University of Virginia Library

V.2.2

ARCHAEOLOGICAL EVIDENCE

Until the end of the second decade of this century the
literary sources discussed on the preceding pages were all
that students of early medieval house construction had to
lean on when discussing the question of Northern parallels
for the guest and service structures of the Plan of St. Gall.
To be sure, some isolated excavations had already been
made in Sweden and Iceland,[100] but this fact was not
widely known; and the procedure for unearthing houses
whose structural members, in many cases, could be identified
only by a shadowy patch of soil discoloration left in
the ground as they rotted away, had as yet not developed
into that highly accomplished technique so successfully
practiced today. But in 1928-1931 this situation began to
change when John and Nils Nihlen laid bare on the island
of Gotland two three-aisles houses of the Migration Period
which looked like physical embodiments of Gudmundsson's
Saga house. In 1930-1934 the Dutch anthropologist Albert
Egges van Giffen initiated a new era of northern house
research with the excavation of an Iron Age dwelling
mound in a hamlet called Ezinge (Groningen) in Holland,
which revealed that a very similar type of dwelling was in
use as early as the fourth century B.C. in the territory of the
Frisians, a West Germanic tribe. In the two following
decades the information gathered from these excavations
was broadened by an increasing number of further discoveries.
At the date of this writing we are able to trace, on
the basis of several hundred excavated dwellings, the
development of the timbered three-aisled house in the
Germanic territories of transalpine Europe from its beginnings
in the Middle Bronze Age through the Iron Age, into
the Early Middle Ages, and through the Middle Ages to
its modern survival form

KÄNNE, BURS, GOTLAND, SWEDEN

The first example of this long line of excavations, as just
remarked, was a three-aisled dwelling, excavated in 1928 by
John and Nils Nihlen, in a place called Känne, in the
parish of Burs, in East Gotland (fig. 290).[101] It was 33 feet
wide (10 m.) and had the extraordinary length—not as
yet matched by any dwelling subsequently unearthed—of
203 feet (62 m.) A recent review of the site has disclosed
that the hall was constructed in two successive phases, and
in its original state was only half as long.[102] Its roof was
supported by two rows of freestanding inner posts, rising
in pairs, at intervals of 9 to 13 feet (3 to 4 m.). Each of
these uprights was firmly secured in the ground by a
ring-shaped wrapping of stones. Over fifty charred beams
and numerous fragments of wood were found on the floor;
among these were the remains of two large beams which
were jointed into each other at right angles. The walls
consisted of solid banks of earth heavily interspersed with
small stones and were faced, outwardly and inwardly, with
a strong lining of heavier stones. The roof must have been
covered with sods of turf, as no other material would have
smothered so effectively the fire that destroyed the house
yet preserved so much of the timbered frame of the roof.
The hall received its warmth from two hearths which lay in
the middle of the center aisle, one of them 33 feet (10 m.)
long. "Longfires" of this kind are well attested from the
Sagas, where they are referred to as langeldar or máleldar.[103]
The general character of the accessories found in the house
pointed to about the year A.D. 200 as the approximate
period of construction.

LÖJSTA, GOTLAND, SWEDEN

The second house, a structure of more normal proportions,
85 feet by 33½ feet (26 m. × 10.5 m.), was excavated in
the summer of 1929 in the vicinity of castle Lojsta in
Gotland (fig. 291A-C).[104] It was the same construction type
except that here the roof-supporting posts were not sunk

[ILLUSTRATION]

298.B EZINGE (GRONINGEN), THE NETHERLANDS. CATTLE BARN OF Warf-LAYER IV, 2nd CENTURY B.C.

[author's reconstruction, drawn by Walter Schwarz]

[ILLUSTRATION]

298.A PLAN
REDRAWN FROM VAN GIFFEN

1:150

[ILLUSTRATION]

299. EZINGE (GRONINGEN), THE NETHERLANDS. CATTLE BARN OF Warf-LAYER IV, 2nd CENTURY B.C.

[excavation photo by courtesy of A. E. Van Giffen]

This building, like those unearthed above and beneath it, owes its magnificent state of preservation to the fact that each settlement stratum in
which houses were buried in the course of successive inundations was sealed by sterile layers of sand and clay deposited after flooding, sealing
their content against the infiltration of air and thus protecting it from decay. The roof-supporting posts of oak, the braided walls and cross
partitions (wattled saplings of birch) were preserved to a height of 4 feet. The manure mats were found to be in such good condition that they
could be walked upon without breaking. The building was 29 feet wide (7.20m) and over 75 feet long (23m) but was never excavated to its full
length. Its construction was identical with that of the houses found in the earlier Warf layers (figs. 293-297).

The systematic division of the aisles into stalls, together with the absence of any fireplaces, suggests that it was used for the stabling of livestock
exclusively. Since every stall had room for two head of cattle, this barn must have been able to hold at least 48 animals, striking evidence of
the economic wealth of these early shoreland farmers. Livestock entered and left the building through doors in the two narrow ends—a feature
found in many other early Iron Age houses (figs. 304, 310, 312, 315, 316), and today in the Lower Saxon Wohnstallhaus and the Frisian
Los-hus, modern descendants of this building type.

[ILLUSTRATION]

PRE- & PROTOHISTORIC CARPENTRY JOINTS

300.A

300.B

300.C

300.D

[after Zippelius, 1954, figs. 1, 2, & 5]

A. Forked posts (Neolithic)

B. Post with slit head

C. Mortice and tenon joint in post and plate assemblage
(Neolithic)

D. Mortice and tenon joint in post and ground sill assemblage
(Bronze Age)

in the ground but rested on slabs of stone. All of these
stones were still in their original position (fig. 291A). The
posts themselves had disappeared. Rising freely from stones
as they did, they could only retain their vertical position by
being framed together at the top by means of cross beams
and long beams. Slight irregularities in the longitudinal
alignment of the posts suggested that the cross beams lay
underneath the long beams. The excavators felt so sure of
their interpretation of these conditions that they undertook
to reconstruct the entire hall on its original site (figs. 291B
and C). Some of the details of this reconstruction have since
been questioned, but the doubts amount basically to no
more than that in the original house the walls were probably
a little higher than they are shown at present.[105] The
pottery found in the house suggests as period of construction
the third century A.D. In the fifth century, for unknown
reasons, the hall appears to have been abandoned.

In the two decades that followed probably more than
sixty houses of the Lojsta type were unearthed on the
islands of Gotland and Öland, on the mainland of Sweden,
in Norway and in Denmark,[106] and, last but not least, in
Iceland, the country whose literary tradition introduced us
to this type of dwelling.

STÖNG, ÞÓRSÁRDALUR VALLEY, ICELAND

Iceland was the subject of an expedition undertaken between
1934 and 1939 by a joint excavation team of Danish,
Swedish, Norwegian, Finnish, and Icelandic archaeologists.[107]
I show as a typical example of the results of this
expedition, the plan and excavation photos of the dwelling
of a farm called Stöng in þórsárdalur Valley (fig. 292A-C)
which was settled during the landnám period at the end of
the ninth century and covered by the ashes of nearby
Mount Hekla in an eruption in the year 1300. The dwelling
unit of this farmstead consisted of a long house 98 feet
long, divided into foreroom, sleeping house, living house
(forstofa, area I; skáli, area II; stofa, area III), milkhouse
(area IV), and cooler (area V). The sleeping hall was
54 feet long and 19 feet wide. Its aisles were raised so
as to form continuous "benches"—the langpallar of
Gudmundsson's Saga house. Inserted into the curbs
of these benches about every 6 feet were large blocks
which served as base stones for the wooden uprights that
once supported the roof of the hall (fig. 292B). The fireplace
lay in the middle of the center floor. Two shallow stone
foundations which bisected the aisles crosswise suggest that
the sleeping hall was subdivided by means of wooden cross
partitions into a sleeping house for men (karlskáli) and
another for women (kvennaskáli)—a distinction also well
known from the Sagas. And judging from the presence of
two rows of stones ranged carefully along the base line of
the two long walls, the hall must have been wainscotted
its entire length (the veggþili or langþili of the Sagas).
There was a "crossbench" (þverpallr) on the entrance
side of the hall, raised like the aisles and screened off by a
wooden cross wall. I am drawing attention to this house not

only because it is the keystone of cumulative archaeological
evidence that established the correctness of Gudmundsson's
literary work, but also because this dwelling may date from
the same century in which the Plan of St. Gall was drawn.
During the Iceland expedition of 1934-39 a total of eight
such houses was unearthed. But by the time these excavations
were conducted, discoveries of even greater significance
were in progress on the Continent.

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.

[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.

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:

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.

[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

[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

[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

[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]

WIJCHEN, MAAS ESTUARY, THE NETHERLANDS

When the Ezinge houses were discovered in 1930-34 they
were a new and entirely isolated phenomenon on the
Continent. But in the five years that followed, before the
outbreak of World War II, every subsequent summer
brought new results. While van Giffen was still at work at
Ezinge, F. Bloemen unearthed under less favorable soil
conditions another group of aisled houses of the first
century B.C. on a mountain range near the estuary of the
river Maas, near Wijchen.[121] The ground plans showed the
transverse alignment of inner and outer posts, which was
typical of the houses of layer V and IV of the Ezinge Warf.
The outer walls consisted of an alternating sequence of one
heavy and two lighter posts; the heavy posts stood in line
with the principal posts (figs. 301, 302). In other aspects,
however, the construction differed. The houses had posts
along their central axes, an arrangement that is in general
interpreted as an indication of the presence of a ridge pole.
The excavation showed that ridge-pole construction, although
unusual, was nevertheless not absent in this territory,
an observation that was confirmed by later finds in
other places.[122]

[ILLUSTRATION]

310.X ISOMETRIC VIEW

[ILLUSTRATION]

310. CROSS SECTION

The house was 102 feet long, 29 feet wide. The nave and one aisle were 10½ feet
wide, the narrow aisle 8 feet wide. The distribution of stones—some for pavement
some for lining or packing of wall-post sockets, others for footing of principal
roof supports—reveals that the house was divided lengthwise into a nave and two
aisles, and crosswise into fourteen bays. In the first ten, only the center floor was
paved, and the aisles were strewn with sand. In the last four bays the pavement
ran across the width of the dwelling. This is the well-known T-shaped floor plan
of the Lower Saxon Wohnstallhaus.

In the house above, bay depth in the stable was 6½ to 8¼ feet. In the living area
the distance between trusses increased, and in the terminal bay containing the
hearth is almost twice as deep as the others. Since the principal inner posts of
the house were footed on stone blocks rather than in post holes, they must have
been framed at their heads by long beams and cross beams somewhat in the
manner shown above. The ANKERBALKEN (cross beams terminating in long
tenons morticed into the main posts a few feet below the tie beams) shown in
Rieck's reconstruction (Reick, 1942, fig. 2) appeared to us to be an anachronistic
feature for so early a structure and for that reason has been omitted in our cross
section.

AALBURG, near BEFORT, LUXEMBOURG

AISLED HOUSE, 5TH CENTURY B.C.

[ILLUSTRATION]

311. PLAN [after G. Rieck, 1942, 27, fig. 1] 1:125

FOCHTELOO, RHEE, SLEEN, AND LEENS,
THE NETHERLANDS

Bloemen's excavation was followed with the discovery by
van Giffen in 1935, 1936, and 1937 of a group of settlements
of the first and second centuries A.D. near the
villages of Fochteloo, Rhee, and Sleen; and in 1938, again
near Fochteloo,[123] of a settlement of the same period which
van Giffen believed to be the farm and residence of a
chieftain (figs. 303-304). This settlement comprised a long
house, protected by fence and ditch, and a nearby hamlet,
likewise fenced in, consisting of three smaller houses and
a couple of open barns. All of these houses were aisled and
were entered broadside by two entrances lying opposite
one another in the middle of the long walls and giving access
to a median crosswalk that separated the stables for the cattle
from the living quarters of the people. The long house of
the chieftain had a third additional entrance at the rear of
the stables, primarily for the use of livestock. This house
was 70 feet long and 21 feet wide (21·40 m. × 6·50 m.).

The great significance of van Giffen's excavations of
Ezinge was that they solved an enigma that had puzzled
students of European house construction for over a century.
They brought to light the prehistoric prototypes of two well-known
and closely related modern house types, namely
that of the Lower Saxon "Wohnstallhaus" and of the
Frisian "los-hus." The oldest surviving specimens of these
two widespread house types date from the early sixteenth
or, at the most, from the end of the fifteenth century.[124]

Van Giffen's excavations demonstrated that this type
was infinitely older than anybody had heretofore presumed
it to be, and their immediate prototypes could now be
traced back as far as the fourth century B.C. It was clearly
only a matter of time for the connecting medieval links to
be found. Once more it fell to van Giffen to lead the way
in this search. A trial ditch run through a Warf in the
vicinity of the village of Leens (Groningen), Holland,
revealed the profiles of a settlement whose life span started
approximately at the point where that of Ezinge ended.
And in a systematic excavation of this Warf conducted in
the subsequent year, van Giffen[125] could trace his aisled
Iron Age house through seven successive layers from the
end of the seventh century A.D. to the beginning of the
eleventh. Altogether some twenty-three houses came to
light: some of them built with wattle walls, others with
walls of turves; but all of them had their roofs supported by
two rows of freestanding inner posts. I reproduce as a
typical example the plan of a house of Layer B (fig. 305),
after van Giffen, and a cross section of this house (fig. 306),
as suggested by Zippelius.[126]

HODORF, SCHLESWIG-HOLSTEIN, GERMANY

Van Giffen's work in Holland was only a beginning. In
1936 Werner Haarnagel launched the first of an equally
exciting series of excavations in the adjacent coastlands of
northern Germany, where he discovered a Germanic flatland
farm of the first and second century, near the village
of Hodorf, Schleswig-Holstein, on the banks of the river
Stör, not far from its confluence with the river Elbe.[127] It
consisted of a three-aisled dwelling with hearth, to which a
one-aisled barn was added axially at a slightly later date
(fig. 307). The construction method employed in this dwelling
was identical, in all details, with those of van Giffen's
houses at Ezinge: six pairs of inner posts serving as principal
roof supports, an outer perimeter of wall posts serving
as footing for the rafters, plus the customary envelope of
wattle walls running in total independence of the supporting
members. The aisles were divided into cattle stalls in the
rearward part of the house, as in Ezinge, except that in
Hodorf this area was entirely matted with wattlework. A
distinctive feature of the Hodorf farm was that its hearth
was framed by four posts which were out of line with the
principal roof supports and also differed from the latter by
being round. They were obviously not part of the regular
structural system. Haarnagel thought that they might have
carried a smoke flue, or that they belonged to a separate
inner armature of poles which carried an elevated section
of the main roof over an opening in the ridge above the
hearth site, serving as light source and as smoke outlet. A
similar arrangement of poles ranged in a square around the
hearth had been observed in other Iron Age houses in
vastly distant places.[128]

Haarnagel has reconstructed the Hodorf house in a handsome
model which is displayed at the Niedersächsische
Landesstelle für Marschen- und Wurtenforschung in the
city of Wilhelmshaven (fig. 308). While many of the details
in the roof section of this model must by necessity remain
conjectural, the concept of the house as a whole is unquestionably
sound. The pottery found in the Hodorf house
indicated as time of occupancy the first and second century
A.D. Toward the close of the second century the site was
imperiled by tidal inundations. Its inhabitants made an
attempt to save the house by filling it up inside with sand,

[ILLUSTRATION]

312. WILHELMSHAVEN-HESSE, GERMANY

EXTERIOR VIEW, AISLED HOUSE, 6TH-9TH CENTURIES

MODEL IN THE NIEDERSÄCHSISCHES LANDESINSTITUT FÜR
MARSCHEN- UND WURTENFORSCHUNG

[after Haarnagel]

The excavation of Warf of Wilhelmshaven-Hesse (trial ditch in
1939 disrupted by World War II, resumed in 1949, continued in
1950) offered the first evidence that the aisled and timbered Iron
Age hall known through the excavations of Hodorf (figs. 307-308)
and Einswarden (fig. 309) continued to be used in the coastlands of
northern Germany in early medieval times. As it was excavated the
Warf revealed, in settlement horizons extending from the 7th through
10th centuries A.D., aisled and bay-divided houses ranging in
length from 39½ to 59 feet, and in width, 17½ to 21 feet.

The house model shown here is a reconstruction of one of the larger
houses of the Warf. Like those of Fochteloo and Einswarden, it had
an axial entrance for cattle in one of the narrow walls and a lateral
entrance to the living area close to the opposite end of the house.

a little more than 2 feet above its original floor level. A
number of posts were reset on this occasion, and the roof
may have been replaced entirely, but in all other respects
the house remained the same, except that now it was used
exclusively as a dwelling. It continued to be used in this
form until the end of the third century when it made room
for a new but smaller house of the same construction type.

EINSWARDEN, GERMANY

The operations at Hodorf had barely been completed, when
in the winter of 1937/38 Haarnagel was called to a site in
the vicinity of the village of Einswarden,[129] on the left bank
of the estuary of the Weser river, where the heavy machinery
of a modern land improvement project had edged into the
core of an ancient dwelling mound. Systematic excavations
were undertaken in the summer of 1938 but remained confined
to only a small sector of this large mound.

They brought to light three post-and-wattle houses of
the period around the birth of Christ and below these
dwellings, in an even earlier settlement horizon which
reached back to the second and third centuries B.C., four
additional houses of the same type. The largest of the
upper settlement measured 56 feet by 21 feet (17 m. ×
6·5 m.); the smallest, 33 feet by 16 feet (10 m. × 5 m.).
The latter, having its wood work practically intact to a
height of 16 inches (40 cm.), was especially well preserved.
Haarnagel could observe that the outer posts of house II
leaned inward. He assumed that the posts that he found
were the lower portions of rafters that rose from the ground
directly, and reconstructed the house accordingly.[130] Albert
Genrich[131] and Zippelius[132] consider it more likely that these
oblique outer posts were short, that they carried an outer
frame of horizontal poles that served as footing for the
rafters, and leaned inward in order to counteract the outward
thrust of the roof, as shown in figure 309 B.

AALBURG, NEAR BEFORT, LUXEMBOURG

With the outbreak of World War II, all of this excavation
ceased. Save for an isolated excavation conducted by Gustav
Rieck during the German occupation of Luxembourg at
Aalburg, near Befort,[133] nothing new was added to our
knowledge of the early history of the three-aisled timber
house. Rieck uncovered the foundations of an aisled timber
hall of extraordinary dimensions (102 feet long and 29 feet
broad [31 m. × 8·8 m.]) which antedated even the earliest
Ezinge houses (figs. 310, 311). Here, it seems, in a dwelling
that had been constructed as early as 500 B.C., in territory
where Celtic and Germanic influences intermingled, the
excavator had come upon a floor plan that anticipated by
one millennium the T-shaped Flet and Dele arrangement
of the Lower Saxon farmhouse. The roof-supporting posts
of this house were not sunk into holes but rose freely from
base blocks above the ground, attesting overhead a solid
frame of cross and long beams. This site raised the interesting
question, whether the aisled West-Germanic timber
house might not have been adopted at a very early date in
the territory of the neighboring Celts.

WILHELMSHAVEN-HESSE, GERMANY

When excavation work could be resumed after the war had
ended, Haarnagel added a number of excavations to his
preceding work, which enabled him to trace the history of
the aisled timber house both further back and further
forward in time. A trial ditch dug in 1939, just as the war
broke out, in Hesse, one of the suburbs of the city of
Wilhelmshaven, had suggested the presence, in a settlement
stratum of the seventh century A.D., of aisled houses of
the Hodorf-Einswarden type, such as he had previously
been able to assert only for the span of 300 B.C. to A.D. 200.
Systematic excavations undertaken in 1949 and continued
in 1950[134] surpassed all expectations by establishing the
existence of this house type in settlement layers not only of
the seventh, but also of the eighth and ninth centuries A.D.
And in 1951-53 this span was further extended into the
eleventh, the twelfth, and thirteenth centuries through the
excavation of a medieval trading settlement in the city of
Emden.[135] The result of these excavations is visually summarized
in a reconstruction model of one of the houses of
Wilhelmshaven-Hesse, here shown as figure 312.

JEMGUM, NEAR LEER, GERMANY

Conversely, in an excavation conducted in 1954 Haarnagel
had the good fortune of unearthing in a place called Jemgum
near Leer,[136] on the left bank of the river Ems, an
aisled house with pottery shards and artifacts ranging from
the beginning of the seventh to the end of the fifth century
B.C. (transition from Bronze Age to Early Iron Age.) The
walls of the Jemgum house (figs. 313-314) were a different
construction type from those of the previously discovered
houses. They were built of horizontal logs of ash, squared
off, and held in place by vertical ash saplings. In the middle
of each long wall there was an entrance protected by a
projecting porch. The roof was carried by four freestanding
inner posts of a diameter of eight inches (20 cm.), dug
sixteen inches (40 cm.) into the ground. The hearth lay in
the middle of the center aisle, in the northeastern half of the
house. On the opposite side of the house the ground was
covered by a wooden floor covering of alder planks, which
suggests that this section of the house was used as a living
and sleeping unit.

FEDDERSEN-WIERDE, NEAR BREMERHAVEN,
GERMANY

Haarnagel's most successful excavation—begun in 1955,
continued every subsequent summer, and still in progress
at the time of this writing—was undertaken on an Iron Age
Warf called Feddersen-Wierde on the right bank of the
river Weser not far from Bremerhaven. As this dwelling
mound was peeled off, layer by layer, it released the remains
of forty-eight houses; the majority were in excellent condition,
reflecting the various stages of growth of a settlement
that had started as a flatland farm at about the time
Christ was born, and was subsequently raised, in seven
stages, to successively higher levels, until around the year
400 it had reached an ultimate height of 13 feet (4 m.)
above its original starting point and a diameter of about
656 feet (200 m.). The results of this extraordinary excavation
are known so far through preliminary reports only.[137]
In figure 315 I reproduce a plan of settlement period IIB,
which shows the Warf in the stage it had reached sometime
during the first century. At this time the settlement consisted
of a principal Warf and a secondary smaller Warf,
both protected by a peripheral ditch. The principal Warf,
some 295 feet long and 98 feet wide (90 m. × 30 m.),
accommodated a cluster of four houses; the smaller, a
cluster of only two. The houses varied considerably in
size, the largest measuring 97 feet by 21 feet (29·50 m. ×
6·75 m.); the smallest, 33 feet by 16 feet (10·00 m. ×
5·00 m.) Each house formed a self-sufficient agricultural
entity, combining under one roof the living quarters of its
owner and the stables for his livestock (fig. 316). The hay
and harvest was stored in separate open sheds to the side of
the house. The layout of the main houses is identical with
that of the contemporaneous houses that van Giffen had
encountered at Fochteloo (figs. 303-304). Like them, the
houses of the Feddersen-Wierde had their principal entrance
arranged in opposite pairs in the long walls, giving
access to a crosswalk which separated the quarters of the
humans from those of the animals. In the smaller houses
where the areas of living quarters of the owner and the
stables for his livestock were more or less equal, this led to
a fairly balanced arrangement with the entrances often
exactly in the center. But in the houses of the leading
families, superior wealth in cattle led to an elongation of the
stables and to the addition in the latter of a subsidiary

[ILLUSTRATION]

C. ELEVATION

principal members only, shown; for more complete
assembly see exterior view on next page.

[ILLUSTRATION]

B. CROSS SECTION

[ILLUSTRATION]

A. PLAN

detail of plan at jamb of doorway

[ILLUSTRATION]

313.A. B, C
JEMGUM, LEER, GERMANY

AISLED HOUSE, 7TH-5TH CENTURIES B.C.

[redrawn from W. Haarnagel, 1957, 21, Plan. No. 7]

The site is on the left bank of the estuary of the river Ems. The house was small, no more than 15 feet wide and 25 feet long. It was used
exclusively as a dwelling and gave no evidence of ever having sheltered animals. In the middle of each long wall, slightly off center, were
opposing entrances protected by projecting porches.

The roof was carried by two pair of inner posts (unscantled oak trunks, dia. 20cm) dividing the house into a central area of roughly 6½ × 13
feet asymmetrically placed, and with aisles all round it. The hearth lay in the axis of this center space, in the western half of the house which
had a simple clay floor and must have served as kitchen.

The floor of the space between the eastern pair of posts and the eastern end wall was covered with wooden planks cut from alder trees; this
area, better insulated from dampness than any other in the house, must have served as living and sleeping quarters. All structural members of
the dwelling that were posted into the ground, or that lay atop the ground, were found to be in good condition, many of the boards forming the
wooden floor of the presumed sleeping area were still in place.

[ILLUSTRATION]

314. EXTERIOR VIEW

JEMGUM, LEER, GERMANY

AISLED HOUSE, 7TH-5TH CENTURIES B.C. [redrawn from W. Haarnagel]

Drawings and models are in the Niedersächsisches Landesinstitut für Marschen- und Wurtenforschung

The walls of the house were made of squared ash logs, of which the bottom course was still well preserved. They were held in place at distances
varying between 5 and 6½ feet, by paired saplings pointed and driven into the ground to a depth of about 2 feet, with five pairs in each long
wall, and three in each end wall. At their meeting points in the corners of the house, the ash logs had rotted away, and for that reason, it could
not be ascertained in what manner they were jointed. It seems reasonable to assume that they were notched into each other at right angles,
since otherwise these timbers would have been subject to displacement from the thrust of the rafters.

Since the free-standing inner posts were only set 15¾ inches into the ground, they must have been framed crosswise at their heads by tie beams,
and lengthwise by longitudinal plates serving as footing for the rafters, or supporting them in midspan. There were no roof-supporting posts in
the end walls, indicating that the roof was probably hipped over the building's narrow ends. The construction of the walls, although common in
heavily wooded areas of Scandinavia and Alpine regions, is atypical for this part of Europe.

[ILLUSTRATION]

315. FEDDERSEN-WIERDE. PLAN [after Haarnagel, 1957, fig. 2]

AISLED HOUSES OF WARF-LAYER II B, 1ST-2ND CENTURIES

Haarnagel's exploration of this Warf, conducted from 1955 onward under the auspices of the Deutsche Forschungsgemeinschaft, was the
German counterpart to van Giffen's excavation of the Warf of Ezinge (figs. 292-99). The site, on the right bank of the estuary of the Weser,
was carefully selected after many sample drillings from a chain of nine dwelling mounds running in an almost straight line south to north over a
distance of 15 kilometers. The Warf encompassed seven settlement horizons, a new one every 50-80 years, to compensate for the steadily rising
innundation level.

The earliest settlement was a flatland farm built around the birth of Christ. The Warf was abandoned around 400 A.D. when it had reached
a height of about 13 feet (4m). The dwellings buried in its various layers were as well preserved as those of Ezinge and for the same reasons
(see caption, fig. 299); and were of the same construction type. The plan above shows the Warf in the stage it had reached toward the end of
the first century A.D.

[ILLUSTRATION]

FEDDERSEN-WIERDE, NEAR BREMERHAVEN, GERMANY

316.B AUTHORS' RECONSTRUCTION (DRAWN BY WALTER SCHWARZ)

SEE FIGURE 175. PAGE 216, VOL. 1, FOR A LARGER INTERPRETATION OF THIS DRAWING

316.A PLAN [after Haarnagel, 1956, Pl. 3]

AISLED HOUSE OF A CHIEFTAIN, Warf-LAYER II B, 1ST-2ND CENTURIES

With the cattle barn of Ezinge (figs. 298-99) this is one of the finest examples of a house type widely diffused in the Germanic territories of
Holland and Northern Germany during the first millenium B.C. and throughout the entire Middle Ages. The house was 97 feet long (28.50m)
and 21 feet wide (6.75m). It combined under one roof the owner's living quarters and the stables for his livestock. In the area used by animals
(eastern 52½ feet of the house) the roof-supporting trusses were more narrowly spaced, leaving in the aisles between each pair of posts a stall
for two head of cattle (32 head altogether).

As in the chieftain's house at Fochteloo (fig. 304) stables and living area were separated by an entrance bay accessible through doors in the long
walls, while animals entered through a gate in the eastern end wall. The walls and all the internal cross partitions were done in wattlework,
daubed with manure. The stable area had the traditional mats of wattlework on which the manure was gathered, with cess trenches beneath to
allow for drainage. The walk between these mats was paved with turves laid between floor beams running parallel with the trenches.

[ILLUSTRATION]

317. FEDDERSEN-WIERDE, NEAR BREMERHAVEN, GERMANY

HOUSE OF THE EARLIEST SETTLEMENT HORIZON. DETAIL

[photograph by courtesy of W. Haarnagel]

These are remains of two of the principal roof-supporting posts of the house, for which the builder used the round trunks of relatively young
and slender oaks without debarking them. The superb state of preservation of both timbers and the wattlework of which the walls and cross
partitions in the aisles were formed owes to the fact that whenever a house was abandoned because of floods and then rebuilt on higher ground,
its remains were soon covered by layers of fine silt deposited during floods, thus sealing its contents against air and bacterial decay.

axial entrance, primarily used for livestock. The house thus
attained the distinctive T-shaped floor plan which later
became the hallmark of the Lower Saxon farmhouse. The
long house in the northwest corner of the main Warf of
settlement period II-B of the Feddersen-Wierde is one of
the finest of this type of Iron Age house known to date. In
figure 316A I reproduce its plan, after Haarnagel, and in
figure 316B a tentative reconstruction of my own. The excavation
photo shown in figure 317 of one of the cattle boxes
of house I of the oldest settlement horizon of Feddersen-Wierde,
gives an idea of the magnificent state of preservation
in which the walls and roof-supporting posts of some
of the older houses of this site were found.

The occupants of settlement-horizon II of the Warf
Feddersen-Wierde were field-ploughing and cattle-raising
farmers. In settlement-horizon III (first to second century
A.D.) the economy, and with it the entire social structure of
the village, begins to change. The dominant architectural
feature now, as well as in all the subsequent horizons (IV,
V, VI, and VII, ranging from the third into the fifth
century A.D.), is a large aisled hall (without stalls for cattle
and carefully fenced in), used as the residence of a person
of conspicuous wealth and prominence. Next to this hall is
a second hall (likewise without cattle stalls) which Haarnagel
believes was used as an assembly place for the entire
community. Animal husbandry and agriculture give way
to industry and trade, and the growth of a new class of
workmen who lived in smaller houses and worked in the
service of their trading chieftain.

[ILLUSTRATION]

ELISENHOF, NEAR TÖNNING, SCHLESWIG, GERMANY

319.

320.

AISLED HOUSE, 9TH CENTURY A.D. [excavation photos courtesy of A. Bantelmann]

The overview (fig. 319) of the Warf shows the remains of the houses; below (fig. 320), the detail shows a portion of the wattled walls of the
house with inclined posts carrying a peripheral course of poles on which the rafters were footed.

The great historical significance of the excavation of this Warf is that it closed the gap between the Iron Age and Migration Period houses
(shown in figs. 293-318) and their medieval derivatives (figs. 339-354). The settlement was started on flatland in the 7th century; its subsequent
development could be traced clear into the 11th century. In layout and construction its houses were virtually identical with those of Ezinge
(figs. 293-299), and Feddersen-Wierde (figs. 316-317). They were in some places preserved to a height of 7 feet.

[ILLUSTRATION]

321. ANTWERP, BELGIUM. UNAISLED AND AISLED HOUSES

PLANS, EARLY 11TH CENTURY A.D.

[after A. Van de Walle, 1961, 128, fig. 35]

The house plans and reconstruction shown here and in fig. 322 are in themselves of no particular architectural distinction. But they do mark the
historical point at which the aisled and bay-divided timber house, the premedieval history of which has been briefly traced in these pages,
attempted to gain a hold in the new and rapidly developing medieval cities.

Excavations conducted in 1955-1957, in what was then the old city of Antwerp (and is now the center of the modern town) brought to light
three medieval habitation levels in an average depth of 5 to 7 feet (1.50-3.50m) beneath the present street level. By pottery and other artifacts
these strata could be dated: the lowest to about 850-976, the middle to about 976-1063, and the top level to 1063-1225. On each horizon the
excavator found three houses in a row, side by side, gable walls facing the street. The houses shown here belong to the middle level. The larger
one to the right is aisled; the others, narrower and shorter, are unaisled. These two are divided internally into a main hall with hearth, and with
one or more partitions to the rear perhaps serving as private or storage rooms.

Aisled houses were well suited to the open terrain of the nonurban countryside. But in the densely built cities, with open land at a premium, the
aisled structure of one story had limited utility and future. Some wealthy individuals or institutions could, to be sure, acquire enough urban land
upon which to build expansive aisled houses on one level, and could afford the expense of their maintenance. Such was the case with ecclesiastical
overlords (see figs. 339-340) or corporate bodies such as the Church (figs. 341-343) or the guilds. But for the most part, aisled dwellings were
impractical in, and proved antithetical to the function of the city. The type came to be replaced by narrower structures of multiple stories
providing space above ground level, set with gable walls toward the street and side walls almost touching, a picturesque and early characteristic
of new urban architecture.

[ILLUSTRATION]

322. ANTWERP, BELGIUM

AISLED HOUSE, EARLY 11TH CENTURY

[the reconstruction illustrated is redrawn from A. van de Walle, 1961, 129, FIG. 36]

This isometric rendering, conjectural in detail yet fairly
certain in general lines, illustrates more persuasively than
the plans of the preceding figure why it was that aisled
houses could not survive the pressures of dense urban
development. The low, aisled house of the open country, in
the struggle to adapt it to urban row-house conditions,
soon proved to be a wasteful use of costly and limited
city space. Therefore this house type was, in the cities
quickly discarded.

In process of adaptation, the remaining nave (after
aisles were eliminated) could, to be sure, have been raised;
but the skeletal construction of the old northwest European
all-purpose house was never intended to bear the load
of superincumbent stories. A new type of timber framing
with strong load-bearing walls evolved to make timber
framing possible in construction of narrow urban houses.
But because of its total vulnerability to fire, the timber
house eventually came to give way, as the cities grew, to
masonry houses.

BÄRHORST, NEAR NAUEN, GERMANY

To the successful excavation work by van Giffen and
Haarnagel in the coastlands of Holland and northwestern
Germany, one has to add the work of others. As early as
1935-37 Otto Doppelfeld had unearthed a palisaded village
of an estimated fifty aisled houses, on the Bärhorst,[140] a
shallow, sandy plateau in a marshy swale near Nauen
(Berlin). The site had been discovered in the course of
trenching operations undertaken before the installation of
a giant sewage disposal plant. Remnants of pottery and
other cultural accessories showed that the village was
constructed around A.D. 250 and that it was held in occupancy
for about a century. Since it lay in an environment
that was utterly unsuited for successful agricultural exploitation
and perished in a fire that seems to have been
associated with a planned and systematic abandonment (no
objects of any use were left), Doppelfeld concluded that it
might have been the temporary site of a wandering Germanic
tribe who discarded the site when they found prospects
for the conquest of more suitable land. While basically
adhering to the same construction type, the Bärhorst houses
showed a great variability in the treatment of their walls
(fig. 318). Some of the houses had simple wattle walls; in
others the walls were formed by boards mounted horizontally
or vertically between the wall posts. Still others
were braided from thin split pieces of straight wood (leftovers
from the hewing of the structural timbers). The
Bärhorst village showed that the aisled pre-medieval timber
house extended into the third and fourth century A.D. eastward
as far as the longitude of the modern city of Berlin.

TOFTING AND ELISENHOF, NEAR TÖNNING,
SCHLESWIG-HOLSTEIN, GERMANY

Albert Bantelmann, on the other hand, pressed the search
northward by excavating, in the summers of 1949 and 1950,
a dwelling mound at Tofting (Schleswig-Holstein)[141] at the
mouth of the river Eider, close by the Danish border. His
excavations showed that conditions in the homeland of the
Anglo-Saxons were identical with those which van Giffen
and Haarnagel had found prevalent in the adjacent territories
of the Frisians.

Tofting was a relatively modest site; but from 1957
onward, in annual excavations as yet not terminated,
Bantelmann peeled off, layer by layer, in the Warf Elisenhof,[142]
near the town of Tönning at the mouth of the
river Eider in Schleswig, the remains of a village that was
founded in the seventh or eighth century A.D. and remained
in continuous occupation deep into the High Middle Ages.

The earliest settlement, which is so far known only
through a preliminary report, was built on natural ground
on one of the banks of the river Eider. Later the site was
raised and peripherally expanded by heavy deposits of
manure and clay, until it finally comprised an area of
roughly seven hectares. The occupants of the earliest settlement
were cattle-raising farmers, which is attested by the

[ILLUSTRATION]

CHEDDAR, SOMERSET, ENGLAND

AISLED PALACE HALLS, 12TH & 13TH CENT.

[courtesy of P. Rahtz, and Ministry of Building and Works, Crown
Copyright]

324.A

The composite plan shows the layout of East Halls I, II, and III. The original ten-bay
building dates to the early 12th century. It was replaced 100 years later by a smaller
six-bay hall 71 feet by 48 feet. At this time the roof supports were set into new square
post holes, some of which overlapped the round ones of the original hall. Toward the
end of the 13th century the second hall was replaced by a third of yet smaller size
66 by 42 feet) and without aisles.

324.B

At the time of Henry I (1100-1135) the hall was an aisled, ten-bay structure, 110 feet
long and 54 feet wide. The course of the outer walls could be identified by large post
holes, at 8-foot intervals, linked by ground timbers. The roof-supporting posts were
apparently rough-scantled logs over 15 inches in diameter, set into round post holes
packed with gravel.

presence of bone deposits of cattle, sheep, pigs, and horses
in, respectively, decreasing magnitude. The predominant
dwelling in all layers was the aisled post-and-wattle house,
giving shelter to humans and animals under the same roof.
Layout and construction were in all essential features
identical with those of the houses of Ezinge and Feddersen-Wierde,
and the state of preservation left no feature in
doubt (figs. 319, 320). Main posts and wattle walls survived
in certain cases up to a height of 7 feet (2 m.). In some
houses the outer posts consisted of inward leaning timbers
of unusually heavy scantling (fig. 320)—no doubt the supports
of a peripheral course of poles that served as footing
for the rafters. The floors of the houses were formed by
turves of clay, heavily matted with roots. The section of
the house that contained the hearth and served as living
quarters invariably lay on a higher level than the part that
contained the stalls for the cattle. From the higher level
the floor gradually slanted down to reach its lowest point
at the end of the stable section, thus affording easy drainage
for the liquid waste of the animals, which was conducted
downward in carefully constructed flues—the same type of
flues (one groundboard and two sideboards) used today
in the farmhouses of the same district, where it is called
Grüpp.

The length of the houses varied with the number of
cattle owned by each farmer. The width amounted uniformly
to about 17 feet (5 m.)—as it did in van Giffen's and
Haarnagel's early Iron Age houses—a dimension obviously
conditioned by the fact that it offers a comfortable minimum
of space for two rows of animals and a central lane of
access with drains for the waste products. As in the Iron
Age houses, the roof received its main support from two
ranges of freestanding inner posts. The cattle stood in
pairs in each stable, their heads turned toward the walls of
the house. The cross partitions by which their boxes were
formed consisted of split logs set into the ground in palisade
fashion. The outer walls were wattled, with the twigs
wound around a sequence of thin posts alternating at
regular intervals with heavy posts which must have supported
the wall plates. Only a small percentage of the
houses was oriented from east to west. The determinant
factor in the choice of the axis appears to have been the
slope of the Warf, as it offered best drainage.

Besides the standard house there was a variety of non-aisled
smaller buildings, some with wattle walls, others
with walls of turf; the latter, apparently used as weaving
houses. Two of these smaller houses were in ridge-pole
construction.

The Elisenhof is altogether a spectacular site. Its full
evaluation, which is forthcoming, will unquestionably give
us new insights into the constructional aspects of its houses.
The same may be expected from the excavation of a ninth-century
village on the Grothenkamp near Neumünster in
Holstein, which came to light in an emergency dig undertaken
in 1962.[143] Also not to be overlooked, in this connection,
are three houses (one of them aisled) which A. van de

Walle excavated in 1955 in the ancient town center of
Antwerp in a settlement-horizon that could be dated in the
eleventh century.[144] I reproduce in figures 321 and 322 a
plan and a reconstruction of this house as proposed by van
de Walle.

The great importance of these settlements, together with
those of Leens, Emden, and Wilhelmshaven-Hesse, is
that they have helped to close the gap between, on one
hand, the Iron Age houses of Ezinge, Jemgum, and Befort
and, on the other, isolated medieval house sites found in
Germany in such places as Wilhelmshaven-Krummer Weg
(eleventh-twelfth centuries), Ramm (thirteenth century),
Hungersdorf (about 1400), Hardesbüttel (thirteenth-fifthteenth
centuries);[145] and in Holland in such fourteenth-century
sites as Lievelde, Waalhaven, and in Boudewijn
Hartsland.[146] Thus, pre- and protohistory were connected
to the modern period with the story of a house type whose
historical life span, it now seemed, would have to be counted
by millenniums rather than by centuries. In this matter,
too, the last six years have brought a sensational surprise.

ELP, PARISH OF WESTERBORK, THE NETHERLANDS

After Haarnagel's excavation at Jemgum (figs. 313-314), it
was generally believed that the aisled hall had been
traced back to the period of its earliest appearance (seventhfifth
centuries B.C.). However, a Bronze Age settlement in
Elp (parish of Westerbork, province Drenthe, Holland)
excavated in 1960-62 by H. T. Waterbolk[147] disclosed that
the aisled long houses of the Ezinge-Feddersen-Wierde
type were in use as early as 1250 B.C. The excavation brought
to light the ground plan of some thirty houses of different
types belonging to a farmstead composed of a main building
and about four subsidiary buildings, all of which were rebuilt
on various occasions during a period of occupation
that lasted from roughly 1250 to roughly 850 B.C. The main
houses vary in length between 82 feet (25 m.) and 135 feet
(41 m.). They are internally divided into two equal parts.
In one half, which was used for living, the posts are
widely spaced. In the other half, which served as stables,
the interstices between the posts are smaller. Some of them
are not sunk into the ground as deeply as the principal
posts and therefore, probably, served as mainstays for stall
partitions. I show as a typical example the ground plan of
house 9 (fig. 323), which illustrates this point particularly.
A similar distinction of the post interstices between living
quarters and the stall sections of the house could be observed
in Feddersen-Wierde (figs. 315-316), in Hodorf
(fig. 307), in two Iron Age houses excavated in 1954 by
P. J. R. Modderman near Deventer (Oberijssel), Holland,[148]
and—although not quite as markedly—in Fochteloo (fig.
304). As in the later Iron Age houses, so in Elp, the narrow
ends of the house were rounded, which suggests that the
ends of the roof were hipped. The houses appear to have
been entered broadside through a passage way that separated
the living quarters from the section that was occupied
by animals—another feature that was to become a characteristic
trait of the later house tradition. Waterbolk had
reason to believe that the house type of Elp already existed
several centuries before the settlement of Elp was founded;
and before the manuscript of his preliminary report on Elp
was finished, he received the news that J. D. van der Waals
had come across a Bronze Age site with an aisled house
118 feet (36 m.) long at Angeloo (Emmen). It had many
features in common with the Elp houses, and, to judge from
its pottery, appeared to be older than the Elp settlement.[149]

The Elp settlement was occupied by an autochthonous
Bronze Age population of Holland,[150] which may or may not
be proto-Germanic. The later Iron Age sites of northwest
Germany and Holland were in the territory of the Frisii,
the Chauci, and the Saxons.[151] The homeland of the Saxons
was east of the river Elbe at the bottom of the Danish
peninsula. Their westward move into the territory of the
Frisians, around 300 A.D., seems to have led to the destruction
of the settlement of Ezinge. Toward the middle of the
fifth century, in several successive waves, they moved
across the channel into England.

YEAVERING, NORTHUMBRIA, AND CHEDDAR,
SOMERSET, ENGLAND

Until the summer of 1956, England, peculiarly enough, had
not yielded a single house site comparable to any of the
Continental finds. A small settlement, of aisled post-and-wattle
houses excavated in 1946 by Gerhard Bersu on a
promontory of the shore of Ramsay Bay on the Isle of
Man,[152] was probably not an Anglo-Saxon settlement but
an outpost of a Norse raiding party which was occupied
only intermittently. In the summer of 1956, however,
Brian Hope-Taylor came upon the site of a royal Anglo-Saxon
palace of the seventh century at Old Yeavering in
Northumbria (Bede's villa regalis ad Gefrin),[153] which contained
the remains of no fewer than twenty-five timbered
houses, most of them aisled; and in 1960-62 Philip Rahtz

[ILLUSTRATION]

WARENDORF, WESTPHALIA, GERMANY

325.A EXTERIOR VIEW REDRAWN FROM WINKELMANN

325.A.1 PLAN. HOUSE 47, LEVEL A, PERIOD 4

HOUSE TYPES OF EARLY MEDIEVAL SETTLEMENTS. 650-800 A.D.

[after Winkelmann, 1958, 500, fig. 5]

Excavations conducted in 1951 by Wilhelm Winkelmann in Warendorf, on the south side of the Ems, brought to light traces and remains in the
soil of no fewer than 186 separate buildings, which could be accurately dated by pottery and other associated artifacts. The settlement was
rebuilt four or five times, in most cases—as the remains of charcoal and fired clay found in many postholes indicate—after houses of the previous
settlement had been destroyed by fire.

Analysis of the successive stages of the site disclosed that each inhabited level consisted of a group of four to five farmyards occupying an area
of about 984 feet (100m) square. Each farmyard held a large dwelling, and fourteen to fifteen smaller auxiliary buildings (barns, stables, sheds,
weaving houses). The variety of these service structures is shown in figs. 326.A-F.

The reason for the boat shape of House 47, a feature very common in early Scandinavian architecture, is obscure. For other examples of
longhouses of this type see below, V. 17. 3.

[ILLUSTRATION]

WARENDORF, WESTPHALIA, GERMANY

325.B EXTERIOR VIEW REDRAWN FROM WINKELMANN

Arrow indicates general direction from which exterior view above is taken

325.B.1 PLAN. HOUSE 7, LEVEL A, PERIOD 1

HOUSE TYPES OF EARLY MEDIEVAL SETTLEMENTS, 650-800 A.D.

[after Winkelmann, 1958, 504, fig. 8]

Plans and reconstructions on these pages are typical of the man or principal dwellings of the Warendorf settlements. By virtue of both size and
architectural distinction, as well as the presence of a hearth, they are unquestionably the dwellings of the owners of the land. They were boat-shaped
(fig. 325.A) or of rectangular plan, as above, varying in length between 45 and 95 feet (14-29m), in width between 14½ and 32 feet
(4.5-7.0m). Their roofs were supported by a perimeter of heavy posts set vertically into the ground, each buttressed from outside by a second,
lighter post rising at an angle to meet the inner post near its head. The triangulation counteracted the thrust of the roof. Wall panels between
vertical posts were daubed with clay.

The axes of these dwellings, and most of the subsidiary structures, ran from west to east. This feature, characteristic of many pre- and
protohistoric buildings in these latitudes, was apparently determined by the desire to expose only a gable wall to the prevailing (here, western)
winds. The houses were entered on their long walls through two opposing entrances in midwall, each protected by a porch. The hearth lay in the
eastern half of the house.

[ILLUSTRATION]

WARENDORF, WESTPHALIA, GERMANY

326.A

326.B

326.C

HOUSE TYPES OF EARLY MEDIEVAL SETTLEMENTS, 650-800 A.D.

[redrawn after Winkelmann, 1958, 504, fig. 8]

These pages give a visual account of the variety of service structures that one might have expected to find in the farmyards of Warendorf.
They have been redrawn from models made under the direction of Wilhelm Winkelmann, and are on display at the Museum für Vor- und
Frühgeschichte, Münster, under the aegis of which the excavations at Warendorf were conducted.

There were forty service buildings of rectangular plan with vertically boarded walls at Warendorf (326.A), used as barns and stables. Small
houses with hearths built in the manner of the larger dwellings presumably provided housing for serfs (326.C). Numerous small sheds with open
walls (326.B) were probably utility buildings for various kinds of storage.

[ILLUSTRATION]

WARENDORF, WESTPHALIA, GERMANY

326.D

326.E

326.F

HOUSE TYPES OF EARLY MEDIEVAL SETTLEMENTS, 650-800 A.D.

[redrawn after Winkelmann, 1958, 504, fig. 8]

The small, walled rectangular shed (326.D) could have served many needs from implement storage to animal shelter. A great number of small
buildings of A-frame construction (326.E) were found in the Warendorf farmyards. They were partly dug into the ground to gain standing
height, and some buildings of this type were identifiable as weaving houses by the artifacts they contained. But a building with below-ground
storage could have served equally well for winter storage of root crops.

The polygonal framework was used to store loose hay or sheaves of grain (326.F); its conical thatched roof could slide down its poles to
accommodate and adequately shelter the diminishing harvest as it was used up through the winter. The floor grid of lashed poles (326.X)
afforded aeration and drainage for the hay or grain; the plans show typical posting patterns for the structures.

[ILLUSTRATION]

326.X HAYSTACK SHELTER, left, and
DRYING PLATFORM, right

Haystacks and aeration platforms of this
type are common in Germany and the
Netherlands even today. For a fine
15th-cent. portrayal, see fig. 368.

[ILLUSTRATION]

326.G WARENDORF, WESTPHALIA, GERMANY

HOUSE TYPES OF EARLY MEDIEVAL SETTLEMENT, 650-800 A.D.

[redrawn from Winkelmann, 1958, 500, 504]

Each of the buildings shown on this and the preceding pages is referred to in the Alamannic, Bajuvarian, and Salic laws discussed above
(p. 26ff), where they are designated in the following manner:

The house of a free man = DOMUS or SALA LIBERI (figs. 325A-B); the house of a serf = DOMUS SERVI (fig. 326.C); a wall-enclosed
barn = SCORIA (fig. 326.A); an unenclosed barn = SCOF (fig. 326.B); a granary = PARC (fig. 326.D); a haystack, or stacked sheaves of
wheat = MITA (fig. 326.F). For a more complete account and the etymology of these terms see above, p. 29, notes 14-18, and the chart in
Winkelmann, 1954, 210, fig. 12.

unearthed at Cheddar, Somerset,[154] another Saxon palace
(sedes regalis aet Ceodre) with the remains of a timbered
long hall from the time of King Alfred (871-900)[155] and an
aisled hall, likewise in timber, from the reign of Henry I
(1100-35), a plan of which is reproduced in figure 324. The
results of these excavations once fully published will put
an end to another enigmatic chapter of early medieval
house research.

THE SINGLE-NAVED HOUSE OF
WARENDORF, NEAR MÜNSTER, WESTPHALIA,
AND IN OTHER GERMAN SITES

If I have given primary consideration to aisled structures
in the preceding account of prehistoric and early medieval
house construction, I have done so because the excavations
conducted during the last three decades seem to indicate,
with mounting conclusiveness, that this was the principal
dwelling type used during the Iron Age and in the Early
Middle Ages in the barbaric territories north of the Alps.
In emphasizing this fact, I do not wish to convey the
impression that it was the only one. Other excavations,
recently conducted, have brought clear evidence of the
existence of a simpler type of house that was not provided
with any aisles.

A good sampling of this latter type was brought to light
between 1951 and 1954 when Wilhelm Winkelmann excavated
a medieval settlement in the vicinity of Warendorf,
near Münster, Westphalia.[156] The pottery found in this
settlement suggested that it was occupied from about 650
to 800 A.D. The leading house type was a structure of
rectangular plan, accessible by two porch-surmounted entrances
facing each other in the middle of the long walls
(figs. 325 A-B and below, pp. 271ff). Similar houses were
encountered in Bucholtwelmen (district of Dinslaken) in
1939, in Haldern (near Wesel, Lower Rhine) in 1938, and
in Westick (district of Unna, Westphalia) in 1935—the
latter a rare example of Continental cruck construction.[157]
Warendorf was of particular interest because on this site
the principal dwellings were surrounded by an entire host
of subsidiary structures of smaller dimensions and lesser
significance, the like of which one would expect to encounter
in all those settlements where the various domestic functions,
rather than being assembled under one roof, are
scattered throughout a variety of separate structures.
Winkelmann has gathered all these types together in a

visual chart which reads like a model book of early medieval
house construction (figs. 325 and 326). He had found an
equivalent for virtually each and every variant mentioned
in the Alamannic, Bajuvarian, and Salic laws.[158]

LACK OF COMPARABLE FINDS IN FRANCE
AND IN THE SOUTH OF GERMANY

The great shortcoming of our present knowledge of
early medieval house construction in transalpine Europe
is that it is based on the results of excavations confined to
the Scandinavian countries and to the northern parts of
Holland and Germany, with the recent addition of Anglo-Saxon
England. France, to this day, has remained a terra
incognita. Bursting with treasures of unsurpassable beauty
created in more advanced and more sophisticated periods
as well as in more permanent materials, she is unlikely
to engage in the near future in any concerted search
for the shadows that the rotting timbers of her humbler
early medieval houses left in the subsoil underneath the
stately structures that replaced them. But even in southern
Germany the soil has as yet been rather unyielding. An
excavation of a Frankish settlement of the sixth to ninth
centuries A.D., made as early as 1937 in Gladbach (district
of Neuwied),[159] brought to light a great variety of smaller
subsidiary structures, all in post-and-wattle work—but no
houses of any primary significance. The same holds true
for an Alamannic settlement of the Early Middle Ages,
excavated in 1947 in the vicinity of Merdingen (district of
Freiburg),[160] and for a Bajuvarian early medieval village
near Burgheim (district Neuburg on the Donau.)[161]

The reason for this scarcity of finds on the mainland is
probably very simple. The dwelling mounds of the coastal
lowlands which yielded such rich information are not only
conspicuous scenic landmarks, but also of a physical
composition (marsh, clay, manure—sealed off by intermittent
layers of silt) that offers unusually favorable conditions
for the preservation of wood, an advantage that is
otherwise only encountered in peat bogs or in sites that lie
below the normal water level. On the mainland these
conditions, in general, are wanting.[162]