (with Harold Brakspear and Charles Peers), 1915-16, "A
Report to the Society of Antiquaries on Certain Repairs Now Being Undertaken at Durham Cathedral," Proceedings of the Society of
Antiquaries,
,
2nd ser., xxvII,
49-53.
1922, "Durham Cathedral. The Chronology of Its Vaults,"
Archaeological
Journal, LxxIx, 101-60.
Bony, Jean, "Le projet premier de Durham. Voitement partiel ou
voitement total?," in Urbanisme et architecture. Etudes icrites et
publikes en l'honneur de Pierre Lavedan, Paris, 1954, 41-49.
Branner, Robert, "Gothic Architecture 1160-1180 and Its Romanesque
Sources," in Studies in Western Art (Acts of the XXth International Congress of the History of Art), Princeton, 1963, 92-104.
Greenwell, William, 1916, "Notes on Durham Cathedral," Journal of the
British Archaeological
Association,
xxII, 1916, 140-42.
, 1932, Durham Cathedral (1881), 9th ed., Durham.
Liess, Reinhard, Der friuhromanischeKirchenbau des II. Jahrhunderts in
der Normandie, Munich, 1967.
Rave, Paul Ortwin, Der Emporenbau in Romanischer und Friuhgotischer
Zeit, Bonn, 1924.
Ruprich-Robert, Victor, L'architecture normande aux XIe et XIIe sikcles
en Normandie et en Angleterre, 2 vols., Paris, n.d. [1884-89].
The Technology of Transition: Sexpartiteto QuadripartiteVaulting in
High Gothic Architecture
WilliamTaylorandRobertMark
Introduction
Ribbed vaulting is often regarded as representing the essence of Gothic architecture. For Henri Focillon, "the
vaulted rib.., became the progenitor of an entire style,"
and, indeed, its introduction marked a distinct change in
the design of church interiors.1 Perhaps most important,
ribbed vaulting allowed greater fenestration of the walls
than did earlier barrel and groin vaults. But there were
constructional limits on how far this advantage could be
carried until the flying buttress was introduced to stabilize
The authors gratefully acknowledge helpful suggestions made during
the course of this investigation by H. Titus of Wake Forest University
and by W. Clark of Queens College of the City University of New York.
1 H. Focillon, The Art of the West in the Middle Ages, Greenwich,
Conn., 1963, II, 3.
580
THE ART BULLETIN DECEMBER 1982 VOLUME LXIV NUMBER 4
2 BourgesCathedral,view of the nave wall and sexpartitevaulting (photo: J. Austin)
1 ChartresCathedral,view of transeptwall and quadripartite
vaulting (photo: ArchivesPhotographiques)
3 Comparativeelevations:Laon,Chartres,Reims, and Amiens
the clerestory wall in the last quarter of the twelfth century.2 By the turn of the century, at Chartres and at
Bourges, clerestory walls were reduced to stone frames enclosing large areas of glass (Figs. I and 2). Furthermore, at
Chartres the height of the clerestory was increased to a
striking degree, with its vaults springing from well above
the clerestory base. The trend toward taller clerestories
and higher vault springing, along with reducing the
solidity of the clerestory walls in seeming defiance of the
thrust of the lofty vaults, was to become the most
definitive feature of the High Gothic cathedral (Fig. 3).
(from Dehio and Bezold, Die Kirchliche Baukunst des
Abendlandes)
2 R. Branner, Gothic Architecture, New York, 1961, 27.
Ongoing studies
in our laboratory of a reconstruction of Notre Dame, Paris (with William
Clark), have confirmed Branner's belief that flying buttresses were indeed necessary in the nave of this tall building.
3 Quadripartite vaulting was often used in early Gothic churches, but
mainly in smaller ones. For an account of the use of sexpartite and
quadripartite vaulting, see Marcel Aubert, "Les plus anciennes croishes
d'ogives; leur r6le dans la construction," Bulletin monumental, xciiim,
1934, 5-67, 137-237; Jean Bony, "Diagonality and Centrality in Early
Rib-Vaulted Architectures," Gesta, xv, 1976, 15-25.
The form of the ribbed vaulting was also altered soon
after the advent of the flying buttress. Prior to the year
1200, vaults sprang from a solid wall at a point below the
base of the clerestory and, with few exceptions, squareplanned vaults of sexpartite figuration were used in the
main bays of the larger Gothic churches.3 In the original
construction of all High Gothic churches after 1200,
THE TECHNOLOGY
OF HIGH GOTHIC VAULTS
581
however, there was a shift to rectangular-planned,
quadripartite vaults, sprung from a point above the base
of the clerestory.4 A causal relationship between the
development of the raised High Gothic clerestory supported by flying buttresses and the shift in vault configuration can thus be accepted prima facie, yet the literature on
Gothic architecture is rather vague on this point. Those
explanations that have been advanced generally fall into
two categories: stylistic and constructional.
Not surprisingly, stylistic explanations predominate in
the art-historical literature. Implicit in all of these is the
understanding that the use of sexpartite vaults arose from
the introduction of alternating nave piers. Since the number of vault ribs that spring from the piers is alternately
one and three for sexpartite vaulting, this system is
claimed to be a more logical visual complement to alternating piers. By the same reasoning, the stylistic theories
attribute the adoption of quadripartite vaulting to the introduction of uniform, non-alternating piers.
Such explanations have their origin in late nineteenthcentury architectural commentary. In response to a
functional theory advanced by Viollet-le-Duc which attempted to rationalize the discrepancy between the uniform nave piers and sexpartite vaults of Notre-Dame,s5
Charles Moore countered that such rationalization "is
hardly a sufficient justification of the whole design as it exists; for this mode of reinforcement does not satisfy the
eye, however adequately it may provide for strictly
mechanical exigencies of the scheme."6 Moore's belief in
stylistic convention was strong. He suggested that
cathedrals in which quadripartite vaults were coupled
with alternating piers - or sexpartite vaults coupled with
uniform piers - were so peculiar that they probably
represented a change from the design of the original
master: "The incongruity thus presented, in the naves of
Paris and Noyon respectively, between the forms of the
vaults and the forms and adjustments of their supports,
constitutes a serious defect in each of these otherwise no-
ble structures, as they have come down to us ... a defect
which so contradicts the logic of the Gothic system as to
leave little doubt that it was in each case the result of a
change from the original project."'7
Reasoning similar to Moore's has also been invoked
more recently by historians to explain the origin of the
sexpartite vault and its subsequent rejection in favor of the
quadripartite system. Focillon, Frankl, Jantzen, Pevsner,
Seymour, Stoddard, and von Simson are among those who
have attributed the form of the high vaults to the configuration of the nave piers.8 If one accepts that alternating piers are visually more compatible with sexpartite
vaults, and uniform piers with quadripartite vaults, it is
problematic to use this observation to explain the adoption
of raised quadripartite vaults in all the High Gothic
cathedrals. First, it would be difficult to show that such
stylistic unity actually was a major concern for the
medieval builder. Noyon, Paris, Laon, and Chartres are all,
in one way or another, exceptions to this schema.9 The
number and importance of the buildings comprising the
exceptions make the foundations of the hypothesis
questionable. Of even greater moment, such an explanation fails to explain why any variation in the pier-vault
system occurred at all. Even granting that uniform piers
required matching to quadripartite vaulting, this
hypothesis still fails to explain the suddenness of the
change that took place at the turn of the century.
Nor do the constructional theories provide an adequate
explanation. These are based on the premise that
quadripartite vaults were easier to construct than sexpartite vaults; and since erecting the centering for the vaulting was one of the most complex, expensive, and
dangerous operations of tall church construction, any
vaulting system requiring less complicated centering
would have been favored by the medieval builders. For
example, Viollet-le-Duc maintained that the primary
reason that the sexpartite plan was abandoned was
because the square sexpartite bay required a diagonal rib
4 After the collapse of the original quadripartite vaulting of the choir of
Beauvais Cathedral in 1284, additional piers were erected which supported a new sexpartite vaulting system. The new six-part vaults, however,
were built on the same rectangular plan as the original four-part vaults
and therefore are not conventional.
Seymour, Notre Dame of Noyon in the Twelfth Century, New Haven,
1939, 69-71, 134-35; Stoddard, 130, 140, 181; O. von Simson, The
Gothic Cathedral: Origins of Gothic Architecture and the Medieval Concept of Order, New York, 1962, 205-06.
was puzzled by the seeming incongruity of the uniform
piers and sexpartite vaults in the cathedral of Paris. He suggested that it
was essential that the arcade piers that supported a transverse rib and two
diagonal ribs receive additional support in some way. Conveniently, the
piers in the aisle that are adjacent to these "strong" arcade piers are encircled by twelve cylindrical shafts. Viollet-le-Duc contended that these
shafts give the necessary support to the aisle pier, because it in turn
provides support, at least indirectly, to the adjacent arcade pier. See
Viollet-le-Duc, 219-221.
5 Viollet-le-Duc
6
Moore, 113.
7 Ibid., 112-13.
H. Focillon, The Art of the West in the Middle Ages, 1963, II, 35; P.
Frankl, Gothic Architecture, trans. D. Pevsner, Baltimore, 1962, 18, 19,
80, 118; H. Jantzen, High Gothic: The Classic Cathedrals of Chartres,
Reims and Amiens, trans. J. Palmes, London, 1962, 13; N. Pevsner, An
Outline of European Architecture, New York, repr., 1978, 100-09; C.
8
SNoyon Cathedral is probably the most outstanding exception. At
Noyon, quadripartite vaults are coupled with alternating piers. Although
the main vessel of the cathedral seems to have been designed for sexpartite vaults, it is probable that the present quadripartite vaults are original.
See: Seymour (as in n. 8), 69-71, 134-35, Marcel Deyres, "Les vofites de
la cathbdrale de Noyon," Bulletin monumental, cxxxIin, 1975, 275-284;
W. Clark, "The Nave Vaults of Noyon Cathedral," Journal of the Society
of Architectural Historians, xxxvI, 1977, 30-33; A. Prache, "Apropos
des voites de la nef de la cathbdrale de Noyon," Bulletin monumental,
cxxxvi,
1978, 73-77.
See also: W. Clark, "The Nave of Saint-Pierre
at
Lisieux: Romanesque Structure in a Gothic Guise," Gesta, xvi, 1977, 2938 and W. Clark, "Spatial Innovations in the Chevet of Saint-Germaindes-Pres," Journal of the Society of Architectural Historians, xxxviiI,
1979, 348-365. At Paris, sexpartite vaults are coupled with uniform nave
piers, while at Laon, sexpartite vaults are coupled with uniform piers in
the western bays of the nave. At Chartres quadripartite vaults are
coupled with alternating piers. Also, many early Gothic churches used
both types of vaults within the same building.
582
THE ART BULLETIN DECEMBER 1982 VOLUME LXIV NUMBER 4
4 Computer-drawnperspectiveprojectionof quadripartite
vaulting.
that was much longer than the transverse rib. This raised
constructional problems because the arches of the transverse ribs had to be very acute or stilted in order to attain
the same heights as the keystones of the diagonal ribs.
Viollet-le-Duc observed that, "in about 1230 [sic] they
gave up this kind of vault upon a square plan, and established the diagonal arches of the naves upon an oblong
plan, or, in other words, each compartment had its complete quadripartite vault. We can thus make the keys of
the diagonal, transverse, and wall arches reach the same
level or nearly so."'o10He then pointed out that "the complex, convex shape of sexpartite vaults made it difficult
for the medieval builders to form the centering.""1 The
simple shape of the quadripartite vaulting, in Viollet-leDuc's view, demanded less complex centering than did the
sexpartite case. John Fitchen and others have expressed
similar opinions.12
In contrast, some historians have concluded that centering was more difficult to erect for quadripartite vaulting
than for sexpartite, but that quadripartite vaults had other
overriding advantages that account for their adoption. For
example, Arthur Kingsley Porter suggested that: "The
sexpartite form in France was probably abandoned
because the great length of the diagonals raised the crown
of the vaults to such a height that it became necessary to
raise correspondingly the walls so that the roof should
clear the vaults. This raising of the walls wasted materials,
and moreover increased the weight that must be carried by
the piers; so the builders concluded that it was better to
return to the uniform system of quadripartite vaulting
even at the expense of erecting two more centering arches
for their vaults."13 More recently, Fitchen suggested that
the High Gothic builders adopted quadripartite vaults
because they were lighter than sexpartite vaults.14Lighter
vaulting also implies less thrust on the walls and buttressing and therefore lighter construction throughout. Lighter
vaulting also offers constructional savings because it
needs less rigid centering for erection, and of course requires a smaller amount of finished stone.
Computer-model studies of ribbed vaulting, carried out
at Princeton, were originally undertaken to determine the
structural role of the rib (Fig. 4).1s In the course of these
studies it was also found that the weight of a sexpartite
vault was significantly less than the weight of quadripartite vaults covering the same area. Indeed, it became evident that the lighter weight of the sexpartite configuration
can be attributed largely to the fact that it carries fewer
ribs than its quadripartite equivalent. In any event, the
finding that the thirteenth-century builders, who
generally favored light construction, would choose to construct heavier vaults over increasingly slender piers and
walls in the tallest churches did nothing to clarify the
enigma surrounding the abrupt change in vaulting form.
Analysis: Construction Forces
The new theory offered here, which explains the need to
employ quadripartite vaulting in conjunction with the
developed High Gothic tall clerestory, is based on a
previously unconsidered constraint that arose during the
high vault construction. Although the computer modeling
of the vaulting supplied essential quantitative data to support the new theory, the basic premise was derived from a
simple geometric interpretation of the nature of the forces
providing support to both types of vaulting at their
springing. Further confirmation of the existence of these
forces and of the sensitivity of the builders to them is
found in the manner in which the vaults were supported
by the clerestory walls of extant Gothic buildings.
The salient structural feature of Gothic vaulting concerns the "focusing" of the distributed forces within the
vaults at the points of vault support on the clerestory wall.
There are three components of this focused force at the
springing in each bay: a downward, vertical component
equal to a fraction of the total weight of the vaulting and
its ribbing (for example, one-fourth of the total vault
weight of one bay of quadripartite vaulting) - supported
by the clerestory and triforium walls which are in turn
carried by the piers of the nave arcade; an outward,
horizontal component (referred to here as outward
''thrust") tending to overturn the clerestory wall, but
resisted in the mature Gothic church by flying buttresses;
and a longitudinal, horizontal thrust component pushing
into the adjacent bay along the axis of the church. This last
thrust is ordinarily stabilized by the adjacent bay of vault75. This argument was also offered by H. Sedlmayr, Die
o10Viollet-le-Duc, 123.
14 Fitchen,
11Ibid., 124.
Entstehungder kathedrale,Zurich,1950, 260.
12 Fitchen,
s15K. D. Alexander, R. Mark, and J. F. Abel, "The Structural Behavior of
181.
13A. K. Porter, The Construction of Lombard and Gothic Vaults, New
Haven, 1911, 29.
MedievalVaulting,"Journalof the Societyof ArchitecturalHistorians,
xxxvI, 1977, 241-251. These studies are summarized in Mark, 102ff.
THE TECHNOLOGY
ing whose longitudinal thrust acts in the opposite direction to that of its neighbor - in effect, the two bays of
vaulting "lean" against each other so that the clerestory
wall does not need to provide any additional support.
From this brief description of the vault supporting
mechanisms, it is evident that the skeletal form of the
developed Gothic church can readily support any
reasonable form of vaulting, sexpartite or quadripartite. A
different condition is present, however, during the construction of the vaults which one may assume was carried
out, one bay at a time, on movable centering.'6 Since the
erection of the vaulting was necessarily preceded by the
erection of the piers, walls, and flying buttresses, the vertical weight and the outward, horizontal thrusts of the
vault bay after the centering is removed were resisted by
the same structural elements as in the finished church. The
longitudinal thrust, though, must at this stage of construction have been supported by the clerestory wall since there
was not yet in place an adjacent vaulting bay to provide
stabilization. And as the springing of the vaults was
carried further upward from the base of the clerestory in
later buildings, coping with this thrust became a more
critical problem during construction.
A qualitative estimate of the relative magnitudes of the
troublesome longitudinal thrust can be determined from
comparing the plans of equivalent sexpartite and
quadripartite vaults (Fig. 5). If it is assumed that the
focused forces at the edges of the bays are directed along
groins and that the outward thrusts are approximately the
same from both types of vaulting, the longitudinal thrust
of the sexpartite vault as illustrated in the vector construction in Figure 5 is considerably larger than that of the
quadripartite vault. In effect, the size of this vector is
related to the angle between the diagonal and transverse
ribs - and the angle of the diagonal within the sexpartite,
square plan is almost twice as great as that between the
diagonal and the transverse rib of the quadripartite, rectangular plan. Indeed, even without the modern concept of
vector force components, the observation of the orientations of the groins in both types of vaulting could well
have provided the key to the Gothic builders to set out in a
new direction.
The computer models allow us now to put these
assumptions aside and to quantify our observations. The
model of the Bourges sexpartite vaulting indicated a
longitudinal thrust of 42,000 pounds (more than the
weight of a twenty-ton truck!); for the slightly larger bays
of Cologne Cathedral's quadripartite vaulting, the
longitudinal thrust was found to be 20,000 pounds, a
16 Fitchen, 171ff. Fitchen describes centering in detail, but he does not address the issue of longitudinal thrust and he does not show the additional
longitudinal bracing that would have been required to support each vault
bay as the centering was advanced.
17 Mark, 117. Note that the outward-acting thrust components are given
at an edge of one bay only; in the completed building the edge thrusts of
two adjacent bays act on the wall, doubling the total lateral thrust into the
buttressing.
OF HIGH GOTHIC VAULTS
583
thrust
longitudinal
outwQUAORIPARTITE
QUADRIPARTITE
SEXPARTITE
SEXPARTITE
5 Comparativeplans of quadripartiteand sexpartitevaulting
showing longitudinaland outwardthrust components
fifty-two percent force reduction compared with sexpartite vaulting. The magnitude of these thrusts might be better appreciated when they are compared with the
calculated, outward-acting thrusts of the vaults: 31,000
pounds for Bourges, and 35,000 pounds for Cologne and also compared with the elaborate buttressing systems
used to contain the outward thrusts.'7 The constructional
problems presented by the intensity of these longitudinal
thrusts does not appear to have been crucial in the early
Gothic churches as the vault springing could be anchored
in the typically massive wall below the clerestory. The
countering of this force only became an acute problem
with the demand for greater clerestory height and the accompanying fenestration which brought an end to the
practice of having the vault spring from a solid wall. In effect, the crux of our argument is that the Gothic builders
were pressed to select a vaulting system with considerably
less longitudinal thrust.'"
The new constructional theory also explains the apparent paradox concerning the use of a heavier vault
system in High Gothic construction. The somewhat
greater outward thrust of the quadripartite vaulting would
cause no special difficulty as it could be easily accommodated by the system of flying buttresses erected on the
exterior of the church prior to the vaulting. On the other
hand, countering the longitudinal force of sexpartite
vaulting during construction would have presented considerable problems. Thin clerestory walls could have displayed cracking, even if braced with temporary timber
falsework, and this might well have convinced the Gothic
builders that the walls were unable to resist forces of such
magnitude.'9 As with the flying buttresses, quadripartite
vaulting was probably adopted in High Gothic churches
for structural reasons: the flying buttresses provided sup-
18 This
change in configuration also helps to explain the adoption of the
tas-de-charge in High Gothic vault construction. See Fitchen, 136ff.
19 In particular, tensile cracking of the newly set, weak lime mortar between the ashlar would have been observed by the builders. Successive
modifications made to the structure to prevent such cracking could
therefore have been an important source of medieval structural innovation. See Mark, 30ff.
584
THE ART BULLETIN DECEMBER 1982 VOLUME LXIV NUMBER 4
port for the outward thrusts of the high vaults while the
quadripartite vaulting reduced longitudinal thrusts to a
manageable degree during construction.
Corroborative Evidence
Support for the new theory is evident in the manner in
which sexpartite vaults were deployed. In every major
Gothic church possessing square-planned, sexpartite
vaulting, the vaults spring from a solid section of wall
below the clerestory. Representative examples might include the cathedrals of Sens (begun ca. 1140), Laon (ca.
1160), Paris (ca. 1160), and Bourges (1195). All display
what may, in light of the new theory, be interpreted as a
disadvantage of sexpartite vaulting. Vaults springing from
below the clerestory partially obscure light from the windows and serve to limit the height of the clerestory wall.
Sens Cathedral, often cited as the earliest entirely
Gothic cathedral, is sexpartite vaulted in both the choir
and nave. Sections of the vaulting closest to the nave wall
were heightened in 1230 to allow larger windows and
hence more light to enter the church. At the same time the
clerestory windows in the choir were heightened and
enlarged and later, after 1310, the windows in the nave
were also enlarged.20 Yet, even after these restorations, the
sexpartite vaults spring from a solid section of wall below
the base of the clerestory (Fig. 6) and it is evident that the
vaults occlude the view of the clerestory windows from
vantage points within the church. Sens is large in comparison to most of the other early Gothic churches (eightyone feet high, almost fifty feet wide); nevertheless its
squat proportions, heavy walls, and conservatively
fenestrated clerestory give the interior a dark, massive
aspect reminiscent of earlier Romanesque buildings.21
Laon Cathedral, like Sens, has square sexpartite bays in
the nave and choir, but the interior of Laon is lighter and
more vertical (seventy-nine feet high, thirty-five feet
wide). The increase in light is owed to the slightly larger
clerestory windows and to the additional light which enters the church through the gallery windows and the
crossing tower.22 As at Sens, the sexpartite vaults spring
from a solid section of wall below the window opening
(Fig. 7). The vaults at Laon are narrower and more steeply
shaped than the vaults at Sens, yet the vault ribs still
obscure light from the clerestory.
At Notre Dame, Paris, which is substantially higher (at
108 feet) and more skeletal in structure than Sens or Laon,
the sexpartite vaults were originally anchored in a section
of solid wall above the oculi and below the clerestory windows (Fig. 8). During the third decade of the thirteenth
century, a more advanced system of flying buttresses was
Stoddard, 113.
The dimensions of the nave of Sens were taken from Stoddard.
22Ibid., 130.
erected and the clerestory windows were enlarged so that
they extended below the springing of the vaults.23 This
modification of the vault placement would not have
presented a constructional problem of the type that we
have discussed since the longitudinal thrust from each
vault bay was already stabilized in the original construction by the thrust of adjacent bays.
The cathedral of Bourges offers the most convincing
evidence for the new theory. Bourges was the largest (the
height from the nave floor to the keystone of the high
vaults is 120 feet) and last of the great Gothic churches
with sexpartite vaulting. As with the other sexpartite
vaulted churches, the vault ribs spring from the base of
the clerestory; but at Bourges, the problem of longitudinal
thrust is further evidenced by the presence of metal,
longitudinal reinforcement. Heavy iron chains, some four
feet in length, are placed in the upper triforium just below
the clerestory in the choir (Fig. 9). Robert Branner observed this reinforcement, but he did not offer an explanation for its presence in the wall.24Since it provides no additional bracing to the masonry in the vertical or lateral
directions, the only reasonable explanation for its placement is that it was intended to help contain the
longitudinal thrust of the vaults during construction.
Branner also noted that the nave of Bourges, constructed
in a later thirteenth-century campaign, had considerably
thicker walls without the metal reinforcement, apparently
compensating for its absence.25
The fact that quadripartite vaulting exhibits less than
half the longitudinal thrust of sexpartite vaulting must
have been understood by the master who raised the vaults
at Chartres. Its rectangular quadripartite vaults, springing
from a point above the base of the clerestory, allowed a
dramatic increase in the height of the clerestory as well as
in the potential for light to illuminate the interior of the
cathedral. The entire clerestory at Chartres is over fortyfive feet high - about the same height as the nave arcade
- and the lancet clerestory windows reached an unprecedented height of more than twenty-two feet. The
vaults spring from about fourteen feet above the base of
the clerestory (Figure 1), a remarkable departure from
earlier designs.26
Our observation that it was impractical to have sexpartite vaulting spring from a point above the base of the
clerestory also sheds new light on the long-discussed comparison of the designs of Chartres and Bourges. The construction
of the two cathedrals began almost
simultaneously and, to some modern commentators at
least, it appeared that the imposing size and beauty of
Bourges, as well as the relative economy of its construc-
20
25 Ibid., 129.
21
26 As
23 Ibid., 138.
24 Branner, 82-83.
a comparison, the clerestory windows at Bourges are only about fifteen feet high and the height of the clerestory is about thirty-three feet.
The dimensions of the clerestories of Bourges and Chartres and those of
Reims, Amiens, and Beauvais were calculated by measuring drawings
reproduced in A. de Baudot, Les cathhdrales de France, Paris, 1905.
THE TECHNOLOGY
OF HIGH GOTHIC VAULTS
585
6 Sens Cathedral, detail of upper choir wall (photo: J. Austin)
7 Laon Cathedral, choir wall (photo: J. Austin)
8 Notre-Dame, Paris, reconstruction of the transept wall (photo: W.
Clark)
9 Bourges Cathedral, sketch of iron
reinforcement below a clerestory
window in the choir (from Branner,
La cathkdrale de Bourges)
586
THE ART BULLETIN DECEMBER 1982 VOLUME LXIV NUMBER 4
10 Reims Cathedral,choir wall. (photo: J. Austin)
tion, could have prevailed over Chartres.27 It seemed
possible that Bourges rather than Chartres might have
become the model for the great High Gothic churches that
followed. The main reason for the ascendancy of Chartres,
according to Branner, is that it was imitable: its design
could be reordered to suit almost any site, whereas the
Bourges scheme could only be adopted whole.28 Now, a
corollary may be appended: the supremacy of the Chartres
model was assured also because the sexpartite vaults of
Bourges would not allow the clerestory to be enlarged on the other hand, the flexibility of the Chartres design,
particularly of its quadripartite vaulting, could well satisfy
the Gothic requirements for additional height and light.
The masters of the cathedrals of Reims (begun 1210),
Amiens (1220), and Beauvais (1225) took advantage of the
quadripartite design to raise the vaulting to progressively
greater limits.29 At Reims (Fig. 10) the height of the
clerestory windows is more than thirty feet and the vaults
spring from a point about twelve feet above the base of the
clerestory. With the further reduction of the walls, the
clerestory windows at Reims are appreciably larger in area
than the windows at Chartres. The dissolution of the
clerestory wall continued at Amiens where the clerestory
attains the remarkable height of some fifty feet. The vault
springing is more than nineteen feet above the base of the
27R. Mark,"The StructuralAnalysisof GothicCathedrals:Bourgesvs.
Chartres," Scientific American,
28 Branner, 168-170.
ccxxvxI,
93.
11 Amiens Cathedral,east wall of the north transept(photo:J.
Austin)
clerestory, seven feet higher than at Reims. The original
quadripartite vaults of the choir of Beauvais Cathedral
probably sprang from the same point as the present sexpartite vaults, about twenty-eight feet above the base of
the clerestory (more than the height of an entire clerestory
window of Chartres), and the clerestory at Beauvais
reaches almost sixty feet in height. With the introduction
of the glazed triforium and the integration of the
clerestoryand triforium in the eastern walls of the transept
and choir of Amiens (Fig. 11) and in the choir of Beauvais,
the attenuation and fenestration of the upper wall were
extended further downward, to the base of the triforium.
The high vaults of these two giants, in effect, sprang from
even more remarkable heights above a solid base. Furthermore, since this modification of the High Gothic wall
elevation is generally taken as being prophetic of later
development in Rayonnant and Flamboyant architecture,30 the structural innovations that led to the glazed
triforium, flying buttresses, and the use of quadripartite
29 Stoddard, 197, 211, 235.
3o Ibid., 218, 239, 279.
vaulting can also be seen as constituent elements leading
to new architectural styles.
The problem of longitudinal thrust and the quest for
greater window size leading to increased light, already a
concern in the twelfth century, must have hastened the
obsolescence of the sexpartite system. The shift to raised
quadripartite vaulting then was an essential technical
development in the evolution of High Gothic design. Its
adoption allowed the structure of the mature churches to
become more truly skeletal. And although the change had
important stylistic implications, it also provides additional
evidence of the Gothic designer's understanding of the
disposition of structural forces during the construction of
the giant buildings - a subject that has only begun to
receive the attention that it deserves in the study of Gothic
architecture.
Princeton University
Princeton, NJ 08540
Bibliography
Branner, R., La cathidrale de Bourges et sa place dans l'architecture
gothique, Paris, 1962.
Fitchen, J., The Construction of Gothic Cathedrals: A Study of Medieval
Vault Erection, Oxford, 1961.
Mark, R., Experiments in Gothic Structure, Cambridge, Mass., 1982.
Moore, C., Development and Character of Gothic Architecture, London,
1899.
Stoddard, W., Art and Architecture in Medieval France, New York, 1972.
Viollet-le-Duc, E., Rational Building, trans. G. Huss, New York, 1895.
SomeManuscriptSourcesfor the Playing-CardMaster'sNumberCards
MarthaWolff
The very few surviving engravings from the first half of
the fifteenth century make it clear that the Master of the
Playing Cards was the most accomplished and influential
engraver during this early period. His playing cards in
particular are one of the earliest instances of the
widespread use of engravings as patterns for artists working in other media. Motifs from these engravings, especially the flowers, animals, birds, and wild men used as
suit symbols on the number cards, crop up in the remainder of the fifteenth century in a wide range of media. As
the new medium of engraving seems to have functioned
here as an important source for artists who needed
decorative designs, the nature of the Playing-Card
Master's own sources and his adaptation of them are of
particular interest.
So far, attention has been focused on the presumed
copies of the Playing-Card Master's motifs.1 Recently
Anne van Buren and Sheila Edmunds showed that the
Playing-Card Master's suit-point motifs appeared in
several centers of manuscript illumination shortly after
1440. They argued persuasively that the context of these
designs within the individual manuscripts and their wide
geographic distribution were evidence that the engravings
were the models for the illuminators.2 Building upon their
work, I would like to indicate some of the sources, in
earlier illuminated manuscripts, for the Playing-Card
Master's number cards, and to suggest that the engraver's
own dependence on earlier manuscript models provides
important new evidence of the role of the engraving
medium as an intermediary. It can be demonstrated that
The following paper is a revision of a chapter from my dissertation, "The
Master of the Playing Cards: An Early Engraver and His Relationship to
Traditional Media," Yale University, 1979. Part of it was presented at the
"Pen to Press" symposium at Johns Hopkins University in 1979. I am especially grateful to Egbert Haverkamp-Begemann for his guidance. I
would also like to thank Alan Shestack, James Marrow, Sandra
Hindman, and John Hand for their comments and support.
1 The surviving impressions of the playing cards are fully catalogued and
illustrated by Geisberg; see also Lehrs, I, 1908, 97ff., Nos. 42-106. Anne
van Buren and Sheila Edmunds made an important contribution to our
knowledge of the editions of the series, van Buren and Edmunds, 12-30.
point motifs and rarer copies after the figure cards. Interest in the appearance of suit-point motifs in manuscripts increased with the discovery of the many correspondences between the cards and the border
decoration in the Giant Bible in the Library of Congress, written in
1452/53; see Miner, passim. and H. Lehmann-Haupt, "Gutenberg und
der Meister der Spielkarten," Gutenberg-Jahrbuch, Mainz, 1962, 360379, and Gutenberg and the Master of the Playing Cards, New Haven,
1966. The frequent appearance of playing-card motifs in Mainz books,
both manuscripts and Gutenberg Bibles, led Lehmann-Haupt to the
highly speculative hypothesis that the copper plates were engraved under Gutenberg's supervision as part of his effort to reproduce manuscript
text and illumination by mechanical means.
2
Lehrs, I, 1908, 142ff., gives an extensive list of copies after the suit-