University of Virginia Library

II. 3

HISTORICAL EVALUATION

II.3.1

SALIENT FEATURES OF THE CHURCH

The salient features of the Church of the Plan of St. Gall
are: its extraordinary length of 300 feet; its extended eastern
altar space (fore choir); its elaborate system of crypts
giving access to the relics of the patron saint without
encroaching upon the space required by the monks; its
disengaged crossing; its nineteen altars; its western counter
apse; its two semicircular atria; its detached towers; and
its square schematism.

II.3.2

LENGTH OF THE CHURCH

The Church of the Plan of St. Gall is chronologically not
the first monastic church of this order of magnitude but
probably the third or fourth. The earliest was the Abbey
Church of Fulda, in the form which it obtained under
Abbot Ratger between 802-817 (fig. 138). It had a clear
inner length of 98.00 meters (321 modern English feet).[211]
The second was probably, although not demonstrably so,
the monastery church of St. Peter's and St. Mary's in
Cologne (fig. 139), founded by Bishop Hildebold (d. 819),
which measured 91.20 meters internally from apse to apse
(300 Carolingian feet, calculated at 1 foot = 30.04cm.)[212]
The third was the original church of the Plan of St. Gall,
as rendered in figure 140 (prototype plan made in 817;
copy for Abbot Gozbert between 820 and 830). The fourth,
if Groszmann's analysis of this building is correct, was the
Abbey Church of Hersfeld, built between 831 and 850.
Together with its west-work, it measured 102.85 meters
(339 modern English feet).[213]

Abbot Ratger's church at Fulda (fig. 138) was a T-shaped
basilica with a continuous transept. The particulars
of its design leave no doubt that it was modeled after the
Church of Old St. Peter's in Rome (fig. 141). Like that
church, its clerestory walls were supported by two rows of
columns which were surmounted not by arches, but by a
straight entablature; also like St. Peter's, the ends of the
transept arms were separated from the principal body of
the transept hall.

The ideological reasons for this emulation of the design
and size of the great Early Christian proto-basilica of Rome
during the reign of Emperor Charlemagne have been
brilliantly analyzed by Richard Krautheimer.[214] The design
was an outgrowth of the general process of Romanization
of the Frankish Church and the Frankish kingdom that
started with the anointment of Pepin and his sons by Pope
Stephen II in 753 and culminated in the coronation of
Charlemagne as emperor on Christmas Eve of the year 800.

[ILLUSTRATION]

142. EPHESOS. FIRST CHURCH OF ST. JOHN (CA. 450)

[after Keil, 1932, fig. 47]

Built over the tomb of John, Bishop of Ephesos (traditionally identified as John the Apostle), this is the largest Early Christian church of
Latin cross plan. It is not known whether the church was so planned, or if it acquired its form by aggregation of four basilicas built separately
in successive stages against the martyrion of St. John.

[ILLUSTRATION]

143. BETHLEHEM. CHURCH OF THE NATIVITY (END, 5TH CENT.)

[after REALLEXIKON ZUR BYZANTINISCHEN KUNST, I, 1966, cols. 603-604, fig. 2]

The most accomplished Early Christian Latin cross church, long dated to the reign of Justinian, is perplexing for its numerous incipient
medievalisms: a nave as wide as the transept and twice the width of each aisle, and, remarkably, a fore choir repeating the dimensions of the
crossing unit—features wholly uncharacteristic for the mainstream of Early Christian architecture. As in Ephesos the eastward extension of the
nave may have resulted from special circumstances, i.e., the desire to encompass in the new church the remains of an earlier sanctuary.

[ILLUSTRATION]

145. THASOS, MACEDONIA

CRUCIFORM CHURCH, 6TH CENT. PLAN

[redrawn after Orlandos, I, 1952, 193, fig. 157, 1]

The intersection of nave and transept masses lends to these cruciform
Early Christian churches an almost medieval appearance. But they are
not designed to any modular concept (cf. p. 208ff).

[ILLUSTRATION]

144. THASOS, MACEDONIA

CRUCIFORM CHURCH, 6TH CENT. PLAN

[redrawn after Orlandos, 1, 1952, 189, fig. 155]

The intersecting nave and transept established before the apse a square
for both altar and benches. The latter, continuing in the apse to form a
synthronon, became standard layout in monastic churches (cf. above
p. 141 and fig. 94).

The ties of the Abbey of Fulda with Rome had been
especially strong. The missionary work of its founder, St.
Boniface (680-754), was closely linked to the papal see.
His successor, Abbot Sturmi (744-769), was an ardent
student of the customs of Monte Cassino on which the
customs of Fulda were based, and Fulda was the first
German abbey to be placed under the direct jurisdiction
of the Roman see.[215] There is no doubt that the return to the
design of the great western Roman basilicas of Constantine
the Great and of Pope Sylvester was an expression of the
renovation by Charlemagne of the universal Christian
empire inaugurated by Constantine the Great. One might
justly conclude that the propensity for colossal dimensions,
embodied in the abbey Churches of Fulda, Cologne, and
the Church of the Plan of St. Gall, was an integral part of
this ideology; but to explain the dimensional boldness of
these churches exclusively in such symbolic terms would
be a gross historical simplification. There are other more
functional and more specifically monastic reasons for the
appearance in transalpine Europe of churches of unprecedented
dimensions. One of them was the need to extend
the altar space in order to accommodate, in addition to the
officiating clergy, an entire community of monks celebrating
the divine services jointly in an elaborate ritual
involving chant and counter chant. Another reason was
the transfer of baptismal rites from a separate subsidiary
building to the basilica; in the Church of the Plan this
function claims one third of the entire nave. A third reason
was that the rapidly increasing veneration of saints resulted
in a multiplication of altars, each requiring additional space.
There also developed the desire to accommodate in a single
oratory a variety of cults that in earlier monastic churches
has been distributed over an entire family of buildings.[216]

But the dimensional enlargement of the church that
these demands generated raised serious economic problems.
Whatever the historical and functional incentives may have
been for building churches of a magnitude of 300 feet and
more, there still remained the question of whether a community
of an average of 100 to 200 monks could afford to
build and maintain such structures. Ratger, the Abbot of
Fulda, thought so. But his monks, who paid for his
ambition with their toil and sweat, were disturbed by his
building program to the point of rebellion. In a formal
petition presented to Charlemagne in 812, they pleaded
that the construction of these "oversized and superfluous
buildings" (aedificia immensa atque superflua) be brought to
a halt or reduced to a normal pace, because it taxed the
brothers beyond endurance, left no time for the lectio
divina, and threatened to exhaust the monastery's economic
resources.[217] The petitioners returned, defeated, to the

[ILLUSTRATION]

146. SALONA, DALMATIA

CRUCIFORM CHURCH, 6TH CENT. EGGER'S RECONSTRUCTION

[redrawn after Orlandos, I, 1952, fig. 4]

This westernmost of cruciform churches of the Thasos type (figs. 144
and 145) is so similar in design to the latter, it could be said that "an
attempted reconstruction of either church is applicable to the other"
(Hodinott, 1963, 181). Of both, only foundations remain.

[ILLUSTRATION]

147. KORNELISMÜNSTER, Inden, Nordrhein-Westfalen

ABBEY CHURCH OF THE SAVIOUR. PLAN

[after Hugot, 1965, 411]

Founded by Louis the Pious for Benedict of Aniane as a model
monastery for thirty monks, according to good contemporary sources
Kornelismünster was intended to be Louis's burial place. It was
consecrated in 817, a few days before the second synod opened at
Aachen. For a full bibliographical account see Vorromanische
Kirchenbauten, 1966-71, 160ff.

monastery: Charlemagne denied their petition. Hildebold,
then arch-chaplain and one of the emperor's closest
advisors, may have had a voice in the negative decision.

By 817, however, the climate had changed. Louis the
Pious was now emperor; sometime between 816 and 817
he received the same delegation with the same petition,
which this time was received favorably. As a direct result
of the petition, Ratger was deposed in favor of Eigil,
leader of the dissenting monks of Fulda. When Eigil was
installed as the new abbot in 817, he was admonished by
Louis "to stop this superfluous work of erecting structures
of inordinate size and to reduce the monastery's building
program to normal proportions."[218] It appears that Louis
made use of the words the monks themselves had spoken,
the first time before Charlemagne and the second time
before him.

Overindulgence in costly building activities was not the
only reason for Ratger's fall, and by itself might not have
brought it about. He was also accused of violations of
sanctioned monastic customs,[219] but the incident shows that
constructing a church 300 feet long was by no means an
easy matter for a monastic polity and could have disturbing
consequences not only for its economic stability but also
for its spiritual health.

The rebellion of the monks of Fulda against the building
activities of their abbot is the strongest historical evidence
to be offered in support of Boeckelmann's theory that the
explanatory title which stipulates a length of 200 feet for
the Church of the Plan is the expression of a programmatic
retrenchment.[220] This measure might have been
directly related to the struggles of Fulda.

II.3.3

EXTENDED EASTERN ALTAR SPACE
(FORE CHOIR)

When the fore choir was introduced between the transept
and the eastern apse of the church, the T-shaped plan of
the Early Christian basilica was transformed into a Latin-cross
plan (crux capitata). The origin and dissemination of
this feature forms one of the most fascinating chapters in
the history of medieval architecture.[221] Contrary to Georg
Dehio's belief two generations ago, the Latin-cross plan
is not a Carolingian invention. It came into use early in
the fifth century[222] as a fusion of the longitudinal basilica
and the cruciform central plan of buildings traditionally
associated with Christian martyria.[223] The cruciform plan
is well established in such buildings as the first church of
St. John in Ephesus, built in the fourth to fifth century
(fig. 142);[224] the Nativity Church in Bethlehem, built by
Justinian in the sixth century (fig. 143);[225] the cruciform
basilicas of Thasos in Macedonia (figs. 144 and 145)[226] and of
Salona in Dalmatia (fig. 146).[227]

In Merovingian France the form appears as early as 577,
when the Greek-cross plan church of Ste.-Croix-et-St.Vincent
at Paris (completed by King Childebert in 558)

[ILLUSTRATION]

CRUCIFORM NON-AISLED CHURCHES WITH
DISENGAGED CROSSING

148.A TOMARZA, CAPPADOCIA. PLAN, 5TH-6TH CENT.

148.B HALVADERE, CAPPADOCIA. PLAN, 5TH-6TH CENT.

148.C PFALZEL, NEAR TRIER. PLAN, ABBEY CHURCH OF ADALA
(BEFORE 715)

The impetus to development of the Carolingian modular Latin cross plan
by small cruciform churches such as Tomarza and Halvadere and their
western derivatives or counterparts, such as Pfalzel, should not be overvalued.
Yet, the undeniable ubiquity of these small buildings spread
throughout the whole Christian world the concept of an arch-framed
—and generally tower-surmounted—crossing created by the intersection
at 90° of two volumes of space, each identical in height and width. Like
that of the quincunx church, this building type had distinct modular
implications, another feature making it attractive to the northern mind.

148.X. Sivri Hissar stands somewhat aside from the other Cappadocian churches. An
aisle added to the northern flank of the nave perhaps served as a sepulchral chapel for
St. Gregory of Nyssa (331-ca. 396) who owned large estates in this part of the world.

was transformed into a Latin cross by the addition of
aisles.[228] The new form thus created was subsequently
copied in several other Neustrian churches, most notably,
perhaps, in the church of Corbie.[229]

Dehio was of the opinion that the Latin-cross plan owed
its rise to practical considerations, namely the need for
more choir space for the worshiping monks. Graf stressed
the commemorative, funerary significance of the centralized
cruciform development of the eastern end of the church.
In a recent review of this controversy George H. Forsyth
has pointed out that these two theories need not preclude
each other and that a vast body of new material, made
available since the time Dehio and Graf discussed these
problems, tends in fact to corroborate both opinions.[230]

In the historical evaluation of this important architectural
motif, a sharp distinction must be made between its origin
and occasional appearance in Early Christian times and its
prevalence everywhere during the Carolingian period.
Practical considerations must have played a decisive role
in its adoption at the time of Charlemagne. The cult of
relics, which had introduced into the church a multiplicity
of altars, made it impossible for the service of the high
altar to expand into the nave or the aisles of the church.
Few monasteries had fewer than 100 monks, and some had
as many as 300 or 400. Without the insertion of a fore choir

[ILLUSTRATION]

149.B

[ILLUSTRATION]

149.A GERASA (JERASH), Palestine

CHURCH OF PROPHETS, APOSTLES & MARTYRS

[after Kraeling, 1938, pl. XLI and XLII]

[ILLUSTRATION]

150. MOUSMIEH, SYRIA

PRAETORIUM (or TEMPLE?) ca. 180

[after DeVoguë, I, 1865, 46, fig. 11.]

between transept and apse, there would have been insufficient
space for the monks participating in the service. At
the same time it cannot be denied that because of its
association with the relics of the Patron Saint of the church,
the high altar had acquired an intrinsically funerary significance—another
historical incentive for the absorption in
the basilican scheme of the cruciform arrangement of the
centralized paleochristian martyria. Lastly, it is also quite
clear that the Carolingian architects who struggled with
the development of the Latin-cross plan could hardly have
been blind to the exciting aesthetic implications of a
fusion between the basilican and the central plan.

Churches with extended altar space preceding the Plan
of St. Gall, as I have already pointed out in my discussion
of Reinhardt's reconstruction of the Church of the Plan,[231] of
St. Gall are the Saviour's Church of Neustadt-on-the-Main,
shortly after 768/69 (fig. 133); the abbey church of St.
Riquier (Centula), between 790-99 (fig. 135); the Carolingian
Cathedral of Cologne, between 800 and 819 (fig.
139); and the abbey church of St. Mary at Reichenau-Mittelzell
(fig. 134), built by Abbot Haito between 806 and
816. Even the church of the model monastery of Inden,
built by Emperor Louis the Pious between 815 and 816 for
Benedict of Aniane and his chosen community of only
thirty monks, had a rectangular space inserted between
transept and apse (fig. 147).[232]

The innovative aesthetic significance of this motif lies
not so much in the addition of the space as such, but in the
modular alliance into which it enters with the crossing
square, the transept arms, and by extension, although at a
slower rate of development, with the square division of the
nave of the church.

II.3.4

DISENGAGED CROSSING

The crossing of the Church of the Plan is completely
separated at floor level from the contiguous spaces by rails
and choir screens (fig. 99); and if it was meant to be
framed by boundary arches at the top, as I believe it was
(figs. 107-110),[233] the crossing would be the only place
where a principal constituent part of the church was segregated
from the rest of the latter's space by dividing arches
reaching up to roof level. No other single architectural
feature had such deep and far-reaching reverberations for
the future.

Like the fore choir, the disengaged crossing did not
originate during the Middle Ages. Crossings framed by
boundary arches, which separated the intersected area from
the nave and the transept, occur in certain cruciform, non-basilican
churches of Cappadocia, such as Tomarza,
Halvadere (fig. 148.A-B) and Sivri Hissar, as early as the
fifth and sixth centuries, as Samuel Guyer has pointed out.[234]
That the type was known in Western Europe is attested by
the Abbey Church of Adala in Pfalzel near Trier, before
715 (fig. 148.C).[235] There is a faint possibility that some

[ILLUSTRATION]

VAGHARSHAPAT (ECHMIADZIN), Armenia

151.A

151.B

CHURCH OF ST. GAYANÉ

[after Arutiunan and Safrian, 1951, fig. 20]

Erected by the Catholicos Ezra between 630 and 646.

[ILLUSTRATION]

BANDE, Orense, south Galicia, Spain

151.Xa

151.Xd

151.Xb

CHURCH OF SANTA COMBA (7th Cent.)

A jewel of Visigothic architecture, small, built in cyclopic masonry, this church attests
the early adoption of the Early Christian quincunx church by the Germanic conquerors
of Spain, and their fascination with cellular space division and the concept of an arch-framed
and tower-surmounted crossing that was to become a key feature of Carolingian
architecture.

151.Xc

rudimentary forms of the disengaged crossing (but not, so
far as I can judge, a fully developed archetype) appear in
certain Early Christian basilicas with tripartite transepts:
in Greece, such as Basilicas A and B in Nikopolis, and the
basilicas of Epidaurus, and Lokris. But it is well to reserve
final judgment until such time as Guyer's schematic reconstruction
of churches of this type with intersecting naves
and transepts surmounted by crossing towers is corroborated
by more tangible archaeological evidence than is
offered in his own study.[236]

If Kraeling's and Egger's reconstructions of them are
correct, the two finest examples of Early Christian churches
with fully developed crossings are the churches of the Holy
Apostles and Martyrs of Gerasa, 464-465 (fig. 149), and
the cruciform church at Salona.[237] However, no less important
for the development of the arch-framed crossing in
medieval architecture may have been such Armenian
central plan churches as St. Gayané at Vagharshapat
(Echmiadzin), 630 (fig. 151) as well as a small but no longer
tracable group of near-eastern quincunx churches, which
one must postulate historically to have formed the connecting
link between such semisacral cross-in-square
buildings as the praetorium (or temple) in the Roman camp
of Mousmieh, Syria, ca. A.D. 180 (fig. 150), an audience hall
of like design outside of the walls of Rusafa, Syria, dating
from about 560, the minute, yet arrestingly beautiful
seventh-century church of Santa Comba de Bande, Spain
(fig. 151.X) and the sophisticated Germigny-des-Prés built
by Theodulf of Orleans, councillor and missus dominicus of
Charlemagne, between 806-810 at his summer residence
on the Loire, near the abbey of St. Benoit-sur-Loire (fig.
152).

Yet the true Early Christian prototype for the Carolingian
disengaged crossing may be found closer to home in
the north basilica of Trier, which Gratian, having chosen

[ILLUSTRATION]

151.Y TRIER, GERMANY

154.A*

NORTH BASILICA

[isometric reconstruction, redrawn from Krautheimer, 1965, 61, fig. 23]

In the conceptual interaction of quincunx and basilica a medieval style of architecture emerges. The north basilica lacks only the ordering principle
of modularity for its full development. Product of two building campaigns (basilica, shortly after 325; quincunx, 380) the result in Trier is a composition
of bold and unusual grouping of masses that must have strongly influenced development of the Carolingian Latin cross church with disengaged
and tower-surmounted crossing.

[ILLUSTRATION]

GERMIGNY-DES-PRÉS

152.A

152.B

152.C

CHURCH OF THEODULF OF ORLEANS, 806-810

[plans after Viellard and Troiekouroff, 1965, 356, fig. 111]

Small, yet of magnificently controlled proportions, the spaces of Germigny-des-Prés step in bold progression to the slender, steeply rising nave and
transept, from whose intersection rises a tower of great elan. This church attests that the quincunx, even in the 9th century, had life enough to capture
the imagination of a great Carolingian churchman. Theodulf was a Visigoth and his acquaintance with such churches as Santa Comba de
Bande (fig. 151.X) may have influenced his choice.

to reside in Trier, rebuilt in 380. In its final Early Christian
form (fig. 151.Y) this building terminated in the east in
what can only be called a grandiose variant of a quincunx
church, consisting of a large, central, tower-surmounted
bay surrounded by eight smaller two-storied spaces: four in
the cross, four in the corners, the latter appearing externally,
like towers. Much masonry of this structure survives
today, incorporated in the fabric of the Romanesque church
which superseded it in the eleventh century. But in Charlemagne's
time it still stood there in its original form for all
his subjects to see, located about seventy-five miles as the
crow flies from the emperor's residence at Aachen.[238]

As one surveys all of this material, it must be stressed
again that sharp distinction must be made between the
occasional appearance of these motifs in the relatively
isolated topographical contexts of Armenia and Spain and
their formal and systematic combination with the basilican
scheme as attested by the sequence discussed in the
preceding chapter: St. Denis, 750-755; Neustadt-on-the-Main,
after 768/69 (figs. 116 and 133); Reichenau-Mittelzell,
806-816 (fig. 134); Cologne, 800-819 (fig. 139); in
some respects even the church of St. Benedict of Aniane at
Inden, 815-816 (fig. 147); and certainly, most decidedly
and programmatically, the Church of the Plan of St. Gall,
prototype plan: 816-817; copy: ca. 820 (figs. 55 and 99).
The tripartite Early Christian transept may have helped
bring about this fusion, but it had in itself none of the
exciting aesthetic and constructional implications which
made the disengaged crossing so important for the future.
Again it is difficult, if not impossible, to say precisely what
triggered the development of this particular solution. There
can be no doubt that the increase in the number of altars
affected the form of the crossing as much as it affected the
fore choir. Nor should it be overlooked that the division of
the congregation into choirs singing in antiphonal response
called for a more distinct architectural recognition of the
two constituent parts of this ritual. The Plan of St. Gall

[ILLUSTRATION]

154.B* ROME. OLD ST. PETER'S. CRYPT PLAN

[after Kirschbaum, 1959, 58, fig. 9]

In the Constantinian basilica the funerary monument marking the place of Peter's
execution stood in the chord of the apse, his altar in the transept before it. Gregory
raised the presbytery floor and beneath it made Peter's monument accessible to
pilgrims by a circular corridor crypt (the first of its kind), thus reserving the altar
space for exclusive use of clergy.

[ILLUSTRATION]

PLAN OF ST. GALL. FORE CHOIR AND ITS SYSTEM OF CRYPTS

155.A

155.B

For our interpretation of how this delineation of the
Presbytery and the Crypt of the Church of the Plan
would have to be translated into a modern architectural
drawing, with walls shown in their full thickness we
are referring to fig. 123, p. 177.

suggests that the fore choir and the apse became the
station for the majority of the monks, while the antiphonal
singing of the psalms by voices trained specifically for that
purpose took place in the crossing. There was, no doubt, a
liturgical need for spatial differentiation; the establishment
of a "house for the counter song" (this is how I think we
must interpret the term chorus psallentium) in the crossing
of the Plan served the purpose of preventing the altar
space from being crowded. Yet it would be foolish to
presume that this need alone brought about the creation of
a new architectural form.

In constructional terms the disengaged crossing established
in the most crucial area of the church a system of
bracing arches which could serve as a sound architectural
base for a superincumbent tower, and thus demonstrated
that arches rising from relatively slender piers could carry
substantial loads at impressive heights without weight-bearing
walls beneath them. This idea, once conceived,
prepared the way for the perforation of the nave walls and
their conversion from weight-bearing sheets of masonry
into a skeletal frame of structural members. Moreover, in
separating the crossing from the rest of the church by
framing arches, a space came into being which could be
used as a unit of measurement for the modular articulation
of the remaining spaces of the church. In the Church of the
Plan of St. Gall this articulation is already far advanced.

II.3.5

CRYPTS

The combination of a circumambient corridor crypt with a
vaulted inner hall crypt, as has been claimed in the preceding
chapter, is a Carolingian innovation that owes its
origin to the need for an area of devotional seclusion
enabling the monks to pray in front of the tomb of their
patron saint without intermingling with the secular visitors
to the tomb.[239] Hall crypts and corridor crypts as separate
entities are well known around 800, both north and south
of the Alps. The former is relatively rare; the latter, very
common.

The most ancient Christian hall crypt known at this date
is the small three-aisled crypt of the church of Santa Maria
in Cosmedin in Rome (fig. 153), built under Pope Adrian I
(772-795).[240] Its columns carry a straight entablature, surmounted
by a flat ceiling made of large slabs of stone. The
earliest known example of an annular corridor crypt, to the

best of my knowledge, is the semicircular passage with
which Pope Gregory the Great encircled the shrine of St.
Peter's when he raised the platform of the Presbytery above
the original pavement of the church sometime between 594
and 604 (fig. 154).[241]

Corridor crypts were ideally suited to relieve conditions
of crowding imposed by the increasing number of pilgrims
and other visitors to the shrines of saints. They permitted
the devout to come into close bodily contact with the tomb
of the saint, while affording to it a central and well-protected
position. The layout assured an easily controllable
peripheral flow of movement for laymen, leaving the central
area free to be utilized (as such areas subsequently were)
for more restricted rites. The corridor crypt became an
integral feature of Christian architecture on both sides of
the Alps from the seventh century onward, as popular
veneration of relics became widespread.

Typical examples of corridor crypts in Rome are those
of the churches of San Crisogono (built under Gregory III,
731-741); San Marco (built under Adrian I, 772-795);
and San Stefano degli Abissini (built under Leo III, 795816).[242]
North of the Alps they are found in the Abbey
Church of St. Emmeran at Regensburg (as early as ca. 740);
in St. Lucius at Chur (mid-eighth century); in the Abbey
Church of St.-Denis (consecrated in 775); in St.-Maurice
of Agaune (end of the eighth century); in the Abbey
Church of Werden (ca. 830); in the Abbey Church of
Vreden (consecrated in 839) and the Carolingian cathedral
of Hildesheim (consecrated in 872).[243] In the ninth century
north of the Alps, the annular corridor crypt was replaced
with increasing frequency by a crank-shaped crypt, consisting
of two straight longitudinal arms connected in the
east by a straight transverse arm. The new form was
unquestionably conditioned by the introduction of the fore
choir, which made the circumambient crypt independent of
the semicircular shape of the apse, thus opening up new
possibilities in crypt design. The earliest occurrence of the
crank-shaped corridor crypt is the Plan of St. Gall (fig. 155)
and here, also for the first time, this type of crypt is found in
combination with a central hall crypt. The designer of the
Plan may very well have been the inventor of this scheme.

Other examples soon followed: St. Philibert of Grandlieu,
847-853 (fig. 156);[244] St. Germain of Auxerre, consecrated

[ILLUSTRATION]

CHURCH OF ST.-GERMAIN

157.A

157.B

Auxerre, Yonne, France

REDRAWN AFTER LAMBERT, 1960, 7, fig. 6

If the inner hall crypt with the coffin of St. Germain was accessible
from the nave, as Hubert assumes (1970, 299, fig. 345), this crypt
would have been, in essence, identical with that of the Plan.

[ILLUSTRATION]

FLAVIGNY-SUR-OZERAIN, Côte d'Or, France

158.B

158.A

CHURCH OF ST. PIERRE. CRYPT

[after Lambert-Jouven, 1960, 27, fig. 25]

The crypt, on level with the nave, was built in two stages. In final
form it consisted of an inner aisled hall and apse, which enshrined
the bodies of SS Prix and Reine. A crank-shaped corridor primarily
used by pilgrims surrounded this hall and gave east access through a
rectangular fore room to a hexagon built in honor of SS Peter and
Paul.

on January 6, 859 (fig. 157);[245] the Carolingian
cathedral of Hildesheim, built by Bishop Altfrid between
851 and 872[246] and St. Pierre in Flavigny, 864-878[247] (fig.
158). None of these crypts are identical in design, yet all
of them have in common the existence, side by side, of a
circumambient corridor crypt (crank-shaped or rounded)
and a highly developed system of inner spaces differing in
form and size from the small confessional chambers of the
preceding centuries.

In St. Philibert of Grandlieu (fig. 156) the inner crypt
consisted of a barrel-vaulted transverse shaft—located at
the base of the apse—that was intersected by three shorter
longitudinal arms. East of this system there was a larger
cross-shaped space accessible through an opening in the
apex of the apse.

At St. Germain of Auxerre (fig. 157) the hall crypt had
the shape of a small basilica with longitudinal barrel vaults
rising from straight entablatures supported by piers and
columns. This space was entered from the east by a straight
longitudinal arm and terminated toward the west in a
polygonal apse which sheltered the tomb of St. Germain.
The latter, if Louis' interpretation of the description of this
crypt by Heiric is correct, was visible from the upper church
by a fenestella.[248] The massif of masonry by which this
confessio is enclosed is 33 feet (10.00 m.) wide and 50 feet
(15.30 m.) long, and thus covers a surface area of 1,650 feet.
The clear inner space of the confessio totaled roughly 670
square feet.[249] This is only slightly less than the area of 787
square feet, assigned in our reconstruction to the confessio
of the Plan of St. Gall. But the crypt of St. Germain is
more elaborate in other respects: by the wealth of chapels
attached to its outer corridor shafts, as well as by the
sophisticated rotunda of St. Maximin at the extreme eastern
end of the system joined to the corridor crypt by an aisled
longitudinal arm two bays deep.

The inner crypt of Flavigny (fig. 158) was of the same
order of magnitude, perhaps larger, than that of St. Germain
of Auxerre. The excavation conducted by Georges
Jouven in 1960 revealed without any shadow of doubt that
the confessio extended westward from the apse for the full
length of the presbytery, with a clear inner width of 20
feet (6.00 m.) and an axial inner length of 34½ feet (10.5 m.).[250]
Its vaults, which carried the floor of the entire presbytery,
were supported by two rows of free-standing columns with
corresponding pilasters in the long walls. Because of the
unco-operative obstinacy of a private property owner only
the southern half of the crypt could so far be excavated for
its full length; but there is no reason whatsoever—as
Jouven points out convincingly—that conditions in the
unexcavated northern part of the crypt were not identical.[251]

There is no doubt in my mind that the designer of the
church of the Plan of St. Gall furnished the prototype for
this combination of an inner hall crypt with a circumambient
corridor crypt. An important link in the dissemination
of this scheme may have been the crypt of the
church of St. Remi in Reims, in which archbishop Hincmar,

[ILLUSTRATION]

159. LEPCIS MAGNA, Tripolitania

BASILICA, SEPTIMIUS SEVERUS (146-211)

[after Ward Perkins, 1965, 23, fig. 8]

One of the finest of the large-aisled Roman market halls with apses
at both ends—perhaps the most outstanding Severan period
architectural monument. It borders on the south on a great colonnaded
court. The layout is reminiscent of the Forum of Trajan, Rome
(fig. 239). For a more modest version built in the transalpine
provinces of Rome, see fig. 202. The influence this hall type may
have exerted upon the Carolingian basilica with apse and counterapse,
as well as on the concept of the medieval cloister, may be greatly
underestimated.

in 852, deposited the relics of St. Remi. Archaeologically
this crypt is terra incognita, but the description "larger and
more beautifully built" (maiori et pulchriori opere factum)
in the Vita sancti Remigii[252] suggests, as Hilde Claussen has
argued persuasively,[253] a more elaborate type of Carolingian
crypt.

II.3.6

COUNTER APSE

The counter apse is not a Carolingian invention. Basilicas
with apses at either end of the nave were in use in Early
Christian times,[254] but appear to have been confined almost
exclusively to the North African provinces of Rome, where

[ILLUSTRATION]

160. ST.-MAURICE-D'AGAUNE, Valais, Switzerland

[after Blondel, 1948, 29, fig. 5]

The plan renders the terminal form (end, 8th cent.) of a succession
of basilicas erected in honor of St. Maurice and his companions in a
monastery founded late in the 4th century under the crag where they
suffered martyrdom. Three preceeding churches, built on the same
site, were of smaller dimensions; each had only one apse.

at least eleven, perhaps twelve, churches of this kind are
known: two in Tripolitania (Lepcis Magna and Sabratha);
three in Algeria (Matifou, Orléansville, and Tipasa); and
six, perhaps seven, in Tunisia (Sbeitla, Haïdra, Henchir
Chigarnia, Iunca, Thelepte, and less well-excavated Mididi
and Henchir Goraat ez-Zid). The apse and counter apse
arrangement of these churches owes its existence to a
variety of reasons. At Lepcis Magna (fig. 159) and Sabratha
it is clearly the heritage of a pre-existing judiciary basilica
put to Christian use. Elsewhere, as at Orléansville, Matifou,
Tipasa, Sbeitla and Haïdra, a square or semicircular counter
apse was added to an earlier single apsed church to
serve as a sepulchral martyrion for a saint, whose growing
importance called for a second place of veneration within
the church. In still other places, the counter apse owed its
existence to the reorientation of an originally occidented
church, when the eastward location of the altar space
became mandatory in early Byzantine times. One cause
does not exclude the other and in some churches the reorientation
of the building coincided with the transformation
of the original apse into a funerary chapel, while the
new counter apse and the area immediately in front of it
became the site for the new high altar (as in the church of
Bishop Bellator at Sbeitla). Whether or not these North
African churches had any influence on the medieval
development is hard to say; but that much is sure, that
when the counter apse was adopted in the north and
became a traditional feature, it was in response to a sharply
rising interest in the cult of relics calling for an augmentation
of the number of stations needed for the veneration of
saints. In purely aesthetic terms one cannot entirely preclude
the possibility of influences from pagan Roman
times, even at this late stage of the adoption of the theme.
I am thinking of such double-apsed judiciary basilicas as
those on the forum of the Romano-British city of Silchester
(fig. 202) or the more recently excavated basilica of the
Gallo-Roman city of Augst in Switzerland. Basilicas of this
type must have been infinitely more numerous in the
Roman provincial territories north of the Alps than would
appear in present-day perspective and the remains of many
of them may still have been visible in Carolingian times.
Their power to influence the medieval development would
doubtlessly have been enhanced by the fact that when the
early Christian basilica entered into a symbiosis with the
concept of a large galleried cloister court, attached to one
of its long sides—as it became standard in Carolingian
times—aesthetic emphasis shifted from the longitudinal
directionalism of the early Christian basilica, to a broadside
orientation that had been an essential trait of the judiciary
basilica of pagan Rome in the first place.

In the north the apse and counter apse motif was not
employed with any consistency until the time of Charlemagne
and its introduction coincided with a renaissance of
the basilican design created for Rome by Constantine the
Great. The fusion established a norm which continued into
Ottonian times and lasted in Germany until the end of the
Romanesque period.

[ILLUSTRATION]

161. CORINTH-LECHAION

CHURCH OF ST. LEONIDAS

[after Pallas, 1962, 142, fig. 142]

The remains of this great church in the harbor
suburb of Corinth, although preserved to no more
than 2 feet above ground are, even in so ruinous a
state, one of the most impressive sights in the entire
Early Christian world, and an expression of the most
accomplished architecture it could produce. The
church dates from 450-460?; its atrium from
518-527. Including atrium and fore court, the full
length of this basilica was 610 feet (186 m. or 600
Byzantine feet).

The basilica itself (450 feet long) consists of a nave
about 60 feet wide and two aisles, a tripartite
transept and an apse. It is preceded by an
exonarthex and a narthex, the latter projecting like
a transept beyond the line of the aisle walls. Four
heavy piers in the eastern transept suggest that its
center bay was surmounted by a timber-roofed
tower—a feature which curiously enough appears at
the same time and thereafter in several Merovingian
churches: St. Martin at Tours (ca. 450), St.
Wandrille (647), and in the Carolingian church of
St. Denis, if Crosby's analysis of its foundations
(fig. 166.X) is correct.

The layout of the liturgical furniture in the BEMA
and the apse of St. Leonidas is very similar to that
in the Church of the Plan of St. Gall, consisting of
a semicircular bench (SYNTHRONON) in the apse
and two lateral benches in the BEMA, allowing the
monks to be seated on three sides around the altar
space—an arrangement that lent itself with
particular ease to monastic use.

From the BEMA of St. Leonidas a raised pathway
(SOLEA) lead to the AMBO, the pulpit from which
the bishop or his representative addresses the
congregation. To the best of my knowledge, the
SOLEA has no counterpart in Carolingian
architecture, but the AMBO is, on the Plan of St.
Gall shown in a similar position west of the
transept, in the axis of the nave of the church.

[ILLUSTRATION]

162. RAVENNA. CHURCH OF SAN VITALE

[after Encyclopedia dell' Arte Antiqua IV, Rome, 1965, 630, fig. 730]

San Vitale dates ca. 532-546. The two detached circular towers are
entered from the narthex. They give access to the gallery of the
church, an area reserved for women attending religious services.
Despite their clearly functional role, even at this early period they
may have had strong symbolic overtones as towers of the fortress of
God. At what point in history they came to be used as bell towers
is not easily ascertained (see above, pp. 129ff).

Probably because of their failure to fill practical needs, the detached
towers of the Church of the Plan of St. Gall had no consequence for
later medieval planning. Single detached towers, associated with
buildings of basilican plan may have been in use in the Exarchate of
Ravenna as early as the 8th century and became a common mark
of the Italian Romanesque and Gothic. North of the Alps, the
preferred solution was to incorporate the towers in the body of the
church—a process beginning with the invention of the Carolingian
Westwork (cf. pp. 206-208) and culminating in the medieval twin
tower façade. Even centrally planned buildings were affected by this
change, as witness the Palace Chapel at Aachen with its towers set
into an avant-corps with raised tribune from which the emperor
attended divine services (for changing stylistic concepts see caption to
Fig. 71.Z).

The oldest transalpine example known to date is the
basilica of St.-Maurice of Agaune, which dates from the
end of the eighth century (fig. 160). Then follow in
chronological order the abbey churches of Fulda, 802-819
(fig. 138); Paderborn, after 799; St. Willibrord in Echternach,
at the beginning of the ninth century; the Carolingian
cathedral of Cologne, traditionally ascribed to Archbishop
Hildebold, who died in 819 (fig. 139); St Remi, at Reims,
consecrated in 852; Auxerre Cathedral, 857-873; and the
Abbey Church at Oberzell on Reichenau, ca. 890.[255]

Liturgically, the counter apse provided a new sanctuary
for the founding saint of the monastery, who had in many
instances become more important in the ritual of the
church than its patron saint. In Fulda (fig. 138), we learn
from the Vita Eigilis the monumental west choir was added
under Abbot Ratger (802-819) to the church of Abbot
Baugulf (790/92-802) as a shrine to St. Boniface because of
the heightened veneration for the relics of the founding
saint.[256] Louis Blondel's excavations of the monastery of
St.-Maurice of Agaune (fig. 160) have shown how a new
church with a counter apse allowed the relics of saints
previously venerated in separate buildings to be housed
together in one church.[257] In churches with west-works, the
monumental western avant-corps of the church served the
same purpose.[258]

The western counter apse had the further advantage of
establishing a close liturgical tie with Rome, as the creation
of a sanctuary at the western end of the church was in
imitation of Old St. Peter's in Rome (figs. 104, 141).

Further, the adoption of this motif marked a decisive
step in the breaking away of Carolingian architecture from
the directional layout of the Early Christian basilica.
Because it was built onto what had formerly served as
the principal entrance to the church, the counter apse
completely eliminated the concept of the traditional
basilican facade. The nave had previously been a great
congregational longhouse designed to channel the worshiping
crowd toward the altar (fig. 81). With the introduction of
the counter apse, the nave became rather a connecting spatial
link between two terminal masses, both of which drew the
worshiper's attention (figs. 55, 107, 109, 111, 112). The
purpose of the nave was changed further by the railing-off

[ILLUSTRATION]

163. WERDEN CASKET (FRAGMENT). LONDON, VICTORIA AND ALBERT MUSEUM.

[by courtesy of the Victoria and Albert Museum]

A reliquary chest of carved ivory, the so-called Werden Casket, formerly believed to date to the beginning of the 5th century, was recently declared
a Carolingian copy (Beckwith, 1958, 1-11). The detail here shows Mary and Anne in the Visitation scene, and to their side the city of Judah,
represented by a building terminating in an apse and flanked by two detached circular towers.

If the Early Christian model of this carving reflects actual building practice, the ivory would bear witness to the existence in Late Antiquity of
detached circular towers flanking a church. So far, there appears to be no tangible archaeological evidence to confirm this conjecture except for the
staircase towers of the church of San Vitale, Ravenna (fig. 162).

[ILLUSTRATION]

164. PLAN OF ST. GALL. ALTAR ARRANGEMENT

                                     

1.	SS Mary and Gall
2.	Holy Cross
3.	SS John the Baptist and John the Evangelist
4.	St. Paul
5.	St. Peter
6.	SS Philip and James
7.	St. Andrew
8.	St. Benedict
9.	St. Columba
10.	St. Stephen
11.	St. Lawrence
12.	St. Martin
13.	St. Mauritius
14.	Holy Innocents
15.	St. Sebastian
16.	SS Lucia and Cecilia
17.	SS Agatha and Agnes
18.	St. Gabriel
19.	St. Michael

For a descriptive analysis of the altars and their identifying titles, see pp.
129-44. The schema shown above does not include altars in the chapels of the
Novitiate and Infirmary, whose patronage is not designated on the Plan (see
fig. 247, p. 302 and p. 311). A total of twenty-one altars is shown on the Plan.
On the number symbolism embedded in this figure and the distribution of altars
within the church see fig. 80.X, p. 124. On the importance of the layout of the
altars in reflecting and stimulating the emergence within the Church of a new
principle of spatial division distinctly different from the spatial directionalism of
Early Christian churches, see pp. 127-28 and caption to fig. 165.

of its terminal bays for the exclusive use of the monks,
leaving only the center of the nave accessible to laymen (figs.
82 and 110). This was the monastic Carolingian response
to the large congregational halls of the age of Constantine.

This architectural change reflects a liturgical one. The
great basilican churches of Constantine had been designed
for large crowds of worshipers, most of whom had only
recently been converted to the new faith. By contrast, the
Carolingian monastery church was designed for the worship
of a small community of men who lived in seclusion. In the
Early Christian basilica the body of officiating priests was
relatively small, the size of the attending crowd, colossal.
In the Carolingian monastery church, the number of
worshiping monks was relatively large (an average of 100 to
150; 300 to 400 in unusual cases), that of the attending
laymen not significantly larger. During the great religious
festivals, and in particular the feast of the patron saint, the
throng of pilgrims could rise to enormous numbers; but for
the rest of the year the lay attendance in the church remained
confined to the serfs who worked within the
monastic enclosure (in general outnumbering the monks
by not more than 30 per cent)[259] plus the tenants who lived
in cottages or on farms immediately around the abbey.

II.3.7

SEMICIRCULAR ATRIA

The two semicircular atria of the Church of the Plan
are a most unusual feature. The customary early
Christian form was a fore court of square or rectangular
plan, surrounded by colonnaded porticoes with a fountain
or water basin in the center and porches in the galleries for
entry and exit. These courts served varyingly as a preliminary
place of assembly for the faithful, as refuge during
inclement weather, as a burial ground, and often also as a
gathering place for those who were as yet not formally
admitted to the Christian community.[260] The architectural
prototypes of these early Christian atria (from Greek:
αίϑςιον; i.e., "a place under the open sky")[261] were the
galleried courts which the ancients interposed as a transitional
zone between the profane world and their sacred
buildings, in a multitude of aesthetic variations including
the use of semicircles, as well as an exuberant combination
of squares and semicircles, such as in the temples of
Baalbek or the Forum of Emperor Trajan in Rome. If
viewed against these lavish architectural orchestrations of
Rome, the early Christian atrium signifies historically a
retrenchment to the somber form of the square. Among
hundreds of well-attested early Christian atria, there are
only three, to my knowledge, that make use of the semicircle:

[ILLUSTRATION]

165. PLAN OF ST. GALL * AXONOMETRIC DRAWING

In aligning altars and altar screens with every second nave column, the architect projected the modular order of the Church (fig. 61) into the layout of its liturgical
furnishings. The interior is divided into a multitude of separate devotional stations, leaving only narrow passages in the aisles for the worshiper to move through the entire
length of the Church (fig. 82).

The emergence in Carolingian architecture of the ordering of space in recognizable, aesthetically modular form (cf. pp 217-23, below) has a striking parallel in the appear-
ance in Hiberno-Saxon and Carolingian book illumination of a decorative scheme whereby the entire book is divided internally into clearly distinguishable parts and sub-parts—
leading, in a crescendo of pictorial emphasis from letter to initial, initial to ornamented page, ornamented to figured page, and thus separating Gospel from Gospel, book
from chapter, chapter from paragraph, and paragraph from sentence. On the emergence of comparable stylistic trends in Carolingian literature and music, see Crocker,
Jones, and Horn in Viator VI, 1976.

[ILLUSTRATION]

166. RAVENNA. SAN GIOVANNI EVANGELISTA

BUILT BETWEEN 424 AND 434

Plan with square grid superimposed [after Petrovič, 1962, 43, fig. 2]

Navenka Petrovič's interpretation of the proportions of the church
of San Giovanni Evangelista in Ravenna is based upon a plan too
small and sketchy to be susceptible to critical evaluation. One is
disturbed to find that in some places the lines run along the outer,
in others along the inner surfaces of the walls of the Church. In the
case of the nave columns one observes with consternation that they
are not even touched by the longitudinal lines of the grid. Petrovič is
probably correct when she concludes that the church of San Giovanni
Evangelista is twice as long as it is wide, but the square grid from
which she thinks these proportions are developed is meaningless, since
the boundaries of the component squares of the grid are in no tangible
relation to the nave columns or the blind arcading of the outer-wall
surfaces. They are even less compatible with the dimensions of the
porch, the apse, or the two rectangular spaces (DIAKONIKON and
PROTHESIS) to the side of the apse.

Similar, if not more blatant discrepancies between Petrovič's modular
grids and the actual course of the masonry as well as the spacing of
the nave columns exist in virtually all of the other dozen odd fifth
and sixth century churches of Ravenna and the Northern Littoral of
the Adriatic Sea with which she deals, although in many of these the
length of the church appears to be indeed, the double of its width.
The proportions of all of these churches should be re-examined with
new plans made on the spot, and at a scale considerably (!) larger
and draftsmanship more precise than those upon which the Petrovič
grids are based.

a large fourth-century church at Damous-el-Karita,
a suburb of ancient Carthage;[262] the domed basilica of
Meriamlik, in Silicia, Asia Minor, 471-94;[263] and the
recently excavated church of St. Leonidas at Lechaion, the
harbor suburb of Corinth, whose atrium dates from 518527
(fig. 161).[264]

The earliest semicircular atrium north of the Alps is the
western atrium of Cologne Cathedral (period VI), which
was added to the enlarged transept of the old Merovingian
church (period V), as is now believed, after the accession
of Hildebold to the episcopal see in 782 and before the
coronation of Charlemagne in 800[265] (fig. 139), and thus
precedes the two circular atria of the Plan of St. Gall by
over twenty years.

It appears to me extremely doubtful that there is any
connection between these semicircular Carolingian forecourts
and those of the just-mentioned fourth- to sixth-century
churches of North Africa, Asia Minor, and Greece;
and I would be inclined to give stronger credence to a
connection with the semicircular courts of the basilica of
Emperor Trajan (fig. 239); their layout may have exerted
the additional influence of stimulating the aesthetic acceptance
in Carolingian churches of the motif of apse and
counter apse. We have other reasons to think that the
Forum of Trajan had some influence on the mind of the
architect who invented the scheme of the Plan.[266]

II.3.8

DETACHED TOWERS

Detached circular masonry towers, like those found on the
Plan of St. Gall, do not occur elsewhere in the ecclesiastical
architecture north of the Alps.[267] The builders of the North
chose to incorporate the towers into the body of the
church, and the preferred Carolingian solution was the
Westwerk, which eventually developed into the two-tower
facade. Detached towers were common enough, however,
in the churches of the Exarchate of Ravenna. Giuseppe
Galassi[268] dates some of these Ravenna churches in the
eighth century: San Giovanni Evangelista, San Pier'
Maggiore, and San Giovanni e Paolo. Others doubt that
they are this early.[269] Still, I am inclined to believe that the
idea originated in Ravenna, where it remained a favorite
motif throughout the Middle Ages. It has been generally
overlooked in this context that the Church of San Vitale in
Ravenna, consecrated in 547, has two impressive circular
towers with spiral staircases, which, for all practical purposes,
may be classified as detached towers (fig. 162).[270]
Therefore, I think it very probable that the trend began
with the towers of San Vitale. The Ravenna towers may, in
turn, have been influenced by Asiatic or Syrian prototypes.
Reinhardt[271] has drawn attention to a church with two detached
circular towers on an ivory panel in the Victoria
and Albert Museum (fig. 163), formerly considered to be of
late antique date but recently declared a Carolingian
pastiche.[272] Another contributing source, as has been pointed

[ILLUSTRATION]

166.X ST.-DENIS, SEINE, FRANCE

ABBEY CHURCH

(Consecrated 24 Feb. 775)

[redrawn from Crosby and Conant]

The plan is based on Crosby's interpretation of the remains of the Early
Carolingian church (Crosby, 1953, 68) as rendered and reproduced by
Conant (1959, 292). The square grid, superimposed in red, is based on
Formigé's analysis of the proportions of the church (Formigé, 1960).

out in the preceding chapter, were the two defensive
towers flanking city gates, shown on contemporary representations
of the two sacred towns of Rome and Jerusalem
(see above, p. 129).

Why the author of the Plan of St. Gall gave preference
to two detached towers over the more common Carolingian
Westwerk remains an enigma. The westwork consisted of
a multi-storied avant-corps on the entrance side of the
church, in which two lateral staircase towers gave access
to a raised tribune from which the emperor and his entourage
could attend the divine services. This innovation—a
tangible architectural expression, it appears, of the protectorate
which the secular ruler exercised over the Church in

[ILLUSTRATION]

NEUSTADT-AM-MAIN. SAVIOR'S CHURCH (768-769) WITH SQUARE GRID SUPERIMPOSED

167.A

167.B

[after Boeckelmann, 1962, 11, fig. 3]

The church was dedicated, according to a later tradition, on 19 August 793. But this date has recently been questioned (see Vorromanische
Kirchenbauten, 1966, 233). For another view of the church, see fig. 116; for its Early Christian prototypes and the roles they may have played
in stimulating medieval modularity, see figs. 144-151 and above, pp. 190ff.

the empire of the Franks—made its first appearance,
perhaps not accidentally, in the royal abbeys of Lorsch,
767-774 (figs. 200-201), and St. Riquier, 790-799 (figs. 168
and 196). The only preserved example besides the Palace
Chapel at Aachen (fig. 71.Za) is the Abbey Church of
Corvey-on-the-Weser (873-885), but others are known to
have existed at Reims Cathedral (founded by Archbishop
Ebbo, 816-835, and consecrated by his successor, Hincmar,
in 862), in the cathedral of Halberstadt (consecrated in
859), Hildesheim (consecrated in 872), and Minden (consecrated
in 952).[273]

Although, on the Plan of St. Gall, the emperor and his
following are given ample space in houses that the monastery
had set aside for reception of visitors,[274] in the church
itself he was not granted a station of his own. Perhaps we
are sensing here, once more, an expression of the touchiness
of the reform movement with regard to the possibility of
secular contamination of the House of God and its servants.[275]
The interpretations of the relation of Church
(sacerdotium) and State (regnum) were still far from attaining
any fixed or permanent form. Under Charlemagne, even
such leading churchmen as Theodulf of Orleans and Alcuin
did not hesitate to refer to the secular sovereign both as
"king in power" and "pontifex in preaching"; and
Charlemagne himself left no margin of doubt that it was
the emperor who installed the bishops and reserved for
himself the ultimate authority in the trial of prelates.[276] Yet
all throughout his reign and, with increasing strength, that
of his sons, there was a clerical reaction that rejected
temporal hegemony, proposed to erect the order of the
church as a second government beside the temporal power,
or even to arrogate complete subordination of the royal
power to the jurisdictional apparatus of the clergy. The
conviction of Louis the Pious that the emperor stood "in
the service of God" (in Dei servitio) gave added impetus
to this concept.[277] This ambiguity was not solved in
Carolingian times; and if it expressed itself in architecture,
we cannot expect it to have led to uniformly acceptable
solutions.

II.3.9

PLURALITY OF ALTARS

During the two centuries after the official recognition of the
Christian faith, most churches had only one altar. When the
need arose for additional altars, they were generally not
installed in the church, but in separate oratories built to the
side of the principal structure. The earliest evidence of
multiple altars within the church itself dates from the sixth
century. In the seventh and eighth, the trend increases. By
the time of Charlemagne, the number of altars in some
cases had risen to as many as thirty.[278]

The Church of the Plan with its seventeen altars (fig. 164)
—nineteen if we add the altars in the towers—is not an

[ILLUSTRATION]

168. ST. RIQUIER (CENTULA). ABBEY CHURCH (790-799) WITH SQUARE GRID SUPERIMPOSED

[as reconstructed by Achter, 1956, 146, fig. 7]

Achter's reconstruction of the plan of the abbey church of St. Riquier appears to us to be superior to that of Effmann (1912) because it takes into
account irregularities in the Gothic church that can only be explained on the assumption that they were conditioned by the layout of the preceding
Carolingian church. It is a refinement, not a contradiction, of Effmann's views. For other aspects of the church see fig. 196.

unusual case. Between 834 and 835 the Cathedral of Le
Mans had fourteen altars; the cathedral of York (766-778),
according to the testimony of Alcuin (d. 804), had thirty;
the abbey of Centula, at the time of Abbot Angilbert
(d. 814), had thirty; there were fourteen in the Church of
St.-Riquier, three in the Church of St. Benedict, and
thirteen in St. Mary's Church.[279]

There are many reasons for this multiplication of altars:
first, the increasing emphasis being placed on the saints
and their relics during this period; second, the introduction
into the liturgical ritual of solitary masses which were
celebrated at auxiliary altars; third, the growing number of
masses held in commemoration of the dead and for other
special occasions; and finally—perhaps the most decisive
cause—a papal ordinance that mass could be celebrated
only once a day at the same altar.[280]

The installation of more altars tended to divide the space
of the church into separate areas for worship. It fostered
processionals, in which all the monks, separated into
choirs, moved from altar to altar throughout the church,
chanting the psalms in antiphon and praying.[281]

The aesthetic implications of this new liturgical development
are clearly marked in the Church of the Plan (fig. 140).
An alignment of altars (or altars and other important
liturgical appurtenances) in nave and aisles at each second
pair of columns establishes transverse divisions on the
floor of the Church, the rhythm of which can only be

[ILLUSTRATION]

169. FULDA. PLAN OF ABBEY CHURCH AS ATTAINED UNDER RATGER (802-817)

[after Vorromanische Kirchenbauten, plate following p. 81] SQUARE GRID SUPERIMPOSED

There is no doubt that the layout of Ratger's church at Fulda was based on that of Old St. Peter's in Rome (fig. 170, and text, p. 221) but the
conceptual method, used in giving nave and transept the same width and using the area of intersection as a module for the proportions of all superordinate
spaces, is fundamentally different.

[ILLUSTRATION]

170. ROME. OLD ST. PETER'S (AFTER 324 CA. 360)

[Plan after Jongkees, 1966, pl. 1. Construction diagram superimposed by Ernest Born]

The architect who planned St. Peter's used a constructional system as classical in concept as the modularity of Carolingian churches is medieval.
He obtained the length of the longitudinal body of St. Peter's by elongating the sides of a square to a length equalling its diagonal. In the same
manner the architect calculated the overall length of the church by using the diagonal of the rectangle thus constructed.

The diagonal of any square does not relate in a ratio of integers to its sides. The metric expression of the diagonal of a square is 1.414, the square
root of 2 (√2). The extension of a square into a rectangle by extending two of its sides to a length equalling its diagonal is a simple task that can
be performed without recourse to calculation and with only the aid of a string.

Spaces designed by aggregation of modules tend to produce such narrow elongated shapes as are exemplified in the churches of Fulda, Cologne,
and the Plan of St. Gall. The trait distinguishes the Romanesque and Gothic architecture of Hildesheim, Jumièges, and Speyer. The extension into
a rectangle of a square by means of its diagonal produces, by contrast, spaces of relatively squat proportions. The Romans made frequent use of
√2 rectangle construction, resulting in the thoroughly un-medieval proportions of many Roman cities and military camps. No doubt strategical
advantages of defending compact fortifications as compared with long, attenuated ones influenced, if not conditioned, Roman application of the √2
rectangle to site layout. But military considerations could hardly have been primary in constructing metropolitan churches. When they came to be
built, the method perhaps had become habit; its simplicity was probably an important cause of its general adoption.

[The linear values √1, √2, √3, √4, √5, etc. are derived by diagonals produced from a generating square of unity. Starting with the initial
square, one unit by one unit, the nth square is formed after n2 steps. The procedure can go on to infinity. But after the √5 rectangle (the diagonal to
the second square) linear or lateral expansion is simplest and most direct by compounds of the square. Hence square schematism. E. B.]

[ILLUSTRATION]

171. REICHENAU-MITTELZELL. HAITO'S CHURCH OF SS MARY, PETER, & PAUL (CONSECRATED 816)

The church is inordinately short for its period. Reisser (1960, fig. 285) held the opinion that Haito intended to extend the church westward to twice
its present length, so that its façade would have been in line with the façade of Pirmin's church. This would explain the shift from square modules
(30′ × 30′) used in the eastern part of the church and oblong ones used in the nave (24′ × 30′). For more detail see fig. 117 Horn and Born, 1974,
453; IDEM, 1975, 372-74; and Erdmann and Zettlar, 1974, 481ff.

compared to that which was established two centuries later
on the clerestory level through the introduction of diaphragm
arches. This is the beginning of the principle of
rhythmic alternation. It starts with the alignment of altars
and columns (Plan of St. Gall), finds structural expression
in the nave walls with the introduction of supports of
rhythmically varying strength (Gernrode and Hildesheim),
gathers vertical momentum with the addition of slender
shafts and arches that carry the rhythm up into the heights
of the clerestory (Speyer I), and enters into its final phase
as arches are thrown across the nave from alternate sets of
piers (Jumièges, Cérizy-la-Forêt).

II.3.10

SQUARE SCHEMATISM

"Square schematism" is a principle of medieval church
design by which the constituent spaces of the church are
calculated as multiples of a basic spatial unit, usually that
of the crossing square. The origin and evolution of this
concept is still one of the great mysteries of medieval
architectural history. The Church of the Plan of St. Gall
represents a crucial stage in the conceptual development of
this principle. The importance of this fact has been blurred
because no consensus of opinion had been obtained, in
previous inquiries, with regard to even the simple question
of whether or not the design of the Church of the Plan had,
in fact, been developed within a system of squares; let alone
the infinitely more complex problem of the historical roots
and the deeper cultural significance of this fascinating
principle of articulating space. I, for one, am convinced
that these questions cannot be solved from within the
field of architectural history. The modular mode of thinking
that underlies this schematism is a general cultural
phenomenon that manifests itself in other spheres of life.
On the following pages I shall try to isolate some of the
converging historical currents that merge in this concept.

PRESENCE OR ABSENCE OF THE GRID OF SQUARES

On the Plan of St. Gall the square and the grid of squares
are used in two different ways: as a method of mensuration,
and as an aesthetic principle. In the first instance the square
grid offered a convenient method of dividing a given area
internally by defining it as a multiple or fraction of certain
modular master units (2½-foot square, 40-foot square, 160foot
square).[282] In the other case, the square grid was used
as an active principle of architectural composition. It is this
latter type alone with which we are now concerned. Reinhardt
categorically denied its presence on the Plan of St.
Gall.[283] Similar views were expressed in 1945 by Samuel
Guyer,[284] but convincingly challenged in 1952 by Albert
Knoepfli[285] in a drawing which shows a grid of 10-foot
squares superimposed upon the Plan of the Church.[286] My
own analysis of the scale and construction methods used in

[ILLUSTRATION]

172. COLOGNE. CAROLINGIAN CATHEDRAL OF SS PETER AND MARY

Like Fulda (fig. 169) and doubtlessly under the same influence, i.e., of Old St. Peter's in Rome (fig. 170), the transept is located in the west. As in
Fulda and in the other Carolingian churches discussed here, the use of modules imparts to the layout of the spaces an aesthetic character wholly
different from the squat corporeality of their Early Christian prototypes.

designing the Plan corroborated this view.[287] In view of the
visual evidence submitted by Knoepfli, as well as my own
analysis of the system of squares shown in figures 61 and
173, I do not see how the validity of this contention could
ever again be questioned.

MEDITERRANEAN OR NORTHERN ROOTS:
A DIVISION OF MINDS

Because of its geographical distribution primarily in the
territory of the Franks, Saxons, and Normans, Georg
Dehio considered the square schematism to be essentially
a "Germanic" contribution.[288] Samuel Guyer,[289] in a complete
reversal of this contention declared this "geometrical
clarity" to be a mark "of the Mediterranean way of thinking,"
and "one that had its roots in classical antiquity."[290]
The square schematism of the Plan of St. Gall, he maintained,
was not one of the new and creative contributions
to medieval architecture that it had been assumed to be,
but "transmits to the West in a rather muddled manner the
thought of the qualitatively superior art" of the Early
Christian period.[291]

These statements are of questionable historical validity—
and the argument does not gain in power when one finds it
supported by such sweeping generalities as "A civilization
in process of just awakening from the darkness of an
a-historical past" and "as yet suspended in a state of
unstable hovering between unconsciousness and awakeness"
could not possibly have produced aesthetic concepts
"of such distinct and clear rationality. . . . The period of
Charlemagne had never the significance ascribed to it so
fervently in recent times. . . . In the time of Emperor
Charlemagne the thoughts of Late Antiquity and Early

[ILLUSTRATION]

173. PLAN OF ST. GALL. PLAN OF MONASTERY CHURCH

SHOWN AT SCALE 1:600

The Church of the Plan of St. Gall is chronologically the last of a triad of Carolingian transept basilicas of monumental dimensions owing
their size to the tide of spiritual and cultural exhilaration that seized the Frankish clergy in the wake of Charlemagne's coronation as emperor,
on Christmas Eve of the year 800, in the basilica of Old St. Peter's in Rome.

Unlike Cologne (fig. 172) and Fulda (fig. 169) which were occidented in imitation of Old St. Peter's, (fig. 170) the Church of the Plan was
oriented. Like Cologne and Fulda, on the other hand, and in contrast to St. Peter's, the Church of the Plan was constructed on a square grid,
in the most elaborate and most consistent application of it, since it encompassed, in addition to the Church itself, the entire claustral complex
and in fact the entire monastery site (figs. 62 and 63).

[ILLUSTRATION]

174. ROME. SANTA MARIA MAGGIORE (432-440)

LOOKING NORTH AND SOMEWHAT EAST TO THE APSE

Despite its coffered Renaissance ceiling (added in 1500, substituting for the original open-timbered roof) this view of Santa Maria Maggiore
conveys persuasively the stylistic quality of the great Early Christian basilicas composed of huge, block-shaped, and internally undivided
voids.

The concept differs fundamentally from that of the square-divided Carolingian churches of Neustadt, St. Riquier, Fulda, Reichenau-Mittelzell,
Cologne, and that of the Plan of St. Gall (figs. 167-69; 171-73), as well as from the bay-divided and arch-framed spaces of the Romanesque
and Gothic (Hildesheim, Speyer, Jumièges; figs. 188-90), the cellular composition of which has primary roots in the Carolingian modular
reorganization of the Early Christian scheme.

For another magnificent view of the interior of a great Early Christian basilica see fig. 81, St. Paul's Outside the Walls, Rome.

[ILLUSTRATION]

175. FEDDERSEN-WIERDE, BREMERHAVEN, GERMANY

175: AISLED HOUSE OF A CHIEFTAIN, WARF-LAYER 11B, 1ST-2ND CENT. (authors' reconstruction)

176: PLAN (after Haarnagel, 1956, pl. 3)

The house belongs to the second settlement horizon of an artificially raised dwelling mound (Warf) which was occupied, on successively higher
levels, from the 1st to the 4th centuries. The house was 28.5 × 7.5m on an east-west axis. The living portion with hearth and the section for livestock
were, respectively, 9m and 16m long. An entrance in the middle of the eastern end wall was primarily used by cattle. (Also see figs. 315-316, II, 58.)

Christianity were taken over in a manner so superficial as
to be incapable of taking any deep root or of being developed
any further."[292]

I propose that we confine ourselves to specific issues
rather than argue the case in such global terms.

INCREASING PROPENSITY FOR MODULAR
SPACE DIVISION IN PRE-CAROLINGIAN AND
CAROLINGIAN ARCHITECTURE NORTH OF
THE ALPS

The emergence of the square schematism in medieval
architecture depended on two crucial innovations in the
interrelation of the component spaces of the basilican
church:

1. The nave and the transept of the church had to be
given the same width, and

2. The width of the aisles had to be fixed to one-half
the width of the nave.

Without the first, the crossing could not form a square;
without the second, the modular division of the nave could
not be carried into the aisles. Both of these features
occurred separately in Early Christian times, but they were
not integrated then into a programmatic architectural
system.

An example of a church with nave and transept of equal
width is the Justinian basilica of the Nativity at Bethlehem
(if Hans Christ's interpretation of its plan is correct).[293] In
several Christian churches of Ravenna—all without transepts—the
width of the aisles is fixed at one half, or approximately
one half, the width of the nave. Yet as we survey
Early Christian church architecture as a whole, we must
conclude that its truly distinguishing feature is not the
presence, but rather the absence of any fixed proportions.
Nevenka Petrović[294] has made an illuminating study of the
proportions in churches of Ravenna and the adjacent
littoral of the Adriatic sea. In attempting to demonstrate
that these churches were laid out according to a system of
squares, as she set out to do, she has de facto illustrated the
fundamental difference between the layout of these later
Early Christian churches and the system of squares employed
in medieval architecture. The salient feature of the
Church of the Plan of St. Gall and its Ottonian and
Romanesque successors is that the squares control the
spacing of the arcades and therefore express the modular
layout of the plan in the elevation of the columns. The
divisions of Petrović's grids (fig. 166), by contrast, have no
relation whatsoever to the position of the arcade columns.
True, in some of the proto-medieval churches of Ravenna,
the length and width of the church exist in a state of

[ILLUSTRATION]

178.

Great Cruciform page

size of original about 33·8 × 24·1cm.

[ILLUSTRATION]

LINDISFARNE GOSPELS

179.A

179.B

LONDON, British Museum, Cotton Nero D. IV, fol. 2v

[by courtesy of the Trustees of the British Museum]

Stages of layout

modular interdependency, but since this module does not
control the spacing of the columns, it is aesthetically of no
consequence.

I tend to agree with George Dehio that the square schematism
is essentially a "Germanic" contribution to Western
architecture for two reasons: first, because it is found
primarily in regions of relatively strong Germanic concentration,[295]
and second, because it is in these areas also that
we may detect its developmental antecedents. An early
medieval church exhibiting an incipient tendency toward
the use of the square as a module was Fulrad's church of
St. Denis, begun after 754 and consecrated in 775 (fig.
167).[296] Its basic layout, if Formigé's interpretation is correct,
was developed within a grid of 6-foot squares which,
in contrast to San Giovanni Evangelista at Ravenna (fig.
166), determined not only the overall dimensions of nave
and transept, but also the interstices of its arcades. The
transept was seven 6-foot units wide, and thirteen long; the
nave was five units wide and fifteen long. The distance from
center to center of arcade columns was two units, and in the
middle part of the transept two cruciform piers establish a
square of five by five units. As yet we cannot speak of
square schematism, because the dimensions of the crossing
square are not mirrored anywhere else in the building, and
in particular not in the intercolumniation of the arcades. A
church that comes closer to this ideal is the Saviour's
Church of Neustadt-on-the-Main, after 768/69 (fig. 167).
The plan of this church together with other cruciform
churches of similar design built in early medieval times,
such as Pfalzel near Trier, and Metlach (both before 713),
may have formed a connecting link between square-divided
Carolingian basilicas of the ninth century and certain
cruciform churches of the fourth and fifth centuries, typical
examples of which are shown in fig. 144-146 and 148-149.
A grandiose variant of this church type, built as early as
380 by Emperor Gratian in his residential city of Trier,
rose in territory that later was part of the very core of the
Frankish kingdom—for every Carolingian to see! (Its
masonry survives to this day, incorporated in the fabric of
the Romanesque church that superseded it.) This is the
only pre-medieval church type where nave and transept
are of equal width, their intersecting bodies forming a
square—and one might indeed regard the fully developed
square schematism of the Carolingian period as a transference
to churches of basilican plan of a principle already
experimented with in pre-medieval times in the highly
specialized context of these Early Christian cross-in-square

[ILLUSTRATION]

180. LINDISFARNE GOSPELS

LONDON, British Museum, Cotton Nero D.IV, fol 2v

Square panel above arm of cross on cruciform page shown in fig. 178

• A. Photo of panel

• B. Photo of square grid visible on corresponding portion of
  fol. 4r.

• C. Square grid with outlines of cross and lozenge pattern
  (first stage of construction)

• D. Final stage of pattern (authors' interpretation)

churches.[297] The first Carolingian church to mark the developmentally
significant moment of the adoption of the
square schematism in a building of unequivocally basilican
design was the abbey church of Centula, 790-799, if
Irmingard Achter's reconstruction of this building is
correct (fig. 170).[298] Because of the scarcity of archaeological
data available on this important building, such an assumption
can neither be fully accepted nor convincingly
rejected. For the same reason it is impossible to ascertain
whether the interstices of the nave arcades were aligned
with these modules.

Modular adjustment between width and length of the
component spaces is clearly visible, however, in the abbey
church of Fulda (802-817).[299] Its nave, measured from the
base of its western to that of its eastern apse, was exactly
four times its width (fig. 169). The dimensions of the
transept were identical with those of the nave. In the vast
body of literature devoted to Fulda—whose authors never
weary of citing the dependence of its design on that of Old
St. Peter's—this crucial aesthetic novelty has never been
pointed out, much less set into proper historical perspective.
We know nothing about the intercolumniation of
Fulda.

On the other hand, it is not possible to interpret Old
St. Peter's as having been developed within a grid of
identical squares—neither each volume by itself, nor any
volume in relation to a neighboring unit or to the whole of
the building mass. The architect who planned St. Peter's
employed instead a constructional system as classical in
concept as the modularity of the Carolingian churches
shown in figs. 144ff is medieval (see Born's analysis, fig.
170). He calculated the length of the longitudinal body of
the church by making use of the diagonal of a square with a
side equal to the width of the church, and developed the
overall length of the church in the same manner, with the
aid of the diagonal of the rectangle obtained by the preceding
method. This configuration, known as a √2
rectangle, is irrational, since the diagonal of a square is not
in any integral relationship to its sides (1: √2 = 1:1.414)
and therefore cannot be defined as an aggregation of an
integral modular value.

Hildebold's church of Cologne (ca. 800-819) was
composed wholly of equal squares: three in the transept,
four in the nave, one in the fore choir (fig. 172). If the
elevation of its nave walls was identical with that of the
church dedicated in 870, the piers of the arcades that
carried the clerestory walls would not have been in alignment
with this system.

The abbey church of Reichenau-Mittelzell, built by
Haito (806-816) is also developed within a modular grid of
squares, but the grid is irregular, and its existence, for that
reason, has been questioned. In evaluating this problem it
is important to distinguish the existence or nonexistence of
the concept of squares at Reichenau from the regularity or
perfection of its execution. The irregularity, in the angular
deviation of the walls from the grid (especially noticeable in
the eastern part of the church) is caused by special topographical
conditions. But no doubt can be entertained that
the concept exists. The shape of the fore choir and of the two
transept arms are almost a mirror image of the shape of the
crossing square, but the squares of the nave are slightly
oblong. Yet the principle of divisions is clearly there, and
the boundary between the two oblongs of the nave is
marked by piers, whose design differs from the columns
standing midway between them. In this feature St. Mary's
Church at Reichenau goes a step beyond even the Church
of the Plan of St. Gall, which has no such alternation in its
main supports.

In Hildebold's church of Cologne (fig. 172) the system
of squares finds clear expression in the west transept and
in the eastern fore choir, both of which are formed of squares
of identical size: three in the transept, one in the fore choir.
The nave is composed of four squares of like dimensions.
We know nothing about its elevation. If it was identical
with that of the church that was dedicated in 870, the piers
of the arcades which carry the clerestory walls would not
have been in alignment with the system of squares.

In the Church of the Plan of St. Gall the square schematism
attains its purest Carolingian form of expression (fig.
173). The basic unit is the 40-foot module of the crossing
square. The transept is formed of three such squares, the
fore choir of one, the nave of four and one-half; and the
dimensions of the crossing square are echoed even in the
Library and Vestiary. In St. Gall, moreover, the interstices
of the columns are in rhythmical alignment with the
squares. It is incomprehensible to me how this fact should
ever have been questioned. What the designer of this
church had in mind were arcades cutting deep into the
masonry of the nave walls (fig. 110) with their supports so
spaced as to give bodily expression to the sequence of
squares on which the Plan was based. This schematism is
a conscious and willed aesthetic principle. It is a fundamentally
different concept from that which produced the
low, narrowly spaced columnar orders of the Early Christian
basilicas of Rome (figs. 141 and 174). Contrary to what
Guyer, Reinhardt, and Reinle believe, it is an ingenious
anticipation of the square schematism of the Romanesque.

What are the historical preconditions of this propensity
for modular organization of space? Some clearly are functional.

[ILLUSTRATION]

181.

Cruciform page preceding the Gospels of St. Luke

size of original about 33·8 × 23·1cm.

[ILLUSTRATION]

182. LINDISFARNE GOSPELS

LONDON, British Museum, Cotton Nero D. IV, fol. 138r

[by courtesy of the Trustees of the British Museum]

Diagram illustrating use of square grid in constructing the layout for
the opposite illustration

Others may have to be traced to vernacular architecture.
For still others we shall have to reach, beyond the
boundaries of architectural history, into the field of book
illumination, where strong expression of modular modes
of thinking can be observed over a century before they
assert themselves in church building. Yet others, and perhaps
the most important of all, may have to be sought in
deeper and more general cultural levels.

NEW LITURGICAL NEEDS
CONTRIBUTING TO MODULAR SPACE DIVISION

I have already drawn attention to a number of contributing
factors that tended to facilitate this development in a functional
sense: the need for an extension of the altar space,
leading to the interposition of a new spatial unit between
transept and apse; the framing of the crossing by means of
arches, creating a square division in the transept, that would
lend itself to being extended to the nave; and most of all,
perhaps, the multiplication of altars, demanding a subdivision
of the spaces of nave and aisles into a sequence of
devotional stations (figs. 164 and 165). We add to this a
feature (which Irmingard Achter stressed in her discussion
of the Carolingian Abbey Church of Centula): circular
towers such as the towers which surmounted the crossing
and the westwork of this church require as base a square-shaped
underpinning. All of these innovations contributed
to the development of a modular scheme, but none of them
alone (and perhaps, not even all together), might have
led to the creation of the modular space division of the
medieval church as a binding architectural principle. There
are other forces to be taken into account.

MODULAR SPACE DIVISION: AN INTRINSIC
FEATURE OF PREHISTORIC, PROTOHISTORIC, AND
MEDIEVAL WOOD CONSTRUCTION

In an article dealing with the origins of the medieval bay
system,[300] I have pointed out that modular design has been
from the remotest periods an intrinsic feature of northern
wood construction. The stability of the timbered Germanic
house required that its roof-supporting posts be joined
together at the top: lengthwise by means of plates, and
crosswise by means of tie beams. This divides the space of
the house into a modular sequence of timber-framed bays
(figs. 175 and 176). Recent excavations have made it clear
that this construction type came into existence around 1200
B.C., and for the next two thousand years it served as an
all-purpose house in the Germanic territories of Holland,
Germany, England, and Scandinavia as well as in all those
areas of Central and Western Europe that were primarily
settled by Germanic peoples.[301] In timber this concept is
old; in stone it is new. In timber it develops as a logical
construction method from the natural properties of the

[ILLUSTRATION]

Canon Tables (183.A)

The Ada Gospels is the first great highlight of the classicizing phase of
illumination of the so-called Court School. It consists of an earlier part
(fols. 6-38) containing the canon tables (fols. 6v-11v) which combine the
decorative tradition of the Hiberno-Saxon school (figs. 178-182) with a tendency
to treat the arcades of the tables in a more architectural manner.

Size of leaves of the manuscript in the present cropped state is 36 × 24.5cm.

Figures 183.A and 183.B are reduced about 12.5 percent.

Originally the leaves were larger.

The later part of the Ada Gospels consists of the remainder of the text, and
portraits of the four evangelists (fols. 15v, 59v, 85v, 127v), one of which is
illustrated in fig. 184 (see overleaf).

[ILLUSTRATION]

183.C ADA GOSPELS (EARLY 9TH CENT.)

DIAGRAM SHOWING USE OF SQUARE GRID IN THE
CONSTRUCTION OF THE CANON TABLES SHOWN IN
183.A AND 183.B (see overleaf)

TRIER, Municipal Library, MS. XXII, fol. 6v

material. In masonry it is an intrusive feature, imposed
upon the material as a willed aesthetic principle—and
therefore ushers in a conflict between style and building
material which, in its ultimate phase, the Gothic, led to a
complete denial of the natural properties of stone. I have
suggested that the modular arrangement of space, which
begins in Carolingian Church architecture, gathers increasing
strength in the Ottonian period and reaches its peak of
expression in the Romanesque and Gothic (fig. 177), has
one of its roots in the fact that these churches were constructed
by men in whose collective memory "to build"
had been synonymous with building in modular sequences
of space.

The validity and importance of this explanation cannot
be appreciated until it is understood that the determining
factor in analyzing the origins of square schematism is not
that it is based on the shape of the square, but that it
establishes a system of binding modular relationships. In
distinguishing between the système des carrés of the Romanesque
and the système des barlongs of the Gothic, we have
lost sight of the fact that both of these systems are members
of the same family. Whether the module is square or
rectangular is determined by secondary conditions, sometimes
functional, sometimes constructional, sometimes
stylistic, and on occasion, even by purely arbitrary reasons.
The house of the Germanic chieftain of the first and second
centuries A.D., which is shown in figures 175-176, employs
both the square and the rectangular module, the former in
the living area, identifiable by the hearth; the latter, in the
section of the house where the cattle are stabled, identifiable
by the manure mats.[302] Here the shape of the module is
conditioned by strictly functional considerations: the roof-supporting
trusses are spaced at intervals of 6 to 7 feet, just
as much space as is needed to stable two head of cattle. In
the living section of the house, on the other hand, the
trusses are set further apart to give greater freedom of
movement. The distinction is very old and can be observed
in Bronze Age houses of the same construction type, dating
from around 1200 B.C., recently excavated by Waterbolk in
Elp, Holland.[303]

In the Carolingian monastery churches discussed in the
preceding pages, the square is the more reasonable form
to be adopted—at least in the liturgically most important
areas—the choir and the transept— which lend themselves
to square division with notable ease. In the nave, this was
more difficult to obtain, since here the square division
conflicted with the narrow intercolumniation inherited
from the Early Christian prototype churches. It required
a strong personality to move the columns apart to the novel
and daring distance of 20 feet, as was done in the Church
of the Plan, and thus to express the module in the bodily
sequence on the columns. The designer of the Church of
Cologne may have struggled with similar ideas (fig. 172), but
abandoned the scheme in actual construction (figs. 15-16).

[ILLUSTRATION]

183.B

[ILLUSTRATION]

183.D ADA GOSPELS

DIAGRAM SHOWING USE OF SQUARE GRID

TRIER, Municipal Library, MS. XXII, fol. 8v

MODULAR AREA DIVISION: AN INTRINSIC
FEATURE IN THE LAYOUT AND DESIGN
OF ILLUMINATED PAGES IN HIBERNO-SAXON
AND CAROLINGIAN MANUSCRIPTS

The modular bay division that governed the construction
of the Germanic house from the first millennium B.C.
onward was not the only source for the appearance of
modular relationships in Carolingian church architecture. It
may, in fact, take second place when weighed against
another influence, which reflects an attitude of mind more
than a constructional necessity. An organization based on
modules is one of the distinguishing features of the layout
of the illuminated pages of Hiberno-Saxon and Carolingian
manuscripts.

Figures 179.A and 179.B show how the artist of the Lindisfarne
Gospels set out to decorate the large cruciform page
that forms the frontispiece (fol. 2v) to this remarkable book
(fig. 178).[304] The principal motif is a square-headed cross
framed by a narrow band and decorated internally with a
key pattern. In the field between the arms of the cross and
the outer frame of the page, there are four panels with step
patterns, two square ones on the top, two of oblong shape
at the bottom. The background is filled with an intricate
design of interlace. The page is framed by a strip of interlaced
birds, held in by narrow bands which terminate at
each of the four corners in an ornamental knot.

An analysis of the construction method used in setting
out the design of this page shows that all the basic divisions
are multiples of the width of the framing bands. The basic
values are 5 · 6 · 7 · 12 (fig. 179.A). The squares of the cross
measure 12 · 12; the panels in the fields above and beneath
the arms of the cross are 10 · 10 and 10 · 25. I feel certain
that a system of linear coordinates, such as is shown in
figures 179.A and B, was laid out on the page, by means of
either lines or prickings before the artist entered the
decorative details. In certain places where the design was
very intricate, such as the panels above and under the arms
of the cross with their complicated step patterns (fig. 180.A),
the illuminator actually drew out the lines with the point
of a fine stylus. This is visible on the opposite side of the
sheet (fol. 2r) as a grid of delicately protruding ridges (fig.
180.B).[305]

I have shown in figures 180.C and D how this system was
worked out. First, the illuminator divided the square
internally into sixteen subordinate squares by the method
of continuous halving. Then he divided each subordinate
square into nine base squares through internal tri-section.
This furnished him with all the desired linear co-ordinates
for the lozenge, cross, and step patterns with which these
squares are decorated (fig. 180.A). The same or similar

[ILLUSTRATION]

184.A St. Luke

The Ada Gospel portraits of the four evangelists framed by arcades
(fols. 15v, 59v, 85v, 127v) depend stylistically on a Late Antique
manuscript tradition combining the sculptural corporeality of Roman
figure style with touches of Byzantine mannerism.

Revived in the art of the Frankish illuminators of the Court School,
this tradition merged with the northern concept of organization of
space. This first encounter of the two traditions is not reflected in the
portrayal of the Ada evangelists, but visibly controls the layout of
the surface in which their images are placed. Later, in a synthesis of
southern corporeality and northern abstraction that parallels the
same development in architecture, these concepts will produce a
figure style that, despite strong dependance on classical prototypes, is
distinctly medieval (see fig. 185).

[ILLUSTRATION]

184.B ADA GOSPELS (EARLY 9TH CENT.)

DIAGRAM SHOWING USE OF SQUARE GRID IN CONSTRUCTION
OF ARCH FRAMING

TRIER. Municipal Library. MS XXII, fol. 59v

methods were used in all other ornamental pages of the
manuscript, and also in the layout of the canon tables (fol.
10r-fol. 17r).

Figures 182.A, B, and C give an analysis of the design
of the great cruciform page on fol. 138v that precedes the
Gospel of St. Luke (fig. 181).[306] This page has as its main
motif a cross with T-shaped arms, filled in with a background
of interlaced patterns; the spaces around the cross
are filled with an animal interlace. The entire decoration of
this page is laid out on a system of squares, each side of
which is four times the width of the framing band. The
page measures thirteen units across and seventeen units up
and down. The transverse axis of the cross is laid out in
the sequence:

the vertical axis in the sequence:

The protruding knots at the corners and in the prolongation
of the two intersecting axes of the page are inscribed
into a marginal area seven units wide.

These principles of modular book design so typical of
Hiberno-Saxon art were inherited by the continental
Carolingian illuminators. Figures 183.C and D are a design
analysis of two of the canon tables of the Ada Gospels, fol.
6v and fol. 8v (figs. 183.A and B).[307] The layout of these
tables varies. Some have four arcades, others have three.
As in the Lindisfarne Gospels all the internal subdivisions of
these pages are calculated as multiples of the width of the
framing bands. In both tables the design is suspended in a
square grid composed of 4 × 4 base units.

On fol. 6v (figs. 183.A and C) the bases of the columns
and their interstices are calculated in the sequence:

the column shafts and their interstices in the sequence:

The columns are inscribed into a grid of 16 × 19 squares,
the arches into a 9 × 19-square grid.

The canon arch on fol. 8v (figs. 183.B and D) has only
three columns. It is based on the same grid pattern. The
bases of the columns are calculated in the sequence:

the column shafts and their interstices in the sequence:

Figures 184.A and B show that the same method of construction
is used in the layout of the arch which frames the
figure of St. Mark on fol. 59v of the Ada Gospels. The basic

[ILLUSTRATION]

185. CODEX AUREUS OF ECHTERNACH

MADRID, Escorial, Cod. Vitr. 17, fol. 2v

[by courtesy of the Patrimonio Nacional]

Emperor Konrad and Empress Gisela prostrate themselves before
Christ in Majesty. School of Echternach, 1043-1046. The Gospel
book was presented to Speyer between 1043 and 1046 by Henry III
(1038-1056) who (folio 3v) is portrayed with Agnes, his consort, in
the act of transmitting the manuscript to Mary, patron saint of the
cathedral. Both illuminations are high points in the synthesis of a
figurative style rooted in Antiquity, with a medieval propensity for
planimetric order and linear simplicity pervading both figurative and
geometric components of each picture (rectangle, lozenge, circles,
semicircles). Byzantinizing mannerisms (cf. fig. 184) are dropped;
the figures have acquired the magnificent blocklike stance that
characterizes much of the contemporary sculpture.

unit is a square, three times the width of the framing bands.
The columnar section is a square, 20 × 20 units; the arch
section, an oblong of 9 × 20 units.

The square grid affects the layout of the page, but not the
design of the figure of the Evangelist. This latter is clearly
patterned after a Byzantine model. The conflict between
the corporeal emphasis of the classical design, and the
tendency of the northern medieval illuminator to subject
the borrowed image to linearism and geometricity provoked
a developmental dialectic in which the ability to absorb
classical influences with increasing strength, in successive
stages, is preconditioned by a partial rejection and successful
transformation of those absorbed in a preceding phase.
In the period of the Romanesque, as a consequence of this
dialectic, solutions are obtained in which southern corporeality
and northern abstraction enter into a state of
balance (fig. 185). In like manner in the field of architecture,
southern masonry tradition fuses with northern frame
construction in a marriage in which the two component
traditions are matched with consummate perfection (fig.
186).

The square schematism is the primary organizing agent
in this development. It helps to disassemble the large
corporeal spaces of the Early Christian basilica, and to
arrange its parts in modular sequences that could be
vaulted. It determines the take-off points for the rising
shafts and arches that were needed to carry the vaults.

MODULAR AREA DIVISION:
AN INTELLECTUAL PRINCIPLE AFFECTING THE
CONCEPTUAL ORGANIZATION IN THE
RELATIONSHIP OF CHURCH AND STATE

It has become a commonplace of historical reference to
speak of the "anthropomorphic" character of Greco-Roman
and Late Antique art and of the "corporeal"
quality of their figurative and spatial composition; and it
has been stressed time and again that this quality grows
out of a way of thinking that interprets man and his metaphysical
environment "in the image of man," a concept so
embedded that even Christianity could not rout it. We
have not yet found any way of describing or explaining
adequately the way of thinking that impelled the medieval
illuminators to submit the classical prototypes to relentless
abstraction and caused the medieval architects to break up
and reassemble their spaces in controlled volumetrical
sequences.[308] Until we have, we shall not be able to understand
fully the meaning of such a phenomenon as the
square schematism of medieval art or, for that matter, any
other schematisms conceptually related to it. Square
schematism is an intellectual principle by which formerly
existent, yet isolated or only loosely connected parts are
brought into an ordered modular relationship. It is a
principle of intellectual alignment that strikes far beyond
the reality of architecture or book illumination into the
realm of literary and musical composition—as Charles W.
Jones and Richard D. Crocker have shown in recent
studies[309] —reflect a cultural attitude that may have had a

[ILLUSTRATION]

186. SPEYER CATHEDRAL (1082-1106)

[after Dehio, GESCHICHTE DER DEUTSCHEN KUNST,
4th ed., I, 1930, plate vol., figure 63]

About 1030, Emperor Konrad II (1029-1039) began to replace the
Merovingian cathedral with a new building (Speyer I) whose crypt
(dedicated in 1041) became a sepulchral sanctuary for the imperial
house. The nave walls of this structure were articulated by a
continuous sequence of engaged shafts rising from the floor to the
head of the walls. The roof was timbered. The aisles by contrast were
covered with shaft-supported and arch-framed groin vaults.

During the reign of Emperor Henry IV (1056-1106, or more
precisely from about 1082-1106), the design of the aisles was
transferred to the nave by the superimposition upon each alternate
tier of a second and heavier shaft, and their connection lengthwise
and crosswise by means of arches capable of carrying vaults. The
view shown above represents the cathedral in the form it had attained
at this point (Speyer II).

direct effect upon even the organization of the relationships
of Church and State, where similar tendencies can be observed
at about the same time. An illuminative reflection of
this mode of thinking is to be found in Walahfrid Strabo's
Libellus de Exordiis, written between 840 and 842. Here
secular rulership and ecclesiastical government are brought
into a system of modular relationships in which each of the
two respective hierarchies is formed by a series of parallel
offices:

Like the "disengaged crossing" or the "extended altar
square" many of the component parts of this system are
old. But the manner in which they were drawn together
into a system of homologous parts presaged a development
which, two to three centuries later, led to the accomplished
and intensely sophisticated metaphysical visions of scholastics.
They envisioned the universe as a triad of structurally
related hierarchies (fig. 187)—each being an identical image
of the other as well as of the system as a whole—that
possessed identical subdivisions into triads of ranks, and
in each of these triads each subordinate rank corresponded
in substance to its equivalent part in every other triad.[311]

SETBACK AND RE-EMERGENCE

On the preceding pages I have shown that the square
schematism appeared in western architecture neither as

[ILLUSTRATION]

187. WILLIAM OF AUVERGNE. LIBER DE UNIVERSO (1230-1236)

HIERARCHIES OF HEAVEN, STATE AND CHURCH

[Author's diagrammatic interpretation]

Components of this concept are Early Christian; their integration into an all-embracing metaphysical scheme is medieval. Similarities in the institutional
organization of Church and State were apparent in the 4th century after the Church began to model its administrative structure after that
of the State. Carolingian awareness of this fact is attested by the passage of Walahfrid Strabo quoted above, p. 231.

Dionysius the Pseudo-Areopagite (5th-6th cent.) speculated that the celestial hierarchies of angels and the orders of the Church were parallel. This
concept became a central theme of Carolingian theology after a manuscript of Dionysius (presented to Pepin I by Pope Paul in 758) had been
translated into Latin by Hilduin of St. Denis. (For more detail see Glossary, s.v. Hierarchy.)

[ILLUSTRATION]

HILDESHEIM. ST. MICHAEL'S CHURCH (1010-1033)

188.B

188.A

Alternating piers and columns at modular intervals is a leitmotif of Ottonian architecture, but has sporadic Carolingian antecedents in Reichenau-Mittelzell
(figs. 117, 134, 171), Werden (Vorromanische Kirchenbauten, 1966-71, 372ff), and the basilica of Solnhofen (see V. Milojcic,
Ausgrabungen in Deutschland, II, Mainz, 1975, 278-312).

abruptly and nor with as few historical preconditions as was
formerly thought. This raises the question: why, once
conceived, did it so suddenly disappear, not to re-emerge
until almost two centuries later?

The answer to this, I think, is relatively simple. The
square schematism, in the highly sophisticated and accomplished
form, which it attained in the layout of the Church
of the Plan of St. Gall, was born within the conceptual
framework of a building that had an overall length of no
less than 300 feet and for that reason could readily be
divided internally into a sequence of 40-foot squares. When
in the revisionary textual titles of the Plan it was suggested
that the church be reduced to a length of 200 feet and that
the columnar interstices be shortened from 20 to 12 feet,[312]
the modular order of the original layout was demolished.
There is no evidence to suggest that this reduction in size
was conditioned by structural or aesthetic considerations.
The change occurred as has been shown,[313] at more or less
the same time—and probably for the same reasons for
which—the abbot of Fulda was deposed for overtaxing the
spiritual and economic resources of his monastery with the
construction of a church considered by his monks as being
outrageously large. In this historical climate the dimensions
of the Church of the Plan of St. Gall, as laid down in the
drawing, could no longer be considered prototypal. The

[ILLUSTRATION]

JUMIÈGES, ABBEY CHURCH (1040-1067)

189.B

The red overprinting supplied by the authors on
Lanfry's fine drawing indicates where certain scars
in the original masonry give evidence of a structural
feature now vanished. This is by some interpreted as
a simple engaged column rising from floor to
clerestory wall-head level, by others as the seat of
abutment masonry of diaphragm arches. The
controversy requires thorough re-examination through
a masonry study made from scaffolds giving access to
full height of nave wall.

As long as such a study is lacking, and until a
structural engineering analysis is made, Ernest Born
and I prefer to keep the controversy alive, Born
favoring the former and I the latter interpretation.

W.H.

189.A

SEINE-INFÉRIEURE, FRANCE

Masonry scars in its clerestory walls (189.C, 189.D) prove that the nave of this Early Romanesque church was spanned by diaphragm arches
rising from engaged columnar shafts attached to every second pier of the nave (begun not before 1052). The square schematism and system of
alternating supports of Jumièges clearly derive from Ottonian architecture (fig. 188).

Columnar shafts introducing modular division into the nave walls first appeared in the cathedrals of Orléans (990) and Tours (ca. 990-1002)
and gained a hold in Germany, after the principle had been established in Speyer I (1030-1061). Jumièges goes further than Speyer through
use of diaphragm arches that carry modular division of nave walls transversely across the space. Diaphragm arches had previously been used in
the abbey church of Nivelles (1000-1046) and the cathedral of Trier (1016-1047). After Jumièges (1052-1067) they are found in other
Norman churches: St. Vigor-de-Bayeux (ca. 1060), Cérizy-la-Foret (ca. 1080), St. Gervaise-de-Falaise (ca. 1100-1123), and St. Georges-de-Boscherville
(after 1114). They become fashionable even in distant Italy: San Pier Scheraggio in Florence (ca. 1050-1086), Lomello (1060?)
and the magnificent San Miniato in Florence (ca. 1070-ca. 1150).

In all these churches the diaphragm arches were placed at intervals too large to allow vaulting between them. This step, the last in development
of the medieval bay system, was made in Speyer II (fig. 190).

[ILLUSTRATION]

189.C SOUTH WALL OF NAVE

Southwest view (toward the Seine and the quarry site for the stones
of Jumièges, showing clerestory windows.

Originally the nave of the church was covered by an open timber
roof, which in 1688-92 was concealed under a vaulted wooden
ceiling supported by sculptured brackets and foliated capitals inserted
on sill level of the clerestory windows.

On this occasion scars were left in clerestory walls through the
removal of some feature, which some believe to have been a
diaphragm arch (Pfitzer, Michon, Horn) and others a simple
engaged column (Martin Du Gard, Lanfry, Born).

[ILLUSTRATION]

189.D DETAIL

A close view shows one of the masonry scars left on the inner face of the
clerestory walls when the original feature for which it formed a seating was
removed, to make room for a vaulted 17th-century ceiling. It is the narrowness
and shallowness of these scars, as well as the height and thinness of the
clerestory walls, that induced earlier scholars to discard the assumption of
diaphragm arches.

Against this view it can be argued that for roughly two-thirds of their total
height, the nave walls are externally buttressed by the gallery vaults of the
church; and that along the lines where the scars occur, the clerestory is externally
reinforced by engoged buttresses rising from the galleries to clerestory wall-head
level. For a good summary of the controversy, see Michon-Du Gard, 1927.
47-54.

[ILLUSTRATION]

SPEYER CATHEDRAL (1082-1106)

190.B

190.A

[redrawn by Ernest Born after plans by
Dehio, Geschichte der Deutschen Kunst, 3rd ed., plate vol. I, figs. 68-69;
Kubach and Haas, 1972, pl. 9; and Conant, 1959, 75, fig. 22]

The great conceptual leap from Early to High Romanesque architecture was made by introducing continuous sequences of arch-framed vaults
springing from shafts that reached from floor to head of clerestory walls. Modularity, now embodied in an armature of architectural members
pervading and framing space in all directions, thus acquired its fully medieval form. The Ottonian "box-space" was transformed into the
bay-divided medieval space. The Gothic changed the vocabulary, but not the fundamental concept of space.

A basilica of magnificent longitudinal sweep and breathtaking verticality (70m. long, over 30m. high), Speyer was the first full embodiment of
this principle of composing churches in continuous sequences of clearly definable modular units of space.

grandiose scale of the original concept had received a
shattering blow in the neo-asceticism of the monastic reform
movement, and, in consequence, was abandoned.

The political chaos that followed the reign of Louis the
Pious offered no opportunities for a return to the earlier
concepts. Their renascence had to await the political and
economic consolidation that was brought about in Germany
by the house of the Saxon kings, and in France by the rising
power and importance of the dukes of Normandy that
peaked in the conquest of England.

The steps that lead to the re-emergence of square
schematism in Ottonian and Norman architecture are well
known and need not be reiterated. They are marked by
such highlights of medieval architecture as St. Michael's
Church at Hildesheim, 1010-1033 (fig. 188); the Abbey
Church of Jumièges, 1040-1061 (fig. 189); and the second
stage of the imperial cathedral of Speyer, ca. 1080-1106
(figs. 186 and 190).

St. Michael's at Hildesheim had a total length of 230 feet
and was internally composed of a sequence of seven
modules 30 feet square plus an apse with a radius of 20 feet
(fig. 188).[314] One could not wilfully construct a more convincing
mirror-image of the modular square division of the
Church of the Plan of St. Gall (figs. 61 and 173).

I do not know of the existence of any accurate measurement
studies of the Abbey Church of Jumièges (10401067).
But from the plans of Martin du Gard[315] and of
Lanfry[316] one gains the impression that it might have been
based on a modular sequence of 35-foot squares, four of
those composing the nave, one the crossing, one the fore
choir, and one half the apse, for a total of six and one-half
squares.

Whether or not the renascence of these modular concepts
at Hildesheim and Jumièges has any direct connection with
the Plan of St. Gall is impossible to say. The discussion of
this subject has suffered from the fact that until very
recently the square schematism even of the Church of the
Plan of St. Gall had been questioned.[317] Yet the similarities
can hardly be overlooked. As in the Church of the Plan of
St. Gall (fig. 61), so in Hildesheim and in Jumièges the
general dimensions of the principal spaces were calculated
as multiples of the crossing square. In both of these churches
this modular division was aesthetically underscored by a
rhythmical alternation of light supports with heavy supports,
the latter marking the corners of the module, the
former rising in the interstices between them. The system
has two isolated Carolingian precursors in the abbey
churches of Werden (dedicated 804)[318] and Reichenau-Mittelzell
(consecrated in 816)[319] but becomes a governing
principle of style only in the Ottonian period, starting with
the abbey church of Gernrode (961-965)[320] and leading
from there in successive steps of refinement through the
magnificent series Hildesheim[321] —Jumièges—Speyer. A
feature of primary developmental implications—completely
overlooked in all authoritative studies on the Abbey Church
of Jumièges—were the great diaphragm arches that spanned
the nave crosswise, rising from shafts attached to every
alternate pier.[322]

Aesthetically this is a first attempt to visually connect the
alternating support articulation of the nave walls with the
aid of a bold transverse member reaching full width across
the space of the nave as well as full height into the roof of
the structure. The diaphragm arch has been variously derived
from Roman,[323] Syrian,[324] Mohammedan,[325] and
Italian[326] sources; but its prototype is much closer at hand;
in the masonry arches that frame the area of intersection in
churches with nave and transept of equal height, and
establish in the transepts of these churches a modular cross
division of space that precedes that of the nave by centuries
(Church of the Plan of St. Gall, 816-17; Hildebold's
Cathedral of Cologne, after 800 and before 819; and perhaps
even the abbey church of St. Riquier, 790-799).[327]
The ultimate prototype of the diaphragm arch is, of course,
the triumphal arch of the Early Christian basilica[328] and

the testing ground for its migration from the transept into
the longitudinal body of the church are the aisles, where
precocious modular cross division by means of transverse
arches appear as early as the beginning of the ninth century
(Werden-on-the-Ruhr, dedicated by Bishop Ludger in 804
and Reichenau-Mittelzell, consecrated by Bishop Haito in
816).

The transept of the Cathedral of Speyer looks as though
it might have been conceived as a triad of 50-foot squares.[329]
The spacing of the piers in the original building (Speyer I,
constructed between 1030 and 1061) did not perpetuate
these dimensions; and when the nave, between 1080 and
1106 (Speyer II) was covered by groin vaults, mounted on
arches rising from shafts attached to every alternate pier,
this resulted in a sequence of oblongs rather than squares.
This variance in modular shape and size is an impurity of
minor importance; the epochal historical advance achieved
in Speyer was that the modular division of the ground floor
was here, for the first time, embodied in an all-pervasive
system of shafts and arches that divided the space lengthwise
and crosswise as well as in its entire height into a
modular sequence of clearly definable cells or bays. Once
this point was reached, the walls between the rising shafts
and arches could be perforated—and were in fact transformed
progressively into that intensely skeletal armature
of shafts and arches that led to the formation of the Gothic.

The self-contained and divisive vaults that covered the
bays of Romanesque and Gothic churches—firmly set off
against each other by their strong relief of framing arches
and ribs—were bound to strengthen the modular organization
of the spaces they covered. Yet they cannot by any
stretch of imagination be interpreted as a technical precondition
of that concept. Modular area division—as has
been made abundantly clear by the examples here cited—
preceded modular vault construction by centuries and
reached far beyond the realm of architecture into the layout
of the decorative pages of Christian service books. It has its
roots in a cultural frame of mind, not in technical conditions.

[ILLUSTRATION]

190.X GENOELS-ELDEREN DIPTYCH

190.Y

Shown same size
as original

BRUSSELS. MUSÉES ROYAUX D'ART ET D'HISTOIRE

[by courtesy of the Musées Royaux]

The monumentality of architecture in concept, execution, and fabric
may tend to overwhelm the scale of, and make distant, those objects that
men once handled and used in their daily pursuits. Tools, books, jewelry,
harness trappings, weapons, liturgical objects—with few exceptions they
are gone from us. The survivors, many of them precious then, as now,
lie in museums, remote from the purposes of their makers and rendered
exotic by their scarcity. Thus, the integration in spirit of such intimate
objects with monuments of architecture is somewhat difficult to achieve.
The many handicrafts that provided embellishment to daily life in a
monastic community such as was proposed by the Plan of St. Gall, has
been but lightly touched upon in this study. That works of art and
adornment were important to the community is undisputed. The Plan
has accommodations for making weapons and associated equipment,
saddlery and presumably other harness tack, and goldsmithing. Silversmiths,
lapidaries, and enamellers may have worked with armourer and
swordsmith. These crafts were housed with other facilities for more
ordinary work, in a pair of buildings in the southwestern tract of the
presumed site. Lay artisans were intended to reside in the community,
as is evidenced by comprehensive housing provided in the Plan.

Crafts that enhanced the praise of God by ornamentation of books,
vestments, and liturgical objects to assist in worship, were proper
activities for monks. Most notable were manuscript copying and
illumination, and ivory carving was likely among them. It is not
referred to specifically on the Plan of St. Gall, probably because its
execution did not require special facilities such as forges, smelters, and a
welter of noisy tools. The work of the ivory carver, silent and delicate,
often closely connected with all aspects of bookmaking, could be done
in a scriptorium, in company with scribes and illuminators.

The illustrated book cover is closely related to illuminations of the
Godescalc Gospels (781-783), earliest of the Court School manuscripts.
It has the same flatness of relief, the same delicate linearity, clearly
distinguishing it from the softly rounded forms and classicizing drapery
style of the later ivories of this school. The model must have been an
Early Christian ivory of Coptic or Syrian origin and representing a
style widely diffused in Merovingian Europe.

The front cover of the diptych shows Christ standing on the asp and
basilisk, flanked by two angels. The back cover displays the Annunciation
(upper register) and Visitation (lower register). Both covers are
pieced from several ivory plaques of different sizes. The work is
perforated and may have been mounted on a foil of gold leaf. The eyes
are inlaid lapis; interlace and step-patterns of the frames are clearly
influenced by insular art and stand in strong contrast to the perspective
illusionism of the two scenes. For references, see Braunfels, KARL DER
GROSSE, WERK UND WIRKUNG (exh. cat.), No. 534, pp. 345-46.

[ILLUSTRATION]

191. PLAN OF ST. GALL. MONKS' CLOISTER, THE CHURCH AND ADJACENT BUILDINGS

When St. Cuthbert built himself a hermitage on Farne Island, where he spent the last eleven years of his life in solitary retreat, he surrounded his
living space "with a wall higher than a man standing upright," and further increased its relative height "by cutting away the living rock so that
the pious inhabitant could see nothing except the sky from his dwelling, thus restraining both the lust of the eyes and the thoughts and lifting the
whole bent of his mind to higher things" (Bede, ed. Colgrave, 1940, 214-17). The Plan of St. Gall achieves a like effect for an entire community
in the sophisticated layout of the cloister with its egress and ingress governed by a body of rigid laws, the open inner court being the monks' only
access to nature and sun—a controlled and ordered island of nature with judiciously selected and carefully tended plants: PARADISUS CLAUSTRALIS.