University of Virginia Library

Search this document 
The Plan of St. Gall

a study of the architecture & economy of & life in a paradigmatic Carolingian monastery
  
  
  
  
 II. 
  
  
  

collapse sectionV. 
  
expand sectionV. 1. 
expand sectionV. 2. 
expand sectionV. 3. 
expand sectionV. 4. 
expand sectionV. 5. 
expand sectionV. 6. 
expand sectionV. 7. 
expand sectionV. 8. 
expand sectionV. 9. 
expand sectionV. 10. 
expand sectionV. 11. 
expand sectionV. 12. 
expand sectionV. 13. 
expand sectionV. 14. 
expand sectionV. 15. 
expand sectionV. 16. 
collapse sectionV. 17. 
 V.17.1. 
expand sectionV.17.2. 
expand sectionV.17.3. 
expand sectionV.17.4. 
expand sectionV.17.5. 
expand sectionV.17.6. 
collapse sectionV.17.7. 
  
  
  
expand sectionV.17.8. 
expand sectionV. 18. 
expand sectionVI. 

EZINGE, PROV. GRONINGEN, THE NETHERLANDS

The low-lying coastlands of the Netherlands and northern
Germany are dotted with man-made circular earthen works
on which the cattle-raising Iron Age settlers of this territory
erected their dwellings in order to protect themselves from
the heavy tides that flooded the surrounding flatlands
during the storms that lashed the shores of the North Sea
in the winter and spring. These dwelling mounds, called
Warfen or Wurten in German,[108] terpen in Dutch,[109] are
the product of the struggle of man against a geophysical
event of major importance which started some ten thousand
years ago, has as yet not subsided, and is even today only
temporarily checked by an elaborate system of dikes. Since
the retreat of the last great glacial cap of ice the shorelands
of Holland and northwestern Germany have gradually
sunk away in the course of a geological action in which long
periods of sinking alternated with shorter and less effective
periods of uplift.[110] The last of these cycles of sinking
started in the centuries immediately preceding the birth
of Christ and is still in progress. Prior to its inception the

[ILLUSTRATION]

302. WIJCHEN (GELDERLAND), THE NETHERLANDS

PLAN [after Bloemen, 1933, 6, fig. 7]

Alternation of heavy posts with saplings in the outer walls of both
houses reveals that the braided wattle walls did not form an
independent envelope, as with the Ezinge houses, but stood in line
with the outer posts.


50

Page 50
[ILLUSTRATION]

303.A FOCHTELOO (FRIESLAND), THE NETHERLANDS

HOUSES OF A WEALTHY WEST GERMANIC FARMER AND HIS FOLLOWERS, 1ST-4TH CENTURIES A.D.

303.A VIEW FROM THE AIR LOOKING NORTHWARD. RECONSTRUCTION BY A. E. VAN GIFFEN, 1954, fig. 85 [drawing based on a sketch by L. Posterna]

303.B PLAN OF SETTLEMENT SHOWN IN AIRVIEW

[ILLUSTRATION]

303.B

This large dwelling was associated with a hamlet of three
similar houses approximately the same width, but only half
its length. It was excavated in 1938 on a sandy elevation
of the Dutch
Geest. The presence of roof-supporting timbers
was determined by discoloration in the ground from where
they had rotted away. By this evidence it was ascertained
that the roof of the main house was supported by two rows
of free-standing inner posts, ten in each row, and that they
were of quarter-split oak sunk, rounded side inward, 0.75m
into the earth. This building was buttressed by a large
number of exterior posts set at an angle to help neutralize
the outward thrust of the roof. The walls were of wattle-daubed
clay; the rounded corners and absence of any timbers
capable of supporting a gable suggest that the building's
roof was hipped over its narrow ends. Both main house and
adjacent hamlet were protected by a pallisaded fence, and
the main house additionally by a ditch.


51

Page 51
dwellings of the coastland farmers of northern Germany and
Holland lay level with the flat land; but as the land began
to sink away, the water of the North Sea rushed in with
steadily increasing frequency and furor, and forced the
settlers to remove their dwellings to successively higher
levels. This they did by packing the floor level of their
houses with thick layers of turves and animal manure and
by re-erecting new dwellings on these mounds above the
inundation level of the heavy winter tides. As this process
continued century by century, it gave rise to a landscape of
man-made dwelling mounds attaining in their ultimate
stage a diameter of twelve or fifteen hundred feet and a
maximum inner height above the surrounding land of as
much as twelve to eighteen feet.

The effects, although not the cause, of this peculiar
geological phenomenon were known to Pliny the Elder,
who visited this territory probably in A.D. 47 and transmitted
his observations to posterity in a derisive yet
highly descriptive passage of his Historia Naturalis:

There, in a region of which one may wonder whether it belongs to
the sea or to the land, a miserable race of people dwell on elevated
mounds or platforms, thrown up by hand [tumulos optinent altos
aut tribunalia extructa manibus
], in houses erected above the level
of the highest tide, resembling men who travel in ships, when the
water floods the surrounding land, and shipwrecked people when
the waters have dispersed.[111]

A vertical profile cut through such a tumulus or Warf
shows as a rule a sequence of several convex layers of soil
in different coloration; the cultural remains reveal layer by
layer the story of the settlement as it was abandoned and
re-erected on each successive level. The physical composition
of these mounds offers unusually favorable conditions
for the preservation of organic materials, such as wooden
uprights, wattled fences or walls, or even objects made of
leather, since each abandoned settlement was covered by a
solid layer of clay which sealed its contents against the
corrosive action of the air.

In 1930 Albert Egges van Giffen dug a trial ditch through
a mound of this type at Ezinge (Groningen), Holland, and
the ensuing excavation (1932-34)—a landmark in the history
of European house research—enabled him to trace the
development of a West-Germanic settlement from its
beginnings in the fourth century B.C. to its end in the third
century A.D.[112]

The earliest settlement of this site (layer VI) was a
single farmstead (figs. 293 and 294), erected early in the
fourth century B.C. on the natural ground of the flatland. It
consisted of a three-aisled house with walls of wattlework,
and a vast enclosure almost entirely taken up by a platform
for the storage of hay or harvest. The timbers of the roof of
the house had disappeared, but the roof-supporting posts
and the braided walls of the house were preserved to a height
of almost a feet (fig. 293). They consisted of five pairs of
freestandings inner posts and a perimeter of thinner outer
posts. The wattle walls ran independently of this system,
slightly inside the ring of outer posts.

[ILLUSTRATION]

FOCHTELOO (FRIESLAND),
THE NETHERLANDS

304.A

304.B

HOUSE OF A WEALTHY GERMANIC FARMER

1ST-4TH CENTURIES A.D.

PLAN AT LARGE SCALE (A. E. VAN GIFFEN)

Plan of the main house (A) shows that aisles of the six westermost bays are cross
partitioned into stalls for 24 cattle. Entrances in the middle of each long wall lead to a
center bay that separates stock from the dwelling
(four eastern bays). B and C: Plans
of the main house, final condition.


52

Page 52
[ILLUSTRATION]

305.A LEENS (GRONINGEN), THE NETHERLANDS

AISLED HOUSE WITH TURF WALLS, A.D. 700-1000

PLAN [after A. E. Van Giffen, 1935-40, fig. 16]

The plan above is at level B noted on the transverse section below
306.A) with horizontal fold shading. Shown at right (306.B) is another
building.

[ILLUSTRATION]

305.B SECTION, EXCAVATION,

scale horizontally & vertically, 1:150

Ground penetration at right is about 3.5m = 11.5 ft.

[ILLUSTRATION]

LEENS (GRONINGEN), THE NETHERLANDS

306.A

306.B

AISLED HOUSE WITH WALLS, A.D. 700-1000

TRANSVERSE SECTION [after Zippellius, 1953, 32, fig. 5f]

Dwellings excavated at Leens were of great historical importance since they offered the
first archaeological proof that the aisled Germanic Wohnstallhaus continued to be built
in the Middle Ages. Fig. 306.A shows a house of Warf layer B of seven strata
spanning roughly 3 centuries. The structure was 38 feet long, 16 wide
(11.5 × 4.8m).
Layer B also held a house with wattlework walls, the soil structure of which indicated
it was almost 72 feet long.

When the water level of the North Sea had risen high
enough to make living on the flatland intolerable, the
single family dwelling of layer VI was buried under a
man-made mound of sods and turves (layer V) which,
after having reached a height of roughly 4 feet and a diameter
of approximately 90 feet, gave birth to a hamlet that
now comprised a total of five houses (figs. 295-297). These
houses belonged to the same construction type as did the
preceding settlement and were equally well preserved.
Three of them were provided with hearths, and hence
must have served as dwellings for people; one was inhabited
by both men and animals, evidenced by the
presence of both a hearth and two narrow strips of wattle-work
in front of the roof posts, which the excavator
interpreted as fodder mats, but which later excavations
proved to be dung mats.[113] The same condition appears to
have existed in the large house in the center, if this house,
as seems likely, had a hearth in its unexcavated eastern
section. Another smaller house, built at right angles against
this dwelling, had neither hearth nor dung mats, and hence
may have served as barn or general storage area. In the
houses that accommodated livestock the aisles were subdivided
into bays, or stalls, by means of braided cross
partitions, each of the thus-created boxes yielding sufficient
space for the stabling of two heads of cattle, facing the
outer wall perimeter of the house. Three of the houses had
their entrance broadside, two were entered axially. Pottery
shards and other cultural accessories associated with this
settlement permit a rough dating of the third century B.C.

In the second century B.C. the hamlet of layer V was
abandoned and the mound on which it stood was enlarged
to more than twice its original diameter and raised to a
level of 6 feet above the natural ground. On top of this
elevation a new village was built in a circle around an
open yard with the longitudinal axes of the house pointing
radially to the center of the Warf (layer IV).

The houses of this layer were of the same construction as
those of the preceding layers, but in general considerably
more spacious, as one may gather by glancing at the
extraordinary cattle barn reproduced in figures 298-299.
It had a length of over 75 feet (23 m.) even in its uncompleted
state of excavation. The posts and carefully
braided walls of this structure (twigs of birch daubed with
cow manure) were preserved in almost original freshness,
in spots to hip and even shoulder height. The building
contained no hearth, but dung mats ran along the inner
roof supports along the entire length of the structure, and
the aisles were systematically subdivided into stalls by
braided cross partitions.

The circular village to which this barn belonged was in
use from the second century B.C. to the first century A.D.,
but the life span of its houses was found to be considerably
shorter than that of the preceding layers. In certain sectors
van Giffen found that five to ten houses had been superimposed
upon one another in rapid succession; and intermittent
stratification of this settlement horizon with sterile


53

Page 53
[ILLUSTRATION]

307. HODORF (HOLSTEIN), GERMANY

AISLED FARM HOUSE, 1st-2nd CENT. A.D.

PLAN [by W. Haarnagel; after Schwantes, 1939, 272, fig. 10]

At the lowest level of the Hodorf WARF lay a flatland farm consisting of a
three-aisled main house divided into living and livestock areas, and an unaisled
barn built in axial prolongation of the house. In the layout of the plan two
measures are clearly discernible, the longitudinal measure of column interval A
and its half measure A/2. This measure and submeasure make up the length of
the house and its
AMBAU (= 111 feet). The width of the house (17 feet) appears
to be uniformly twice that of the center aisle. While the observation is simple and
even superficial, it hints, at this period and in this region, of an emerging
awareness of systematic measure in simple building practice and agriculture. All
trace of bulging curvature of wall line, or boat-like plan, has disappeared in
favor of a rather uniform rectangular geometry. Discipline of measure prevails
over scattered spacing and casual positioning of posts. A knot tied midway
between the ends of a braided rope could graphically solve the problem of
division by 2 for men unversed in the mystery of abstract arithmetic. It would,
too, lead to successive halving in series.

courses of sand gave evidence that this village, in its
initial stages at least, was still dangerously exposed to the
destructive action of the heavy winter tides.

In the centuries that followed, the second and third
centuries A.D., the Warf had to be raised again on two
successive occasions (layer IV-III). The house type remained
the same, except that in the later stages the wattle
walls were frequently reinforced externally by heavy layers
of turf. Toward the close of the third century, finally, the
village perished in a fire—an event that van Giffen connected
with the intrusion into the Frisian territory of the
first westward-moving Anglo-Saxons. The spacious three-aisled
houses were now superseded by small rectangular
huts which are of no interest to this study.

The excavation photos shown in figures 293 and 299 convey
in persuasive terms the unusual state of preservation in
which the Ezinge houses were found. They furnished
conclusive evidence about the construction of the walls and
the nature of the principal roof-supporting members (in
places preserved to a height of 4 feet above the ground),
but they told us nothing about the manner in which these
members were framed together at the top into a stable
roof-supporting system, nor how the roof itself was constructed.

There are, nevertheless, a few inferences that can be
made with relative safety from the conditions of the walls
and the placement of the posts. One of these is that the
roof must have been hipped over the narrow ends of the
house. This must be inferred from the fact that the two
end-walls of the house are not provided with posts that
could have carried a gable. The reconstruction of the roof
shapes shown in figures 295 and 297 render this condition
correctly.[114] Second, the principal posts must have been
framed together lengthwise by long beams which were
needed for the support of the rafters. There is no unity of
opinion, however, on whether the posts were in addition
connected transversely by crossbeams. Van Giffen felt
that, provided the posts were set sufficiently deep into the
ground, no such cross-connections were needed; and this
was also, in part at least, the opinion of Joseph Schepers.[115]
The technical soundness of this view, however, was questioned


54

Page 54
[ILLUSTRATION]

308. HODORF (HOLSTEIN), GERMANY

EXTERIOR VIEW, AISLED HOUSE

1st-3rd CENTURIES A.D.

[by courtesy of W. Haarnagel]

Although conjectural in many details, this model in the
Niedersachsische Landesstelle für Marschen- und Wurtenforschung,
Wilhelmshaven, is nevertheless a very convincing reconstruction of
the main house of the flatland farm of Hodorf unearthed in 1936-37.
It demonstrates that the Lower Saxon
Wohnstallhaus,
surviving examples of which date only to the 15th century, is in fact
a modern derivative of a prehistoric building type.

As in the similar Ezinge houses, the rafters of the roof were carried
by a row of posts placed slightly outside the independent wattle
walls. The rounded corners of these walls, and the absence of any
strong support at the building's narrow ends, suggest that its roof
was hipped. Four round posts around the hearth
(fig. 307) and
unaligned with the principal posts, are correctly interpreted as
supports for a canopy raised slightly above the main roof with lateral
openings for light, and smoke escape—a device well known through
the Sagas
(see p. 23ff) and crucial for interpretation of the guest
and service buildings of the Plan of St. Gall
(see below, pp. 117ff).

by T. Hermanns and Adelhart Zippelius[116] who
pointed out that if the posts were only connected by long
beams, the roof-supporting frame would still be exposed
to the danger of bending and buckling under the strain of
heavy loads of snow or the thrust of the wind during
storms. Moreover, cross connection is suggested by the
extremely accurate transverse alignment of the post, as it is
found in all of the Ezinge houses, as well as in the majority
of Warf dwellings subsequently unearthed. Whether the
cross beams lay beneath the longitudinal timbers, or above
them, must remain an open question.

There appears to be general agreement that the peripheral
row of posts—standing either within the walls of the house
or at a slight distance away from them—consisted of
short uprights terminating in a fork and carrying in that
fork a course of horizontal timbers which served as footing
for the rafters. The wattle walls themselves would have
been too weak to carry the roof. In some of the Ezinge
houses the outer posts were found to lean inward in close
adjustment to the angle of the roof thrust—a feature that
was encountered again in many houses subsequently
unearthed.[117]

The construction of the roof itself has been the subject
of some penetrating, yet careful and equally cautious,
observations made by Adelhart Zippelius.[118] Zippelius feels
that the layout of the Ezinge houses suggests that they were
covered by a continuous sequence of coupled rafters
(Sparrendach). The absence of any trace of posts along the
central axis of the house precludes the assumption of a
ridge pole. In primitive ridge-pole construction the two
sides of the roof were, in general, formed by means of
poles (in German called Rofe) which were hooked into the
ridge piece with their heavy ends upward and suspended
in the pole by a hook formed by the stub of a former
branch. This type of roof construction (Rofenkonstruktion),
ideal for houses of relatively smaller dimensions, could also
be employed in connection with aisled houses, but only if
the width of the nave was not much greater than the width
of the aisles.[119] Zippelius contends that in the Ezinge houses,
where the nave is generally twice the width of the aisles,
this system would not have worked, since the overhanging
portions of the roof poles (over the nave) would have outweighed
the lower portion of the roof, which covered the
aisles. The structural stability of the Ezinge houses required
that the roof poles were laid upon the supporting frame
with their light ends upward. Conjectural as all this may
be, it is based on sound speculation, and in the absence of
more tangible archaeological evidence provides us with as
good a working hypothesis as can be found at present.

Zippelius made some further, no less persuasive, assumptions
about the manner in which these timbers might have
been jointed. The easiest, simplest, and oldest method of
carrying a horizontal log is to lay it upon a row of timbers
terminating in a natural fork (fig. 300A)—a method that
continued to be employed long after more sophisticated
forms of joining had come into use, and is practiced even


55

Page 55
[ILLUSTRATION]

EINSWARDEN (NORDENHEIM), GERMANY

309.B

309.A

POST AND WATTLE HOUSE, around the birth of Christ

PLAN AND RECONSTRUCTION [after Zippellius, 1953, 38, fig. 8]

The site is on the estuary of the river Weser. The construction of
this house is virtually identical in all respects with the houses of the
cluster settlement of Layer V of the Ezinge Warf
(figs. 295-97 and
fig. 327
). Like most of those, as well as the chieftains's house at
Fochteloo
(fig. 304), its living quarters (two westernmost bays) are
separated from the stables
(two easternmost bays) by a center bay
entered through a door in the middle of the southern long wall,
while the cattle enter through a door in the eastern end wall. The
house is 33 feet long and 16½ feet wide.

today by primitive men throughout the entire world. When
natural forks of the desired height could not be found
among the available logs, the fork had to be shaped with
tools. The closest man-made imitation of the natural fork—
and here again I think Zippelius is correct—is a joint to
which he refers as Pfostenzange and which is obtained by
simply cogging the notched portion of a large beam into a
corresponding slit in the head of the upright beneath it
(fig. 300B). Another way of locking posts into horizontal
timbers (either at the top, bottom, or in-between) is by
means of mortice and tenon joints (Verzapfung), as shown
in figure 300C and D, or by halving them into one another.
Halving would also appear to be the most sensible joint for
the tips of the rafters, the connections being given additional
strength at this point, perhaps, by some braided strands of
willow. The reconstructions shown in figures 297 and 298
attempt to conform with this thinking.[120]

 
[108]

Warf: Old Frisian: warf, werf; New High German: werfen, "to
throw," but originally perhaps in the sense of "to whirl" ("a circular
mound created by the whirling action of the sand"); cf. Grimm, XIII,
1910, cols. 2012ff. Wurt: Old Frisian wort, related to Middle High
German worfen; cf. Heyse, 1849, 1990.

[109]

terp: Old Frisian thorp; New High German: Dorf; related to Greek
τύρβη; Latin: turba, "a gathering of small people in the open field," and
hence "a rural settlement;" cf. Franck's Etymologisch Woordenboek,
1929, 695 and 127, where it is related to the Indo-European word
*tereb- "to cut, to hoe;" cf. also Grimm, II, 1860, cols. 1276ff, under
"Dorf."

[110]

With regard to these geophysical events see Reinerth, I, 1940,
75ff; and Haarnagel, 1950.

[111]

Plinius, Historia naturalis, Book XVI, chap. 1; cf. Pliny, Natural
History,
ed. Rackham, 1952, 387, 389. (The English translation, here
quoted, is my own).

[112]

Van Giffen, "Der Warf in Ezinge," 1936; and idem, "Die Siedlunge
in de Warfen Hollands," 1936.

[113]

Van Giffen's interpretation of these mats as "fodder mats" was
questioned by Helmers, 1943, who interpreted them as "manure" mats,
in analogy with the later Frisian farmhouse, where the cattle invariably
stood with the head to the wall of the house. His interpretation was
confirmed when, in subsequent excavations, sewage trenches were
discovered in the place of, or running parallel to, the wattlework mats
(Wilhelmshaven-Hessens, Elisenhof; see below, p. 59, n.85 and p. 69.

[114]

The reconstruction shown in fig. 295 is taken from Reinerth, I,
1940, 88, fig. 25. The others are my own.

[115]

Van Giffen, "Der Warf . . . ," 1936; and idem, "Die Siedlunge . . . ,"
1936, 191: "Ankerbalken dürfen noch nicht angenommen werden,
Kehlbalken mögen dagewesen sein." Schepers, 1943 (Plate 9, fig. 58)
published a reconstruction of one of the Ezinge houses which shows the
terminal pairs of posts connected by tie beams, the ones farther inward
not so connected.

[116]

Zippelius, 1953, 37ff.

[117]

Most markedly so on the Elisenhof near Tönning (figs. 319 and 320
below, and Bantelmann, 1964, 233, as well as plate 62, figs. 1 and 2);
but also in Einswarden (fig. 309 below) and Haarnagel, 1939, 269; and in
Warendorf (see Winkelmann, 1954; and idem,).

[118]

Zippelius, 1953; and idem, 1954.

[119]

A typical example of a house making use of this type of construction,
according to Zippelius, is house 22 of a Celtic Hallstatt settlement on
the Goldberg, dating from about 800 B.C. (Zippelius, 1953, 19, fig. 2).

[120]

Both reconstructions were made before I had an opportunity to
familiarize myself with Dr. Zippelius' thinking. Fig. 298 is a revision of
and supersedes, an earlier reconstruction of this cattle barn which I
had published in an article dealing with the origins of the medieval bay
system (see Horn, 1958, 6, fig. 9).