ROOF
There can be no doubt that the frame of the roof of the
Church of St. Gall was meant to be constructed in timber.
It took two more centuries in the development of western
architecture before basilicas of major dimensions were
vaulted in stone. Since no timbered Carolingian church
roofs survive to guide us in our reconstruction, the details
of the carpentry of the roof of the Church must remain a
matter of conjecture. The earliest extant medieval church
roofs date from the twelfth century. They consist of simple
sequences of coupled rafters of uniform scantling rising
from the ends of a tie-beam and made rigid by collar beams
and an elaborate system of bracing struts. The cross section
of the twelfth-century roofs of the churches of St.-Germain-des-Prés
and St.-Pierre-de-Montmartre in Paris (figs. 124125)
are typical examples.
[170]
Although perfectly feasible for churches of moderate
dimensions, this roof design would not have been solid
enough, in my opinion, to safely span the vast interstices
between the clerestory walls of the larger Carolingian
churches. The nave of the Abbey Church of Fulda, 802819
(fig. 138) had an inner width of 17m (calculated by
Vonderau as corresponding to 60 Roman feet),[171]
and thus
was narrower than the nave of its Early Christian prototype,
Old St. Peter's in Rome (fig. 141), by only a small margin
(18.80m, listed by Volbach as corresponding to 61 feet,
8 inches).[172]
I am inclined to believe that the roof that
covered the basilica of Fulda derived its design from the
same source that inspired the entire building.[173]
That the
roof of Old St. Peter's was well known to Frankish architects
may be inferred from two letters of Pope Hadrian I to
Charlemagne (one written between 779 and 801; the other,
between 781 and 786), in which the pontiff asks the
emperor not only for the beams for the repair of the roof
of St. Peter's but also for a magister to supervise the work—
clear evidence of the high esteem Frankish carpenters and
builders enjoyed in Rome in those days. The Pope asked
for the services of no lesser man than Wilcharius, Bishop of
Sens, to direct the restoration.[174]
Carlo Fontana made an engraving in 1694 of the roof
trusses spanning the nave of Old St. Peter's (fig. 126); these
he considered to be an authentic record of the original
(Early Christian) roof of the church.[175]
The nave span of the
Church of the Plan is only about one-half (40 feet) of that
of Old St. Peter's, and therefore, would not have required a
roof of such heavy design. In our reconstruction of the roof
of the Church (figs. 108-110) we have been guided by a roof
design which, it is believed, was a standard Early Christian
type, and which, to judge by a description in Vitruvius'
Fourth Book,[176]
must also have been standard for broad
spans in Roman Imperial times. Vitruvius distinguishes
between two roof types, one suited for "spaces of relatively
small dimensions" (commoda spatia), the other for buildings
involving "broader spans" (majora spatia). The former,
according to his description, consisted of two simple
rows of rafters converging at the top in a ridge beam and
extending downward all the way out to the eaves of the
building (columen et cantherii prominentes ad extremam
subgrundationem); the latter was made up of a sequence of
vertical trusses which supported the covering of the roof
by means of purlins. Vitruvius lists the different parts that
make up this frame and tells us that their names express
their different functions (ea autem uti in nominationibus ita
in re habet utilitates); "Under the roof, if the span is
broader, there are tie-beams (transtra) and bracing struts
(capreoli). . . . Above the principal rafters (cantherii) there
are the common rafters (asseres) extending outward sufficiently
to protect the walls with their overhang."[177]
This terminology is indeed highly descriptive and typical
of the classical habit of defining the functions of inanimate
objects by imagery borrowed from animate life. Cantherius
(a beast of burden) is an appropriate term for the load-bearing
action of the rafters; capreolus (a wild goat) expresses
vivdly the butting action of the diagonal timbers
locked in the center and at the bottom of the king post like
the horns of two fighting goats; transtrum, derived from the
preposition trans (across) is equally expressive of the purpose
of the large crossbeam that forms the base of the truss.
Vitruvius fails to furnish us with the name for the king
post, which in this type of construction rises almost
invariably from the center of the tie-beam to the ridge pole.
The primary function of this post is not to support, as has
been frequently thought, but to serve as a base of departure
for the diagonal bracing struts which prop the rafters midway
in their span, and thus prevent them from sagging
inward under the load of the roof covering.[178]
The early
translators and commentators on Vitruvius have interpreted
these descriptions of the two basic Roman roof
types correctly; and such reconstructions based upon these
interpretations as are found, for instance, in Barbaro's
Italian translation of Vitruvius, published by Francesco
Marcolini in 1556 (fig. 127)[179]
or in the 1827 edition of
Vitruvius published with commentary by Joannes Polenus,[180]
cannot, in my opinion, be improved upon.
The correctness of such interpretations has recently been
confirmed by George H. Forsyth's extraordinary discovery
of the original timbers of the roof of the sixth-century
church of St. Catherine at Mount Sinai, which is the
earliest surviving example of its kind.
[181]
But the design of
others, of even earlier date—exhibiting the classical truss
formed by tie-beam, rafters, and bracing struts
[182]
—is engraved
into the masonry gables of the porches of certain
Syrian churches of around 400 (figs. 128 and 129). In Italy
this roof type survived unchanged throughout the Middle
Ages. One of the finest extant examples is the magnificent
fourteenth-century roof of the church of San Miniato al
Monte in Florence.
[183]
In conformity with this well-attested Roman and Early
Christian roof tradition, we have reconstructed the roof of
the Church of the Plan as a trussed timber roof with purlins
supporting an outer set of rafters (figs. 108-110). The proportions
of the Church suggest that these trusses were
placed at a distance of 20 feet from one another over the
center of each nave column, or at intervals of 10 feet, if we
assume intermediate trusses over the apex of each arcade
midway between the columns. Our roof pitch is of course
purely conjectural. Since the Church of the Plan lies
stylistically midway between the Early Christian and the
Romanesque, we have constructed it at an angle of 45
degrees—a pitch considerably more obtuse than that of the
average Early Christian roof, yet substantially more acute
than the average roof of the transalpine churches of the
tenth and eleventh centuries.