Richmond, 4 Roman Camps at Cawthorn

THE FOUR ROMAN CAMPS AT CAWTHORN, IN THE
NORTH RI DI NG OF YORKSHI RE
By I. A. RI C HMOND
The excavation of Cawthorn camps by the Roman
Antiquities Committee of the Yorkshire Archaeological
Society was made possible by Mrs. . E. Mitchelson,
of Pickering, who owns the site and generously refrained
from shooting over it from 1923 to 1929. We also
thank Mr. Payne, of Elleron Lodge, who lent us
accommodation on the exposed moor. The moving
spirit of the investigation was Dr. J. L. Kirk, F.S.A.,
whose local knowledge and enthusiasm lay behind all
our efforts. General and financial propaganda was
undertaken by Mr. William Newbold and Prof. H. A.
Ormerod, secretaries of the Committee, and Lieut-Col.
E. Kitson Clark, F.S.A., Hon. Treasurer. Generous
grants were made by the Haverfield Trustees of the
University of Oxford, the University of Leeds and the
Leeds Philosophical Society. Trial trenches were dug
in 1923 by Mr. F. G. Simpson, Hon. F.S.A. Scot.,
who first perceived the unique value of the site. Then
the late Mr. H. G. Evelyn White, Lecturer in Ancient
History at Leeds University, superintended a first
season's work, assisted by the writer, by Leeds Univer-
sity engineering students, who laid down a surveyor's
grid, and by Mr. F. Manby, photographer to the
Agricultural Department. After this, the writer
worked for five seasons ; and among colleagues in the
field, he wishes to thank F. G. Simpson, for invaluable
advice, timely photography and spade-work on the
tribunal; Dr. J. L. Kirk, for trenches in camp D ;
S. N. Miller, for friendly devil's advocacy; R. G.
Collingwood, for inspiration and for help in the draw-
ings ; T. W. Woodhead, for reports on wood remains ;
and E. Hitchen and C. Bird, for conspicuous devotion
as workmen.
l 8 T H E F O U R R O MA N C A MP S A T C A WT H O R N
T H E S I T E
In the Parish of Middle ton, near Pickering, at the
top of the long slope which forms the northern edge
of Rydale, lies the farmstead of Cawthorn, giving its
name to four Roman camps (Fig. i ) .
1
Nowadays, the
main roads that lead north from Rydale do not pass
this way : but folk who travelled thence to Eskdale
in ancient times would use this slope, avoiding the
narrow Seven valley and Fylingdale ; and a rapid
FI G. I . GENERAL PLAN OF THE CAWTHORN CAMPS
reconnoitre would show that the easiest route on the
slope passed through Cawthorn, aiming for the western
shoulder of Wheeldale (Fig. 2). Cawthorn, more-
over, is a natural halting-point : for the gentle rise
from the south is here broken by a great east-to-west
fault, well-known to northern geologists, which creates
1
The camps were first planned by-
Drake (Eboracum, 1736, p. 36),
who shows an imaginary Wade' s
Causeway leaving the Nort h Gat e
of C, and a way t o the river, out of
the north-east angle of D. He also
shows Porter Gat e at the north-east
angle of C, and in marks vestiges
of barracks where the turf mounds
are most prominent to-day. General
Ray (Mil itary Antiquities of the
Romans in North Britain, PI. xi)
produce an excellent plan of the
camps, connecti ng them wi th Deal gi n
Ross and the Ni nt h Legi on (see
Roy, p. 65), in 1755 (see Macdonal d,
Archaeologia, l xvi i i , p. 172 for this
date). Young, History of Whitby,
vol . ii (1817), p. 699 makes Wade' s
Causeway f orm the via principalis
of D, an unf ounded guess. Th e
best modern plans are those of the
Ordnance Map and Victoria County
History, Yorks., vol. ii, pp. 1 4- 1 5,
whi ch also includes some sections.
To face page 18.
P L A T E I.
Yorks. Arch. Journ.
B. DI TCH AT N. J UNCTI ON OF A AND B.
P L A T E II.
To face page
B. KEY TO ABOVE.
I N T H E N O R T H R I D I N G O F Y O R K S H I R E 1 9
a northward-facing cliff, 150 ft. high (PI. I A). These
are the facts which explain the position of the camps.
It is, however, usually assumed that their position
is also determined by the Roman road, known as
Wade's Causeway,
1
which ran from Malton to a
FI G. 2. MAP OF THE CAWTHORN DI STRI CT I N ROMAN TI MES
destination, still obscure, near Dunsley Bay. But this
is far from certain (see Fig. 2). In the first place, the
course of the road near the camps is unknown. Even
1
The name has no connection wi th of folk-lore. The supposed inscrip-
General Wade, bei ng older than his tion f ound on this route at July Park
days of road-making fame, see . ; is not Roman, as Mr . R. G.
it represents a much older piece Col l i ngwood assures me.
20 THE FOUR ROMAN CAMPS AT CAWTHORN
two centuries ago,
1
when it was clear from Great
Barugh, near Malton, to Middleton, there was a gap
between Middleton and the north side of Sutherland
Beck, below the Cawthorn escarpment ; and excavation
has failed to add anything to our knowledge. Secondly,
the history of the road is quite uncertain. It was
clearly not an important highway, for it seems never to
have been heavily metalled. Thus, it could be regarded
either as a first-century line of penetration, analogous
to the pontes longi
2
of Germany, or (perhaps less
probably) as a fourth-century cavalry-road, connecting
Malton with the northern Theodosian signal-stations
(Fig. 2). Anyhow, it is clear that neither the road
nor the camps throw light upon each other, and no
proof has yet been obtained that they are in any way
connected, except in so far as they lie upon the same
natural route.
Tactically, the site has one serious drawback,
despite its splendid outlook. The water-supply is bad.
The two steep tracks which lead to the Sutherland
Beck are entirely unsuited to vehicular traffic ; and
the Saintoft Beck lies half a mile away to the south.
Both streams are well out of range of the camps, and
the geological formation of the escarpment, which is
the rim of an anticline, precludes the digging of wells.
Permanent occupation of the site is thus excluded.
Temporary occupation could, however, be undertaken
with safety in peace-time or with risk in war. And
there were, in fact, two distinct occupations of the
plateau, each presenting, as will be seen, the same odd
characteristics.
THE FIRST OCCUPATION
(1) ITS EXTENT
Field-observation long ago, and air-photography
in 1925, revealed that the four entrenchments
(A, B, C and D) were not contemporary. It was
obvious that D overlay C, and therefore was the more
recent work. On B, however, simple observation
1
See Drake, Eboracum, p. 36.
2
Taci tus, Annals, i, 63 : angustus
is trames . . . a L. Donatio aggeratus.
IN THE NORTH RI DI NG OF YORKSHI RE
2<)
threw no light, and it was alternatively regarded as a
separate entrenchment or as an addition, or annexe,
to A. Anticipating the detailed account, it may be said
here that excavation proved to include the whole
area of A ; three sides of A' s rampart having been
refurbished, and the fourth side levelled and thrown
back into the ditch, while B' s new rampart passed across
the filled ditch at two points (see Fig. 3 ; Pis. i B, ii).
So it becomes evident that A and C are respectively
earlier than and D. Further, and D are connected
in purpose and time, since a lightly-metalled road joins
the two, avoiding the north end of C on its way. No
road connects A with C, yet C is unquestionably related
to an occupation of the plateau upon which A stands,
since all its three gates are on that side, a most ex-
ceptional arrangement ;
1
and, further, it cannot be
connected with B, since it is older than D, which linked
with by a road. The general relationship is thus
clear ; goes with D, and C with A. The significance
of this twin arrangement will be discussed later.
(2) EARTH- WORK A
(a) The Ditch. The earth-work is completely sur-
rounded by a great ditch (Pis. ii and xx), 15 ft. wide
by 7 ft. deep, including a channel at the bottom
1 ft. square. The ditch is uninterrupted in front
of all four gates, but at the east gate, amid a belt of
rather harder rock, it is dug only to half
2
depth (Fig. 5) ;
nor was this half-scale ditch considered unfinished,
since it was provided with the regulation channel at
the base. Under normal conditions, in the soft rock,
the cutting of the ditch is good and the sides are
smooth ; but, owing to the orientation of the camp, the
cutting happens everywhere to be diagonal to the
bedding of the rock, and, where the harder belts of
rock occur, the surface is necessarily not wholly smooth
(see PI. i B). It is, nevertheless, remarkable what a
degree of smoothness has been achieved, considering
the difficulty of the work and the simple tools employed.
The only bad piece of work discovered on the circuit,
apart from a slight error in general lay-out, caused by
sighting without the groma, was the curve at the
north-east angle (see folding plan), of which the sym-
metry was spoilt, by carrying the deeper part of the
east ditch too far northwards, a discrepancy which
would be explicable if two gangs had been working
1
The only parallels for the con- sence of a small tutulus at the point
centration of camp gates upon special gave sufficient security. For the
objects come f rom siege-works, and tutulus does not cover the whol e
even these are not unsupplied wi th a length of ditch, and the arrangement
gate for retreat, j ust because serious is not adopted in connection wi th the
war is in question. other tutuli. Nor is the difference
to be explained in the same way as
2
Thi s seems a better explanation that at Raedykes ( P. S. A. Scot.,
than the assumption that the pre- 1, p. 344) where the cause is contours.
To face page 22.
P L A T E III.
Yorks. Arch. Jourti.
A. Tutulus OF A COVERED BY TURF
clavicula OF AT S. GATE OF A.
Yorks. Arch. Journ.
B. Tutulus AT E. GATE OF A.
P L A T E I V.
To face page
UNFINISHED tutlllus. W. GATE OF A.
IN THE NORTH RI DI NG OF YORKSHI RE 2<)
towards the angle, and the eastern sector had been
pushed on too far. Some slight changes of alinement
in the course of the ditch also suggest that we have to
do with working-parties which literature describes
1
in
connexion with camp-building. At all the gates except
the north, where the escarpment provides a sufficient
obstacle, the ditch is fronted by a second ditch,
placed so as to cover exactly the gateway opening.
Following tradition,
2
the name tutulus has been adopted
for this obstacle, though it must be observed that the
connexion with shortness is wrong, and that the name
came not from titulus but from the same root as tueor,
since the little ditch was a safety-device (see page 43),
like the flaminical hat called by the same name. The
tutuli at the east (PI. iii B) and south (PI. iii A) gates
were beautifully cut. Neither contained much silt, and
both had clearly been filled up again very soon, leaving
on the surface no trace of their presence. The west
tutulus was incomplete (PI. iv), and its half-finished
shape demonstrates how such small ditches were dug.
The form arrived at was clearly evolved by widening
(Fig. 4) the top of a narrow trench, dug to the depth
and length required for the bottom of the completed
ditch, and wide enough for a man to work in. The
trench from which the process started could be rapidly
1
Caesar, B.G. i, 42. See also over their prowess in basketing
Cichorius, Trajan's Column, Taf . xlvi, earth.
for a scene where two soldiers wrangle
2
De munitionibus castrorum, c. 50.
cut with the pick, and the rest of the work could be
done with an excavating tool, of the type discovered at
Newstead,
1
the earth being carried away in baskets.
2
This method, however, suits small ditches only, and
large ditches may have been dug in another way (see
p. 72). Finally, the fact that one tutulus was un-
finished supports the implication of the small amount
of silt in the east and south tutuli. The gateway
defences of A had a short life.
(b) The Berm. No space separated the outer edge
of the rampart from the lip of the ditch, for the rampart
was not heavy enough to crush the lip.
(c) The Rampart. The rampart (Fig. 5) is formed
of the upcast from the ditch, which varies from
rough lumps of rock, or ' brash,' to roughish sand,
that sets sharp and firm. Such material could not
stand alone safely, and was therefore reinforced with
wood. This reinforcement was not the temporary form,
of brushwood,
3
but was a solid erection, of permanent
type.
4
The first element therein (Fig. 5) was a line
of small vertical post-holes, on the lip of the ditch,
spaced roughly ten feet apart on the straight, and
five feet on curves. Only slight remains of timber were
found in them, consisting of scraps of birch from the
north-west and south-east angles. The holes were
filled with fine silted rampart-material, very soft and
and usually dark-colouredless often, if the filling
had been rapid, hardly discoloured at all. They held,
it is clear, a rather small upright post; and their
spacing would suggest that this held in turn a strut,
thus taking directly the whole weight it had to bear,
rather than some frontal line of bratticing which would
give attackers a hold. Certainly, the holes are too
slight for the posts to have held a front line of upright
boarding, with the whole weight of the rampart behind
it, and birch would not have been chosen for such work.
The real front line of the rampart must therefore be
1
Newstead ; A Roman frontier
1
I am particularly indebted, in
post, PI. lxi, Fi g. 9. writing the following section, to
2
Vegeti us, ii, 25 ; Cichorius, Mr . S. N. Miller' s detailed dis-
Trajan's Column. Taf f . xv, xxx, cussion wi th me of the nature of this
xlii, xlvi. rampart.
8
cf. p. 33, n. 4.
sought six feet further back, where was found the second
element. This was a continuous line of wooden up-
rights. They were fixed below ground, for about
twenty feet on either side of each gate, in a continuous
palisade-trench, filled with big wedging stones, like
E A ST FVAM/ PAKX
DKOlLSHEJ> iw B PimjoD
the better-known and larger example of the German
limes. The feature was particularly easy to follow,
since the trench was cut below the rampart, through
the original turf-covered ground surface, which was
bleached white, by chemical action described else-
where (see p. 54). At the east gate (see PL vii), the
ground being rocky, the trench contained a post-
hole at every ten feet ; but, after the second post-hole
northwards, it sloped up and disappeared, suggesting
that, as it got further away from its firmly-fixed ter-
mination, it had been planted rather higher up in the
rampart-mass : while on the south side of the gate
it stopped abruptly at the corresponding point, as if
it had stepped up all at once. The east rampart,
being almost completely demolished, did not provide
quite everywhere evidence for the continuation of the
trench at higher level, yet it was possible in most places,
by cross-trenching, to detect the heel of the trench,
with some of the filling (see Pis. xiv A and xv B) ; and
all these points lined out with great accuracy. The best
details of all came from the firm sand of the south
rampart, which preserved not only a continuous line of
trench (PI. A), but complete evidence about what the
trench held. For the rampart-mass had here been firm
enough to retain the impression of the boarding against
which it had been piled, giving thus, from the south
gate to the south-west angle, a slot so uniform and so
continuous that it was possible to form a perfectly
accurate estimate of the timber which it had contained.
The fact emerged that the timbers were flat on both
front and back, and from two to three inches thick.
In other words, a continuous line of flat boarding was
inserted in the rampart at this point. Boarding of
this kind is unsuitable for a support ; and its unbroken
continuity shows that it must have been the front
palisade of the rampart. Further, it was not hammered
into a completed rampart, but the rampart, as the
layers of material clearly showed (see Fig. 5), was
heaped on either side of it. Elsewhere, in the west
and north ramparts, the traces were less perfect. The
slot became a wider thing in stony upcast, and was
also much discoloured, where the stones had jammed
r
A
. ' CLOSE-UP ' OF A'S SOUTH RAMPART, SHOWI NG
PALI SADE-SLOT. A AND MARK THE TWO
SI DES OF THE SLOT.
P L A T E VI .
To face page 27.
B. OVEN A2, VI EWED FROM RAMPART-TOP.
IN THE NORTH RI DI NG OF YORKSHIRE 2<)
and it had not filled up solid: but only at one point,
in the south rampart just west of the south-east
angle, was it not to be found at all : and there the
stones composing the rampart were so large and so
tumbled, that no impression could be expect to survive.
At the angles, the trench swung round in an easy curve,
parallel to that of the rampart and ditch, and only at
the south-east angle does it seem to have been inter-
rupted. Further, two points which bear upon the
length of life of this timber-work are clear. The wood
was withdrawn from this slot, and did not decay
therein ; for, had the decay taken place, there would
have been found either traces of the wood, or great
discolouration of the sides of the hole and a much less
well-consolidated filling. Secondly, to fill a vacant hole
so tight as on the south side, the upper part of the
rampart must have been loose and not long consolidated
by time. In other words, the wood was not only
purposely withdrawn, but dismantled fairly soon after
it had been put in for the first time. The question,
when this was done, will be discussed later, in connexion
with the second occupation (see p. 48).
The two elements so far described, frontal post-
holes and palisade-trench, ran all round the earth-
work (see plan, PL xx). They were sought in many
places, and, as explained, the frontal post-holes never
failed to appear, while the palisade slot was only
missing at one point. The third element was a line of
rearward post-holes, stout and deep, set roughly five
feet apart at the rear of the rampart, some six feet
behind the palisade line. These were found between
the east and south gates, with a break at the
south-east ballistarium. Like the other holes, these
contained no timber, but a mass of discoloured packing
stones on edge, just like those of the palisade-trench :
and in rockier ground, where they had been particularly
difficult to cut small, they became veritable pits.
The closer interval and size of these post-holes com-
pared with the frontal line, and the relatively small
mass of collapsed rampart behind them, combine to
suggest that they held strong uprights, supporting a
vertical back for the rampart. But all attempts
failed to find these posts in any other sector than that
already described. They did not exist elsewhere on
the circuit ; and thus it would seem that the rampart,
like the west tutulus, was never completed.
What was the final, and what the intermediate,
form of this rampart ? The strong line of posts at the
rear, six feet behind the palisade-trench, demands a
vertical back; and this is also warranted by the
relatively small amount of debris behind the line of the
post-holes. Granted this back, it is not in doubt that
above it must have lain the rampart-walk. Six feet
further forward, comes the continuous line of upright,
thin planks. This cannot be an intermediate support
for the rampart-walk : it must be a continuous front.
And the only point which remains in doubt is whether
it had a parapet only, or merlons as well. Finally, the
frontal post-holes will carry struts for the palisade,
in the manner of Remagen
1
and Alteburg,
2
but
without their complicated underground ties, of which
there was not the slightest trace. There may have
been a low brattice, to prevent debris from trickling
into the ditch, but this seems doubtful, since the posts
appear too far apart to hold such a structure safely.
The intermediate form has no vertical back ; and
it is hardly possible to assume that the rampart was
finished in this guise, for a reason which is as follows.
To provide a steady rampart, upon which it would be
possible to walk behind the palisade, it would be
necessary to have at least twice as much material
behind the palisade as now : and to get so much earth,
even from so large a ditch as here, it would be necessary
to have the palisade on the front line, in order to pack
every bit of earth behind it, as at Munningen.
3
But
this was not done : and it is, therefore, evident that the
stage reached was that when earth was heaped on
each side of the palisade, holding it tight, but leaving
sufficient room to fix the rearward posts of the rampart-
walk, and fill in the casing thus formed, so as to create a
1
Remagen, Bonner Jahrbuch, 1 4 -
3
Munni ngen, O.R.L. 68a, pp. 3-4,
115 (1906), p. 229, pis. vi i -x. pi. i, ii.
' Al t ebur g, Bonner Jahrbuch, 1 1 4-
115 (1906), p. 249, pis. xi i -xv.
horizontal top to the rampart. Thus the stages in
construction are as follows :
(1) Lignatio : the preparation of wood (sudes) for
the palisade, composed of timber quite
different from the stakes (valli) which the
ordinary legionary carried.
1
(2) (a) Ditch-digging; with which is contem-
porary
(b) The making of the agger, heaped up on
either side of the palisade,
(3) The completion of the rampart, by constructing
a permanent high-level rampart-walk behind it.
The whole rampart was brought as far as stage
2b, and only a small sector was brought to stage 3.
This fact emphasises a point of importance. Not
only is the whole order quite inevitable, but the
character of the work is quite different from a tem-
porary camp, or aestivalia, where ditch or rampart
must be completed as soon as possible, and cannot
be delayed by the preparation of wood, which in this
case must have been a lengthy business, occupying
some days. And since there is no sign that A replaced
a slighter fortification, that grew into A, it follows that
its builders must have lived elsewhere while making it.
(d) The Gates. The position of A' s gateways is
curious (folding plan, PL xx). Only the north and south
gates are opposite each other, and these are almost
central, leaving only a slightly larger part of the camp
to the west : also, the north gate is a small postern.
Yet, if the ordinary rules of castrametation obtain
here, the north and south gates must be the portae
principales, since only they provide a straight run
through for the via principalis and allow sufficient
space for the praetorium and retentura. The next
question, seeing that the camp is not tertiata, is the
position of portae praetoria and decumana. If general
principles of castrametation count for anything, the
east gate,
2
facing the enemy, should be the porta
praetoria. But it is not central, and in this respect
1
These would be the quadrifidas
3
Vegetius, i, 23.
sudes of Vergil, Georg. ii, 25.
breaks the normal rule. Yet the displacement has
parallels, for example, the early fort at Ambleside;
1
and here the device gives additional room for deploying
between the gate and the escarpment. To choose the
west gate for porta praetoria only introduces further
difficulties. It is not on the offensive side of the camp ;
only the back of the entrenchment would then have a
ballistarium; the back rampart would be completed
before the front ; and the porta praetoria would have
an unfinished tutulus. Accordingly, it seems easier
to accept the abnormal position of the east gate, and
to identify it as porta praetoria, than to satisfy the
augural requirements, leaving the tactical difficulties
unexplained.
The names of the gates, however, matter less than
their form, and about this much evidence came to
light (PI. vii). When the tutuli were discovered, it
became clear that the early gates of A must have been
much narrower than the wide openings between the
claviculae, for tutuli are intended to cover at least the
whole gateway opening.
2
Again, the openings, to
tally in character with the rampart, must have had
straight, boarded sides.
The first verification of these inferences was
obtained by finding that the ends of the palisade-
trench coincided with those of the tutuli. Post-holes
were then sought and found to the front of the palisade,
terminating the frontal line of post-holes, but very
much larger than these, and confirming the view that
the latter had not a great weight to support. This
indicated a line of boarding on each side of the gate,
at right-angles to the rampart and supporting its
butt-ends. But it was clear that this boarding could
not have stopped at the palisade, but must have
continued backwards, to the rear of the rampart.
Further search then revealed, twelve feet further back,
another pair of great post-holes, defining the length
and width of the entry, between boarded sides. At
the north, east and west gates (PI. vii) the plan thus
outlined was rectilinear ; at the south gate (PI. vii)
1
C. & W. A. fif A. Trans., Ser. 2,
2
De mun. castr., c. 49.
xxxi, p. 5.
To face page 30. PLATE VII.
^ 4 / ^ B: SOUTH-EAST GATE
DITCH
A'sfmisked ascer\sus
A: SOUTH GATE.
B: Sourttwisrtkre.
A:NORXH.<J*TE,
B:NOTM-wtr GATSL
mWTWl WTO
lil<liiililiiilililil(/ \
B:NOWH-EAST GATE.
\ \
V *
/K
Metres.
P Feet.
o o i ^ Y "
1
r *" ^ 1: '
1

PaliSOiic--trer\ca j J |
n i i n i i n mu j j j JliUIHH m i n i i B
.!., ' :. 1;;.,. ,r. t i , L . I i , :
Tutulus A: EAST GATS, OEMOUSMSO
o - *
Tutulus
x - \ i =rO
v
' funimished)
A I WSST GATE.
B: \ VITST GATX
GATES OF A AND B.
IN THE NORTH RI DI NG OF YORKSHI RE 31
the frontal post-holes curved inwards, forming a
recessed entrance like the stone gateways of Plumpton
Wal l
1
(Voreda), but only the east side was completed.
All the gates had a uniform width, except the north,
which was a narrow six-foot postern. There was no
indication of doors.
The twelve-foot extension of the entrance behind
the palisade raises a further question. Clearly, it
implies a special treatment of the back of the rampart
at this point, since the normal depth of the rampart
behind the palisade is about six feet. Again, if the
rampart possessed the vertical back which the rear-
ward post-holes imply, there must have been stair-
cases or ramps (the ascensus
2
of literature) leading
thereto ; and, wherever else these may have been
situated, there must have been at least one at each
gate. No sign of these was visible on the surface.
But, on peeling off the turf, a great mass of rampart-
material could be seen extending inwards at right-
angles to the rampart (PL vii), under the inner
claviculae of the south and west gates. This mass
was not related to the claviculae above it, but was
obviously going to be used to make an ascent. At the
east side of the north gate (see Pl. vii), a like mass
seemed to lie parallel with the rampart, and similar
parallel mounds could be seen at the west and north
sides, respectively, of the south and west gates. The
east gate, being demolished, provided no evidence.
It was not to be expected, however, that, where the
vertical rampart back did not exist, the ramps or
stairs could have been fully constructed. Accordingly,
only on the east side of the south gate was it possible to
trace, by the aid of post-holes, the plan of an ascensus.
This was (Pl. viii) a rectangular extension of the back
of the rampart,
3
six feet square, which would provide
for a rise of about six feet, with the usual nine-inch
treaders and ten-inch risers of Roman staircases : and
!
Pl umpton Wall, C. &-W.A.& A.
Trans., Ser. 2, xiii, p. 179, also
Haltwhistle Burn, Arch. Ael. 3rd ser.,
vol v, p. 224 : in civil fortifications
the device appears at Colchester and
Silchester, but is very rare.
2
De mun. castr., c. 58.
3
The material for this work woul d,
no doubt, be derived from the ditch
in front of the Gate, from whi ch the
upcast woul d not be needed for the
rampart: hence the need for digging
a ditch here.
the relative shortness of the projection, compared with
the ramps of Saalburg I I
1
which measure as much as
six, seven or eight metres in length, suggests that stairs
rather than a slope were employed. Similar stairs
were no doubt planned at the west gate : but the
mounds (see Pl. vii) at the other sides of the west and
south gates, and at the east side of the north gate,
show that it was intended also to build the alternative
type of ramp, parallel with the rampart, whence
the duplices ascensus, or twin ramps,
2
of literature
were derived. One such single ramp was discovered
at the south-east ballistarium (see Pl. viii and p. 34),
but the absence of post-holes elsewhere prevented the
identification of more.
This lack of detail may be regretted, but there are
compensations. For the gates thus provide the same
evidence for unfinished work as the ditch and rampart.
And there is a further important point. The shape of
the mounds at the unfinished gates clearly indicates
that the south and west gates were going to be pro-
vided each with two different types of ascensus, the
sloping ramp parallel to the rampart and the staircase
at right-angles to it. But this difference, as between
one side and another, is quite contrary to the normal
Roman arrangement, which provides identical stair-
case arrangements at each gate, in order to avoid
confusion : and it becomes of importance later, in
considering the purpose of the camp.
One more question remains. Was there a tower, or
an overhead gangway, above these entrances ? The
size and depth of the post-holes proves clearly that
there was no tower. Again, a gangway would
require support in the centre of a twelve-foot opening,
and holes for any such support were not to be found ;
and if there were a gangway, there would be no need
for ascents on each side of the gate. These remarks
apply to the south, west and east gates. The six-foot
postern at the north gate appears to have had only one
ascensus, and may therefore have been bridged by a
gangway a continuation of the rampart-walk : but this
1
Saalburg II, Saalburg Jahrbuch,
2
De mun. castr., loc. tit., duplices
iv, 1913, i, p. 14, Abb. 4, 5, 6, 9. ascensus.
s
1 i
S METS-ES
- t
3 FEET
Gun-pit
Moi a r r v t u r f
Skeets of
vegetable matter
BALLI STARI UM OF A, PLAN AND SECTI ON.
is by no means certain, since the post-holes at the end of
the palisade, which should have held one of the main
supports of the gangway, were not at all deep. It
seems wiser to restore all the gates in the same manner,
as long narrow passages, the angustiae portarum,
1
with high boarded sides, sloping down with the rampart
from the palisade towards the front, but keeping
uniform in height from the palisade backwards, to the
base of the stairs. So those who were mounting the
stairs would have cover, while the defenders of the
gate could stand shoulder to shoulder against the high
sides, like Caesar's hero, Baculus.
2
Such a gate
could also be rapidly closed
3
with a sham wall or with
a spiked barricade (ericius).
(e) The Ballistarium. One of the prominent
features of A is the great mound at the south-east
angle (Pl. viii). When trenched in 1924 it was found
to be built up with alternate layers of turf and very
stony upcast, like the alternate clay and brush-wood
layers of the Birrenswark ramparts.
4
This com-
position, meant to give resiliency to the platform upon
which the powerful spring-gun recoiled, indicated at
once the real purpose of the mound. On top of it
was a small gun-pit, much defaced by a large tree
root, but, nevertheless, recognisable. In front of the
mound the place of the frontal post-holes was taken by
a great pit, whose edges were cut to a very definite
shape, indicative of a series of beams to which the
palisade and its struts had been fastened underground
5
. (see Pl. viii and Fig. 6).
No system of post-holes was discovered, to suggest
that the back of the mound was retained. On each
side of it (Fig. 8) the rearward line of posts belonging
to the east and south ramparts ceased when they
reached the mound, and the mound itself seems to
have sloped downwards at the angle of rest. It was
not ascended by climbing this slope, however, since
1
Li vy, xxxi v, 46, xl, 25 ; Taci tus,
4
Birrenswark, P.S.A. Scot., 1898-
Agricola, 26. 99, p. 224.
2
Caesar, B.G., vi , 38.
5
Caesar, B.G vii, 73, 2, hue
demissi et ab infimo revincti : also
3
Caesar, B.G., v, 51 (wall of sods) ; Spartian, vit. Hadr. xii, 6, stipitibus
B.G., iii, 67 (spikes). . . . funditus iactis.
the post-holes for a parallel ascensus were found at the
west side of the mound. Three of these ran along the
back of the south rampart, denoting a ramp twelve
feet long, rising at a gradient of about one in four.
This gives a fairly intelligible gun-platform. But three
other post-holes in the body of the rampart were hardly
explicable on the assumption that they belonged to the
ballistarium. They look more like posts connected
with propping B' s rampart over this awkward hump.
The machine used on the mound was a fairly large
FI G. 6. BALLI STARI UM OF A, RESTORED
one : but the curious sheathing at the front, com-
bined with the absence of any substantial sheltering
tower, gives the same impression of temporary
occupation as at the gates.
(/) Interna] structures. The only structures inside
A, apart from pits, which can be assigned to the first
occupation are earth-ovens. All others, roads, tribunal,
stone ovens, cook-hole, latrine and ' dug-outs,' go
unquestionably to B, as will be explained when is
described. The earth-ovens divide into three distinct
groups, those of the west rampart, that of the south-east
quarter, and those of the centre.
(i) The ovens of the west rampart.
This group consisted of three ovens (Fig. 7 and
Pis. and vi), carved in the back of the rampart,
below ground-level, and disposed in a row, some 26 feet
apart. Each oven was round and shallow, with a flat
FI G. 7. OVENS I N A, PLANS AND SECTI ONS
floor of natural earth ; and in front of it was a great
stoke-hole. The southernmost stoke-hole had no step,
the other two had one step cut in the back of each.
From the distribution of the ashes, it was clear that
those using the ovens had stood at one end of the
stoke-hole, and had raked out the ashes into it. The
oven doors were about one foot wide, with a special
notch in one side, as if for inserting a portable door,
and there was, as usual in these structures, no flue
to create a through draught. It is also evident that
they were not used for long, since the burning within
them had not baked the floors very heavily ; nor,
indeed, can such structures have been strong enough
to endure much usage.
Some light was shed upon the structure of this type
of oven, and its history at Cawthorn, by the northern-
most oven (Fig. 7 A I, PL VB). This was quadrilateral,
with rounded angles, a shape ill-adapted to carry a
domical roof : and the roof, composed of layers of
burnt earth (perhaps turf and clay, but the burning
had disguised their nature) had collapsed. About one
half of this ruined oven had been left as it was, with
remains of the fuel, oak-scrub, inside it : the rest had
been cut away, and a new oven had been dug, opening
out from the north side of the old one. This oven was
complete and had never been fired. But it went deeper
into the rampart, and therefore its sides and back
were carved deeper into solid earth than the others.
It was thus possible to roof the structure entirely with
large turves, and a great unburnt mass of these was
found, covering the virgin floor and choking the oven.
The history of the structure thus becomes plain. After
short usage, the first oven collapsed, and a second was
built, leading out of it and using the same stoke-hole ;
but this one was never fired. No clearer proof could be
afforded of a brief occupation.
Ovens of this type are not common. The position
is perfectly normal, and the early north rampart of the
fort at Malton
1
has yielded very similar earth-ovens,
now collapsed. But no example of the elaborate
stoke-hole has yet come to light. It may be thought
1
Roman Fort at Malton, p. 40 ; Fi g. 49, T5.
that at Cawthorn three factors led to the making of
the stoke-hole. In the first place, the site is very ex-
posed and windy, and it would be much easier to deal
with the rakings from the oven in the shelter of a
stoke-hole than in the open. Secondly, it was neces-
sary to cut the oven as much underground as possible,
in order to have solid sides of the kind that the loose
rampart-material could not provide. Thirdly, it was
difficult to rake out such low and frail structures with-
out destroying them, unless it were possible to rake
horizontally, on a level with the floor. These needs
could only be met by digging the stoke-holes which we
see. It must be noted, however, that these ovens
will not fit a rampart with a wooden back, such as
was built in the south-east angle of A. They would
come too close underneath the wood-work, and would
certainly set it on fire. Yet it is quite clear that these
ovens belong to the first occupation of A, since the
intravallum road of was laid on top of their filled
stoke-holes. It is thus evident not only that no
wooden back was fitted to the rampart here, but that
it was not intended to fit one. Yet the rampart as
constructed is useless without such a revetment.
In other words, by the time the ovens were constructed
it was known that the rampart was going to be left
incomplete. This implication will be discussed afresh
when the purpose of A is considered.
(ii) The south-east quarter. One oven of the
type which has just been described was found in the
south-east quarter of the camp (Fig. 7, A4). As might
be expected, it lay away from the wooden back of the
rampart. There are no constructional peculiarities
to record : but the position (see folding plan) is quite
abnormal, if the encampment is considered to have
been designed upon permanent lines.
(iii) The central ovens. While the exploration
of pits in the interior of A was in progress, a depression
in the ground near a rather prominent hummock of
firm sand was examined. Starting with a long trench,
this presently led into a perfect oven, cut in the soil,
with an intact roof of ordinary soil. It was also clear,
from the plan (Fig. 8), that at least one other oven,
whose outline is completed in broken lines, had been
cut through in making this one : and there was another
incomplete one to the right. There were thus at least
three successive ovens, of which none had been fired,
nor was there any stoke-hole.
Two interpretations of this complex are possible.
Whichever is chosen, it is clear that the work was
incomplete, since the stoke-hole was not dug, and its
place was occupied by a narrow trench, from which it
would be quite impossible to operate the oven when
fired. But the fact that the ovens were dug in succes-
\
e
/
1

A5 1
1 \ \
c

a(vb uj b
V
/
/
y /

5
?
5
rrr-r
IA R
FI G. 8. PLAN OF OVENS NEAR CENTRE OF A
sion may mean either that three attempts were made
to dig an oven for use, and that the operation only
succeeded after two failures while the stoke-hole was
not yet dug : or that a series of ovens was constructed
for practice, and scrapped deliberately, one after the
other. The latter hypothesis seems the more attrac-
tive. For the position of these ovens fits in with no
normal plan, either temporary or permanent. Rather,
it may be thought that the hummock has been selected
because it seemed a good place for practice, while the
ovens were not fired because they were not meant to
be used. This view will be considered afresh when the
purpose of the camp is discussed (see p. 46).
(g) The Pits. Two great pits, lined with some sort
of vegetable matter, perhaps moss, were discovered
behind the south rampart, on the intravallum, west
of the south gate. The purpose of these is obscure.
They contained neither rubbish nor the remains of
sewage, and would seem to have been used for storage,
since the western one was provided with a step for
getting down into it. But the position is a damp one,
and it is difficult to see what they could have held
except water. Three points suggest that this inter-
pretation is right. The vegetable matter with which
they were lined may indicate some sort of proofing
to retain water in them. Their gently curving sides
suggest that they were not ordinary storage pits.
If there were baggage-animals in the train of the little
army, as might be expected,
1
some such pools would
be useful for their watering, upon so completely dry a
site.
So much for A, its defences and internal structures.
Its first occupation yielded no relics, and its relation
to the second occupation is discussed on a later page.
The discussion of its purpose is also postponed, since
it is intimately connected with that of earthwork C,
now to be described.
(3) EARTHWORK C
C is a remarkable camp, whose coffin-like shape
(plan, Pl. xxi) has attracted attention ever since
it was first planned by General Roy.
2
It has been
explained as a cattle-compound, or, again, as a camp
of auxiliary troops, less well versed in castrametation
than the legionaries : and its close connexion with A
has always been assumed. Whatever the value of
these explanations, there can be no doubt that the
last assumption is correct, since all the gates that the
camp possesses (see Fig. 3) face A, and have wide
passages, facilitating ready access thereto. Our new
evidence, however, carries a little further than this.
1
For baggage-animals, see De
2
Military Antiquities of the Romans
mun. castr., 5. The y were essential in North Britain, Pl. xi.
for transporting the tents.
For, when it is proved that and D belong together,
and that C is considerably older than D, just as A is
considerably older than B, the inevitability of the
connexion between A and C is greatly strengthened.
For the position of C, in the low ground between the
two plateaux occupied respectively by D and A (or ),
would never have been chosen unless A (or B) was
already there or was planned to be put there. Now B,
which goes with D, was certainly not there. Thus,
only A remains, though the question whether it was
there or not must wait until C has been described.
The excavations did not extend to the interior of
C. For it was always plain, from surface indications,
that this camp had contained pits and temporary
erections of turf, just like those of B. Thus, as time
was limited in the last season, when A and had
been thoroughly examined, the interior of C remained
unexplored.
The defences, however, were examined in some
detail. They consisted of a turf rampart and a ditch.
(a) The ditch. The beautifully cut ditch (Fig. 9)
was of small dimensions, with the regulation channel
at the base. It was carried all round the camp (see
folding plan), including the claviculae at the gates.
But it was not finished on the short north-eastern
sector, north of the northernmost gate, where a great
pit interrupted its course. Between this gate and
the central east gate there were also three more
pits. These pits were not later than the ditch, for
the latter died into them, and was not cut through
by them : and the finding of a fragment of rough
native pottery in association with one of them sug-
gested that they were earlier. But their purpose
remains obscure, though it appears not impossible that
they may have been quarry-pits connected with the
tumuli (see Fig. 1) which exist in and about C. Yet,
whatever their purpose, it seems highly likely that they
influenced the lay-out of the north-east end of C, and
partly account for its odd shape. The upcast from
the ditch was spread in a very low mound outside it.
(b) The Rampart. Immediately behind the ditch,
separated from it by no berm, came a small rampart
To face page 40.
P L A T E I X.
B. OUTER DI TCH OF D, CUTTI NG THROUGH W. DI TCH OF C (HI GHER AND TO
RI GHT), AT THEI R S. MEETI NG.
A. B'S SOUTH DI TCH ENDI NG AGAI NST THE WEST
SI DE OF THE SOUTH porta quint ana.
"built in turf and very well preserved (Pl. ix A). There
were no post-holes in connexion with it, and it was
clearly neither wide enough nor high enough to carry a
rampart-walk. In fact, it is a simple breast-work
originally about five feet high, which weather has not
denuded much. It is also clear that there was no sort
of breast-work on the top of it, such as might have given
to the defenders the shelter of parapet and merlons.
For such a structure, had it existed, would certainly
have been embedded in the rampart, and the turves,
t Z 3 4 6
11 1 1 1 1 1
a I S.METRIS
FI G. 9. DI TCH OF C, SECTIONS
with their particularly plain black lines (see Pl. ix A),
representing the grass-faces, were quite unbroken.
Thus it is possible to be quite sure that the rampart
was very like its present remains in the original form,
though rather larger in bulk, and sharper in outline.
The rampart and ditch of C are thus remarkably
different from those of A. They could be built and
completed with great rapidity, and they are exactly
the kind of structures prescribed by literature
1
for
1
De mun. castr., 49 (ditch). 50 (rampart).
aestivalia, granting the fact that the upcast from the
ditch is too loose to make a rampart, and also too small
in quantity. The temporary character of the work
would thus be explicable if the builders were there
for a short time only, on the march or resting for a
night or two. But it is evident that this was not so,
since the enclosure is covered with temporary structures
which imply a rather longer stay. The explanation
must therefore be that the builders had other work to
do, and were anxious to spend as little time as possible
over the defences of the earthwork in which they lived.
The odd shape and position of the camp, the fact that
its occupants decided to live in it while their other duties
were in progress, and the very slight nature of the
defences, indicate beyond doubt that there was no
fear of an enemy or of surprises : whatever the work
in hand, it was to be carried out in peace and tran-
quillity.
(c) The Gates. The arrangement of the gates (see
folding-plan, PI. xxi) is quite abnormal and un-
paralleled. There are only three, and all of these are
on the east side of the camp, equidistant from one
another, and from the angles. The arrangement thus
breaks all the basic rules of castrametation, and in-
dicates that the occupants of the camp were concentrat-
ing all their attention towards the east. But that
attention cannot have been concerned with war-like
operations, since, in addition to the points already
mentioned, it is clear that no commander in wartime
would have laid out his camp without portae principales
or, more especially, without a porta decumana, which is
essential on this site in order to cover the D-plateau.
The gates themselves are all three of exactly
the same pattern and dimensions. The opening
is wide, 38 feet across, and direct entry thereto
is precluded by an external clavicula, running out, as
regularly, on the right-hand side of defenders of the
gate. The ditch also follows round the clavicula, and
terminates, as discovered in the northernmost gate,
in a bevelled square end (cf. PI. A), neatly finishing
the whole work. There was no trace whatever of any
wooden structure closing the entrance, although post-
I N THE NORTH RI DI NG OF YORKSHI RE 43
holes were sought: and, indeed, it is clear that the
presence of the clavicula renders useless such a barrier,
which would leave the clavicula outside the gate and
thus provide an approaching enemy with cover.
The external clavicula is a more effective barrier
1
than the tutulus: for the latter bars the main line
of entry, but does not prevent an oblique rush from
the left by a force provided with shields. The real
merit of the clavicula is to hinder this variety of
attack, by leaving only an enfiladed entry from the
right. Thus, it improves upon the tutulus by being,
in effect, a tutulus linked with the rampart on its
vulnerable side. The internal clavicula is designed
on the same principle, and both can be combined, as
in B. In time of danger, the entrance is closed, as at
Dealgin Ross,
2
with a barrier.
The following conclusions about C, therefore, follow.
The camp was rapidly constructed, with the minimum
amount of labour, and with the minimum regard for
defensive advantages. It contained a rather small
body of troops, whose duties were somehow connected
with the ground east of the camp. The troops were
there for some time, since they rigged up temporary
buildings, of the type which will be studied more
closely in B. But they can hardly have been engaged
in active campaigning, since they neglected the
elementary precautions which a Roman army on
campaign habitually took. It cannot be they did not
know these precautions, or that they were unskilled,
for their work, so far as it went, was highly efficient.
Rather, they ran up their bivouac as quickly as pos-
sible, and began the work which they had come to do.
What was this work ? The conclusion now becomes
irresistible, that it must have been the occupation
of the plateau upon which A stands. No other sup-
1
Compare its use by Caesar at satisfactorily. The combination of
Mauchamp (Napoleon, Atlas, Pl. 9) inner clavicula and tutulus, as at
and by Flavius Silva at Masada Featherwood and at Gl enwhel t Leases,
(Hawkes, Antiquity, l une, 1929, woul d be another method, but this
Pis. i - i x, p. 198) in situations exactly is rare ; and a doubt may be ex-
supporting this view. pressed whether, as von Domaszewski
2
Military Antiquities, Pl. x i ; it and Collingwood suggest, the two
seems doubtful whether digging would things are meant to be used together,
now reveal these peculiar gateways as described in De mun. castr. 50, 55.
position will account for the awkward position and the
abnormal lay-out of C. For, if the work was further
afield, then the army should either have occupied the
A plateau, if A was not there, or have re-occupied A,
if A was there already, or have occupied the D-plateau.
Now A was indeed re-occupied, but the earthwork of
that re-occupation was B, and has nothing to do
with C, but is closely connected with D, which is
some years later than C. Thus, neither a re-
occupation of A, nor an occupation of D is possible,
and it follows that the occupants of C came to
undertake the original building of A. They did
not occupy the splendid position on the A-plateau,
because they were intending to build A there. And
it must be emphasised once more, at the risk of
reiteration, that they came in peace.
How will this view of the case suit the evidence
collected in A ? In the first place, it will be observed
that the explanation adopted for C is just what is
needed for A. It has already been pointed out that A
was unfinished when it was abandoned, and that
considerable time had been involved in getting it even
so far built. In other words, the troops who were
building A must have lived somewhere else con-
venient while they were at work. Thus, the conclusion
that they lived in C now becomes really cogent, since
it is demanded by the peculiarities both of A and of C,
and fits both sets alike. In brief, C was laid out in
relation to Anot vice-versa : C was finishedA was
not ; C was intensively occupiedA was not ; C was
rapidly constructedA took some time to build.
Thus, while our evidence supports the close connexion
which had always been held to exist between A and C,
it reverses the generally accepted relation between
the two. For, hitherto, it has always been supposed
that C was some sort of annexe to A. This remains
true, but in the sense that the dwelling of a building-
staff is the annexe of the building under construction.
(5) THE PURPOSE OF A
The facts thus elicited have an important bearing
upon the purpose of A, but they result in an apparent
contradiction. It is quite evident that C was built
under peaceful conditions, as its plan and defensive
arrangements show. But A is very strong ; it has all
the air of a permanent erection, and its ditch and ram-
part would rank with those of most Flavian forts.
If it had been completed and equipped with permanent
buildings, it would have been a regulation fort.
Thus, it might readily be thought that A was a per-
manent fort, for some reason never finished, and that
C was the bivouac of those who came to build it.
Gellygaer
1
provides a useful parallel of fort and
bivouac side by side.
It is just at this point where the contradictions
begin. In the first place, the site is not such as the
Romans were wont to choose for the erection of a
permanent fort, where the water-supply was always a
prime concern ; and here no water was obtainable,
even if an attempt had been made to get it by digging
wells. The position is also exposed to an abnormal
degree : and, even supposing that permanent occupa-
tion was required in the Cawthorn district, which is
by no means clear, there are several sites quite close
which would have provided far better conditions.
Finally, granting all this, why was the work not com-
pleted, when it started amid conditions of peace ?
It might be argued that the work started in peace and
that opposition developed. But it was not the Roman
way to yield before such opposition to permanent
schemes anywhere, and it is quite out of the question
so near the legionary fortress of York.
Another line of explanation would be that C was
built in peace for some other work than A : that war-
conditions then developed, and that the occupants of
C entrenched themselves strongly in A, the crisis passing
before A was finished. But this implies that A was
not envisaged when C was built, whereas the very
plan and position of C prove that operations on the
adjacent plateau were intended from the moment
when C was first laid out. Again, inside A, the
occupants were spending valuable time in building
1
Gel l ygaer: Collingwood, Archaeology of Roman Britain, Fig. 6.
ovens, when, on this hypothesis, they ought to have
been devoting their whole attention to the defences.
Finally, the defences of A, though very strong, are the
result of long and deliberate work : they could not
be erected rapidly, as this view demands.
What explanation, then, will fit a permanent
work, built in time of peace upon a site quite unsuitable
for such a work, and then left unfinished ? There is
just one possibility that remains unexplored, namely,
the view already advanced to explain the central
group of ovens in A, that the whole work was done for
the sake of practice. In other words, that the camps
were built for training during manoeuvres, and were
then abandoned.
It is clear that this view explains C better than any
yet advanced ; its odd lay-out ; the calculated con-
centration of its occupants upon the special work in
hand upon the A-plateau ; and the semi-permanent
character of the buildings inside it. But how will
it fit A ? This may be demonstrated by recapitulat-
ing the different problems. The defences become
completely comprehensible. Any amount of time may
be spent on their erection, since as much practice as
possible is wanted for all kinds of work, lignatio,
ditch-digging, and the erection of a wooden palisade
and rampart-walk. Only one angle need be completed,
however, since this is enough for practice. For the
number of troops involved, on this hypothesis, cannot
have been very large. All were contained in C, which
would hold at the most two cohorts. Again, the
plan of the gates (see Pl. vii), with different types of
ramp, explains itself : as much practice as possible
is obtained by building gates of mixed plan, quite
unsuited to serious campaigning, but useful in man-
oeuvres for teaching both methods of ramp-building
or entry-revetment at once. The ballistarium (see
Pl. viii) also fits as a practice platform, unprovided with
a permanent tower, perhaps for first lessons in using
the machine. Finally, the ovens turn out to be
practice-ovens (see Fig. 8), which can well be built in
abnormal positions (see folding-plan, Pl. xx), and are
not always kept after being used, but demolished one
after the other, as in the central mound (see Fig. 8), or
begun again, after a collapse, as in the west rampart
(see Fig. 7, PI. vb) .
A further implication must now be mentioned
about the character of the troops involved. The
validity of this conclusion is quite independent of
the manoeuvre theory, but it must be mentioned here
because it goes far to strengthen that view. The
camps are early in date. This is suggested by the
claviculae of C and by the fact that A is not tertiata ; and
it is conclusively proved by the pottery associated
with the B-occupation, described on a later page
(p. 76, Fig. 20), to belong to between A.D. 90 and 120.
The fact that A has a ballistarium then becomes of
great importance, since ballisiae as field-equipment
were not yet being issued to auxiliary cohorts, but
were entirely the equipment of legionaries. Legion-
aries are thus involved at Cawthorn, both in C and A.
And with legions, so near their own quarters as
Cawthorn is to York, there can be no question of
changed plans or abandoned purposes. The occupation
falls into place, in its first phase at least, as the
manoeuvring-bivouac of the York legion, where one or
two of its cohorts, perhaps those containing new
recruits, learnt to build entrenchments, and to use the
ballista. On this exposed site they also underwent that
hardening process which was the glory of an efficient
Roman legion.
1
(6) THE EVACUATI ON OF A
Evidence already cited in connexion with the
timbering of the rampart of A suggested that the
occupation of this entrenchment was not long ; and
the character of the ovens and their use also hinted
at the same conclusion. Much more positive evidence
upon this point was given by the content of the ditch.
Wherever the ditch was examined, the lower half
thereof (see PL ii) was filled with a mass of rampart
material, disposed so as to show that it had been shot
down into the ditch and had not drifted thither with
the action of weathering. Below it, there was very
1
See . 1, p. 77, for references to these activities in the Roman Army.
little natural silt, barely what would fill the channel at
the bottom of the ditch. This evidence, then, would
suggest that very soon after the ditch had been dug,
it was half-filled with material from a demolished
rampart ; and the inference is that the demolition
occurred when the wood was being removed from the
rampart at the time when A was dismantled.
It might be suggested, however, that this dismant-
ling took place not when A was evacuated, but when
it was occupied for a second time, and refurbished
in different form. But this notion was disproved at
two points. When the northern junction of A and
was examined, it was found (see Pl. ii) that B' s turf
rampart had not been built directly upon the returned
mass of rampart material lying in A' s ditch. It was
laid down upon a mass of fine marl filling, and below
this, sandwiched between it and the returned ram-
part material, there was a thick layer of vegetable
matter. Part of this was undoubtedly laid turf,
but the last two inches looked like growth in the lowest
part of the ditch. A clearer discovery was made at
the north gate (see Pl. vii), where the ditch had been
filled up to carry the causeway of the B-period occupa-
tion. The lower half of the ditch was filled with the
usual mass of returned material. Above this came a
thick blue-grey band of vegetable matter, and on top
of it again lay a further mass of returned material,
forming the causeway of the second period. Here
there was no suspicion of the admixture of turf, and
no mistaking the line of growth. In other words,
the occupation of was separated from that of A by
such time as is required for a band of vegetable matter
two to three inches thick to grow in the hollow of A' s
half-filled ditch ; and the fact becomes clear that the
filling was done at the close of the occupation of A,
and that the growth took place between this operation
and the construction of B.
Exactly the same type of evidence was yielded by a
couple of sections (Pl. ii) in the east side of A, where
the whole of the rampart had been demolished in the
second period and thrown back into the ditch, so far
as the latter would hold it. The two returned masses
were separated from each other by a band of vegetable
matter, from two to three inches thick.
It is not at all easy to estimate how long such growth
may have taken to form. But the following facts which
bear upon the point may be recorded. Some of our
own ditch sections were left open for two years. At the
end of one winter, in damp positions, a thick growth of
moss had established itself right up the side of the
ditch, and the bottom channel was well filled. In two
winters the growth had greatly increased, and strong
coarse grass had established itself, in vigorous tufts.
Thus, it would seem that about six to ten years, and
not more, should be allowed for the production of
conditions which would give the strong line of vegetable
matter on this particular site.
A check upon this calculation is provided by the
relation of C and D. At the southern junction (Pl. ix B)
of the ditch-system of these two camps a detailed
examination proved that D' s outer ditch had been
laid out so as to intersect C's ditch, and to cut right
through it, regardless of its presence. At the point
of junction C's ditch was therefore covered up with
the yellow sandy upcast from D' s ditch, and all
the deposit which had got into C's ditch up to that
moment was thus safely sealed. C's rampart, which
was also in the way, was not returned into C's ditch,
but spread in a broad mass as the glacis of the new
ditch of D. Now the filling of C's ditch was very
distinctive (Fig. 9), being composed of the wash-down
from C's turf rampart : and it filled the whole of the
channel of C's ditch and about one quarter of the
ditch above that level, almost as high as filling of
this class reached in any ordinary section-. Im-
mediately above it came the equally distinctive
yellow sand upcast from D' s ditch. Here there was
no vegetable growth separating the two masses. In
other words, by the time D' s ditch was made, weather-
ing had not yet finished its work on C, and was still
depositing enough material from the rampart into the
ditch, to prevent settled growth taking place there.
In this case, a time-limit is a little easier to judge.
When a turf-rampart is first built, weathering con-
solidates it, and in the process deprives it of a great
deal of loose material. This denudation goes on for
two or three seasons. But by that time growth is
beginning to counteract the weathering action, and
steadily reduces it. This has an immediate effect
upon the conditions in the ditch into which the turf-
rampart is silting. For as soon as the rate of deposit
becomes small, vigorous growth can begin in the ditch ;
and, once growth is established, the character of the
ditch filling completely changes, as other sections of
C's ditch demonstrated to perfection. Thus, it is
evident that D' s ditch was dug about the time when C
had almost ceased to weather markedly, but before
growth had become established in its ditch. In this
case, a time-limit distinctly less than a decade seems
demanded ; on the other hand, it cannot have been
less than six years. And so, since C goes with A and
D with B, there is substantial agreement between the
time-limits, suggested by different conditions, in the
two groups. If, in either case, the interval between
the two occupations is suggested to be about a decade,
the error one way or the other will not be a large one.
Closer than this it is quite impossible to get, on present
evidence.
THE SECOND OCCUPATION
( I) ITS EXTENT
From the facts already considered, it is possible to
see what the Romans found on the spot when they
occupied the Cawthorn site for a second time. The
defences of C must have been in excellent order.
The rampart somewhat weather-beaten, but well
consolidated ; the ditch about half-full with silt, but
not yet obscured by growth ; the interior still exhibit-
ing in good order the turf mounds and pits of the
previous occupation. It would not have been impos-
sible to refurbish such an entrenchment, and use it
again, after a very few hours. A, on the other hand,
would have required a far greater time for its recon-
struction. A much greater mass of material was
filling its ditch, upon which growth had already started,
- J
and its rampart would have needed a drastic recon-
struction. Without a complete rebuilding, involving a
great expenditure of timber, it would have been
impossible to restore this camp to its original form.
On the other hand, it was not difficult to refur-
bish A so that it might form a bivouac. The ditch
was quite deep enough for this, even in half-filled
condition, and the rampart could be made to serve
by constructing a new crest. And this was, in fact,
what was done. For the new force was too large to
occupy C without extending it, and the position of C
was not well suited to an extension, especially if work
were planned upon the D-plateau. They therefore
re-occupied A, but A also was too small for their needs,
and was enlarged (folding-plan, PI. xx) by demolish-
ing the eastern side, and extending towards the east
until the camp became a large bivouac, twelve acres
in area. On the Scottish scale, which seems to allow
thirty acres to the marching legion, this would be big
enough to take four cohorts.
After this force had thus dug itself in, it began to
turn its attention to the opposite D-plateau, just as
the occupants of C had turned their attention to the
-plateau upon a former occasion. And they built
there the earthwork D, connecting themselves with it
by a lightly-metalled road (see Fig. 3). This earth-
work, like A, has all the features of a permanent fort
of small size
1
(3 J acres). As has been seen (PL ix B),
its south-east angle impinges upon C, and it is clearly
the earthwork which the occupants of came to
build. But they left it unfinished, just as A had been
left unfinished at the close of the previous occupation.
(2) EARTHWORK
(a) The Ditch. The west side, and one half of the
north and south sides, were already surrounded by the
half-filled ditch of A. This was made to serve without
alteration as the ditch of the new earthwork. The
eastern half of the enclosure was then surrounded by a
new ditch (Pis. i , A, xi ; Fig. 10) 8' 9" wide by 3' 9"
1
A distinction noted long ago by Young, History of Whitby, vol. ii,
p. 698.
deep, with a deep little channel, 9" by ' o", at the
bottom. This was a much more formidable obstacle
than the dimensions suggest, for it was very difficult
to get out of it unaided, and an incautious or unexpected
descent into it would certainly lead to a twisted or
broken ankle, if to nothing worse. It is, in fact, a
very skilfully designed obstacle, dug with the minimum
amount of labour and providing the maximum amount
of hindrance for its size. At the gates, which had both
external and internal claviculae, the ditch was inter-
rupted. It did not continue round the external
clavicula, but stopped at its inner end (see Pl. xi A) with
a neatly bevelled rectangular termination. It began
FI G. . N. DEFENCES OF
again in the same line, at the end of the inner clavicula,
but with a round bath-like end, better suited to drain-
ing the water from the gateway opening. At the angles
(see folding plan) it swung round in an easy, open
curve, parallel with the rampart.
The only points where peculiarities presented them-
selves (see Pl. i ; plan, Pl. xx) were the junctions
with the old north-east and south-east angles of A.
Here it would have been very difficult, from the
engineer officer's point of view, to obliterate entirely
the curved angles of A, and to continue its ditch to-
wards the east in exactly the same line as its north and
south fronts take. It was easier to recede slightly,
and to make a right-angled junction with the east side,
where the curve of the angles was just beginning.
In this way also the rampart could be carried more
easily, and with greater likelihood of stability, across
A' s filled-in ditch. The only disadvantage was that
this arrangement of the junction created a slight
re-entrant in the line of the rampart; but this did
not matter on the north side, where there was very
little room between the camp and the edge of the
escarpment ; and on the south side, as will be seen,
the fault was compensated by the provision of a
ballistarium. The ballistarium, however, was also
needed in order to meet a further difficulty. At
the southern junction it would have been unwise to
link the old and the new ditch-systems, since the point
of junction is a low one, and the gathering ground of a
considerable area. To have united the two ditch-
systems here, would have resulted in the presence of a
standing pool of water. The new ditch was therefore
finished in a squarely-bevelled end just short of the
side of the old one ; and there was thus left a gap which
had to be guarded by some special feature, like a
ballistarium.
(b) The Berm. There was no space between the
ditch and rampart.
(c) The Rampart. This followed everywhere the
same general type, a breastwork of turf, placed just
behind the ditch, in exactly the same manner as the
rampart of C. But in detail, its construction varied in
at least three ways.
In the first manner the rampart was continued
along the half-demolished rampart of A as a turf
capping (Fig. 5). The remains indicate that it was
not built to the usual height or width, but was
designed to give the older rampart a new, well-defined
crest. At the gates (Pl. vii), as will be described
in detail later, the new turf-work was made to form
new claviculae, the turf-work became a complete
rampart. At the north-west angle it was also being
laid on top of a wash-out in A' s old complete
rampart.
The normal form measured 8' 9" wide at the base.
It was composed entirely of turves, laid face down-
wards, and cut, in many cases, with feathered edges.
The turves were all very thick in live vegetation
when they were cut, and this abundance of organic
matter has left an extremely fine definition. The
thick vegetation faces had carbonised, and the acid
set up in the process had very effectively bleached
the earth and root matter forming the sod proper,
so as to render it extremely white, very sandy in
consistency, yet firm and ice-cold to the touch. These
observations cover the main bulk of the rampart. But
towards the outer edge, where much damp had had
opportunity to penetrate, the carbonisation had been
superseded by oxidisation, the black carbonised matter
being substituted by a very thin layer of oxideferric
oxide at Cawthorn, since the local soil bears iron.
This process was very marked indeed where the
rampart had sunk down over the ditch of A at the
north junction (see PL ii). The distinctive black
lines had entirely disappeared, and their place was
taken by thin red ones, not always so continuous or
definite. Two years later a wet position produced
similar results at High House,
1
on the Turf Wall that
runs behind Hadrian's Wall. But Cawthorn pro-
vided the first case where it was possible to note the
phenomenon in detail and to explain it adequately.
The third form which the rampart took was a
composite one (Fig. 10), of turves and ditch-upcast.
Two cheeks of turf were built up, on the back and front
of the rampart, and a base of turves on which it might
stand. The space between the two cheeks was then
filled in with upcast, rather like filling a faced wall
with grouting. This feature also was noted on the
Turf Wall, but in rather less definite form, in the section
cut at Appletree for the Hadrian's Wall Pilgrimage of
1930. At Cawthorn, so far as is known, it only occurs
at the north-east angle of B.
(d) Gates. There were five gates in all (Pl. vii),
two on the north and south sides, and one on the west
side. The east side had no gate, and it is therefore
clear that it represents the back of the camp. The
three front gates, porta praetoria (west), porta principalis
1
J.R.S. xxi, Pl. iii.
dextra (north) and porta principalis sinistra (south),
are built on the site of A' s gates : only the two rear-
ward gates, the portae quintana dextra (north) and
quintana sinistra (south) are new. They may be
examined first, since they give a clue to the type.
The southern quintana gate was completely ex-
cavated (Pl. xi). It was found to be very accurately
set out, with a radius of 21 ft. to the centre of
the rampart, the centre being the inner tip of
the opposite rampart for the inner clavicula, and the
outer tip for the outer clavicula. The tips of the
claviculae themselves seem to have been rounded
off, but weathering made it impossible to be quite sure
whether they had not originally approached a square
plan more closely. The ditch associated with the
inner clavicula ended in a round bath-like curve ;
that which ran up to the outer clavicula, without
following round it, had a square bevelled end. No
trace was found of any device to close the opening.
The road passing through it was planned obliquely,
but had been much damaged by rain-storms. It
did not extend beyond the limits of the gate, towards
the exterior. Traces of burning raised the question
whether the gate had at any time been assailed with
burning missiles. The burning is in isolated patches,
and when first recognised it suggested some such
explanation. But further examination revealed such
patches, varying in heaviness and extent all over the
camp, and never occurring in association with un-
questionably Roman wood-work. They are there-
fore better explained by an entirely modern cause,
the burning of sticks and brushwood when the site
was finally cleared of litter in 1923.
The porta quintana dextra was of the same type as
that just described, and it guarded the one possible
descent from the camp to the Sutherland Beck. In
order to place it better in relation to this break in
the escarpment the line of the north side was slightly
recessed at this point, and the inner clavicula formed a
much elongated shallow curve (see Pis. vii and xx).
The porta principalis dextra differed from all the
rest in having no outer clavicula, because of its prox-
imity to the edge of the escarpment, just as its fore-
runner, in A, had possessed no tutulus. The inner
clavicula was identified and found to be of the usual
radius (twenty-one feet), but it was not examined in
detail. Like all the others, it was built in solid turf.
The road through this gateway was well marked, and
it has already been noted that it crossed A' s half-
filled ditch by a new causeway (see Pl. vii).
The porta principalis sinistra was also examined
in detail. The remarkable feature here is the small
amount of trouble taken to disguise the presence of
an earlier gate. The butt-ends of A' s much narrower
opening were left projecting, and the new road-way
built across their ruined tips. A' s ditch was filled up
to carry the road and the outer clavicula, and the outer
clavicula was also laid out over A' s filled-in tutulus
(see Pl. iii A). In order to attach the outer clavicula
to the front of A' s ruined rampart, a series of three
steps was cut in the earthwork, into which the turf-
work was bonded. But this was the only feature in
the adaptation showing forethought. The ditch-filling
and the road-bottoming was not rammed, and it
presently sank, leaving the deep hollows that gave the
clue to the history of the gate.
The porta praetoria was damaged by the medieval
pack-horse track known as Roger Gate or Porter Gate,
which passed through the camp in an oblique direction
(see plan, Pl. xx). Both the tips of the claviculae
were eroded by the hollowing out and gradual widening
of the track. The chief feature was the very marked
return of A' s rampart under the internal clavicula,
marking the site of an earlier ascensus (see Pl. vii).
If the tutulus of A' s gate had been finished, the outer
clavicula would have overlapped i t ; actually, it only
ran along the edge of the narrow, filled-in trench.
At both these gates it is quite evident that the filling-
in of the earlier tutuli, in contradistinction to the filling-in
of the main ditch, took place after a short interval of
time. The deposit of silt at the bottom of each of them
was so small as to form the very thinnest line, just
sufficient to distinguish the filling from virgin soil.
This is important, since it hints that on the abandon-
To face page 56.
P L A T E XI .
Yorks. Arch. Journ.
B. S. QUI NTAN GATE OF ; LOOKI NG S.E. FROM I NTERI OR.
PLATE XI I .
To face page 57.
B. B-PERI OD TURF MOUND, WI TH TWO-FOOT SCALE.
I N THE NORTH RI DI NG OF YORKSHI RE 57
ment of A some intentional demolition or obliteration
took place.
(e) The Ballistaria. At six points in the retentura
of the back of the rampart was observed to be
thickened, by the addition of mounds of built turf.
So far as could be ascertained, these mounds (Pl. xii A)
were semicircular, with a diameter of eight feet.
Neither their form (Fig. I I ) nor their disposition (see
plan, Pl. xx) suggests that they were ascensus, which
FI G. I I . TYPI CAL BALLI STARI UM I N B, SECTI ON AND PLAN
this type of rampart does not require. They must
therefore be ballistaria, or tribunalia, which ancient
literature describes
1
as built of turf, designed to take a
small type of ballista. The distribution of the northern
set of three is of especial interest. They guard the
space between the north rampart of the camp and the
escarpment, and, at first sight, it looks as if they might
be useful in repelling a rush of attackers towards the
1
Ammi a nus Marcel l i nus xxi i i , 4, 5.
north gate. But the field which they have to cover is
so small that they would not be effective except in
dealing with a crowd; and a crowd large enough to
rush the camp by this means could attack with much
greater effect elsewhere. It may therefore be concluded
that the ballistae were placed here in order to practise
shooting into or across the valley below, rather than
to deal with any objective in the immediate foreground.
The other three ballistaria occur in the east and
south rampart. Two divide the eastern rampart into
three equal sectors, and the third is situated at the
south junction of the ditches of A and B. The last-
named is the one example which can be said to have a
sound defensive object. But the disposition as a whole
does not support the view that they belong to a care-
fully considered defensive scheme. The very facts
that no gate is provided at the back of the camp, and
that the scale of the defences is in no way commensurate
with a defence that demands the use of artillery,
suggest very strongly that the ballistae were there for
practice rather than for serious duty.
(J) Internal Structures. These were numerous,
but all, as will be seen, were of temporary character.
They consist of four classes, turf structures, ovens,
dug-outs and pits.
(i) Turf structures, including the tribunal. The
earliest notices of the site refer
1
to the presence inside
of prominent mounds, suggestive of buildings. These
are conspicuous to-day during the winter, not only in B,
but in parts of A, and in C. That they were composed
of turf was always evident in B, where rabbits have
greatly disturbed them and scattered the characteristic
grey soil; but most of them are too small and too
saturated wi th moisture to exhibit the regular car-
bonised lines indicative of built turves. Many were
examined before it was possible to secure a good
example, but eventually this (Pl. xii B) came to light
in the retentura of A. It was also impossible to make
an accurate and satisfactory plan of these mounds,
since most were worn down, especially at their butt-
1
See Drake, Eboracum, 1736, p. 36 : also Roy, Pl. xi.
IN THE NORTH RIDING OF YORKSHIRE 59
ends or angles, and, when the laid turf was not there
to serve as a guide, it was impossible to trace their
course by trenching. Nevertheless, a diagram (Fig. 12)
is appended of the best preserved mounds in B, in
order to show that these were once related to tents
or to rough buildings of the type that appear in not
FI G. 12. TURF MOUNDS I N
dissimilar stone mounds at Masada,
1
in Palestine
(Fig. 13.) Most of the Masada mounds also defy
precise interpretation owing to their scattered con-
dition, although it is quite evident that all once were
disposed in orderly and comprehensible fashion, visible
in some. But, while on the one hand it is clear that at
Masada buildings were involved, it is not clear that at
1
Hawkes, Antiquity, June, 1929, Fi g. 2, p. 207 ; Pis. v, vi , vi i i , ix.
Cawthorn these turf mounds ever formed the walls of
buildings. Their size and shape make this extremely
doubtful. In some places the lines are too isolated,
in others too close, to form part of a building-system.
Twice at least (see Fig. 12) they form a square or an
oblong too large to stand without the support of
wooden posts. Yet everywhere post-holes or sleeper-
100 200 ioo 400 S 00
1 n > i t ' ' ' 1
F E E T
FI G. 13. ROMAN CAMP AT MASADA, PALESTI NE
(Reproduced by permission f rom Antiquity, iii.)
trenches failed to appear in connexion with the turf-
mounds, although they were not absent, as will be
seen, in connexion with the ' dug-outs.' Thus it
seems clear that these lines of turf have to be con-
sidered much rather as screens against wind, or as
dams against wet, than as the walls of regular buildings,
and that they were arranged round tents or camp-fires,
or even inside tents, serving as benches or tables.
1
Even on this view they do not all now become coherent;
but the south-east group clearly marks two lines of
tents, back to back, with a pair of centurion's tents
next to the gate. Fortunately, there can be no question
of their Roman date. They occur only inside the
camps, and they are carefully orientated with them.
One was connected with a stone oven (see Fig. 12),
of indisputably Roman character ; others border the
Roman streets ; finally, one of them (Fig. 14) screened
a pit and supported a fire-back with which was con-
nected a characteristic group of Roman potsherds.
This important group is shown in Pl. , and it is
evident that the object of the oval pit is exactly
the same as that of the sunk stoke-holes of the
A' s ovens, namely, to avoid the wind on an exposed
site. The turf screen serves the same purpose, and,
in order that the fire may be laid up against it without
spreading by smouldering, it is faced with a little
stone hearth-back, as is done to-day in the Orkneys.
The pottery, which was the most important group that
the site produced, is described elsewhere.
The other structure of turf which calls for especial
notice, is the tribunal. It has long been known that a
prominent mound existed in the centre of A, at the
junction of the viae praetoria and principalis. This
FI G. 14. COOK-HOLE OF B-PERI OD
1
Cf . Caesar, B.C. iii, 95, recentibus caespitibus tabernacula constrata.
had been recognised as a tumulus by Bateman in
1860, and was then examined for an interment. But
the examination did not find the circular interment-
pit immediately, since the mound (Pl. xiii ; Fig. 15)
had been altered in shape by the addition of built
turves on its northern side, which disguised the position

Trineliw 1*23-2?

Trtrvc-St I92 -Z?
I
1.1_ y
of the true centre. This digging, and the fortuitous
work of rabbits, had done much to harm the mound,
which was left alone after a preliminary examination
by Mr. F. G. Simpson and Dr. J. L. Kirk in 1923,
until 1929, when Mr. F. G. Simpson completed the
work for the writer. Even then the presence of a
hive of wild bees somewhat delayed operations.
Eventually, there emerged a platform (Fig. 15),
12 by 22 feet in size, with two small expansions, like
the type of tribunal shown on Trajan' s Column.
1
There can be no doubt that this platform is in fact a
tribunal, for its position in relation to the B-period
street plan (see folding-plan) shows that it can be noth-
ing else. The turf-work also connects it with B, but
the conclusive proof of its connexion with as against
A is its relation to the general plan. It would not
fit with the praetorium of the -period, supposing that
there was any such erection inside A at all. On the
other hand, it fits nicely into the space that may be
presumed in front of the commander's quarters of B,
and borders upon the via principalis. It is also on the
left-hand side, where tradition
2
insists that it should
be : parte laeva tribunal statuitur, ut, augurio accepto,
insuper ascendat, et exercitum felici auspicio adloquatur.
(2) The ovens. The ovens of are illustrated as a
group in Fig. 16. It will be seen that all, with the
exception of nos. 4 and 5, were situated behind the
rampart and were arranged in groups of two. All were
rather carefully built in stone, with low bee-hive domes,
probably turf-covered, over the top. They show signs
of considerable but not lengthy use. All except the
nos. 7 and 8, at the north-east angle of A, have a pit in
front of them for use as a stoke-hole, but these pits
are not carefully constructed, like the stoke-holes of A.
No. 5 (Pl. xiii A) is situated among a group of turf
mounds, and, as falling between the viae quintana and
principalis, may be considered to form part of some
officers' quarter. This would explain their exceptional
1
Cichorius, scenes x, lxxvii, civ, Provincia Arabia, iii, p. 227, Fig. 1108,
cxxxvii. The tribunal at Masada has and Hawkes, Antiquity, 1929, pp.
a projecting ascent at the back (Fig. 206-7.
13), ; see von Domaszewski, Die
2
De mun. castr., 11.
Bl.
aea, r
? Metres
3 Ft.
&7 s.
To face page 64.
P L A T E XI I I .
Yorks. Arch. Journ.
A. STONE OVEN B5, AMONG TURF
MOUNDS I N B.
B. B'S tribunal, THE BURI A L -PI T OF THE TUMULUS I S I N THE FOREGROUND,
THE LAYERS OF THE TURF-BUI L T tribunal APPEAR I N SECTI ON.
PLATE XI V.
To face page 65.
A. B-PERI OD DUG-OUT 2. THE OLD SURFACE LI NE BELOW A'S DEMOLI SHED
RAMPART APPEARS BELOW THE POST-HOLES. THE I NCI SI ON TO THE LEFT OF
THE FURTHEST POST-HOLE IS A'S PALI SADE-TRENCH.
Yorks. Arch. Journ.
. N. I NNER DI TCHES OF D, SHOWI NG MI D-RI B.
position. Finally, two pairs of ovens, nos. 3 and 4
(Fig. 12) and 7 and 8 (Fig. 16), are set in different
positions, at right angles to one another. The ex-
planation suggested by Mr. F. G. Simpson for a similar
pair at the Goldsborough signal-station
1
may be
adopted here : on an exposed site high wind might
render one or the other unworkable, but hardly both at
once. It will be noted that no oven was found in the
retentura, so that it is likely that at least another pair is
to be postulated. No. 9-10 is of exceptional interest
(see Pl. xvi A). It started life with a southward-facing
flue, giving on to a large rectangular pit of the ' dug-
out ' class (see Fig. 17). This was an awkward arrange-
ment, since it must have meant that the ' dug-out'
was filled with a cloud of ashes every time the oven was
used, quite apart from the danger to its temporary
roof. After some little use the oven was therefore
rebuilt, and made to face east. This does not imply
that the group was used for a long time, but it demon-
strates that the occupation was long enough for some
discomforts to be noted and rectified.
(3) ' Dug-outs.' Five of these pits were identified
(Fig. 17), in the area of the praetorium. All were
of the same general type, an oblong pit with ridge-pole
set above it on the long axis, and held by a post-hole
at each end, while in the two back angles were cut
slots for holding the struts of a wigwam roof. Owing
to the angle of these strut-holes or slots, the type
can be restored without difficulty (Fig. 17, 1).
No. 2 (Pl. xiv A) had a sloping back-end, with a
post-hole at the base of the slope.
Nos. 3 and 4, which were rather smaller in size,
had an additional feature (Pl. xv). The front end was
continued forward for half its width, in order to form a
convenient entrance to the narrow space, and the
front post for the ridge-pole was situated at the junc-
tion between the pit and the rudimentary passage
thus formed. Only small pits required this feature.
1
Personal information from Mr. appear in a subsequent number of this
F. G. Simpson. The report will Journal.
10
FEI
^MUSXee-TiUMCH
jgg{ ~ -S F
RAAMART OF l Uf A ^
OLD "SUttFAcT LINE

1
FI G. 17. " DUG-OUTS " OF B-PERI OD
A. B-PERI OD DUG-OUT 3. THE THI N WHI TE LI NE
LEVEL WI TH THE TOP OF THE POST-HOLE IS THE OLD
SURFACE LI NE BELOW A'S DEMOLI SHED RAMPART.
A. B-PERI OD DUG-OUT 5, WI TH SUPERI MPOSED OVENS
9 AND 10, TO NORTH. A METRE ROD IS I N THE PI T.
No. 5 (Pl. xvi A) associated with oven no. 9-10,
possessed no post-hole or slot.
No. 1 (Pl. xvi B) had very finely developed post-
holes and slots, and was the largest and most solid of all
the ' dug-outs.' It was situated not far away from the
tribunal, and, owing to its specially solid construction,
it seems most likely to have been the ' dug-out' of
the officer commanding the little army. Close to the
entrance there was a hearth, near which was found
part of a red carinated bowl, and further fragments of
the same occurred in the bottom of the pit. This is
important, since it proves the Roman date of the pit.
The fact that these pits belong to the second occupa-
tion is proved by the position (see Fig. 17 and plan,
Pl. xx) of nos. 2, 3 and 4, which are cut in the demolished
rampart of A, and destroy its palisade-trench.
1
Also,
No. 5, as has been seen, is associated with one of
the distinctive stone ovens belonging to B. It is note-
worthy that all these pits are situated in the praetorian
area. For it clearly implies that they are officers'
' dug-outs.' Both the praetentura and the retentura
were well searched for similar features, which were
well-marked depressions on the surface: and it
can be stated with some confidence that they do not
occur outside the zone described. Further, we feel
equally sure that there are no more pits of the kind
within the praetorium area. Thus, it follows that there
were in camp five officials of sufficient importance to
have ' dug-outs.' Three of them were grouped to-
gether, two had separate quarters, of which one must
be connected with the praetorium itself. The signifi-
cance of this will be discussed later.
We know of no precise parallel to these remarkable
pits. Rectangular storage pits are not uncommon
upon Roman sites, as at Newstead,
2
Slack
3
and
Old Kilpatrick,
4
but they are not provided with wig-
wam roofs, nor do they occur in association with
1
Thi s might seem to invite an
inrush of water, whi ch most of the
occupants of the plateau were trying
to avoid. But it will be remem-
bered that this dug-out was cut in a
high-standing piece of the mound
formed by the demolished rampart
of A.
2
Newstead, pp. 117-18, pits vi , ix.
3
Slack, Y.A.J., xxvi , p. 23-25.
4
Old Kilpatrick, pp. 18, 19.
occupations so ephemeral as this. The Cawthorn pits
are more like the shallow and larger pre-Agricolan
pits, used as sleeping-quarters, between the viae
principalis and quintana at Margidunum.
1
It is also
clear that they are connected with a military occupa-
tion, as their precise alinement and the association
of no. 5 with an oven proves. There can thus be no
doubt that we have here a unique class of dwelling-
pit, a Roman military ' dug-out,' or half-underground
shelter, built to serve as officers' sleeping-quarters on
an exposed site. The idea of luxury-quarters is not
strange in a Roman army occupying a site for some
time, and it is illustrated several times in Roman
literature.
2
In actuality, it occurs at Numantia
3
and at Masada.
4
But the form which it takes at
Cawthorn is unique, and may therefore be welcomed
as particularly interesting.
(4) Pits. During the search for dug-outs and ovens
many pits were identified, but few were of interest,
and fewer still contained objects. The pottery from
the cook-hole shelf and a fragment of glass from
no. B2 are the notable exception. Most of them
appeared to be storage-pits, or pits for the collection
of water. No. B4 was linked up with a drainage chan-
nel. No. B6 was dug so close in behind B' s rampart
that it may be suspected to have served as a latrine.
No. 1 was supplied with a step.
No. B5 merits a more detailed description. It was
oblong (Pl. xvii A and Fig. 18), five feet wide, nine feet
long and five feet deep. Its long sides were not ver-
tical, but sloped sharply down for one foot, leaving a
trench at the base nine feet long by three feet wide
and four feet deep. Opposite the sides of the narrow
portion, at each short end of the pit, were two post-
holes. The purpose of this pit can hardly be in doubt.
It must certainly be a latrine-pit,
5
and it is noteworthy
1
Margidunum (Nottingham Art
5
Latrine-pits are not, so far as we
Museum publication) p 19, Pis. i, vii, know, mentioned in Roman litera-
xxii. t ure: but the ordinary latrine, as
2
Caesar, B.C., iii, 96, Li vy, xii, 2. existing at Housesteads, Ostia or the
Vita Hadriani, 10. Palatine, is a simple oblong pit, wi th a
3
Numantia, i, p. 369; i v, Taf . 6, xxv. drain in the base of it, and seating
4
Hawkes, Antiquity, June, 1929, ranged above it. It probably derives
p. 207. from the temporary form.
A. B-PERI OD L A TRI NE-PI T : THE CANES MARK
THE SI TE OF POST-HOLES : A METRE ROD IS I N
THE PI T.
PLATE XVI I I .
To face page 69.
AI R-PHOTOGRAPH OF THE J UNCTI ON OF C AND D.
that the type corresponds closely to that dug by the
modern army in the field. It was well filled with
turf, not silted but packed; thus it would seem that
the pit had been systematically filled up before being
abandoned. This also would conform to the best
practice. The size of the trench, however, and its
position in the praetorian area would suggest that it
was not meant to accommodate more than a few
officers : and although search was made for the men's
trenches, in the hope of locating rubbish therein, it
did not meet with success.
Pll|Hilll|lim
J
Kim} IlllltlU^
u
FI G. 1 8. L A TRI NE-PI T, B-PERI OD
(g) Streets. The viae praetoria, principalis, quintana
and the intravallum were clear. The lines of other streets
are less well defined, but at least two are clear. The
effect of this upon the planning of the camp (see folding-
plan) is to make its main lines certain, but its details
highly uncertain. At the south porta principalis the
street was laid upon a foundation of laid turf, giving a
camber.
(3) EARTHWORK D
The westernmost camp (see plan, Pl. xxi, Pl. xviii),
which goes by the name of D, is the most distinctive
and best preserved of the four. In type, it very closely
resembles the Roman camp (Pl. xix) on Hod Hill,
1
Dorset, being protected by two sets of ditches,
separated by a wide platform or ravelin and backed by a
strong rampart. Its size is just over three and a half
acres. The proportions are not tertiate, and there
is no north gate. A well-marked mound crosses the
earthwork on the site of the via principalis, and it
becomes clear that the south gate is the porta praetoria,
while the east and west gates are the portae principales
sinistra and dextra respectively.
2
Neither the gates
nor the rampart were examined in detail, but the
defences were cut at three points by sections (Fig. 19),
and the following facts were revealed.
The rampart is composed of turf, and is twenty feet
wide at the base. But the turves do not seem to have
been laid, since they do not exhibit the regular lines,
either in perfect or debased form. There is no berm.
The ditch system is divided by a platform, thirty-one
feet wide, into two sets, inner and outer, linked by
ditches returning sharply at right-angles on each side
of the gateway openings. At the edge of the escarp-
ment, the platform is reduced to twenty-seven feet
in width, in order to make the outer ditch larger and
wider.
The inner ditch system differs markedly in charac-
ter. On the north side, section A produced a double
ditch (Pl. xiv B), sixteen feet wide, each ditch being
seven feet wide by three and a half feet deep. Section
produced one central lock-spit. Section C produced
only a shallow depression. On the east side, section D
produced a similar shallow depression, with a little
heap of upcast therefrom. On the west side, section F
gave the same result as section D. But a further
section on the east side, E, gave the same result as
1
Crawford and Keil l er, Wessex the view that right and left were
from the Air, Pl. i. reckoned as if looking out f rom the
praetorium, whi ch seems most prob-
* Thi s depends upon accepting able.
To face page 70.
P L A T E XI X.
R i v e r Stour
l l
anford
l l
\
/
\
\
We s t G a t e
= G a t e
I - v ; 7}y - v; Vi v jy; i.vi.i;;;
P l o u g h e d i n
y # V , i d 1 9 ^ C e n t u r y i
fbepressions'
Leigh Gate
'' H o me
' % % G a t e
'--,'//////", >,
Ashfield Gat
Slight traces of
cultivation
IjllllWS-
Steepleton Gate
zoo 600 /ooo
2000 FEET
PLAN OF THE PREHI STORI C AND ROMAN EARTHWORKS ON HOD HI LL, DORSET, FROM AN AI R-PHOTOGRAPH.
(Reproduced by permi ssi on f rom O. G. S. Crawf ord and A. ICeiller, Wessex from the Air.)
section A. It thus becomes evident that the scheme
was carried forward much further in some sections
than in others.
The platform exhibits the same lack of uniformity.
Section D produced only two mounds of upcast :
section produced a regular well-formed mound,
level from back to front : section F gave the same
result ; while the northern half of the west side exhibits
the plainest traces (see Pl. xviii) of heaps of earth which
have not been levelled or disposed.
The outer ditch was traced on the north, east and
west sides. Section A produced a fine rock-cut
Punic ditch,
1
with vertical inner side, making the
escarpment even more difficult to climb. Sections
D and F contained shallow depressions. Section
yielded a finished Punic ditch, with vertical outer side
(Pl. xvii B) . The difference between the two completed
sections is to be explained by their situation, since
reversing the normal arrangement of the ditch makes
it much more formidable and much less likely to
collapse on the edge of an escarpment. But the
depressions of sections D and F are obvious examples
of unfinished work.
It thus becomes clear that the defensive system
of D was never finished. Its ditches remained incom-
plete, the platform between them was only perfected
at one point, and, elsewhere, the scattered heaps show
that the work was not even tidied before it was aban-
doned. This raises the question whether even the
double ditch is complete as we now see it. If it was
designed as two contiguous ditches, then the ditches
were dug in an uncommon way, and could have been
rendered twice as effective by digging them deeper at
the bottom, in the regulation manner that appears in
C and B. On the other hand, it is possible, as Mr. R. G.
Collingwood suggested to me, that the intention was
not to dig two small ditches, but one large one, and that
the two small excavations which are now visible
represent the half-way stage in digging a great ditch,
working from each side and finally working out the
central rib left. This would explain why the work
started in some places as a shallow excavation, without
1
It may be noted that De mun. castr.
(c. 49) does not say whi ch side should
be verti cal ; but, since it is treating
of an ordinary camp ditch, we may
assume the outer side, where pressure
from the rampart woul d not cause the
vertical side to cave in.
any sign of the central rib, which would have to be
preserved, if two ditches were designed, from the very
first. Finally, one great ditch rather than two small
ones is demanded by the device of the ravelin, or
defensive platform, now to be examined.
It has been suggested that such pl atforms
1
were,
in fact, designed as a first defence, from which the
defenders would retire only upon being hard pressed.
But the fatal weakness of this view is the fact that no
provision is made for retirement : it is more difficult
for defenders to leave the platform than for the
enemy to reach it, since the enemy land upon it by
leaping one ditch, while the defenders have either to
leap the side ditches in confusion and crowd back
through the gateway openings, or cross the rear
ditch and scale their own rampart. In short, this
theory placed the defenders in a weaker position than
their enemies, and therefore cannot be the right one.
A closer consideration of the outer ditch suggests a
better explanation. This is a Punic ditch (Pl. xvii B),
which differs from the normal fossa fastigata in having
one vertical side ; and the effect of this, in jumping,
is that while it is easy to land safely from the vertical
side on to the sloping side, it is extremely difficult to
reverse the action. The tendency is to miscalculate
the distance owing to the difference in angle of the two
sides. The effect of this upon the use of the platform
is as follows: on the north side, where the platform
reaches to the edge of the escarpment, and where
attackers are not invited, the Punic outer ditch is
placed with vertical face warding off approach : on
all the other sides it is placed with vertical face unseen
by the attacker, who is tempted to undertake the easy
j ump on to the platform. Once there, not only does
he find retreat difficult, but he has come into full
range from the defenders on the rampart ; thus he has
either to come forward, and face an impregnable
defence, or to return and expose himself as a living
target while trying to cross the outer ditch. In other
1
Collingwood, Archaeology of at Ardoch and elsewhere, that they
Roman Britain, p. 46, suggests at last prevented easy escape, as the Punic
an alternative theory for the ravelins ditch was perhaps meant to do here.
words, the platform is not an advanced line, but a
trap designed to bring the enemy into the worst possible
position.
This explanation seems to meet the facts best.
But it demands that the ultimate defence of rampart
and ditch shall be as strong as possible, since the enemy
trapped on the platform is likely to become desperate,
and must therefore be met by a very difficult obstacle.
This would be supplied by the twenty-foot rampart,
and either a sixteen-foot single ditch or two really deep
seven-foot ditches ; but the ditches in their present
state are inadequate, and if it must be conjectured
how they were to have been finished, the writer inclines
to choose the sixteen-foot single ditch as the more
l i kel y.
1
It is, however, clear that the upcast from these
ditches would in itself be insufficient to complete the
platform. Additional earth was therefore derived
from pits outside the earthwork. Fi ve of these were
located on the west side (see plan, Pl. xxi), and it may
be that some of the pits in the ditch of C, which are
situated very close to the roadway between and D,
were also dug for the same purpose.
2
Inside the earthwork no trenching was done,
excepting one cross-trench in the southern, unploughed
half. This revealed no trace of buildings, wooden or
stone, and no sign of occupation. The north rampart-
section produced a single scrap of mortarium-rim,
discussed below. Otherwise, our experience entirely
confirmed that of Sir Nathan Bodington and Mr.
S. D. Kitson in 1908,
3
who found the interior barren
when they trenched it.
(4) THE OCCUPATION OF AND D
It has now become clear that the occupation of
and D presents a set of circumstances strikingly
similar to those of the occupation of A and C. The
1
It should be noted, however, that alternative solutions to a rather
the north ditch of the fort at High similar problem.
Rochester (Brememum) seems to , s however, p. 40 for an
present both types, a large single ,t , _, ; e w
ditch east of the north gate, and a
a l t e ma t i v e v i e w
"
double one west of the gate, as forming
3
See Y.A.J, xxviii, pp. 29-30.
only difference is that on this occasion the force
employed was bigger, and that we know much more
about than opportunity permitted to learn about C.
Further, there is no question which came first, since D
clearly took some time to build and yet was never
used for a dwelling. A force of some size arrived at
Cawthorn, re-occupied and enlarged A, making of it
the entrenchment which we call B, and lived there for
some time, long enough for officers to demand ' dug-
outs,' and for an adjustment to be made to one of the
ovens. As a whole, the army lived in its leather
tents, but the exposed nature of the site and (as our
own experience would suggest) bad weather, led it to
screen its tents and hearths from wet and wind with
mounds of turf.
The nature of the troops that arrived is also clear.
They were legionaries, since they brought with them
ballistae. Their numbers cannot have been very large.
One or two cohorts have been suggested for C : here
there is just over twice as much accommodation,
suggestive of from three to four cohorts. This agrees
with the ' dug-out ' accommodation. The three ' dug-
outs ' situated together would belong to the three
tribunesthe only officials important enough to require
quarters of this kind ; while the larger isolated ' dug-
out,' in the praetorium, would belong to the praefectus
castrorum, whose special duty it would be to accom-
pany troops on a constructional expedition
1
such as
this. The inferior dug-out, no. 5, not far away, may
have belonged to one of his special subordinates. This,
at least, is the natural arrangement, corresponding to
the officers who would accompany such an army.
Finally, it must be emphasised that, as in the case
of the former occupation, the army came in peace.
The defences of D may be warlike, and would have
been formidable had they been compl eted; but they
could not contain the army occupying B, which was
building them. And the rampart and ditch of is of
the simplest type, quite unsuited to stern needs, but
perfectly adequate for the ordinary encampment of a
1
Cf. Tac. Annals, i, 20, where the gineering works, was turned upon by
praefectus castrorum, supervising en- his troops, near Nauportus.
force not in danger. The disposition of its ballistaria
tells the same story, for they are not arranged so as to
defend the camp best, but so as to shoot into a field
of fire that could interfere with nobody. Thus, while
the front of the camp is directed towards D, and
connected with it by a road, the back is used for
artillery practice with the ballista.
(5) THE DATE OF THE OCCUPATION OF AND D
A clue to the date of B' s occupation is given by the
pottery. The dating depends upon the mortarium and
the carinated bowls (Fig. 20), though the other types,
tiny jars, dish, and mortarium-like bowl, are such
as fall between A.D. 80 and 120. The first two types,
however, take us a little closer. Comparison with
the pottery from the Fort at Malton will show that
neither the mortar nor the bowls have the sharp
outline or the hard fabric which distinguish Agricolan
ware ; on the other hand, they have not reached
m

FI G. 20. POTTERY-SECTI ONS FROM B. (| )
To face page 76.
P L A T E XX.
CAWTHORN : CAMPS A AND B.
Royal Archaeological Institute.
the simplified types current under Hadrian, as exem-
plified by the pottery of Hadrian' s Wall. Wi th this
opinion Messrs. R. G. Collingwood, F. G. Simpson
and . B. Birley concur. The date indicated falls
not much more than ten years on either side of the
turn of the first century, during the period of quiet
which followed Agricola' s conquest and preceded the
troubles at the end of Trajan' s reign. Thus it would
not involve great error to choose A.D. IOO as the mean
date of the second occupation of Cawthorn. The first
occupation will precede it by about ten years, and will
also fall in that time of quiet.
This dating is of some importance, since it entirely
confirms the reading of the evidence concerning the
purpose of the site. This period is one during which
we should expect the legion at York to be indulging
in manoeuvres without fear of interference, and,coming
on to the wild moors to do them, just as our own army
from Catterick and Newcastle invades the uplands of
the Pennine or Cheviot, or as Hadrian' s African ar my
1
went out into the desert. Further, the fact that the
period was peaceful accentuates all the difficulties of
accepting the alternative view, that two attempts
were made to throw a permanent garrison into
Cawthorn, and that both f ai l ed; or, again, that the
praefectus castrorum of York was so stupid as to
attempt to occupy twice a site that he had once learnt
to be waterless. In short, both the date and character
of the camps fit the manoeuvre-theory better than
any other. Al l other views introduce unavoidable
contradictions.
SUMMARY OF RESULTS
The results of our inquiry may, therefore, be
tabulated as follows.
(1) Cawthorn was occupied twice.
(2) The occupations were separated by an interval
1
C.I.L., viii, 2532. Also Appian , rots , re
Iberica, 86, ^ :
, , & Cf. Corbulo' s action in Syria, Tac.
\\ / re /,... Ann. xiii, 35 ; and Vegetius, i, 25 on
rots , training recruits.
estimated at six to ten years : the mean date of the
second occupation is A.D. IOO.
(3) On each occasion the force lived in one camp
and built another ; thus, the occupants of C built A,
while the occupants of built D. This is proved by
the character and lay-out of the works.
(4) On the first occasion, A was systematically
demolished when abandoned, though not yet com-
pletely finished. D, on the other hand, was abandoned
incomplete, without demolition.
(5) The forces employed on both occasions were
legionaries, since they possessed ballistae, weapons
issued at this period to no auxiliary troop.
(6) The character of the entrenchments in which
these forces lived tells us that they expected no danger,
and that they were entirely occupied in drill and in
building A or D.
(7) The fact that the site chosen was unsuitable
for permanent occupation, coupled with the abandon-
ment of A and D when unfinished, leads us to conclude
that the operations were manoeuvres, and not serious
warfare.
(8) Granting conclusions 5 and 7, the force involved
will be the Ninth Legion from York. Thus, by a
completely different line of argument, we reach the
conclusion drawn by General Roy.
1
(9) Numbers are hard to calculate. But we venture
to estimate the first force as one or two cohorts, the
second as three.
(10) The site has no connexion with the permanent
occupation of Yorkshire, and therefore need not be
connected with the Roman road usually associated
with it.
(11) These camps are unique, not only in Britain,
but in the Roman world as at present known. They
are at present in excellent hands, but their future
should be watched with attention by a nation that cares
for its antiquities.
1
Roy , Military Antiquities, p. 65, pl. xi.
-^sasmnir,,-
R-OAD
To face page 78.
PLATE XXI.
SOO FEET
ISO METRES
CAWTHORN : CAMPS C AND D.
Royal Archaeological Inslitu

Richmond, 4 Roman Camps at Cawthorn

Uploaded by

Copyright:

Available Formats

Richmond, 4 Roman Camps at Cawthorn

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Richmond, 4 Roman Camps at Cawthorn

Uploaded by

Copyright:

Available Formats

THE FOUR ROMAN CAMPS AT CAWTHORN, IN THE

You might also like