V. 6
CRITERIA OF RECONSTRUCTION
II
V.6.1
ROOF CONSTRUCTION:
ASSEMBLAGE OF THE SUPPORTING
FRAME OF TIMBER
The aisled and timbered medieval barns and houses discussed
in the preceding pages give us a fairly good idea of
the kind of carpentry we might expect to have been employed
in the guest and service buildings of the Plan of St.
Gall. Although we could identify two distinct classes of
roofs, we found that the roof-supporting trusses were of
very similar design in both cases. In the majority of buildings
the arcade plates from which the rafters spring were
tenoned into a recess in the head of the posts. The tie
beams were laid upon this assemblage from above and
were locked into the posts by means of mortice-and-tenon
joints, into the arcade plates by dovetail joints (fig. 356).
One building only, the Barn of Great Coxwell, Berkshire,
departed from this rule by putting the plates on top of the
tie beams (fig. 357). We have chosen the more common
form of Chichester (fig. 356) and Parçay-Meslay (fig. 354B)
as the standard form for our reconstruction of the post-and-plate
assemblage of the houses of the Plan of St. Gall.
The bracing beams were either straight, bent, or curved;
sometimes long, sometimes short; sometimes single, sometimes
double; but in all cases of relatively heavy scantling.
We may safely assume that all of the simpler variants of
this group were in use in Carolingian times, excluding such
types, of course, that owed their design to influences from
Romanesque and Gothic masonry architecture, such as the
round arches of Hereford Palace (fig. 340) or the Gothic
arches of Nurstead Court, Kent (fig. 345) and Little
Chesterford, Essex (fig. 348).
From the side of the aisles, the principal trusses were
steadied by aisle ties, tenoned into the freestanding posts
on the level of the wall head. The outer ends of these ties
served as springing for inner rafters of heavy scantling,
running parallel to the roof slope, a short distance inward,
and butted into the heads of the principal posts some two
or three feet below the arcade plate. On the wall side these
aisle ties rested either on two parallel courses of wall plates,
one running along the outer, the other along the inner edge
of the wall head, as in Great Coxwell (fig. 350B); or they
were tenoned into the heads of posts set against the long
walls on the inside, a method by means of which the timber
frame was, structurally, held virtually independent of the
masonry walls, as in Parçay-Meslay (fig. 354B). In buildings
constructed entirely in wood the post-to-plate assemblage
of the outer walls would, of course, have been a repetition
on a smaller scale of that of the principal posts and the
arcade plates (as in the barn of Little Wymondly, discussed
below, fig. 434). In reconstructing the St. Gall outbuildings,
we have chosen freely, from among all these different possibilities,
whatever the condition of a particular building
suggested as the most logical solution.
V.6.2
ROOF CONSTRUCTION:
SOME ALTERNATIVE ASSUMPTIONS
So far we are on fairly safe ground. The range of possibilities
widens drastically, however, as we move from the
main stage of the trusses to the assemblage of the roof itself.
Here, in light of the available historical evidence, we are
forced to make a choice between two radically different systems:
the roofs of the guest and service buildings must
either have belonged to the family of the Sparrendach or to
the family of the Pfettendach. As we have no way of knowing
to which of these two they belonged—rather than decide
this issue in an arbitrary manner—we have chosen to use
both systems. Our reconstruction of the House for Distinguished
Guests (figs. 397-399)[215]
is a typical example of the
purlin roof; that of the Granary (figs. 435A-F),[216]
of the
rafter roof. It is not at all impossible that both types
were used together—and there is also, of course, the possibility
of early hybridization.
We have fashioned our reconstruction after English and
French models rather than Dutch and German, because
the English and French material is older than the earliest
Dutch and German parallels, which do not antedate the
beginning of the sixteenth, or at best the end of the fifteenth,
century.
There is one further alternative to be taken into consideration.
Among the excavated pre- and protohistoric
houses previously discussed are found a small number of
buildings that had their ridges supported by a median row
of posts (see Wijchen, figs. 301-302).[217]
They are few, true
enough, but their existence forces us to take them into
account, the more so since in the territory of the Bajuvarians,
at least, this house was common enough to merit legal
codification (see above, pp. 27ff and figs. 289A-B).
I do not believe, however, that the guest and service
buildings of the Plan of St. Gall should be reconstructed
in this manner, since the houses of the Plan of St. Gall have
vital passageways at the very spot where the construction
of the Bajuvarian standard house calls for ridge-supporting
center posts. I draw special attention to the doorways in
the House for Distinguished Guests connecting the dining
hall with the bedrooms of the royalty housed in this structure.[218]
As one inspects the remaining houses one by one,
one finds that the entrances from the hearth-room to the
two end rooms invariably lie in the longitudinal axis of the
building. This arrangement is incompatible with the ridgepole
construction of the Bajuvarian standard house.
Our choice for the roof skin of the guest and service
buildings is shingle. Thatch and reed, while common in
the sixth, seventh, and eighth centuries, would have been
an anachronism in the ninth in a monastery of paradigmatic
significance. I remind the reader of the passage already
quoted in the Life of St. Benedict of Aniane where we are
told that he covered the houses of his monastery at Aniane
first with thatch (non tegulis rubentibus, sed stramine) and
then completely redid them in tile (non iam stramine domos,
sed tegulis cooperit).[219]
In the southern, more Romanized
parts of the Carolingian empire, tile was doubtlessly the
customary material. Farther north it is more likely to have
been shingle, probably the larger variety which in vernacular
American English is referred to as "shakes." Numerous
medieval sources could be quoted in support of widespread
use of shingles.[220]
I confine myself to one, the well-known
passage in Ekkehart's Casus sancti Galli, where we are told
how the fire set by a pupil to the roof of the Outer School
of the monastery of St. Gall ignited "the dry shingles"
(tegula arida) of the school and from there was blown by
the north wind to the roof of an adjacent church tower,
which had "a shingled roof superimposed upon a stone
roof" (tegulis ligneis super lapideas tecta).[221]
V.6.3
CARPENTRY JOINTS
I am intentionally not entering into a discussion of carpentry
joints, but refer the reader to the masterful review
of pre-medieval methods of jointing timbers, published by
Adelhart Zippelius in 1954,[222]
which discloses that all later
known methods of assembling timbers, such as halving
(Verkämmung), joining by mortice and tenon (Verzapfung),
and lapjoining (Verblattung) were in full use in the Iron
Age, some being attested for the Bronze Age and even for
the Younger Stone Age (fig. 300), including the very sophisticated
dovetail joint. In our reconstruction we have not
used any carpentry joints that are not well attested for the
Middle Ages. We cannot prove that all of these were in use
at the time of Louis the Pious. Some of them possibly were
not. But to anyone who is inclined to underrate the skill of
Carolingian carpenters we recommend a study of the intricate
carpentry joints discovered in the two fortified ninth-century
sites of Stellerburg[223]
and of Husterknupp.[224]
V.6.4
PROCEDURES FOLLOWED IN
RECONSTRUCTION
In what follows, then, we are entering upon the task of
reconstructing the guest and service buildings of the Plan
of St. Gall, structure by structure. All buildings are drawn
to scale. But certain liberties of interpretation are inevitable
when the constructional elements of a house whose
vertical elements are rendered in simple line projection have
to be converted into three-dimensional entities. All vertical
values will by necessity remain a matter of speculation. In
calculating our elevations we have followed the same method
that we used in our reconstruction of the Church,
assigning to each component space in the house the comfortable
minimum proportions required by its function. We
have assumed that the roof slopes follow, in general, an
angle of 45 degrees. This is a reasonable, yet by no means
compelling, assumption.
The working out of the constructional details of some
forty-odd buildings is not only an arduous, but also a very
costly task—so costly and complex, indeed, that we might
never have accomplished it had the Council of Europe
exhibition Charlemagne (held in Aachen in the summer of
1965) not afforded us the unique opportunity of testing our
views in the construction of a three-dimensional model of
the monastic buildings shown on the Plan of St. Gall.
Although in the majority of cases this model called for
exterior views only, it was obvious enough that even this
objective could not be attained unless the internal constructional
problems of each house were settled. This required
for each individual building a complete set of work-drawings,
sufficiently detailed and comprehensive so as to allow,
if need be, for reconstruction of the houses in their actual
size.[225]
It is on these drawings that the reconstructions in
the present study are based.
We cannot dare to claim that our interpretations are the
last word on this challenging subject, but we believe them
to be based on sound historical assumption. We are certain
that they are sound in all of their constructional implications.
Before turning to the discussion of the individual
structures, however, an important detail remains to be
settled, i.e., the question of the heating, lighting, and ventilation
devices of the guest and service buildings.
