Excavations on the White Cliffs Experience site, Dover, 1988-91
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE
SITE, DOVER, 1988-91 *
DAVID R.P. WILKINSON
With contributions by Leigh Allen, Anthony J. Barham, Martin R.
Bates, Paul Booth, Cathy E. King, and Catherine Underwood-Keevill1
GENERAL INTRODUCTION
This paper presents the results from a number of small excavations
carried out by the Oxford Archaeological Unit on the site of the White
Cliffs Experience, Dover. All excavation and post-excavation work was
funded by Dover District Council.
During the 1970s and 80s a number of different developments were
proposed for this archaeologically-important town centre site,
prompting the rescue excavations discussed below, but it was not until
1988 that a project - the construction of the White Cliffs Experience2 -
was finally agreed. In September of that year the developers, Dover
District Council, asked the Oxford Archaeological Unit (O.A.U.) to act
as archaeologists to the scheme and set up the Dover Archaeological
Advisory Board to monitor both this project and the general
archaeological situation in the town.3
• This paper has been published with the aid of a grant by the Dover District Council.
1 All contributors from Oxford Archaeological Unit except for C. Underwood-Keevill,
5, St. Peter's Rd, Didcot, Oxon, OXll SEP; Dr C.E. King, Heberden Coin Room,
Ashmolean Museum, Beaumont St., Oxford; Dr A.J. Barham and Dr M.R. Bates,
Geoarchaeological Service Facility, Institute of Archaeology, University College,
London, 31-34 Gordon Square, London, WCIH 0PY.
2 The White Cliffs Experience heritage centre, and the adjoining Dover Museum, were
opened on 1st May, 1991.
3 The Dover Archaeological Advisory Board consists of Professor B. Cunliffe
(Chairman), Professor D. Harris, Professor P. Rahtz and Professor P. Salway. Acting on
their advice, Dover District Council commissioned the O.A.U. to produce an
archaeological implications survey of the town (Wilkinson, ( 1990)) to facilitate the
integration of archaeology into the planning process.
51
DAVID R.P. WILKINSON
A total of ten evaluation trenches were dug in November 1988 and
February 1989 to assess the condition of the archaeology and to
pinpoint the position of some structures. Normally, only backfill from
previous excavations was removed but the results from Trench 1, where
a small amount of stratification was investigated, are published here.4
In particular, the information from the evaluation trenches allowed the
supporting piles for the building to be positioned so as to minimise
damage to the underlying archaeology.s
With the design of the building finalised, only a few small areas
remained where damage to the archaeology could not reasonably be
avoided and these were the subject of mitigation excavations carried out
either before or during construction work. All of these areas, Trenches
11-14 (Fig. 2), are reported on here. Finally, the design of the heritage
centre element of the scheme provided for three archaeological display
areas which were cleared of backfill by the O.A.U. These were (see Fig.
2 for locations) the Time-and-Tide Theatre undercroft ( Class is
Britannica fort internal buildings, Code: DTT), the North Crescent (west
end of St. Martin-le-Grand church) and the South Crescent (east gate of
Classis Britannia fort, Saxon Shore fort bastion, Code DSC). Only in the
latter case were any archaeological contexts removed, and these are
reported on together with the adjoining Trench 14. The excavation
archive, including all finds, will be deposited with Dover Museum.
THE SITE: TOPOGRAPHY AND GEOLOGY
The site, N.G.R. TR 31904137 (centred), lies in the town centre of
Dover, to the south-west and west of Market Square (Fig. 1). It is
bounded by Queen Street to the south, by York Street to the south-west
and is crossed by the 10 m. contour. The steep slope of the Western
Heights rises immediately west of the site to some 90 m. O.D., while to
the east is the flat, built-up Town Centre which overlies the infilled
estuary of the River Dour. A number of buildings have stood on the site
in the recent past, including housing, a school and a warehouse but
these were demolished in the 1970s. By 1988, much of the area was in
use as a car park with only a few buildings left standing in the east and
north of the site.
4 The evaluation phase of the work, and also the excavation of Trench 11, was directed
by Gill Hey of the O.A.U.
5 The intention throughout the project was to leave the archaeology preserved in situ
below the new building wherever possible. 167 piles, each 0.60 m. in diameter, were
placed across the site in positions agreed by consultation between the engineers and the
O.A.U. Damage to the archaeology from this process was kept well below 1 per cent.
52
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
Fig. 1. The White Cliffs Experience site, Dover, location map, showing extent of the
harbour/inlet in the Roman period.
53
DAVID R.P. WILKINSON
Geologically, on the west side of the site, the chalk bedrock is
overlaid by Coombe Rock which here takes the form of a buff, chalky
clay with flints. On the east side of the site, being closer to the bottom
of the river valley, the Coombe Rock lies below estuarine sands and
gravels, themselves in turn overlain by alluvial silts.6
ARCHAEOLOGICAL BACKGROUND
The White Cliffs Experience site lies in an archaeological area which is
unquestionably of international importance. The stratification in this
part of Dover varies from four to over six metres in depth and all major
periods from the Neolithic to the present day are represented. Two of
the best-known elements of Dover's archaeology - the Classis
Britannica ( Cl. Br.) and Saxon Shore forts - lie superimposed on the
site, which also encompasses a Roman military bath-house and the
remains of the Norman church of St. Martin-le-Grand.
Observations during building work from the late nineteenth century
onwards revealed the presence of substantial Roman remains on the
site,7 but the first specifically archaeological excavations, by the then
newly-formed Dover Archaeological Committee, followed the largescale
damage inflicted on the town during the Second World War. One
of the sites excavated during this campaign was at Fox's bakery, north
of Queen Street (Fig. 1), and consisted of Roman walls below medieval
deposits. The Roman walls have since been shown to be part of a
building within the Cl. Br. fort. Between 1951 (when the Excavation
Committee dug their last site) and 1970 very little archaeological work
was carried out in Dover, and only one excavation lay close to the area
relevant to this paper. s
6 A.J. Barham and M.R. Bates, The Holocene Prehistory and Palaeoenvironment of
the Dour Valley Catchment: a Geoarchaeological Assessment Report for Dover District
Council, Geoarchaeological Service Facility Technical Report 90/04, (1990), 21-25,
55-68, see especially Fig. 13.
7 A local man, Mr E.G.J. Amos, was a particularly keen observer. The evidence was
summarised by himself and R.E. Mortimer Wheeler in 1929, when the presence of a
Saxon Shore fort was correctly supposed: Arch. Journ., lxxxvi (1929), 47-58. Their Site
6, ibid., 49-50 was a piece of the Shore fort wall seen in 1915 and lay within the area
discussed in this paper.
s Arch. Cant., !xiv (1951), 133-4. For positions of these walls see P hilp 1981: 9, 46,
Fig. 16. For the work of the Dover Archaeological Committee, op. cit., in note 4,
130-149 and Arch. Cant., lxxi (1957), 14-37. For summaries of the work up to 1970, see
Arch. Journ., cxxvi (1969), 79-99 (S.E. Rigold's classic paper on the Roman harbour)
and P hilp, (1981), 7-12. In 1958, P.A. Rahtz observed part of St. Martin-le-Grand church,
and underlying Roman structures, on a site between the Market Square and eastern
boundary of the White Cliffs Experience, Arch. Cant., lxxii (1958), 111-37.
54
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DO VER, 1988-91
From 1970 to 1988, the Kent Archaeological Rescue Unit
(K.A.R.U.) executed a large programme of rescue excavations in
central Dover, and much of our background knowledge of the White
Cliffs Experience site is a direct result of their efforts; detailed
information is currently available only for the earlier Roman (Cl. Br.)
period, but an outline history for other periods can be pieced together.9
The earliest known use of the area is represented by ditches, pits and
flint scatters of the Neolithic period which were concentrated in the
northern part of the site, forming part of what the excavator has
described as 'an extensive farm' .10 An apparent gap in occupation was
then succeeded by an Iron Age settlement of c. 500 B.C. characterised
by wooden huts and storage pits.11
In the Roman period, a fort was laid out c. A.D. 115-20 but was
never completed; its north-east corner occupies the western half of the
site. A second fort, built c. A.D. 130-140 on a similar alignment, was
completed. It was occupied, though probably not continuously (see
below), until the early third century. The large quantity of tiles stamped
CLBR leave little room to doubt that this was a major base for the
Classis Britannica, the Roman fleet in British waters.12 North of the
fort, and still within the area of the site, was a substantial military bathhouse
built c. A.D. 140-60.13 The latter structure is immediately south
of the Roman building known as the 'Painted House' which lies just
outside the site boundary; this may have been part of a mansio. 14 To the
south of the fort, and just east of the southern roundabout on York
Street (Fig. 1), part of the wooden Roman waterfront was recently
discovered. IS
Following the abandonment of the Cl. Br. fort in the early third
century, soil accumulated over the remains of the fort buildings until
the construction of the Saxon Shore fort (A.D. 275 or later, see below).
The Shore fort, trapezoidal in shape, is both off set from and
considerably larger than its predecessor, enclosing an area of at least
9 A summary and discussion of Dover's archaeology is contained in Wilkinson,
(1990), 13-22.
10 Kent Arch. Rev., 28 (1972), 237; Wilkinson, (1990), 13.
11 Kent Arch. Rev., op. cit. in note 8; Philp, (1981), 80. The recent discovery of a
Bronze Age boat shows that in this period, too, the Dour estuary was of considerable
importance: K. Parfitt 'The discovery of the Bronze Age boat', Canterbury's Archaeology
1992-93, Canterbury Archaeological Trust, 1993, 15-18.
12 Philp (1981), is the definitive report on the Classis Britannica forts. Data on the fleet
are well summarised by H. Cleere, 'The Classis Britannica', in Maxfield, ( 1989), 18-22.
13 Philp, (1989), 275, Fig. 103.
14 lbid., 281-2.
15 K. Parfitt, 'The A20/Dover Sewers Project', Canterbury's Archaeology 1992-3,
Canterbury Archaeological Trust, (Canterbury 1993), 15.
55
Vt
0\
'
'
'
'
'
'
'
'
'\
N
t
0 25 m
l::::===--====-===
'
'\
'
'
Borehle 1 -$
'\_
Key
Saxon Shore fort
ffiffil Excavated/exposed 1988-91
fmWLineof
Classis Britannica fort Phase II
- Excavated/exposed 1988-91
Line of
lJ Saxon Shore fort terrace
\.=:•=•=====•======:=:==·:=·:·::;;·;;.;·: •••. .. ·.·.,=:=:::.:x.-.·;· ,k=Tr .1·2·
-as---
\slope
Fig. 2. Trench positions and major archaeological features; fort layouts after Philp (1981: Fig. 5).
t)
e
:;o
:-0
I
er.,
0
z
EXCAVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DO V ER, 1988-91
1.5 ha. The south and west walls of the Shore fort cross the White
Cliffs Experience site while the south-west corner lies just outside the
site boundary, below modern York Street (Fig. 2).
With regard to the early medieval period, much evidence, as might
be expected, has been recovered from inside the walls of the Shore fort,
including sunken-featured buildings, a multi-phased timber hall and
huts, and a multi-phase timber church with burials. There was evidently
occupation from at least the seventh century onwards, possibly
preceded by a break, and activity continued into the late Saxon and
Saxo-Norman periods. The timber church and hall can perhaps be
linked with the ancient Minster of St. Martin, founded in the seventh
century.16
For the Saxo-Norman and later medieval periods the archaeology of
the site is dominated by the remains of the stone church of St. Martinle-
Grand. Construction probably began in the late eleventh century and
the church was originally intended as the centrepiece of St. Martin's
Priory, but later became a parish church when a new priory was
founded outside the town in 1130. The ruined west end of the structure
was exposed in the 1970s by the Kent Archaeological Rescue Unit17
and is now on display in the grounds of the White Cliffs Experience
(Fig. 2).
DESCRIPTION AND DISCUSSION OF THE ARCHAEOLOGY
INTRODUCTION
The following section is not intended to be an exhaustive, context-by context
description of the archaeology. Instead, key contexts or groups
of contexts are discussed under a number of thematic headings.
Discussion of wider issues (e.g. the date of the Saxon Shore fort) is also
contained within this section. Numbers given without any prefix refer
to contexts while those prefixed by Number or No. refer to the finds
catalogue which has a single unique sequence for all classes of
material. Frequent reference is also made to building numbers (e.g.
Building 2 or B2) and these are the numbers allocated to structures
(including roads) by Philp in his publication of the Class is Britannica
16 For a more detailed discussion of the archaeological evidence, and its relationship to
the historic sources, see Wilkinson 1990: 17.
17 Kent Arch. Rev., 45 (1975), 119-20. In addition to the church, a lime kiln relating to
its construction has also been excavated and a possible monastic masonry building
(Wilkinson 1990: 22).
57
DAVID R.P. WILKINSON
fort.18 All references, unless otherwise stated, are to the completed and
occupied Cl. Br. II fort rather than its unfinished predecessor.
INTERIOR OF THE CL. BR. FORT
Trenches 11 and 14 provided evidence from the fort interior. Trench 14
lay just inside the east gate of the fort, near the south-west corner of
Building 23, a barrack block, and within the road B39 (Fig. 2). A thin,
hard-packed layer of water-worn pebbles (917) on a base of chalk (918)
was succeeded by 0.10 m. of clean silt (915) and then by a north-south
drain built of chalk blocks (914). The drain was overlain by a second,
metalled surface (913; Figs. 10, 11). The silt 915 contained the only
dating evidence, a single sherd of samian (c. A.D. 150-200). On
balance, the surface 913 probably belongs to Philp's Phase II of the
Cl. Br. occupation (c. A.O. 190-200 until c. 210). The overlying
stratification may have been truncated during the construction of the
later Saxon Shore fort (see below). Both drains and surfaces of the
types described above are typical of the open areas within the Cl. Br.
fort - Philp identified a complex sequence of metalled surfaces and
chalk-built drains running north-south along the same road, B38.19
Trench 11 was situated between the north end of the barrack block
B20 and the south wall of Building 32. The earliest situation revealed
by excavation is shown on Fig. 3: two short sections of rough, chalkblock
wall (362) bonded with brown clay, clearly represent the east
and north walls of B20, along with a clay surface (414) inside the
building and a surface of water-worn pebbles ( 417, 402) outside. The
latter was associated with two chalk-block gutters (421, 418) which
both apparently emptied into a third drain (423) of which only the cut
was seen. A patch of the metalled surface was repaired using chalk
fragments (420). The next development of the area (Fig. 4) has the
walls of Building 20 unchanged, more resurfacing (403) of the
metalled area, and a new gutter (370), which ran under the corner of
the building. The latter feature, perhaps intended to resolve drainage
problems within the building, probably emptied into a new
north-south drain (410), which itself joined a drain along the
east-west road B37.
The pottery recovered gives a general date of mid second century or
later for both of the phases described above; the size and nature of the
assemblages do not allow any further refinement. Thus, these situations
1s Philp (1981), Fig. 5.
19 Philp (1981), 84-6, Fig. 16.
58
Ul
\0
0
Building 20
/:/oaa
838
-·-
.-·
.-
4 m.
cG-
Q;j
b
Q:'
Fig. 3. Trench 11, earliest excavated phase.
-·-
I
I
/
_____ .---•-i
I
I
I
I
I
I
I
I
/■ opus signinum
I
I
I [Z] clay
I
I
■ I burnt clay
I
I
metalling
,,,,
0
z
V,
@
I
i
trl
V,
t:1
0
<
-
00
00
0\
0
\
\
o 4m
Building 20 ----
'":J
Aoact 838
I.
'
\
;a-,
..,''7.:-'
' J
. ,, .......... ,
'--·-·-·-·-·-·-·-·-L: 402 ,_:.---·-·
-- . ...,. ·-r .---· - . - .---i
I
I
I
/
/
/
I
I
I
I
I
I
I
I
I
I
I
I
I
Fig. 4. Trench 11, second phase of metalled surfaces and chalk-block drains, with Building 20.
s
:-ti
!'1:1
0\......
0 4 m.
-=-ea::-=-<=----====----====
------------ -
·
-•-i
I
.--- ---· I
-----· ---·
Building 20
. '
\ ti&333 :
--- ,
:--,/ 1395_, ,' ...........-1 -.,
"' 395
,,{-1r-.
,/ 389-, I ,
I /
I , '--,--f
:#/
1,'
\
EJ:38
\ /:i0€/c:t
\ 359
•a
\_ ____ - ·-·-·- -·-·---·- ----------
Qi
l:),
o"'
/
/
/
I
I
I
I
I
I
I
I
I
I
I
I
I
Fig. 5. Trench 11, third phase of metalled surfaces and chalk-block drains, with Building 20 extended to the north.
p
en
tI1
en
....
\0
CX)
CX)
DAVID R.P. WILKINSON
could relate to Philp's Period I (A.D. 130-40 until 154-55) or Period II
(A.D. 163-65 until 180); this aspect is discussed further below.
Fairly deep layers of silt (401) and rubble (400) overlay the external
and internal situations just described, and date to the second half of the
second century or later. Following this Building 20 was extended
northwards by the addition of wall 385 (Fig. 5). No return wall was
found, so the length of the extension is unknown. A floor of opus
signinum (397, 399) was laid inside the building, while externally,
more metalling (333) was put down together with a chalk-block drain
(350), which must have drained the newly-extended north end. The
dating evidence for this phase is very sparse, and only a general date of
mid to late second century or later can be assigned with confidence.
Finally, deep (up to 0.45 m.) rubble was dumped across the area,
succeeded by a further phase of chalk-block drains and metalled
surfaces in the area of Road B37. This latest phase (not illustrated) had
been heavily cut about by later graves and only fragments remained.
The fate of the building at this point is unclear - wall 385 did not
survive although wall 362 may have done so. In terms of dating, four of
the dump layers (334, 381, 389, 391) produced sherds of grog-tempered
ware for which a late third-century date is normally suggested (see
below, P. Booth, The Roman Pottery, Fabric R95, No. 23).
Relating this evidence from the interior of the Cl. Br. fort to the
1970s excavations is not without its problems. Four phases of
development have been described above, compared with the three main
phases proposed for the fort20 by Philp, but this (even without the later
pottery) would not be abnormal for an external area, and Philp notes
five successive road surfaces just to the south, as well as a complex
20 Philp sees the three phases as being periods of occupation, with abandonment in
between. By postulating that the garrison supported campaigns in the north when not in
Dover, he adds historical and archaeological evidence together to arrive at the phase dates of
A.D. 130-40 until 154-55; A.D. 163-65 until 180; A.D. 190-200 until 208; Philp (1981),
97-99. In reviewing the excavation report, D.J. Breeze rightly points out that a Cl. Br.
presence in the north is only proven in one case (Hadrian's Wall, A.D. 122-28), while other
campaigns quoted are very poorly documented; on archaeological evidence alone, the dates
of the three phases are A.D. 119 or later, A.D. 154 or later and A.D. 157 or later; Britannia,
xiv (1983), 372-5. The rebuilding events seem too widespread to represent damage caused
during occupation as Breeze ibid., 373-4, suggests, so that there is still a reasonable
argument for the garrison being absent and then returning. In this context, little bas been
made of the fact that the known rebuilding (except for one reconstructed gate-house) is
limited to the barrack blocks, which are built in chalk blocks bonded with clay. Gate-houses,
granaries and the fort wall are bonded in mortar, use more tufa and flint, and seem to survive
the entire life-span of the fort. It is much easier to imagine either a care-and-maintenance
staff coping with the latter structures, but not the flimsier barrack blocks, or even, that the
chalk and clay wall stubs of the barrack blocks supported timber-framed walls, which were
removed to be used elsewhere.
62
EXCAVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DO V ER, 1988-91
sequence of drains, all within the same B38 road as was investigated
here.21 The sheer quantity of drains and gutters, and their frequent reordering,
is striking, and shows the difficulties caused by the fort's
position at the bottom of the Western Heights slope. It is also the case
that chalk blocks, a relatively rarely used building material in the
Roman period,22 were not a very suitable choice for drain-building, as
the soluble chalk caused rapid silting. The main point of interest as
regards dating is the presence of late third-century grog-tempered ware
in contexts below the final phase of Road B37. The implications of this
are discussed below (see The End of the Cl. Br. Fort). Extension of the
building B20 was previously recorded during the 1970s excavations, as
well as extensions (always towards the fort wall end of the building) to
two other structures, all in Philp's Phase III. One of these latter, like
B20, was given an opus signinum floor. The extensions of these barrack
blocks can be interpreted either as being for officers previously
quartered elsewhere, or as an improvement of officers' quarters which
had formerly (and unusually) been of the same size and shape as the
men's accommodation.23
THE DITCH AND EX TRA-MURAL AREA OF THE FORT
Trench 13, which was situated immediately outside the north wall of the
Cl. Br. fort (Fig. 2), provided a small amount of information on the extramural
arrangements. A limited area of stratification at the northern edge
of the trench (Fig. 6) consisted of a loam layer (861) covered successively
by rammed chalk (860) with a surface of water-worn pebbles (857); a silt
layer (858) had formed over this surface. Layer 861, at the bottom of the
sequence, can be dated to the mid second century or later, while 858, at
the top, contained Antonine sarnian. All of these deposits were cut by 863
(Figs. 6, 7) which ran parallel to, and 4 m. north of, the Cl. Br. wall. The
cut, which sloped away southwards at c. 45° , should be the north edge of
the recut Cl. Br. fort ditch. This recut, which from the 1970s excavation
could only be shown to be post c. A.D. 70-100,24 can now be seen as post
c. A.D. 140-150.
21 Philp (1981), 71.
22 Britannia, ii (1971), 166-95.
23 Philp (1981), 110, prefers the first interpretation, with the officers quartered in
separate buildings along the Via Principalis. For a detailed argument of the second
interpretation, and its ramifications for the size of the garrison, see D.J. Breeze's review
of the excavation report, Britannia, xiv (1983), 372-5.
24 Philp (1981), 24-5; the evidence comes from just north of the work reported on here,
and is therefore likely to be the same recut.
63
°'
.j:s.
s
see figure 8 for key to conventions
9.59 m. O.D.
7'
section A
•::;,';'tl 1970 excavation
"!.£."•f- - . -· - . - . - . --
north wall of
CJ. Br. fort
section 8
w
......
. 855'
--/.-..;,,;,;,;;·_
grave .,/,/'
..
· -•,. ·. B3.s7 T·.;;., .:.-a- ..-
I
-".i. 8WQ---:::::::=1
- - T- - - .,.- - - # j\
- - - -
.- .- ·. / ;II 853 # · '
.,,...- ..... 1i51 '!'."." .. ss2-==- -----1
- .,., ..........-!' ,...:<.;,, .,,,.#_---
.
/ .,,,. $?':::'.::?'- - - # ;II I
_.,,,. ./ # - # ii! #
.
...,. __ _ ...... # #---
# # 856 #
. ;-, ·.;-y.:- 85 /\:.: ,.,.·"":,;;
I
9.59 m.
7' O.D.
858
857
cut 863 CJ. Br. fort ditch
section B
intersects
834 835 --1-- E
U:::n"'"'· 9.59 m. O.D.
Sh
a20
ore fort
1 0
graves
2 m.
wall
Fig. 6. Trench 13. Sections through rampart tips of the Saxon Shore fort rampart (for position of sections see Fig. 7).
s
::0
:-0
i
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
- Cl. Br. II fort wall
Saxon Shore fort wall
Concrete pile
841 grave
1970 excavation
0
854
856
.-----slop,
-----------
4 m.
-=-c::a::-=----====----======t
Fig. 7. Trench 13, plan, showing rampart tips against Saxon Shore fort wall.
65
l
DAVID R.P. WILKINSON
AN EXTRA-MURAL STRUCTURE NORTH OF THE CL. BR. FORT
This structure was identified in Trench 1, situated 35 m. east of the
fort's east gate (Fig. 2). A short (1.20 m.), north-south length of chalkblock
wall set in mortar was revealed (36). The footing of the wall was
off-set to the south by 0.20 m., giving a total width of at least 0.58 m.;
two courses of walling survived (Figs. 8, 9). The footing was covered
by a metalled surface formed of small water-worn pebbles, in turn
covered by a layer of silt (37) which had formed against the south side
of the wall, and a layer of rubble (38), probably wall-collapse. Both of
the latter two deposits contained many small fragments (total= 110) of
painted plaster. The few larger fragments preserved (i.e. c. 0.05 x
0.05 m.) showed a background of reddish-pink onto which was
imposed some small areas of dark reddish-black and some of dark red
(not illustrated). This pattern suggests that at least some of the plaster
was painted as imitation marble, a technique used commonly for both
the dado and middle zone of Roman painted walls.25 Munsell numbers
for the colours are as follows: 2.5YR 6/6 (reddish-pink), l 0R 2.5/1
(dark reddish-black), lOR 3/6 (dark red).
The structure described above cannot be independently dated, but
the use of chalk-block construction recalls structures within the Cl.
Br. fort, and the metalled surface is identical to that used for roads
and other external areas within the fort (see above: The Interior of the
Cl. Br. fort). More compellingly, the building almost certainly
precedes the construction of the Saxon Shore fort (c. A.D. 275 to
early years of fourth century, see below) and the accumulating silts
(37, 35) with fragments of plaster suggest at least considerable
dilapidation, if not abandonment, well before this date. On balance,
then, while no construction date can be established, the building
should have been in use during the final phase of the Cl. Br. fort and
for at least part of the intervening period, prior to the construction of
the Shore fort.
The context in which the plaster fragments were found, in a silt layer
accumulated overlying a metalled surface, strongly suggests that they
had fallen from the external face of the wall, as metalled surfaces of
this type seem, at least on the evidence of the Cl. Br. fort, to be used for
outside spaces. It is, of course, possible that the wall was fronted by a
portico which protected it from weathering to some extent, as has been
proposed for a painted courtyard wall from Verulamium.26 The building
discussed here may be the same as that found in 1982 by the Kent
25 N. Davey and R. Ling, Wall-Painting in Roman Britain, Britannia Monograph Series,
3, (London, 1982), 31-2.
26 Ibid., 66.
66
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
s TARMAC N
=
3.44 m.O.D. 7'
ot::::11-=-==-=--=.t::========2
.,_____ .i3 m.
tile
clay
m
sand
IDilll] silt
·-· ___ limit of
excavation
c; ,,
C.')
e:;
,)
chalk
flint
"oO 0 tufa 0 0 0 0
0 0 00
11 limestone
_____ edge of deposit
[ill] mortar
CJ. Br. II fort wall
[ill Saxon Shore fort
+ wall
pile
edge of cut or
solid feature
Fig. 8. Trench 1, west section (see Fig. 9 for position).
67
DAVID R .P. WILKINSON
Archaeological Rescue Unit,27 and the painted wall could have faced
out onto the main approach road to the east gate of the fort.
THE END OF THE CL. BR. FORT- ABANDONMENT AND CONTINUITY
Coin and pottery evidence from the 1970s excavations indicate that
the Cl. Br. fort was abandoned in the early third century.28 The rubble
from the buildings appears to occur directly over occupation in some
buildings, so that there may have been deliberate destruction of the
walls.29 A fine soil then accumulated over the rubble, and preceded
the construction of the Shore fort, so that there was clearly a substantial
time gap between the two events.30 Thus far, the story seems clear, but in
the area of Trench 11 just inside the west wall of the fort (Fig. 2) no trace
of an abandonment loam was found. It has also already been noted
that the road B37 was resurfaced, and a new chalk drain built, in the late
third century or later. This latter operation probably relates to the
Shore fort (see below) but it increases the plausibility that earlier road
surfaces continued in use following the abandonment of the Cl. Br. fort,
particularly if the road provided a useful thoroughfare. In this case,
the road may have led westwards to the north gate of the abandoned fort,
thus giving access northwards to the area of the 'Painted House'. To the
east, the road could either have followed round the east wall of the fort,
or gone through a break in the fort wall. This evidence, along with
possible continued use of an extra-mural building east of the fort (see
above) and the 'Painted House' itself are all reminders of the continuity
of life in the vicus following the abandonment of the fort. The possibility
that part of the road continued to be used during the life of the Saxon
Shore fort is discussed below (see The Shore Fort Walls ... ).
CONSTRUCTION OF THE SAXON SHORE FORT
Evidence relating directly to the Shore fort construction period was
recovered from Trenches 1 and 13. Trench 1 (Figs. 8, 9), from which
some of the deposits have already been described above, would have
27 Kent Arch. Rev., 71 (1983), 10-11. The structure is described as 'a substantial Roman
building, with plastered walls painted in many colours'.
2s A general early third-century date is preferred here to the precise A.D. 208 which
Philp gives, as the latter date supposes that the fleet had left Dover to support campaigns
by Septimius Severns in the north, and this is not proven; Philp (1981), 99; Britannia, xiv,
(1983), 373-4.
29 Philp (1981), 43, Fig. 25, S13.
30 Philp (1981), 94-6.
68
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
---
---
---
---
-
, ·
\
\ ,.,,,. . ..,,--·-. ___ . ----
\ ' (UNDISTURBED)
\ \ 26
\ \
\ \
N
\_.,,.,..---·-
.-·-
\ BACKFILL
oI::, :11--=i--=:11--=i--=:11•.i::
1
=======i
2
..., ____ ....ii
s m.
Fig. 9. Trench I, plan, showing wall (36) of Roman building with painted wall-plaster.
been 7 m. inside the east wall of the Saxon Shore fort, and close to the
projected line of the Shore fort terrace.
The destruction rubble (38) from wall 36 (see above) was covered by
a deep (up to 0.40 m.) layer of clay (34), with very few inclusions, and
sloping downwards from south to north (Fig. 8). Layer 34 contained
four coins, Nos. 11, 12, 14, 15, all dated to c. A.D. 260-280 and 27
sherds of pottery (e.g. No. 34) which indicate a date from the late third
century onwards. Some caution must be exercised here, given the very
small area of excavation, but it seems likely that layer 34 represents the
tail of the Saxon Shore fort rampart - a feature which is known to have
existed and which in other areas of the fort extended at least 7 m. back
from the fort wall.31 The clay which constitutes layer 34 could easily
31 See e.g. Philp (1989), Fig. 22, Section E.
69
DAVID R.P. WILKINSON
have been derived from excavation of the Shore fort terrace into the
natural clay with flints.
Further evidence of the rampart was found in Trench 13 (Figs. 6, 7),
just inside the western wall of the Shore fort. As discussed above (see The
ditch and Extra-mural Area of the Fort), the cut for the Cl. Br. fort ditch
was identified in this area, but only its uppermost fills were seen. The
1970s excavations showed, however, that the ditch was being used for the
dumping of domestic rubbish in the second half of the second century, and
did not serve its original function in the final phases of the fort's life.32 In
Trench 13, two layers of loam, 856 in the top of the ditch, and 854, which
covered 856 and spread northwards, can both be dated from their pottery
as mid to late third century or later. The relationship of 856 to the Shore
fort wall is unknown, so that it would be unwise to rely on this layer for
dating purposes. In the case of 854, however, more careful consideration
of its origin is required, and the shape of the deposit in section (Fig. 6,
Section B) would be consistent with it being derived from excavation
of the foundation trench for the Shore fort wall (820). At the same time,
854 spreads some distance (at least 4.60 m.) back from the wall,
and therefore might also include material being dumped from elsewhere.
Altogether, layer 854 is of considerable interest for dating purposes, as a
primary rampart tip, and this is discussed below in more detail.
Above 854 were two sequences of tip layers, many of which appear on
Fig. 6, and the majority of which also contained significant quantities of
pottery. The sequences, which were recorded separately, are: Sequence
A (in order of deposition) 822, 821, 819, 818, 812, 811, 805; these
deposits varied from fine rubble (including tufa, chalk, tile and shell) to
silty loam. Sequence B, also in order of deposition, comprised 835, 852,
851, 850, 849, 848, 847, 846, and (probably) 834; these deposits were
mainly silty or clay loams, with some sand or mortar lenses. Layer 835
contained a large quantity of broken roof tiles (see below, L. Turner, The
Roman Tile). The character of these rampart tips seems consistent with
that found elsewhere although clearly there is a change where substantial
buildings were demolished prior to the Shore fort construction.33
THE DATE OF THE SAXON SHORE FORT
Although the stratification relating to the Shore fort from previous
(1970s and 80s) excavations is not fully published, some discussion of
32 Philp (1981), 23-5.
33 For similar deposits, see Philp (1981), Fig. 23, Section 11 - layers above 73; for the
rampart constructed from demolition rubble, see Philp (1989), Fig. 22, Section E - Layers
54, 64.
70
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
the evidence is still possible. Deposits published with the Cl. Br. fort
include the soil which had formed over the rubble of the fort following
its abandonment (perhaps about A.D. 210). These layers contained a
number of coins of which the latest to definitely precede the Shore fort
(the coin of Allectus is from well outside the fort) is a coin of Salonina
(A.D. 253-59). Furthermore, the context containing this coin can be
matched by level and description with a context cut by the Shore fort
wall foundation trench, giving a terminus post quern of A.D. 253 for
this part of the wall at least, which lies about 20 m. south of the
rampart section excavated by the O.A.U., and described above.34 For
the length of rampart built over the demolished 'Painted House' and
other structures, 50 m. north of the O.A.U. excavations, a date of
A.D. 250-270 is suggested. As the provenance of four crucial coins is
not entirely secure (the latest of these is A.D. 238- 44) the terminus
post quern is best regarded as A.D. 222-35, from a coin of Julia
Mammaea in a demolition layer.35 Pottery from the demolition layers
includes groups comparable with those from the Trench 13 rampart tip
layers (see below).
To the above we can now add the new evidence, namely, the four
coins all dating c. A.D. 260-280 from the clay layer (34) which was
probably the tail of the rampart on the south side of the fort. We also
have the rampart tip layers (822 etc., 835 etc.) from the centre and front
(i.e. just behind the fort wall) of the west rampart (see above, Figs. 6,
7), for which a date of at least c. A.D. 275, but possibly as late as the
early fourth century is proposed, based on the large and coherent
pottery group which was recovered (see below, P. Booth, The Roman
Pottery, Nos. 67-87). The overall result of this review of the evidence
is to push the construction date for the Saxon Shore fort at Dover
slightly later than had previously been considered. The original
proposed date of A.D. 250-70 now looks untenable unless parts of the
fort (such as the west rampart investigated in Trench 13) were
completed considerably later than others. Such an explanation cannot
be written off completely, but would probably have to involve a break
of some years in construction activity; as such it verges on special
pleading and in any case can only be explored further when the rest of
the evidence from the site has been published.
A date of c. A.D. 275+ is, of course, perfectly reasonable, and would
fit Dover in to the second phase of Shore fort building - Stephen
Johnson suggests these sites (Burgh Castle, Bradwell, Walton Castle,
34 For the coin see Philp (1981), 122, No. 83, and pp. 94-6 for a discussion of soil over
the demolished buildings. For the contexts compare ibid., Fig. 25, Section 13 - Layer 3
(west end of section) with Fig. 25, Section 15 - Layer 33 (east end of section).
35 Philp (1989), 50-1.
71
D AVID R.P. WILKINSON
Richborough, Dover, Lympne) should fall within the decade
A.D. 275-85. A slightly later date for Dover would place the fort with
those at Portchester and even Pevensey.36 With regard to the problems
of dating the Shore forts, Johnson has also examined the typological
development of their defensive architecture. Within this scheme he
points out that Dover, having at least one 'early' feature, i.e. an internal
rampart, and some 'later' features, i.e. non-rectangular plan, sharp
corners and projecting towers (some integral, some added), could be
seen as transitional.37 This is not the place to explore this intriguing
idea in detail, but the date proposed above does not fit particularly well
- on current knowledge (and many of the forts are poorly dated at best)
a date of A.D. 250-60 might better suit a transitional structure. The
decision to construct the 'early' feature at Dover, the internal rampart,
could also have been influenced by other factors, such as the desire to
protect the vulnerable inner chalk face of the wall from weathering (see
also below - The Shore Fort Wall and External Towers). The rampart
also provided a dump for the rubble from demolished buildings,
excavation of a very large levelling terrace within the fort and
(possibly, depending on the construction sequence) the excavation of
the massive external ditches (Fig. 2), so that it did not have to be
transported off site.
THE SHORE FORT WALLS, EXTERNAL TOWERS AND A POSSIBLE POSTERN
GATE
The Shore fort wall was constructed on a trench-built foundation which
was c. 0.25 m. wider than the wall on each side, and contained four or
five courses of large chalk blocks (up to 0.30 x 0.30 m.) with some
flint. Only one narrow and heavily-robbed section of foundation (820)
was investigated in 1989 (Fig. 6, Section B), but previously-published
sections clearly show the construction detail.38 The wall itself (504)
was examined in Trench 12 (Fig. 12) where six courses were seen - the
foundation level was not reached here. The wall is 2.50 m. wide and uses
well-faced tufa blocks averaging 0.10-0.12 m. high x 0.18-0.30 m.
long for the exterior (southern) face, large unshaped chalk blocks,
flint nodules and very sparse small tufa fragments for the core, and
36 Johnson (1989, 39-44) points out that the dating for Portchester (after A.D. 286-7) is
more secure than that for Pevensey where the date of A.D. 330 or later rests on a single
coin from a beam hole in one of the towers.
37 Johnson (1989), 39-44.
38 Philp (1981), Fig. 24, Section 11, Fig. 25, Section 15; Philp (1989), Fig. 22, Sections
E,H.
72
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
shaped chalk blocks for the interior face. The coursing is continuous
throughout the whole wall, and courses were made up to an average
height of 0.14 m. by a layer of mortar - this was coarse, creamcoloured
and contained frequent large inclusions, including flint
fragments and water-rounded pebbles up to 0.02 m. in diameter.
Previous excavations in this area by the Kent Archaeological Rescue
Unit have shown that the wall still stands to a height of 4.50 m. in
places, with no change in the upper part from the use of chalk for the
inner face and tufa for the outer face.39 This aspect of the wall
construction merits some further discussion, for it is clear that the use
of chalk on the inner face only means that this material, which is highly
susceptible to weathering (particularly frost), could be covered and
protected by the rampart behind the wall. It may seem that it would
have been easier to simply use tufa for both faces, but tufa of suitable
quality for building was probably a relatively rare commodity. It is
noticeable that in the wall core it occurs rarely compared with flint and
chalk, and the available geological data also suggest a comparatively
rare resource.40 The necessity of exposing a minimum amount of chalk
to the elements must have been a continual concern for the builders at
Dover, and the use of an internal rampart may have originated with the
earlier Cl. Br. fort. 41 The Cl. Br. fort also used some tufa in the external
face of the fort wall, often alternating with chalk blocks.42 Finally, we
should not ignore the possibility that the external wall faces of either or
both of the forts could have been covered by a protective layer of either
whitewash or mortar at some point in their lives,43 although no
evidence for this was found at Dover, and the piece of wall described
above, sealed behind an external tower (see below) was apparently not
covered.
39 A large (c. 9 m. long) fallen section of the wall was also found, and it is suggested
that its collapse was caused by an earthquake, Kent Arch. Rev., 78 (1984), 187-90. It
seems at least equally possible that it was caused by the unusual combination at Dover of
a relatively narrow wall backed, and therefore under pressure from, the massive rampart,
particularly considering the extra material, in the form of dumped rubbish, which was
later added to the rampart (see this article, The Occupation of the Shore Fort).
40 Tufa occurs only under quite specific conditions (precipitation of calcium carbonate
near spring lines) and the early to middle Holocene marine transgression (stabilising near
modern sea level about 6000 years B.P.) eroded existing tufa deposits and buried others
below marine/estuarine deposits; see Barham and Bates (1990), 26, 101-7.
41 The implications of a rampart, however small, within the Cl. Br. fort are not
discussed by its excavator, but such a feature clearly existed: Philp (1981), 72-3, Fig. 24,
Section 9 (Road B40); ibid., 20 (Structure B2); ibid., Fig. 23, Section 7 (Structure B4,
Layer 55).
42 Philp (1981), 23.
43 Current Archaeology, 109 (1990), 119; ibid., 122 (1990), 94.
73
DAVID R.P. WILKINSON
The rear of one of the external towers was also found within Trench
12 (Fig. 12, 505), constructed almost flush to the outer face of the fort
wall, with the narrow intervening space being filled with mortar. The
mortar within this space, and in the fabric of the tower, was visually
identical to that used in the wall, but a very faint line (visible only in
section) down the line of the wall face indicated that it was built before
the tower. In terms of construction detail, the face of the tower was not
seen, but the core was broadly similar to that of the wall. It differed in
that the coursing seemed less regular, and a clear band was formed by
three successive courses of flint. The tower investigated here is
probably the same as that revealed by excavations in 1984, in which
case it stands over 4 m. high, and the facing includes tile stringcourses.
44
Of the known towers from the Dover Shore fort some are of integral
construction with the wall, while others, like that described above, are
built flush. Most of the known towers appear on Fig. 2, where on the
south wall they alternate between flush and integral. On the west wall,
working northwards from the integral corner tower, the next tower
(shown on Fig. 2) is flush, and the next two (not illustrated) are
uncertain, and then flush.45 The pattern may thus be the same as on the
south wall. No date for the addition of the extra towers has yet been
published, but two published sections both suggest the intervening
period between the construction of the wall and the additional towers
was short. In one case, the substantial wall stubs of an earlier structure
in front of the Shore fort wall are used as foundations for the tower -
had the wall existed for some time without a tower it would be
reasonable to expect the wall stubs to have been cleared away. Also, in
neither case has any intervening stratification formed between the
construction of the two elements although this could, of course, be
accounted for by truncation prior to building the towers.46 On present
evidence, then, the decision to add extra towers could actually have
been taken during the construction period.
Evidence has been discussed above for the use of the road B37,
within the Cl. Br. fort (Fig. 2), after the fort had been abandoned (see
The End of the Cl. Br. Fort ...) .I t is also possible that this road, or at
44 Kent Arch. Rev., 78 (1984), 190.
45 For the tower just north of the Cl. Br. fort the wall/tower interface was concealed
below the wall of the former joinery works in this area . Following demolition of the
joinery works superstructure, the archaeology in this area was reburied under O.A.U.
supervision, and is now within the grounds of the White Cliffs Experience. The next
(built flush) tower to the north overlay the 'Painted House', see Philp (1989), Fig. 21,
Fig. 22, Section E.
46 Philp (1981), Fig. 23, Section 6; Philp (1989), Fig. 22, Section E.
74
EXCAVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
least part of it, served some purpose within the Saxon Shore fort.
Firstly, dump layers below the latest phase of metalling in the road
contained late Roman grog-tempered ware which is thought not to have
reached Kent before the late third century (see P. Booth, The Roman
P ottery, Fabric R95). Secondly, although the stratification was much
disturbed by later graves and robber trenches, no rampart tips were
identified in this area (Trench 11), nor does the 1970s excavation plan
show that the Shore fort wall was found here, although the fort ditch
apparently continued.47 It seems, then, that while the southern part of
the road B37 would have been cut away by the Shore fort terrace, a
short northern length could have survived, and may have led to a
postern gate in the fort wall. A gate in this position would have given
access to the berm, or, if a wooden bridge existed, to the area beyond
the ditch. A postem gate was found during excavations in the early
1970s but its position is currently unclear.48
OCCUPATION OF THE SHORE FORT
The only evidence for activity within the Shore fort came from deposits
in Trench 1, above the rampart tail, 34 (see above, Abandonment of the
Cl. Br. Fort. .. ). A 1.50 m.-deep sequence of silts and clay-loams was
recorded in section (Fig. 8, Contexts 16-17, 21-32). Lenses of darker
silt and charcoal occurred frequently, as did shell fragments, and even
from the limited area of investigation, it is clear that a large number of
these deposits were pit fills (see Fig. 8). Section cleaning produced a
high concentration of coins, 17 in all; these are listed in Table 1.
In general, these deposits would seem to have been formed by the
dumping of rubbish, and by the excavation of pits for further rubbish;
this is confirmed by the quantity of pottery recovered - 48 sherds from
section cleaning alone. The depth of deposit accumulating over the
rampart is of interest, particularly as the coin evidence suggests this did
not begin, at least in this part of the fort, until the late fourth century.
Unfortunately, it is difficult to define how long this process continued,
as none of the coins or pot-sherds is definitely later than A.D. 402, and
there is a high degree of residuality as might be expected when rubbish
deposits are being frequently dug over and redeposited. To add to the
difficulty, no finds were recovered from contexts above lay er 16
(Fig. 8) and we can therefore only say with confidence, allowing time
for the coins to circulate, that deposition continued at least into the first
47 Philp (1981), Figs. 5 and 15 (area of drains B56, 58, 59).
48 Current Archaeology, 38 (1973), 88.
75
Context (latest at top)
17
23
26
25
24
27
28
32
DAVID R.P. WILKINSON
TABLE 1
Date of coin(s) (A.D.)
c. 330-368
c. 330-368
c.250-368
c. 250-368
c.348-360
260-268
260-268
330-348
c.330-368
260-285
c.260-285
c.260-285
268-274
c. 330-368
348-360
348-360
388-402
Catalogue No.
No. 28
No. 25
No. 5
No. 6
No. 31
No. 8
No. 7
No. 24
No.27
No.10
No. 16
No. 17
No. 20
No.26
No. 29
No. 30
No. 32
Table 1. Coins from the Saxon Shore fort occupation deposits (Trench 1).
quarter of the fifth century. Elsewhere in the fort, from the evidence of
material dumped in the fort ditch, Philp has suggested occupation into
the sixth century at least. 49 In general, the dumping of rubbish over the
rampart from the late fourth century onwards would fit with a model of
an initially well-organised life within the fort, surely military or under
military influence, followed later by a more rough-and-ready
occupation. This may well be pushing the small amount of evidence too
hard, but it is nevertheless interesting that a similar model has been
proposed for the Shore fort at Reculver, although here the whole
process is some 50 years earlier at least.50 The detail on the interior
arrangements of the Dover Shore fort has yet to be fully published, but
a circular wattle and daub hut was found in the south-west corner.
Other elements known to have been found are: at least eleven timber-
49 Op. cit. in note 34.
so Op. cit., 87; for the dating of the foundation of Reculver to A.D. 200-225, see P hilp
(1981), 18.
76
EXCAVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
built structures (circular, square, oval and sub-rectangular), metalled
internal roads, a postern gate with footbridge, ovens, pits and stakehole
complexes. The military bath-house, built in the second century
outside the Cl. Br. fort, was apparently re-used within the Shore fort.51
THE EARLY MEDIEVAL EV IDENCE; AN EXTENSNE SAND DEPOSIT SOUTH
OF THE SHORE FORT
Within Trench 14, just outside the south wall of the Shore fort, a
metalled surface (913) has already been discussed as part of the Cl. Br.
phase (Fig. 10). It was overlain by a silty loam (912) which could not
be dated. However, as the next layer in the sequence (911) was clearly
a partial collapse of the Shore fort wall, it is difficult to interpret the silt
layer confidently; perhaps, the most likely scenario is that the metalled
surface was in use during the life-time of the Shore fort, forming part
of the berm, and that the silt formed when it fell out of use. A metalled
surface which has been assigned to the late Roman period was found at
the Zion Chapel site, c. 40 m. to the south (Fig. 22).52 The layer of
Shore fort collapse was succeeded by a silty loam (910) with relatively
few inclusions, the accumulation of which suggests a period of little
activity, and followed by a sand layer (919), a second rubble (902)
derived from the Shore fort wall, and then a much deeper (at least
0.80 m.) sand layer (903; Fig. 10). Rubble 902, apart from some
residual second-century Roman sherds, also contained a single
sixth/seventh-century sherd (see The Medieval Pottery, Fabric 13, No.
95).
The sand layer 903, which can thus be dated as sixth/seventh-century
or later, has been identified across a wide area of the town south of
Shore fort (Fig. 1) and is over 3.50 m. deep in places. Observations in a
trench dug in York Street53 during roadworks show that in this area
(just south of the White Cliffs Experience site) there was no occupation
on the sand until at least the later eleventh century. Deposits below the
sand in Borehole 1, just south of Trench 14 (Figs. 11, 21, 22) were rich
in 'anthropogenic elastic material' (see below, Barham and Bates, The
Sedimentology) and are probably rubbish deposits within the Saxon
Shore fort ditch. A section running 140 m. southwards from close to
Trench 14 (Figs 1, 21) shows how the depth of the sand deposit
51 Current Archaeology, 38 (1973), 87; Wilkinson (1990), 17.
52 For an interim report on the Zion Chapel site, see Mynott (1981).
53 O.A.U. Watching Brief, 20/2/91; report lodged with the Kent Sites and Monuments
Record.
77
-.J
00
7.694 m. O.D.
0 2 m.
lllllc=-c:::::a,c::,-==-=-------======-;
Fig. 10. Trench 14, east section; extensive sand deposits (919, 903) and rubble layers (902, 911) from the Saxon Shore fort wall (for position of
section see Fig. 11).
s
:;,;i
i
@
i 2m.-8
-+
b-+
+V
A+
-.4
+_
4+
:;
ground surface
well sorted
sands
Borehole 1 $
(position)
- Borehole log
sllty sands with angular
gravel and clasts
.--•-.........
/ / "'l i
(
.
. I
903
poorly sorted sandy slits
with ceramic/building
stone gravel clasts
\
\
I
fining with depth
well sorted silts
well sorted clay silts
--
Barrack Block
823
0
/-- ·--·--,"! t.,.J....
I '·---- .' ·--
., D \ 7--.. ......., . ..........,, / , SC 91 \ .
-
-- -. / ' ' I ""' .....
I ', ' · 902 , ·'. .' , ' I I
I ' '· 910 ', 919 _!'-... -. I
I ' I ------/ .... _
I c- --, -. , '?"---r ----:foHc -- --- -') , , / 89 -- 911 .
I -..
scale 1:100
\. ..... -- -- /
/ ------/
I
'•-/
I
tv
-i--1970 excavation
I
5 m.
Fig. 11. Trench 14, plan. Position and log of Borehole 1.
79
DAVID R .P. WILKINSON
increases towards the sea-front, and also how at the Zion Chapel site it
overlay the late Roman surface (see above). This sand deposit is
discussed below in detail by Barham and Bates (see The
Sedimentology, and Appendix 1), who conclude that the onset of sand
deposition was rapid, quickly sealing deposits, and that fast deposition
occurred thereafter. The deposit, of which a significant proportion was
wind-blown, was derived from a beach or back-beach environment
which lay south of the York Street roundabout area, i.e. close to the
confluence of the River Dour with the outer estuary.
The extensive sand deposit has implications for both the interior and
exterior of the Shore fort, which was by now clearly in a state of some
disrepair, its ditch infilled, but with much of the wall still likely to have
been standing to a considerable height. Taking the exterior area south
of the fort first, the presence of a beach, and of the sand layer itself,
should imply that the Roman water-front, part of which was recently
found just east of the York Street roundabout, was now disused and
buried.54 This has some bearing on Biddle's suggestion that Dover,
where the place name 'Wyke' is known to have been used for the later
water-front, could have been the site of a middle Saxon trading
settlement like those known at Norwich, Ipswich and Sandwich.55 I
previously suggested that such a settlement might reasonably be
expected to lie on the water-front close to the Shore fort, but extensive
work by the Canterbury Archaeological Trust has now shown this to be
unlikely, at least for the southern area.56 Other possible locations are
within the Shore fort itself, with boats being brought up on the beach,
or in the area of the 'Inner Harbour' north of the promontory on which
the fort is built (Fig. 1). Neither of these possibilities can as yet be
explored much further, as the Shore fort evidence is not yet published,
and the 'Inner Harbour' is insufficiently explored. However, we do
know that a river rather than a sea harbour existed at Dover in 1066,
perhaps making the second possibility more likely.57
In the interior of the Shore fort, we have already noted occupation
into at least the sixth century58 and extensive later remains have also
54 Barham and Bates, 'The Sedimentology', (below); K. Parfitt, 'The A20/Dover
Sewers Project', Canterbury's Archaeology 1992-3, Canterbury Archaeological Trust,
(Canterbury 1993), 15; M.R. Bates, 'Recent Work by the Geoarchaeological Service
Facility (UCL) in Dover, with Particular Reference to the Discovery of the Bronze Age
Boat', ibid., Fig. 13 .
55 For discussion, see Wilkinson (1990), 18.
56 K. Parfitt, 'A20 Dover Sewers Project', in Canterbury's Archaeology, 1991-92,
Canterbury Archaeological Trust, 1992, 11-16; and K. Parfitt,pers. comm.
57 Wilkinson, (1990), 18.
58 Current Archaeology, 38 (1973), 88.
80
EXCAVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
been found, including a timber church and burials, other timber
structures, sunken-featured buildings, hearths, pits etc. Detailed dating
for these remains is not available, although some of the occupation is
certainly seventh- to ninth-century. We thus cannot y et establish
whether there were any breaks in occupation, or what the situation was
within the fort during the period when the sand deposits were forming
against its south wan.s9
Other evidence of early medieval (i.e. pre-eleventh century) activity
is limited to residual pottery. No. 95, from the sand layer, was the only
probable non-residual sherd. Otherwise less than a dozen sherds were
recovered, mostly eighth- to ninth-century; only three sherds of North
French Black ware were definitely imports, but little can be read into
the results from such a small collection (see C. Underwood-Keevill,
The Anglo-Saxon, medieval and post-medieval Pottery, e.g. No. 94).
THE MEDIEVAL PERIOD (ELEVENTH TO FIFTEENTH CENTURIES)
With the exception of a single context (363) in Trench 11 (which
contained twelfth-century pottery), and some residual sherds in later
contexts, all of the medieval evidence came from Trench 12, situated
on the south wall of the Saxon Shore fort (Fig. 2). The level to which
this trench was excavated is shown in Fig. 12, which makes it clear that
a dense complex of pits had been cut into the Shore fort wall and
bastion. Disturbed backfill over this area indicated that some truncation
of later deposits had taken place, and there may also have been some
excavation here during the 1980s.6°
The earliest excavated pit (556; Fig. 12) was cut right into the core
of the Shore fort wall. The labour required would have been
considerable, as the mortar in the wall remains sound to this day, and
the decision to go on digging the pit after the wall was encountered
could indicate that space for rubbish disposal was limited. It is,
however, difficult to imagine that the pit would have been dug if the
wall was visible at the time, and later operations (e.g. stone settings
510, 528, see below) may therefore have involved some clearance and
truncation. The eleven excavated fills of the pit consisted of mainly
silt-loams and thin lenses of sand, suggesting intermittent filling. No
great quantity of domestic rubbish was recovered from any of these,
and much of the fill may have been cut away by later pits. Three of the
fills contained small quantities of eleventh- to twelfth-century pottery.
59 Op. cit., 17-19.
6° Kent Arch. Rev., 71 (1983), 10-11.
81
DAVID R.P. WILKINSON
,·-·-·-·-·-·-·-·
I
----
' '
\
1 516
1980's
excavation
\ '
,, ', '
', ',
0
·-·-·-·-·-·-·,.. ·-·-·-·I
I j j
2 m.
Fig. 12. Trench 12, plan of intercutting medieval pits dug into Saxon Shore fort wall
(504) and back of added tower (505).
The uppermost excavated fill (507) of pit 556 was cut by another pit
of irregular shape, 519 (Fig. 12), which retained some traces of a burnt
wooden lining. The lower fills were similar in character to those
described above, but some, such as 522 and 533, contained
considerable quantities of shell, fish bone and animal bone, and clearly
represent domestic rubbish. Pottery recovered from the pit included a
residual sherd of Ipswich Ware (No. 94) from Context 546, but was
otherwise mainly late eleventh- to twelfth-century. The best
representations of pottery of that date were in Contexts 520, 544 and
549 which contained 24, 16 and 15 sherds respectively, including
82
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DO VER, 1988-91
imports from northern France. A single Flemish import, a fragment of
Aardenburg-type ware, is also the only thirteenth-century sherd and
came from near the top of the pit (Context 536). For further detail on
the pottery, see below, C. Underwood-Keevill, The Anglo-Saxon,
medieval and post-medieval Pottery.
The fills of pit 51 9 were cut by a sub-rectangular pit (528) and
a shallow post-hole (542). The post-hole was at least 0.35 m. in diameter,
and filled with sandy silt - no finds were recovered. Post-hole 542
was then cut away by another large, irregular pit (510) which extended
beyond the northern edge of the trench; this situation is illustrated in
Fig. 13. It was immediately apparent that neither of the cut features 528,
510, was used for rubbish disposal. Six large water-worn limestone
pebbles were placed in the bottom of feature 528, with some burnt and
unburnt flint (530) placed around them (Fig. 13, Phase I). Gaps between
the stone and flint were filled with clean yellow sand (531). Over this
stone setting was a layer of crushed chalk (529) followed by a second
similar setting (526) topped with crushed chalk (525; Fig. 13, Phase II).
Feature 510 contained a spread of small, shaped chalk blocks and
unshaped chalk fragments up to 0.30 x 0.50 m., covered by a layer of
silty loam and chalk (517). There then followed two successive layers
of roughly-shaped chalk blocks up to 0.30 x 0.40 m., 511 and 509
(Fig. 13) overlain by crushed chalk rubble (502) and then silty loam
503. The latter context contained a pottery handle, possibly from northwest
France (No. 107) while 509 and 511 contained residual eleventhand
twelfth-century material.
Both of the 'stone-packed' features described above must be thirteenthcentury
or later, but neither can be dated more closely. Beyond noting
that an area previously used for rubbish disposal was now surely the site
of a structure of some kind, it is not easy to interpret these features
within such a small area of excavation. One possibility is that 528
represents the foundation, or post-pad for a gate or large door-post, while
510 is the (separately excavated) foundation for an associated wall.
The careful laying of these foundations, as well as the possible
replacement of the packing in 510 (if this does genuinely represent
two phases) could both be explained by the loose fills of rubbish pits
beneath, which would have been liable to subsidence. Any further
interpretation must await the publication of other excavations in this area.
THE POST-MEDIEVAL PERIOD AND LATER
For the most part, evidence of these later periods was represented by
pottery from grave fills and other disturbed contexts in Trenches 11 and
13. The majority of the pottery was seventeenth- to eighteenth-century
83
1980's
excavation
PHASE I
DAVID R.P. WILKINSON
507
Shore fort tower
Shore fort wall
·-·-·-·-·-·-·-·-·-·-·-·-·-·-·-·-·-·.J
PHASE II
t::::11-=--==-=-,c:::,•oi:::======::::i1 m.
528
Fig. 13. Trench 12, medieval stone settings 528 (two phases) and 510.
84
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
(see Fabrics 4, 5, 40-1, Group 6, 100-1; Nos. 108, 109). Human bone
from the cemetery, which was still in use in the mid nineteenth century,
was bagged and reburied in the local Charlton cemetery, where a short
service was held. Aside from pottery, the only find of note from the
cemetery was a diamond-shaped copper alloy coffin plate, measuring
0.33 m. x 0.33 m. and bearing the following inscription, picked-out in
punched holes:
CONCLUSION
ARABELL A WYAT T
WIDOW OF C APTAIN
FRANCES WYAT T DYED
THE 17TH OF MARCH
1789 AGED 75
The work at the White Cliffs Experience site has shown that small-scale
excavations of this kind, within a general programme of preservation
in situ, can contribute valuable evidence. In the case of this particular
site, the presence of a body of published evidence for parts of the site
(albeit incomplete) has considerably aided interpretation.
The new detail obtained on the interior arrangements of the Cl. Br.
fort is valuable, but the most interesting information from the pre-Shore
fort period concerns an extra-mural building with a painted wall,
possibly external, facing the approach to the Cl. Br. fort's main east gate.
This building went out of use when the Shore fort was built, if not
before, but a road alignment within the abandoned Cl. Br. fort
apparently continued in use, possibly leading to a postern gate in the west
wall of the Shore fort. New evidence for the Shore fort's construction
suggests a date no earlier than A.D. 275, and possibly as late as the early
years of the fourth century. The presence at Dover of an 'early' feature,
the internal rampart, may be a local response to difficulties in
obtaining sufficient quantities of good building stone. Finally, the most
important evidence from the early medieval and medieval periods
concerns the extensive deposits of sand which formed south of the Shore
fort after the sixth to seventh century. The true significance of this
apparent abandonment of a large water-front area of the settlement will
only become apparent when more evidence is available for the nature of
early medieval settlement in other parts of the town.
85
DAVID R.P. WILKINSON
ACKNOWLEDGEMENTS
It would be impossible to list all of the people involved in the project,
but my thanks go to all who have helped. From Dover District Council,
the assistance of the following people was invaluable: John Moir, John
Clayton, Roger Madge and particularly Christine Waterman in her
capacity as Curator of Dover Museum. The project management
consultants, Richard Ellis, provided continual support throughout,
particularly Chris Richards, Mike Warner and Jane Male. Co-operation
during the difficult on-site phase of the work was excellent, and I am
grateful to Keith Emery, Alistair Gardener, Linsay Lloyd and Bob
Walker of Bovis Construction; the engineers, Anthony Hunt
Associates, particularly John Walker; Graham Powell of Michael Jones
Associates; and to the architects - Ahrends, Burton and Koralek. John
Sunderland (Designer), assisted by Allan Cobbold, was always willing
to discuss and consider the archaeology. Later in the project I received
help from Alan Smith, Paul Pinnock and the late Tony Gregory of
Heritage Projects, and from Danny Andrews of Carden, Godfrey,
MacFaddyen and Sturgis, conservation architects. Advice and guidance
on the site's archaeology came from Roger Thomas and Amanda
Chadburn of English Heritage, the Dover Archaeological Advisory
Board (Professors B. Cunliffe, D. Harris, P. Salway and P. Rahtz) and
John Williams of Kent County Council. Finally I would like to
acknowledge the help of my colleagues, Gill Hey, David Miles, Paul
Hughes, David Jennings, Ellen MacAdam and Anne Dodd, and of the
O.A.U. site teams, particularly Chris Bell, Cecily Cropper, Paul
Murray, Mike Napthan and Mick Parsons, with help from Tony Barham
and Martin Bates of G.S.F., and from Alan Massey and team (Dover).
As always, Janet DeLaine has supported me throughout everything.
THE COINS
Cathy E. King
Thirty-four coins were recovered from this site of which 32 were
Roman (see Table 2). Three were imperial bronzes, the first an illegible
as or dupondius of the first or second century A.D. (No. 1), the second
a bronze of Antoninus Pius (No. 2) and the third a sestertius of
Postumus (No. 9) which will be discussed more fully below. The
remainder of the pieces consist of debased silver and copper issues of
the third or fourth century. The largest concentration of coins (15) were
minted in the years between A.D. 260 and 280. While the majority of
British sites have significant numbers of coins minted between these
years the White Cliffs Experience site is unusual in having so few coins
86
EXCAVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOYER. 1988-91
datable to the fourth century, particularly from A.D. 330-48 and
A.D. 364-78. These periods are normally well-represented on sites in
Britain and northern Gaul. In part the small number of fourth-century
pieces is due to the low total number of coins recovered during the
excavations and in part it results from the relatively high proportion of
coins (5) which are not sufficiently legible to be assigned to a specific
period within the fourth century.
The number of ancient imitations from the site is also small; there
are two third-century antoniniani copying pieces minted between
c. A.D. 260 and 285 (Nos. 15, 17) and a fourth-century FEL TEMP
REPARATIO falling horseman type (No. 31) whose prototype was
minted between A.D. 353 and 360.
The non-Roman coins consist of a silver penny of Edward I, II or III
from the Canterbury mint produced between 1300 and 1400 (No. 33), a
Queen Victoria halfpenny and an unidentifiable coin or token minted
between c. 1700 and 1900 (the latter two are not catalogued here).
The most interesting coin recovered was a sestertius of Postumus
(No. 9) with the obverse legend IMP C POSTVM VS [P F AVG] and a
left-facing bust with the emperor raising his right hand and the reverse
VIRT VS AVG S C. The piece is rare owing to the special bust type,
and Bastien cited only one very worn specimen in the Bibliotheque
Nationale in Paris (Bastien 1967, No. 123). It belongs to a group of
coins which share this special obverse and which, according to Bastien,
were minted in A.D. 261. At some time after the coin was minted, an
attempt was made to turn it into a double sestertius (a denomination
minted by Postumus) by scratching lines across the hair to simulate a
radiate crown (see Plate I). Other examples of this practice are known
(Gricourt 1990, Pl. X, No. 4, Pl. XII, No. 8, Pl. XIII, No. 14 ).
PLATE I
Bronze coin (sestertius) of Postumus, from the White Cliffs Experience site, Dover
(No. 9, A.D. 260-68).
87
00
00
Cat.No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Obverse Emperor
II.LEG.
II.LEG. A.Pius
MIVL Philip II
PHILIPPVS
CAES
II.LEG.
ILLEG.
ILLEG.
ILLEG. Gallienus
II.LEG. Gallienus
IMPC Postumus
POSTVMVS(J
ILLEG.
ILLEG.
II.LEG.
ILLEG.
ILLEG.
TABLE2
Reverse Mint Date
11..LEG. Rome 44-193c.
ILLEG.Stg. Rome 138-161
female figure
PRINCIPI IVVENT Rome 244-246
ILLEG. 250-285c
ILLEG. 260-368c
II.LEG. 260-368c
□CONS □animal Rome 260-268
ILLEG. Rome 260-268
VIRTVS AVG SC Gaul 260-268
ILLEG. 260-285
ILLEG. Stg. figure 260-285c
ILLEG. 260-285c
ILLEG. Stg. female 260-285c
figure
ILLEG. Stg. female 260-285c
figure
ILLEG. Stg. female 260-285c
figure
Denom. Mat. Comments
As/Dp AE
As/Dp AE
Ant. AR RIC 216c
Ant. AE Empress
AE
AE In fragments
Ant. AE
Ant. AE
Sest. AE Bast 123
Ant. AE
Ant. AE
Ant. AE Broken
Ant. AE
Ant. AE
Ant. AE 14mm
Obje ct No.
16
181
44
20
8
5
186
4
97
(same dies?)
188
30
28
34
36
37
Context
1
U/5
335
33
26
26
27
24
U/5
32
34
34
154
34
34
6
::,.,
0
z
00
\0
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
II.LEG.
[DNO
CLAVDIOJ
ILLEG.
[JPFAVG
ILLEG.
[ JV TETRICVS
CAES
ILLEG.
ILLEG.
ILLEG.
ILLEG.
ILLEG.
ILLEG.
ILLEG.
ILLEG.
ILLEG.
ILLEG.
ILLEG.
ILLEG. 260-285c Ant. AE
Claudius II [CONSECRATIOJ 260-285c Ant. AE 12mm
altar
ILLEG. 260-368c AE
Stg. female figure Gaul 268-274 Ant. AE
ILLEG. Salus? Gaul 268-274 Ant. AE
Tetricus II SPESAVGG Gaul 270-274 Ant. AE Cun 2647
ILLEG. 310-313c Fol. AE
ILLEG. 320-368c Fol. AE
ILLEG. 330-348 Fol. AE
ILLEG. 330-368c Fol. AE Chipped
ILLEG. 330-368c Fol. AE
ILLEG. 330-368c Fol. AE
ILLEG. 330-368c Fol. AE
ILLEG. 348-360 Fol. AE
ILLEG. 348-360 Fol. AE
[FEL TEMP 348-360c Fol. AE 12mm
REPARATIO]
Falling horseman
ILLEG. Vi ctory 388-402 Fol. AE
Edward I, II Canter- 1300-1400c Penny AR
or III bury
Table 2 Coins from excavations at the White Cliffs Experience site.
189
190
192
33
187
16
153
15
12
9
184
11
10
25
191
24
185
40
32
32
261
U/5
32
103
827/2
101
27
23
32
28
17
32
32
25
32
315
Cl)
0
z
j
I
i
n
tn
Cl)
-
t;1
0
......
\0
00
00
DAVID R.P. WILKINSON
Bronze coins of Postumus are not common on British sites although
they have been recovered in modest numbers at Bath (4),
Colchester (4), Richborough (7), Verulamium (2), Harlow (1) and
Leicester (2).61 They also occur in a few British hoards, again in small
numbers, and can be found in mixed hoards of silver and bronze coins
or finds of bronze coins only (See Table 3).
TABLE 3
HOARD62 AE AR TOTAL POSTUMUS
Bourne End (Herts.) 35 5 40 1
Leysdown (Kent) 504 504 2
Ramsgate (Kent) 27 27 1
Combe Hill (E. Sussex) 4 140 144 1
Gare (Cornwall) 1037 47 1084 4
Alchester (Warwickshire) 51 95 146 2
Ham Hill (Somerset) 1066 1066 2
Table 3 British coin hoards containing bronze coins of Postumus.
The geographical distribution of Postumus bronzes both in hoards
and on sites suggests that they are more likely to be recovered in
61 D.R. Walker, 'Roman Coins from the Sacred Spring at Bath', in (Ed.) B. Cunliffe,
The Temple of Sulis Minerva at Bath, II: Finds from the Sacred Spring, (Oxford, 1988),
281-358; (Ed.) N. Crummy, Colchester Archaeological Report 4: the Coins from
Excavations in Colchester, 1971-9, (Colchester, 1987), 84-92; R. Reece, 'The Roman
Coins from Richborough - a Summary', Bull. of the Inst. of Arch. of the Univ. of London,
xviii (1981), 49-71; R. Reece, 'The Coins', in S.S. Frere, Verulamium Excavations, Ill,
(Oxford, 1984), 3-17; N.E. France and B.M. Gobel, T he Romano-British Temple at
Harlow, Essex, (Gloucester, 1985), 67-70; K. Kenyon, Excavations at the Jewry Wall
Site, Leicester, (London, 1948), 279-82.
62 A. Burnett, 'Bourne End Find (1976)' Coin Hoards III, (1977), 77-8, No. 174;
R.A.G. Carson, 'Leysdown (Kent) Hoard of Early Roman Imperial Bronzes', Numismatic
Chronicle, xi.7 (1971), 189-197; R. Merrifield, 'A Roman Coin-hoard from Ramsgate,
1969', Numismatic Chronicle, xi.7 (1978), 199-201; D.R. Rudling, 'A Hoard of Roman
Coins from Combe Hill, East Sussex', Coin Hoards from Roman Britain, V I, Brit. Mus.
0cc. Pap. No. 58, (1986), 147-55; R.A.G. Carson, 'Gare (Cornwall) Find of Roman
Silver and Bronze Coins', Numismatic Chronicle, xi.7 (1971), 181-8; R.A.G. Carson,
'Alchester (Warwickshire) Find of Roman Ant oniniani and Sestertii', Numismatic
Chronicle, ix.7 (1969), 23-128; W.A. Seaby, 'Coinage from Ham Hill in Taunton
Museum', Numismatic Chronicle, ix.6 (1949), 166-79.
90
EXCAVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
southern Britain and they do not seem to be restricted to a narrow
geographical area. However, it would not be surprising if they proved
ultimately to be more dominant in the south-east.
Comparison of the pattern of coin loss from the White Cliff
Experience site with that from the Cl. Br. and the 'Painted House'
excavations shows distinct chronological differences between them.
While the majority of the White Cliffs site coins is concentrated in the
period between c. A.D. 260 and 296, coins from the second century,
and bronzes in particular, are dominant in the published lists from
Cl. Br. and 'Painted House' sites and they show virtually no coins after
A.D. 260 (Philp 1981, 119-20; 1989, 56-7). This is at variance with the
normal British pattern where late third- and fourth-century coins are
recovered in large numbers, but it should be noted that 'hundreds of
late-Roman coins' (Philp 1989, 283) are still to be published from
within the Shore fort, so that the true pattern may not yet be apparent.
The low representation of second-century coins at the White Cliffs site
can probably be explained by the very limited number of contexts
excavated from the Cl. Br. fort. The Cl. Br. and 'Painted House' sites
are also unusual in yielding a relatively high proportion of silver coins
produced in the period between A.D. 193 and 260. Viewed in the
context of the total numbers of coins recovered from the various Dover
excavations, the late third-century concentration of coins from the
White Cliffs site is less anomalous than it first appears.
THE ROMAN POTTERY
Paul Booth
INTRODUCTION
The various excavations produced in total 2040 sherds (c. 31.5 kg.,
39.05 EVEs) of Roman pottery, which was examined at varying levels
of detail. The majority of this was from the various DHC trenches. The
total includes pottery from the small-scale clearance operations DTT
and DSC, which produced 25 and 12 Roman sherds, respectively, and
are not commented on further (i.e. the discussions of fabrics and vessel
types below exclude these figures). It does not include a further 3.5 kg.
of pottery discarded by the Kent Archaeological Rescue Unit (marked
with various DV context numbers) which was recovered from the
backfill of some of their trenches. This material is not considered here.
The pottery spans the period from the early second century to some
time in the fourth, with a very small number of earlier sherds,
presumably residual in the contexts in which they occurred. A wide
91
DAVID R.P. WILKINSON
range of sources of supply is indicated for the Classis Britannica fort,
but the assemblage is most important in providing a good group of
material from the rampart of the Saxon Shore fort.
The pottery was recorded using the system currently employed by
the O.A.U. on sites within the Oxford region and elsewhere. The
pottery was divided into a series of fabric and ware groups (see below)
and quantified by sherd count, weight and EVEs (Estimated Vessel
Equivalents, based on rim percentages - more strictly rim-equivalents,
see e.g. Pollard 1990, 76). Most of the discussion of variations in fabric
proportions is based on the sherd count figures. Discussion of the
vessel types represented is based on the figures for EVEs.
FABRICS
Fabrics were assigned to one of a number of 'ware groups' (e.g. F=Fine
Wares, O=Oxidised Coarse Wares, etc.). The system is organised
hierarchically so that sherds can be recorded at one of several interrelated
levels of detail. The primary level is that of the ware group itself ( e.g. F,
0, R=Reduced Coarse Ware, etc.), the secondary level is that of the major
subdivisions of the ware group (e.g. R80=grey wares in which sand is
the principal tempering agent, BlO=all BBl fabrics, B20=all BB2 fabrics)
and a third level can be used to identify a specific fabric or ware source
(e.g. F51=Oxfordshire colour-coated ware). The fabric/ware codes
used here represent an extension of the system as employed in
Oxfordshire. Direct overlap with the Oxfordshire codes was avoided
where it was not appropriate; many of the latter were clearly not
applicable to Kent. Many of the codes for fine wares, amphorae, mortaria,
black-burnished wares, etc., are, however, universally applicable.
Limitations of time meant that it was not possible to identify every
sherd down to the level of individual fabric/ware. For many of the
reduced coarse wares, for example, attribution was at the intermediate
level of precision. As far as possible, however, fine wares, amphora and
mortarium sherds were assigned to specific fabrics. Definition of two
important groups of coarse wares (here fabrics R95 and R98) follows
the usage of Pollard (1988, 126, 129; see below).
FABRIC DESCRIPTIONS
Only brief descriptions are given here, and widely-known fabrics are
referred to by their common names. Full descriptions, where appropriate,
are contained in the excavation archive. The total number of sherds in
each fabric is also given here and tabulated below (Table 4).
92
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
S. Samian ware.
This has not been examined in detail. The overall designation S was
used for all fabrics regardless of source. 344 sherds.
S45. Argonne ware (tentative identification). 2 sherds.
F. Fine wares (i.e. colour-coated, lead glazed, mica dusted, etc., but not
fine oxidised and reduced wares in the sense used by Pollard (e.g.
1988, 59-60).
F35. Fine sandy oxidised, mica-dusted. 9 sherds.
F36. Sandy oxidised, mica-dusted, probably Canterbury? 1 sherd.
F40. Fine buff/white fabric, red-brown colour-coat, ?Continental.
2 sherds.
F43. Central Gaulish 'Rhenish' ware. 1 sherd.
F44. Trier(?) 'Rhenish' ware. 4 sherds.
F46. Fine, white fabric, colour-coated, 'Lower Rhine fabric 1'
(Anderson 1980, 14). 23 sherds.
F49. 'Ceramique a l'eponge' (Fulford 1977, 45-46). 2 sherds.
F50. Fine oxidised, red-brown colour-coat, source uncertain, probably
British. 1 sherd.
F51. Oxfordshire colour-coated ware. 8 sherds.
F52. Lower Nene Valley colour-coated ware. 38 sherds.
F54. New Forest reduced 'stone ware' colour-coated ware. 1 sherd.
F55. ?Colchester colour-coated ware. 9 sherds.
F56. ?Much Hadham colour-coated ware. 2 sherds.
F60. Oxidised, red-brown colour-coat, source uncertain/?various,
probably British. 3 sherds.
F66. Fine sandy oxidised, red colour-coat, cf. F35, ?local. 3 sherds.
F67. Coarse sandy oxidised, red colour-coat, ?Pompeian red ware.
Source uncertain. 4 sherds.
A. Amphora fabrics.
AlO. Buff-brown ?South Spanish Dressel 20. 1 sherd.
A l l . Buff-brown South Spanish Dressel 20. 12 sherds.
A l 3. Fine-buff ?South Gaulish (e.g. Pelichet 47 (Peacock and Williams
1986, class 27) etc.). 25 sherds.
A20. Fine oxidised, source uncertain. 4 sherds.
A22. Fairly fine sandy, oxidised, probably South Gaulish as A 13.
2 sherds.
A25. Sandy oxidised, carrot amphora (exact type uncertain). 1 sherd.
M. Mortarium fabrics.
MIO. Buff, coarse sandy, probably Rhineland. 1 sherd.
Mll . Fine buff, Hartley (1977) group 1? North Gaul/South-east
England. 2 sherds.
93
DAVID R.P. WILKINSON
M12. Fine buff, Hartley (1977) group 2? North Gaul/South-east
England. 3 sherds.
M22. Oxfordshire white ware. 1 sherd.
M23. Mancetter-Hartshill white ware. 1 sherd.
M29. Buff, fine, Kent/south Essex/Colchester group (Pollard 1988,
213, cf. Hartley 1982, 151, fabric lB). 11 sherds.
M35. Fine oxidised, white calcareous inclusions, white slip, source
unknown. 1 sherd.
M36. Sandy oxidised, white slip, ?east Kent. 1 sherd.
M41. Oxfordshire oxidised, white slipped fabric, Young (1977) fabric
WC. 1 sherd.
M55. Buff sandy fabric, Canterbury. 7 sherds.
M56. Fine sandy similar to M56, ?Canterbury but possibly imported.
5 sherds.
W. White wares.
W. General category, undifferentiated. 1 sherd.
W lO. Fine white ware, uncertain source. 1 sherd.
W20. Sandy white wares, ?various (uncertain) sources. 5 sherds.
W2 l . Sandy buff/white ware, Verulamium region. 1 sherd.
W40. Fine 'pink-buff' wares, probably local. 16 sherds.
Q. White-slipped fabrics, except mortaria. Mainly oxidised 'flagon'
fabrics.
Q40. Fairly fine oxidised, cream or white slip, ?various sources.
4 sherds.
Q41. Sandy oxidised with white slip, probably Canterbury. 16 sherds.
Q42. Fine sandy oxidised with white slip, cf. F35. 5 sherds.
Q43. Fine oxidised, micaceous, with clay pellets, white slip, probably
local. 4 sherds.
Q44. Fine, hard oxidised with white calcareous inclusions, buff-white
slip. Source uncertain. 28 sherds.
0. Oxidised coarse wares.
0. General category, undifferentiated. 4 sherds.
010. Fine oxidised wares, ?various sources. 7 sherds.
019. Fine red-brown, very micaceous fabric. ?source. 1 sherd.
050. Sandy oxidised wares, ?various sources, most probably ?local.
23 sherds.
051. Canterbury sandy oxidised ware. 84 sherds.
053. Slightly sandy oxidised, micaceous, with clay pellets, probably
local, cf. Q43. 8 sherds.
055. Slightly sandy oxidised, white calcareous inclusions. Similar
to 053 but less micaceous. ?Local. 2 sherds.
94
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
057. Fine sandy oxidised. Perhaps from Much Hadham (Herts.).
4 sherds.
061. Fine buff-orange, source unknown. 1 sherd.
080. Coarse (usually ?grog-tempered) oxidised fabrics. 1 sherd.
085. Coarse oxidised, grog inclusions. 2 sherds.
R. Reduced coarse wares.
R. General category, undifferentiated. 7 sherds.
R30. Fine reduced wares, ?various sources. 44 sherds.
R35. 'Upchurch type' fine grey ware. Most, if not all, from north Kent.
130 sherds.
R80. Sandy reduced wares, ?various sources. 401 sherds.
R81. Canterbury sandy reduced ware. 49 sherds.
R82. Similar to R81, but less uniformly sandy. 63 sherds.
R83. Buff-grey, fine sand inclusions, ?North Gaul. 2 sherds.
R85. Highgate Wood sandy reduced ware (Brown and Sheldon 1974,
224). 3 sherds.
R89. North-east Gaulish sandy reduced ware. 11 sherds.
R90. Coarse (mainly grog?) tempered reduced wares, various sources.
19 sherds.
R95. 'Late-Roman' grog tempered ware (Pollard 1988, 129). 67 sherds.
R96. Brownish-grey with common flint inclusions. Possibly prehistoric?
1 sherd.
R97. Hard, dark grey, cf. R98, but with moderate flint inclusions.
3 sherds.
R98. Hard, dark grey fabric with coarse grog and quartz inclusions,
'Native Coarse Ware' (Pollard 1988, 126). 9 sherds.
B. Black-burnished wares.
BU. Black-burnished ware category 1 (BB l ). Dorset. 107 sherds.
B15. Black-burnished ware category 1, non-Dorset source. 5 sherds.
B20. Black-burnished ware category 2, various sources. 330 sherds.
B22. Black-burnished ware category 2, possibly Colchester? 19 sherds.
C. Calcareous tempered fabrics/wares.
C lO. Shell tempered fabrics, various sources? 5 sherds.
CU. Midlands late Roman shell tempered ware. 2 sherds.
P. ?Prehistoric fabrics.
P. General category, black, with flint and sand inclusions, hand-made.
2 sherds.
Z. ? Anglo-Saxon fabrics.
Z. General category, black, sand tempered, hand-made. 2 sherds.
95
\0
O'I
s
331
F35
9
AIO
l
MIO
l
w
l
Q40
4
0
4
R
7
B11
107
ClO
5
p
2
z
2
S45
2
F36
l
All
12
Mll
l
WIO
l
Q41
16
010
7
R30
42
Bl5
4
Cll
2
TABLE 4
FABRIC AND NUMBER OF SHERDS
F40 F43 F44 F46 F49 F50 F51 F52 F54 F55 F56
2 l 4 23 2 1 8 37 l 9 2
Al3 A20 A22 A25
25 4 2 1
Ml2 M22 M23 M29 M35 M36 M41 M55 M56
3 1 l 11 1 l l 7 5
W20 W21 W40
5 l 16
Q42 Q43 Q44
5 4 28
019 050 051 053 055 057 061 080 085
l 23 84 8 2 4 l l 2
R35 R80 R81 R82 R83 R85 R89 R90 R95 R96 R97
127 401 49 62 2 3 11 19 66 1 3
B20 B22
321 19
Table 4 Roman pottery: total sherds in each fabric type.
TOTAL
SHERDS
333
F60 F66 F67
3 3 4 110
45
33
24
57
137
R98
9 802
451
7
2
2
per.
cent.
16.6
5.4
2.2
1.6
1.2
2.8
6.8
40.l
22.5
0.3
0.1
0.1
s
:;d :-0
0
z
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
The assemblage was dominated by reduced coarse wares and blackburnished
wares, with samian ware also forming a substantial
component. The major ware group proportions expressed in terms of
EVEs gave quite closely comparable figures, with black-burnished
ware slightly better represented ( 24.5 per cent), and reduced coarse
wares and samian slightly less common (respectively 37.3 per cent and
14.0 per cent). The total of so-called 'fine and specialist' wares (e.g.
Booth 1991, 5), i.e. samian, other fine wares, amphora, mortarium,
white and white-slipped fabrics, is 29.8 per cent (3 2.1 per cent EVEs).
Such a relatively high representation of these wares is what would be
expected on what is essentially a high status site.
The samian is drawn from all sources, but South Gaulish material is
very rare, which is consistent with the generally accepted early secondcentury
foundation date for the Classis Britannica fort. East Gaulish
material seems relatively well-represented, and two sherds of Argonne
ware were tentatively identified. There was a wide range of fine ware
fabrics, including mica-dusted wares as well as colour coated fabrics.
Of the latter the 'lower Rhine' fabric F46 was most common in the
earlier phases (though several significant occurrences of this fabric
were residual in later contexts). Of the principal Romano-British colour
coated ware sources, the Nene Valley was the best represented,
constituting exactly one-third of all the fine ware sherds. Oxfordshire
and probable Colchester and Much Hadham products were all less
common. There were four sherds which were perhaps of Pompeian Red
ware (which has been noted at Dover (Pollard 1988, 8 2)), and two of
'ceramique a l'eponge, never particularly common in Kent but also
already identified at Dover (ibid., 14 2 and 216).
Amphorae were relatively scarce, only one rim being represented
(No. 33 below). The most common fabric was Al 3, attributed to
Southern Gaul and used principally for types such as the wine
amphora Pelichet 47 (Peacock and Williams 1986, 14 2-3, class 27).
Fabric A2 2 (see No. 66) may have originated in the same area. The
ubiquitous south Spanish olive oil amphora, Dressel 20, was rather
less common, but sherds in fabrics AlO, Al l and A20 were probably
all of this type.
Of the wide range of mortarium fabrics the most important were
relatively local products; M 29, which may represent several sources in
Kent and Essex, including Colchester (No. 50, below, has a typical
Colchester herringbone stamp), and the probable Canterbury fabrics
M55 and M56. Sherds of imported vessels (fabrics M10-M1 2) formed
only a small proportion of this group. There were single sherds of
Mancetter-Hartshill and two Oxfordshire fabrics, all of which occurred
in late-Roman contexts. Most of the fabrics of the white and whiteslipped
ware groups are much less easily sourced, though the only
97
DAVID R.P. WILKINSON
likely non-local example was a single sherd of fabric W 21, a
Verulamium region product.
Oxidised wares formed only a relatively small part of the assemblage
(6.8 per cent). Most of these were sand-tempered, the exceptions being
a few fine fabrics (010, 019 and 061) and the grog-tempered 080
group. 051, much the most important oxidised fabric, is a Canterbury
product, and a fairly local origin is likely for most of the other fabrics
in this group. 057, of which four sherds were identified, can, however,
be attributed to Much Hadham (Herts.) (along with the colour-coated
fabric F56). While these sherds are unlikely to have arrived in Dover
before the late third century (Pollard 1988, 119), the bulk of the
oxidised wares will have been of second-century date since the
Canterbury industry is thought to have been in decline by the later
second century (ibid., 178-9).
Reduced wares, the dominant ware group, were also drawn largely
from Kentish sources, though the importance of Canterbury is less
clear. Sherds attributed to Canterbury with some confidence only
amounted to 6 per cent of all the reduced wares, and although the
general sandy reduced ware group (R80, which comprised half of all
the reduced ware sherds) may have contained many more examples,
Canterbury products were a much smaller component of the reduced
than of the oxidised wares (see also discussion of vessel types below).
Sand-tempered fabrics amounted to two-thirds of all the reduced
wares, but the number of possible sources is unknown. Three sherds
(fabric R85) were identified as coming from the kilns at Highgate
Wood (R. Pollard, pers. comm.). A further 11 (Fabric R89) and
possibly two in Fabric R83, are likely imports from northern France,
where several different sources produced vessels in a common
tradition (Richardson and Tyers 1984, see also Richardson 1986,
106-9). Some of these sherds have characteristic horizontal burnished
line decoration ('bandes lustrees', cf. ibid., 134) and the fabrics are
sufficiently distinctive to allow confident identification. Rims of
several vessels occurred, including a possible jar, a beaker ('vase
tronconique' - Tuffreau-Libre 1980, 96-9), bowl and ?dish types, but
all were fragmentary.
The other principal categories of reduced wares were fine and grogtempered
fabrics. The former, Fabrics R30 and R35, amounted to
21 per cent of all reduced sherds. A north Kent source is likely for
these fabrics, and indeed some of the R80 sherds may also have
originated here. R35 was very fine, often with a characteristic
sandwich, and is consistent with Monaghan's fabric N2/lb (Monaghan
1987, 252), which may reasonably be regarded as genuine 'Upchurch
Ware' (ibid., 173). Sherds of R30 were generally similar, but usually
not quite as fine; they cannot be regarded as certain Upchurch products.
98
EXCAVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
R90 fabrics, all grog-tempered, included a small quantity of 'Native
Coarse Ware' (Fabric R98; Pollard 1988, 126) and a higher proportion
of late Roman grog-tempered ware (R95, which comprised 3.3 per cent
of the total assemblage; Pollard 1988, 129).
Black-burnished wares formed the only remaining ware group of
significance. The majority of this group was composed of BB2 fabrics
(B20 and B22). Sherds of B22, with very high quality burnishing, were
separated because it was felt that they might represent a distinct
production site (e.g. Colchester, as suggested by Williams (1977, 208)),
but this remains uncertain. No other attempt was made to subdivide
B20, which must include vessels from a number of sources, probably
chiefly the Upchurch and Thameside kilns. Both of these centres were
of little significance after c. A.D. 250, however (Monaghan 1987, 221
and 225). B l l , which totalled a little less than a quarter of all blackburnished
ware, shows a complementary pattern in that it was most
important in the late third and ?early fourth centuries. The majority of
recognisable BB 1 vessel types can be assigned to this period. While
BB 1 probably occurred at Dover in small quantities from the
foundation of the Classis Britannica fort onwards, in this assemblage,
at least, it can be seen as succeeding BB2 in widespread use. The
majority of the BB 1 seems to have been the standard Dorset product,
but there were five sherds of a distinct fabric (B 15) with fine white
inclusions which were presumably from another source.
VESSEL TYPES
Vessel types were recorded in a manner similar to the fabrics, i.e. major
vessel classes (flagons, beakers, jars, etc.) were defined and then
subdivided hierarchically. Detailed recording of rim profiles potentially
allowed extensive subdivision of the major classes, but this was not
done for the Dover assemblage except in a few cases. Most vessels
were therefore recorded at the level of the principal subdivisions of the
major vessel classes, though some rims (particularly jars) were not
defined more precisely than at the level of the major class. Table 5
shows the breakdown of fabric and type at this level of detail. A more
detailed breakdown using the subdivisions of the major vessel classes
was not particularly informative and is retained in archive. Some of the
information derived from that data is, however, discussed below.
The vessel type codes used for the major classes and found in
Table 5 were as follows:
A, Amphora; B, Flagon; C, Jar; D, Jar/beaker (not used); E, Beaker;
F, Cup; G, Tankard (not used); H, Bowl; I, Bowl/dish; J, Dish; K,
Mortarium; L, Lid; M, Miscellaneous (not used); Z, Uncertain.
99
DAVID R.P. WILKINSON
TABLE5
TYPE A B C E F H I J K L z TOTAL per
WARE cent
s 1.18 3.69 0.13 0.28 0.08 5.36 14.0
F35 0.23 0.12 0.35 0.9
F36 0.04 0.04 0.1
F40 0.05 0.05 0.1
F46 0.37 0.37 1.0
F51 0.09 0.09 0.2
F52 0.91 0.91 2.4
F55 0.75 0.75 2.0
SubF 2.08 0.36 0.12 2.56 6.7
A22 0.25 0.25 0.7
MIO 0.12 0.12 0.3
Mll 0.04 0.04 0.1
M12 0.13 0.13 0.3
M23 0.14 0.14 0.3
M29 0.36 0.36 0.9
M35 0.17 0.17 0.4
M41 0.02 0.02 0.1
M55 0.18 0.18 0.5
M56 0.08 0.08 0.2
SubM 1.24 1.24 3.2
W40 0.88 0.88 2.3
Q41 0.64 0.64 1.7
Q43 0.33 0.33 0.9
Q44 1.00 1.00 2.6
SubQ 1.97 1.97 5.2
100
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
TYPE A B C E F H I J K L z TOTAL per
WARE cent
050 0.35 0.02 0.37 1.0
051 1.02 0.31 0.03 0.10 1.46 3.8
061 0.22 0.22 0.6
SubO 1.02 0.66 0.05 0.10 0.22 2.05 5.4
R30 0.26 0.54 0.08 0.88 2.3
R35 1.38 0.59 0.54 0.05 0.06 2.62 6.8
R80 4.77 0.06 0.65 0.06 0.05 0.05 5.64 14.7
R81 2.04 0.09 0.05 2.18 5.7
R82 0.62 0.55 0.08 0.04 0.02 1.31 3.4
R83 0.06 0.06 0.2
R85 0.10 0.10 0.3
R89 0.30 0.10 0.04 0.44 1.2
R90 0.18 0.12 0.30 0.8
R95 0.42 0.17 0.08 0.67 1.8
R98 0.03 0.05 0.08 0.2
SubR 9.76 2.04 1.78 0.39 0.11 0.05 0.15 14.28 37.4
Bll 0.50 0.60 0.04 0.65 1.79 4.7
B15 0.03 0.03 0.06 0.2
B20 1.42 0.07 0.73 3.14 1.30 6.66 17.4
B22 0.25 0.16 0.32 0.14 0.87 2.3
SubB 2.20 0.07 1.49 3.50 2.12 9.38 24.5
Cl0 0.25 0.25 0.7
TOTAL 0.25 3.87 12.87 4.19 1.18 7.37 4.01 2.36 1.52 0.15 0.45 38.22
per cent 0.7 10.1 33.6 11.0 3.1 19.3 10.5 6.2 4.0 0.4 1.2
Table 5 Roman pottery: ware types represented in each fabric.
101
DAVID R.P. WILKINSON
The vessel assemblage from the various DHC trenches totalled 38.22
EVEs (sites DSC and DTT excluded). Jars were, as usual, the principal
vessel class, but they only amounted to one-third of the vessels. Bowls,
dishes and uncertain intermediate bowl-dish types were respectively
19.3 per cent, 6.2 per cent and 10.5 per cent. Almost half the bowls
were in samian and fineware fabrics, emphasising the importance of
these fabrics in the assemblage. The relatively high representation of
flagons and beakers (10 per cent and 11 per cent) also underlines the
importance of the fine and specialist wares in the assemblage, since
these types occurred principally in such fabrics. Cups (3.5 per cent)
were entirely in samian ware (forms 27 and 33) and mortaria (4.0 per
cent) also included a small samian ware component. The low
representation of amphorae and, in particular, the almost total absence
of lids ( despite the fact that they were relatively common products of
the Canterbury kilns (Jenkins 1960, 160) is striking. Unidentified types
only amounted to 1.2 per cent.
Jars occurred principally in reduced and black-burnished fabrics,
with small quantities in 051 (Canterbury oxidised) and ClO (shelltempered)
fabrics. The sandy reduced fabrics (R80, etc.) were the most
important here, but probable jar types, including the only narrowmouthed
jar in the assemblage, were found in the fine 'Upchurch'
fabric R35. Jars were apparently also important in the grog-tempered
fabric R95, but there were insufficient vessels to allow firm
conclusions to be drawn here. Detailed recording of particular jar types
was not possible, but most jars probably fell into a general mediummouthed
category. Specific types noted included 'cooking-pot type'
vessels (particularly important in black-burnished ware) and those with
lid-seated rims. The latter amounted to one-third (expressed as EVEs)
of the rims identified as Canterbury sandy fabric R81 (almost all of
which were jars). In contrast to this less than 7 per cent of the jars
recorded in fabric R80 were of lid-seated type. This distinctly different
pattern probably supports the conclusion (above) that R80 represents
products from a variety of sources and contains only a relatively small
amount of Canterbury material. (Reduced ware lid-seated jars were not,
of course, exclusively produced at Canterbury, they occur, e.g., in the
Thameside industry (Monaghan 1987, 108-11), but the characteristic
form at Dover (cf. e.g. Macpherson-Grant 1982, 136, No. 257) is
typical of Canterbury types).
Jars comprised respectively 28 per cent and 22 per cent of BBl and
BB2 vessels, together amounting to just under 20 per cent of the total
of jars from the site. It is unclear if these slight differences were
significant.
Bowls, dishes and the intermediate bowl-dish types can be discussed
together. A small number of carinated 'bowls' occurred in reduced
102
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
fabrics, the best example (No. 55 below), in fabric R35, presumably a
north Kent product, is not paralleled in Monaghan's corpus but is
broadly of his Class 4H (Monaghan 1987, 129). The great majority of
all bowl, dish and intermediate vessels were straight-sided; curving
sided vessels in these groups were rare, except for samian ware types
(particularly Drag. 37 and 38). Fine ware occurrences of bowl-dish
types were limited, with three examples ( one of uncertain type) in the
mica-dusted fabrics F35 and F36 (see Nos. 45, 49 below) and a single
example probably of type C51 (Young 1977, 160-61) in Oxfordshire
colour coated ware (F51).
The bowl-dish continuum comprised 18 per cent of reduced ware
vessels, most of which were bowls. Confidently identified dishes were
only 0.9 per cent of all reduced wares. The figures for BB 1 and BB2,
however, show contrasting patterns. In BBl the representation of bowls
and dishes was closely comparable, each amounting to a third of all
vessels, and the intermediate category was only 2.2 per cent. In BB2,
bowls constituted c. 12 per cent of the vessels, dishes 21 per cent, and the
indeterminate group amounted to 44.5 per cent (the remaining vessels
being jars). These differences can be explained in part in terms of
chronology and rim typology. The majority of the BB 1 in this
assemblage is of late date (mainly late third-century, see above), by
which time the typological distinction between bowls and dishes is clear,
even when the surviving sherds are small. In BB2, conversely, there is no
such clear-cut distinction between the rims of bowls and dishes (a
problem which would also have applied to BB 1 had its second-century
forms occurred widely). It is impossible therefore to determine what
breakdown of BB2 bowls and dishes was represented by the numerous
indeterminate group. Since the overall representation of bowl-dish types
in BB 1 and BB2 is fairly closely comparable (respectively 72 per cent
and 78 per cent), it may be tentatively suggested that the ratio of
bowls:dishes was also broadly similar, and that the approximately 1: 1
relationship observed for BB 1 also applied to BB2. In any case, with the
exception of three examples (one samian, two reduced wares) all the
dishes in the assemblage were in black-burnished ware.
The remaining important vessel classes require little comment.
Flagons occurred chiefly in white slipped fabrics (50.9 per cent), the
buff-white fabric W40 (22.7 per cent) and the Canterbury oxidised
sandy fabric 051 (26.4 per cent) accounting for the rest. B eakers
(represented by rims) occurred in fabrics F40, F46, F52 and F55, of
which F52 (Nene Valley ware) was most important, and there was a
single example in B20. They were also important in reduced fabrics;
the possible and probable Upchurch products (R30 and R35), the fine
sandy R82 and the north Gaulish sandy ware R89, which together
amounted to 48.7 per cent of all beakers in the assemblage.
103
DAVID R.P. WILKINSON
CATALOGUE AND STRATIGRAPHIC DISCUSSION (Figs. 14-18)
The illustrated vessels (Figs. 14-18) are presented in stratified
sequence within the individual trenches of the excavation. No attempt
has been made to correlate deposits in one trench with those in another
for the purpose of identifying precise ceramic phases.
In all entries the fabric code is followed by that for the vessel type
with the context number last.
Trench 1
34. Fabric F52, type ED. Context 34.
35. Fabric R80, type HB. Context 32.
These vessels occurred in contexts dated by coin evidence to the late third
century (Context 34) and later (Context 32) and are consistent with this
date. No. 35 is in a sandy fabric rather than the grog-tempered or blackburnished
wares which provided most examples of the form at this date.
Trench 11
36. Fabric B20, type JA. Context 416.
37. Fabric B20, type CH. Context 402.
38. Fabric B20, type CM. Context 402.
39. Fabric B20, type CK. Context 406/2.
40. Fabric B20, type HB. Context 406/2.
41. Fabric B20, type IA. Context 406/2.
42. Fabric R35, type EH. Context 401.
43. Fabric M56, type K. Context 409.
44. Fabric R81, type CK. Context 409.
45. Fabric F35, type HC. Context 360.
46. Fabric Q44, type BB. Contexts 350/2, 315 and 334.
47. Fabric M35, type KA. Context 333.
48. Fabric B20, type JA. Context 333.
49. Fabric F35, type IB. Context 364.
50. Fabric M29, type KA. Context 387.
51. Fabric M55, type KA. Context 372.
52. Fabric B20, type IA. Context 372.
53. Fabric 051, type C. Context 357/3.
54. Fabric R80, type JA. Context 355.
55. Fabric R35, type HA. Context 342/1.
56. Fabric R95, type IA. Context 337.
Nos. 36-56, arranged in stratigraphical order, are from a sequence
which, although not completely excavated, spans most of the main
periods of occupation of the Classis Britannica fort.
104
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
)
,, 38
)-
39 ,,
44
, ..
,, \'
-.
43 ... ,
,t
Trench 2
J
7
I'
-
,,
37
, 11' f
42
1, I '\ J
46
Fig. 14. Roman pottery from Trench 1 (Nos. 34-5) and Trench 11 (Nos. 36-46).
105
DAVID R.P. WILKINSON
The source of the mica-dusted vessels Nos. 45 and 49, apparently not
Canterbury, is uncertain. No. 45 is reminiscent of the L ondon fineware
type 14 (Marsh 1978, 148-50) but No. 49 is not closely paralleled
there. The mortaria are in a variety of fabrics, mainly from local
sources (including Canterbury). No. 50, a Colchester product, can be
dated c. A.D. 130-170 on the basis of the stamp (cf. Hartley 1987,
100). The source of the flagon No. 46 may also have been relatively
local. The pulley-wheel rim form is found at Canterbury Castle
(Macpherson-Grant 1982, 148-149, No. 413) in a 'colour-coated' pink
buff sandy ware, and amongst the material from the Whitehall Road
kilns where, however, the parallels are not particularly precise (Jenkins
1960, 157-9, nos. 4 and 5). A very close parallel at Chelmsford is in an
'unspecified buff ware' (Going 1986, 32 and 52, class J2 2.1), but the
type is rare. The remaining coarse wares are mainly Canterbury
products and BB2 of second-century date. No. 56, from the top of the
stratigraphic sequence, seems to be in late Roman grog-tempered ware
and as such was perhaps intrusive, but M. Lyne (pers. comm.) suggests
that this is an East Sussex form for which an early third-century date
would be acceptable. No. 54 may be from the same source and of
similar date. A single small sherd of New Forest purple colour-coated
ware in a fairly coarse, sandy fabric may likewise have been an early
third-century piece, though the rubble deposit (363) from which it
derived is dated to the medieval period.
Trench 11
57. Fabric Ml2, type KD. Context 324.
58. Fabric M12, type KD. Context 324.
59. Fabric MIO, type KD. Context 312.
60. Fabric 061, type Z. Context 312.
These vessels occurred in the fills of Victorian graves which cut the
Roman deposits. They supplement the range of vessels from stratified
contexts. No. 58 is similar to Hartley (1981) No. 383, but No. 57 is not
precisely paralleled among the published Dover material, although its
affinities with No. 58, both in fabric and form, are clear. No. 59, from the
Rhineland, is paralleled, e.g., by Gose (1950) No. 453. All these vessels
belong to the later second to early third century and may be presumed to
have derived from the latest phase of the Classis Britannica fort. No. 60 is
unique in fabric and form and is possibly not of Roman date.
Trench 13
61. Fabric R80, type HA. Context 861.
62. Fabric F46, type EC. Context 829.
63. Fabric F55, type EC. Context 827.
106
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
Trench 11 J
47
-
50
-
I )
t---------<< 53
Fig. 15. Roman pottery from Trench 11 (Nos. 47-55).
107
DAVID R.P. WILKINSON
64. Fabric RSI, type CK. Context 827.
65. Fabric R35, type JB. Context 827.
66. Fabric A22, type A. Context 859.
Nos. 61-5 are from Classis Britannica fort contexts, Nos. 63-5 being
from a salvage-recorded drain and No. 62 from a layer beneath it. No.
66 is from a fill in the top of the fort ditch, which could have been
deposited some time after its disuse.
67. Fabric F52, type E. Context 822.
68. Fabric F52, type E. Context 819.
69. Fabric F52, type ED. Context 850.
70. Fabric M29, type KA. Context 847.
71. Fabric M56, type KA. Context 854.
72. Fabric 051, type BA. Context 821.
73. Fabric R95, type IA. Context 854.
74. Fabric Bll, type CK. Context 849.
75. Fabric Bll , type CK. Context 850.
76. Fabric Bll, type HB. Context 818.
77. Fabric Bll, type HB. Context 848.
78. Fabric B ll, type HB. Context 849.
79. Fabric Bll, type HB. Context 849.
80. Fabric B ll, type JA. Context 848.
81. Fabric B ll, type JA. Context 849.
82. Fabric B ll, type JA. Context 849.
83. Fabric B ll, type JA. Context 849.
84. Fabric B 11, type JA. Context 850.
85. Fabric B20, type HB. Context 848.
86. Fabric B22, type HB. Context 850.
87. Fabric B22, type JA. Context 819.
This group is from component layers of the Saxon Shore fort rampart
and is dated to the last quarter of the third century or slightly later. The
homogeneity of the material and the occurrence of sherds, particularly
of BB I, probably from the same vessels in more than one contexts
(albeit non-joining) justifies its treatment as a single whole. The
rampart contexts together produced 245 sherds (4.10 EVEs). The
principal characteristics of the group were high representations of fine
wares and black-burnished ware in comparison to the overall
assemblage. Samian and oxidised coarse wares were of reduced
importance, both obvious trends in view of the date of this group. The
representation of reduced coarse wares was much the same as the site
average (as a percentage of sherds; it was much lower when expressed
in terms of EVEs).
108
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
Trench 11
.
., ' '
.
.
.
,
. '
'.,, ., . · .,."{ :.. · ./ -;
/ 57
7 ,,
': 65
)
.:\ I
I I I
)
61
-✓
5
58
Trench 13
)
I i'
" 64
.. ,
I I I I
.J\. IA\ . .- 69
,,/,, r\
70
,n
1
1 68 I \
)
Fig, 16. Roman pottery from Trench 11 (Nos, 56-60) and Trench 13 (Nos, 61-70).
109
DAVID R.P. WILKINSON
The fine wares consisted almost exclusively of Nene Valley colourcoated
ware (F52) which totalled 8.6 per cent of the sherds. Samian
ware was now 7.3 per cent. The main reduced ware was still R80. R81
(Canterbury sandy fabric) did not appear at all, but R35 was only
slightly less common than in the overall assemblage, though it is likely
to have been out of production by this time (see above). R95, late
Roman grog-tempered ware, was now a significant component (7.8 per
cent) of the group. BB 1 comprised a quarter of the assemblage, and
was now more than twice as common as BB2. A high proportion of the
BB2 sherds was in the fine fabric B22, but it is uncertain if this is
significant.
The date of the group rests primarily on the Nene Valley ware and
BBL The mortaria and flagon (Nos. 70-72) are likely to have been
residual in this group. This was probably also the case with the BB2,
particularly Nos. 85-6, which are considered to be 'primarily a second
century form' by Monaghan (1987, 144, his type 5D). This conclusion
may be supported by their almost total absence from New Fresh Wharf,
London (Richardson 1986, 127, with discussion of fabrics), but the
comparable type IV.H.4 at Southwark, with a date range of A.D. 140+,
is considered 'probably commoner after 170' (Hammerson 1988, 208).
In any case, Nos. 85-6 are likely to have been residual in later thirdcentury
deposits.
There were four Nene Valley beaker rims, and body sherds from
further vessels including at least four indented examples, three with
large indentations (cf. Howe et al. 1980, 21, Nos. 51 and 52, fourthcentury)
and one with narrow indentations (ibid., No. 53, also fourthcentury).
The character of the barbotine decoration of No. 69 suggests a
third-century date (cf. ibid., Nos. 48-50), but the indented sherds and
all the rim forms seem later. For example the closest parallels for the
rim form of No. 67 (Howe et al. 1980, 21, Nos. 54, 56) are dated to the
fourth century), though a late third-century date may be acceptable for
this vessel.
The BB 1 constitutes a tight, internally consistent group within this
assemblage, of which the most important component is the flanged
bowls (Nos. 76-9), and includes several almost identical examples of
the same types (e.g. Nos. 74-5). The absence from this group of BBl
bowls with 'incipient flanges' is noteworthy. There are some
similarities between this group and that from New Fresh Wharf,
London (Richardson 1986, 124-125), dated not later than c. A.D. 245
(ibid., 96), but the range of forms at the latter site includes vessels of
late second-early third-century date which are absent in the Dover
group. Whether or not there are problems with the stratigraphy at New
Fresh Wharf (ibid., 125), the combination of types seen at Dover
suggests a date after the mid third century rather than earlier.
110
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
Trench
: - ( j ,rr
It I \
72
--.---
t-
---_-_-_ ------) / I (74 \ , . ,,2
( 75
.-E>
' ' 77 17'-7 ,, I I 78
80
82
Fig. 17. Roman pottery from Trench 13 (Nos. 71-82).
111
DAVID R.P. WILKINSON
Evidence from northern Britain supports this (though it is possible that
particular vessel types did not reach all parts of Britain synchronously).
At Vindolanda, Bidwell concluded in discussion of Gillam (1970) type
228 that 'the earliest contexts for bowls with true flanges ... were ... no
earlier than c. 275' (1985, 177). Evidence from other Saxon Shore forts
is lacking. At Portchester, there are no significant groups dated before
A.D. 300 (Fulford 1975a, 46); by which time the absence of BBl bowls
other than fully flanged types is to be expected (Fulford 1975b, 336, type
85). At Lympne, the small assemblage contained jars and dishes in
BBl , but no bowls (Young 1980, 280). Elsewhere in Dover fully-flanged
bowls in BBl are recorded from deposits assigned to Period V II (dated
A.D. 210-70) at the 'Painted House' (Willson 1989, Nos. 217-8).
Illustrated flanged bowls in the Period VIII deposits from the same site
are all in late grog-tempered fabrics, but it may be presumed that
BB 1 examples were also present. Other aspects of the Period VIII groups
from the 'Painted House' closely parallel the evidence outlined above,
as would be expected in groups which were deposited to prepare the site
for the construction of the Saxon Shore fort. These include the BB 1 jars
(e.g. ibid. Nos. 268, 270) and the presence of Nene Valley products
including indented beakers (ibid. Nos. 272-3).
In summary, the forms of the Nene Valley colour-coated vessels, the
range of types in BB 1 and the relatively high representation of late
Roman grog-tempered ware, the first appearance of which in east Kent
is in the late third century (Pollard 1988, 129) combine to suggest a
terminus post quern of at least c. A. D. 275 for the construction of the
Saxon Shore fort rampart. The date of the assemblage could be as late
as the early years of the fourth century but the earlier date is preferred
here (see above, The Date of the Saxon Shore fort).
88. Fabric Q43, type BB. Context 842.
This vessel is from a post-Roman context. The form occurs in the
material associated with the Dane John kiln at Canterbury (Kirkman
1940, 129-130, no. 53, a precise parallel) and at Canterbury Castle
(e.g. Macpherson-Grant 1982, 144-145, no. 373, in 'buff-pink ware' in
a probable late second-century context (ibid., 149)).
Trench 15
89. Fabric F46, type EC. Context 924.
90. Fabric F55, type EC. Context 924.
91. Fabric R81, type C. Context 924.
92. Fabric R81, type C. Context 924.
93. Fabric R82, type C? Context 924.
112
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
Trench 13
].,
Trench 15 w.
,, I \
..
,,
?:,;;:·;-:, 89
II
\
91
,, \93
Fig. 18. Roman pottery from Trench 13 (Nos. 83-88) and Trench 15 (Nos. 89-93).
113
DAVID R.P. WILKINSON
This group, from a small salvage-recorded area near Trench 13, is from
a layer beneath a wall of a probable Classis Britannica fort building.
The date of the group, which includes?Cologne and Colchester colourcoated
vessels (Nos. 89-90) and Canterbury sandy reduced wares (Nos.
91-2) is entirely consistent with this and probably lies in the second or
third quarter of the second century.
ACKNOWLEDGEMENT
The help of Richard Pollard in identifying certain sherds and discussing
wider aspects of Roman pottery in Kent is gratefully acknowledged. He
is not, however, in any way responsible for the deficiencies of this
report.
THE ANGLO-SAXON, MEDIEVAL AND POST-MEDIEVAL POTTERY
Catherine Underwood-Keevill
A small assemblage of Anglo-Saxon, medieval and post-medieval
pottery was recovered from Trenches 11, 12, 13 and 14 and from small
assessment trenches (including Trench 1). All of the pottery was
considered together initially to try to establish a preliminary fabric and
form series for medieval and post-medieval Dover.
A total of 270 sherds weighing 3.76 kg was recovered from stratified
contexts in Trenches 11, 12, 13 and 14. The material was grouped into
fabric types on the basis of the identification of the predominant rock
and mineral inclusions within the clay body. Fabric types were defined
according to paste colour, hardness, texture and fracture as well as
inclusion type. Details of inclusions are recorded according to the
Peacock system (Peacock 1977b). A type series of typical sherds has
been retained for reference purposes and cross-referenced, where
possible, to equivalent fabric types recovered in Canterbury.63
All the sherds were divided into groups according to their main
inclusion type as recognised by macroscopic and microscopic analysis
with reference to the Peacock guidelines. These were assigned a
numerical code: Group 1, sandy wares include 1, 10-19, 100-199;
Group 2, shelly and limestone gritted wares 2, 20-29, 200-299, etc. It
should be noted that not all the numbers have been used within each
63 All sherd counts, weights and drawings of other diagnostic sherds have been retained
with all other Level II archive material at Oxford Archaeological Unit, to be transferred
to Dover Museum.
114
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
group. Inevitably with such a small assemblage, some fabrics are
represented by very few sherds, including only one example in several
cases. The later post-medieval material has been assigned to Group 4,
tin-glazed wares 4, 40-49; Group 5, salt-glazed and stonewares 5,
50-59; Group 6, red earthenwares 6, 60-69, 600-699 and Group 7,
factory produced wares 7, 70-79. For the very fine sandy wares in
Group 1 and for the red earthenwares in Group 6 a x20 microscope was
used to aid the identification of fabric types. All fabric types were
counted and weighed for each context. Rim and base types and
decorated body sherds in each fabric were noted in each context and
each new type drawn for reference purposes and retained in archive.
DISCUSSION
The majority of the medieval pottery was recovered from Trench 12,
129 sherds weighing 1.9 kg. The main fabric type was fabric type 1
which comprised 68 per cent of the assemblage from this trench by
number and 61 per cent by weight. The imported, mainly Northern
French, wares made up 14 per cent by number and 18 per cent by
weight. The other major types were the shelly wares forming 18-19 per
cent by number and weight, with the predominant type being the soft
shelly ware, fabric type 20.
The majority of the material could be dated to the late eleventh to
twelfth century on the basis of the local sandy wares, fabric type 1.
Contexts such as 536, due to the presence of a twelfth- to thirteenthcentury
cooking bowl form and a small sherd of Aardenburg type ware,
fabric type 15, may be thirteenth-century in date. It should be noted
that several contexts have only one sherd in them and the more
productive contexts such as Contexts 520, 544 and 549 have 24, 16 and
15 sherds in them, respectively. It is noticeable that these contexts also
contain the larger quantities of imported material which date to the
eleventh and twelfth centuries.
The Ipswich ware sherd was recovered from Context 546, which
otherwise has eleventh- to twelfth-century fabric type 1 sherds. Other
contexts have earlier material in them, such as fabric type 22, a late
eighth- to ninth-century coarse shelly rim in Context 520 and fabric
type 25, a ninth- to tenth-century fabric in Context 541.
Trench 11 has a total of 102 sherds weighing 1.3 kg. The only
context from this trench with a medieval assemblage is Context 363.
This context also has the largest pottery assemblage by number and
weight of all the contexts in this trench. The only other contexts with
large amounts are Contexts 312 and 315. Context 312 has some
residual medieval and early post-medieval fabrics such as fabric
type 12, but has mainly seventeenth- to eighteenth-century red
earthenwares and seventeenth-century Surrey white wares. Context 315
115
DAVID R.P. WILKINSON
is very similar in composition, but it is notable for a possibly late
medieval sherd, fabric type 103 which is unusual in fabric and
decoration. Context 363 is dominated by fabric type 1, but also has one
sherd of fabric type 119, Normandy Black ware and a residual eighthto
ninth-century sherd, fabric type 26.
Trench 13 contexts are mainly grave fills and the ceramic
assemblages tend to reflect this with mainly seventeenth- to eighteenthcentury
earthenwares being present. The numbers of sherds, however,
is extremely limited, with a total collection of 36 sherds weighing 0.4
kg. Trench 14 is also very limited with 3 sherds weighing 21 gr.
recovered from Contexts 902 and 908. The two sherds from Context
908 are body sherds in a dense coarse sandy version of fabric type 1.
The sherd from context 902 is a sixth- to seventh-century rim sherd.
This assemblage indicates strong affiliations with fabric types from
Canterbury, with some early middle-Saxon wares which were unknown
as far afield before. The dating from the sandy ware fabric, fabric
type 1 and of the Northern French imports tends to suggest that the
assemblage is predominantly late eleventh- to twelfth-century in date,
especially the main medieval contexts from Trench 12. Occasional
cooking bowl rims from Contexts 535 and 536 suggest some late
twelfth- to early thirteenth-century material. Very little thirteenth- to
fourteenth-century material is present, with the exception of the few
Aardenburg-type sherds.
The main imported wares appear to come from Northern France and
Flanders, although the small quantity of material available may be
misleading. It does suggest that, like Exeter and Southampton, there
was a strong commercial link with northern France. Imported fabrics
occur in Exeter in the tenth to eleventh century, but most of the
stratified finds are in deposits dated to the twelfth century. In
Southampton, some of the richest groups are attributed to mid to late
twelfth century, and this trade expanded during the thirteenth century.
The small amounts of imported material and the lack of diagnostic
sherds make it difficult to ascertain with any degree of certainty the
main period of trading activity from this assemblage. Other imports
published from Dover, such as Hamwih type 15, grey wares from
Flanders of eleventh- to twelfth-century date (Hodges 1977, 250), and a
Normandy pitcher of late eleventh-century (Dunning 1945) could
support the suggestion that this was the main period of trading activity
in Dover.
There is, however, a small amount of residual mid-Saxon material
especially in Contexts 363 and 520. The one sherd that may be in situ is
a sixth- to seventh-century sherd (fabric type 13) from Context 902,
Trench 15. The Ipswich ware sherd from Context 546 is residual, but
adds greatly to our knowledge of the distribution of Ipswich ware
116
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
especially within Dover since only a few examples have been recorded
from the town previously.
A brief description of identified fabric types, the main form types for
each fabric, their possible derivation and equivalents in Canterbury and
elsewhere are listed below. Illustrated examples (Nos. 94-109) appear
on Figure 19-20.
Sandy wares
Local: Group 1, Fabric types 1-12
Fabric type 1
This is the most common fabric type in the medieval contexts. The
fabric is highly variable in colour from orange-red to grey surfaces and
margins, with a dense sandy matrix consisting of common-abundant,
clear sub-angular quartz and occasional coarse haematite and ironstone
inclusions. The fabric is equivalent to Canterbury type EM l , dating
from the eleventh to the twelfth century and is probably a product of
the Tyler Hill kilns (Blackmore 1988). The forms are mostly limited to
bevelled everted-rim cooking-pots with sagging bases, with knife
trimming around the base (Nos. 99-102) and a few examples of widemouthed
cooking bowls similar to No. 103.
Fabric type 10
Dense sandy fabric with sub-angular common, medium-fine white quartz,
occasional medium-coarse red-brown grog inclusions and fine red sand.
Burnished exterior, rough interior. Late eighth- to mid ninth-century
Canterbury local sandy ware (N. MacPherson-Grant, pers. comm.). This
fabric was represented by one sherd from an unstratified context.
Fabric type 13
This is a compact very dense sand-tempered fabric with clear mediumfine
quartz and very occasional white chalk in a dark red-brown clay
with burnished brown-black surfaces. Only one rim sherd of this fabric
type exists in Context 902 (No. 95). The fabric is equivalent to
Canterbury local sandy ware of the sixth to seventh century
(N. Macpherson-Grant, pers. comm.).
Fabric type 11
This fabric type is represented by two very small sherds and, therefore,
cannot be confidently attributed to a source. The fabric has common
clear angular white quartz inclusions, and a grey colour with whitegrey
margins under a thick olive-green glaze. It is possibly a
Canterbury local fabric.
117
Ill
I I
'
I I
'
I I
DAVID R.P. WILKINSON
'· .-
--
. --
-=·· 98
I I
07.,
Fig. 19. Early medieval and medieval pottery (Nos. 94-102).
118
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
Fabric type 12
This is a fine sandy fabric with very fine quartz and occasional red iron
ore in a soft light orange-pink clay. The decoration consists of a very
distinctive bright green, frequently patchy glaze. Rim forms are
infrequent but indicate open bowl forms and pancheons. This is
probably a late medieval to post-medieval fabric and could possibly be
a south coast imitation of Hants.-Surrey Border ware, a fourteenth- to
fifteenth-century product.
Sandy Wares
Non-local: Group 1, Fabric types 14, 16, 103-119
Most of the imported material is limited in the number of sherds
represented in each fabric type group. There is also a distinct lack of
diagnostic sherds such as rims and bases to establish any corpus of
dated vessel types.
Fabric type 14, Ipswich Ware
A soft, light grey fabric with pink/grey core. Inclusions are very fine,
with clear mica the only visible inclusion type especially on the
surfaces. Rough fingering and wiping on the interior indicate a very
pliable, plastic clay with ridges and indentations very apparent.
Decoration on this example consists of punched decoration of a series
of eight dots laid out in a roughly rectangular motif set diagonally in a
line across the vessel. Burnishing is also evident below the line of
stamps. This example (No. 94) was from Context 546, in Trench 12.
Ipswich ware is dated to c. A.D. 650-850, but in Canterbury it is
mainly found in eighth-century contexts. Other examples of Ipswich
ware have been recovered in Dover from Yewden Court (Dunning
1957).
Fabric type 16
This fabric type is a very hard grey ware, with dark red-brown core and
very fine abundant sub-round clear quartz and moderate honeycoloured
quartz which gives a rough granular feel. The body sherds are
of an even thickness with finger dimpling especially on the interiors
indicating a hand-made construction. The decoration comprises fine
combed detail as illustrated by No. 97, Context 847 and simple vertical
combing as in an example from Context 312 (not illustrated). An outturned
elongated bead rim with linear punched-dot decoration in the
same hard dark grey fabric from Context 837 is the only other example
of this fabric (No. 95). It is possibly imported northern French/Flemish
(Hodges 1977, 249-52), on the basis of its fabric type, although there is
as yet no parallel for the decoration.
119
l- ,,
104
DAVID R.P. WILKINSON
I- ,,
rl'-W...
107
Colour key (No. 109)
II Pale blue
ll2] Yellow ochre
§ Mid grey
o==--==--=::=115
_____ ....,
10 cm. (1&2)
o=-==--==-----
1o
t::::::=====-----
20 cm. (3-16)
Fig. 20. Medieval and post-medieval pottery (Nos. 103-109).
120
EXCAVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
Fabric type 103
Only one sherd of this fabric has been isolated from Trench 11, Context
315. The fabric is dark orange to red in colour, very hard with very fine
quartz and occasional medium ill-sorted white limestone. The
decoration type consists of combed oval motifs with a thick green
glazed exterior and light orange-brown partial dripped glaze interior.
The form suggested is of a high shouldered angular pitcher, imitating
metal prototypes (No. 106). Possibly late-medieval Dutch by
comparison with material from Norwich (Jennings 1981, 32).
Fabric type 114 and fabric type 115, Aardenburg type ware
A very hard, light orange coloured fabric with fine-medium wellrounded
white quartz with a white slip and clear yellow-green glaze.
Fabric type 115 is softer with a sandier texture with moderate fine
rounded white quartz and a white underslip and green-orange glaze.
Two sherds came from topsoil, Context 301, and one sherd from
Context 536, Trench 12. This fabric dates from c. A.D. 1250-1400 and
in the assemblage from Dover Castle was dated to c. A.D. 1300 or later
(Cook et al. 1969, 97-8).
Fabric type 105
Only two sherds of this fabric type could be observed, in Context 549,
Trench 12. The fabric is white to pink-white with visible red-orange
tinged quartz, 0.25-0.5 mm. in size. A bright yellow to light green
glaze is also evident as is prominent rilling or wheel-marks. This fabric
is similar to the Normandy gritty ware fabric type 113, but the sand
grits are not so large. This may be the same fabric type described as
Normandy gritty glazed ware (14), which is dated normally from the
twelfth to the thirteenth century but can date back to the eleventh
century (Hodges and Mainman 1984, 14).
Fabric type 112
Northern French red-painted ware
Only two sherds of this fabric type were present, from Context 520 in
Trench 12. The fabric is very fine, soft and light pink to light orange in
colour with very fine sub-angular quartz and black very fine possibly
dark mica inclusions. The decoration consisted of a light red-brown slip
wash with thicker red diagonal lines painted over it. Beauvais type
ware has been dated to the twelfth to thirteenth century (Barton 1970).
Fabric type 113, Normandy Gritty Ware
Only two sherds of this fabric type were identified: one from Context
520, Trench 12, the other from topsoil Context 301. The fabric is buff
to light pink, with prominent red-orange sub-angular quartz grits. This
121
DAVID R.P. WILKINSON
is dated as twelfth- to thirteenth-century, but can date back to the
eleventh century (Hodges and Mainman 1984, 14).
Fabric type 117, Northern French Blackware
This fabric type has dark grey to black surfaces with a dark red-brown
core with moderate well-rounded, fine, ill-sorted clear quartz and
coarser (1-2 mm.) red-stained quartz. Only body sherds were recovered
from Context 261, an assessment context and Context 544, Trench 12.
This is possibly dated to the eighth to tenth century (N. MacPhersonGrant,
pers. comm.).
Fabric type 118, Northern French Grey Sandy Ware
A hard, dark grey fabric with very fine clear sub-angular quartz and sparse
very fine white limestone inclusions. The decoration is limited to rouletted
bands and occasional roulette impressions on the main body of the
vessel (Nos. 104-5). Three examples of this fabric type were recovered
from unstratified contexts in Trench 13. This may be comparable with
fabric and decoration types at Exeter attributed to the Loire Valley and of
probable eleventh-century date (Hodges and Mainman, 1984, 16).
Fabric type 119, Normandy Black Ware
This is a cream-buff fine sandy fabric with very fine clear and honey
coloured, sub-angular quartz and black rounded inclusions, and dark
grey surfaces. The exterior surfaces are partially knife trimmed.
Limestone and Shelly wares: Group 2
Fabric type 2
Limestone and sand-gritted ware, with common rounded ill-sorted
limestone, moderate angular grey flint and common red-brown wellrounded
quartz. The fabric varies in colour from dark grey to pinkorange
to light grey. A similar fabric type has been recognised at
Folkestone (N. MacPherson-Grant, pers. comm.).
Fabric 20, Soft Shelly Ware
This is similar to St. Neots-type ware, with dense medium to coarse
plate-like angular ill-sorted shell in a soft pink-orange to grey clay. The
one rim observed is of a cupped everted rim cooking pot form (No. 98).
Fabric type 21, Shell-dusted sandy ware
This fabric has been recovered and described in Canterbury, type coded
EM/Ml (Blackmore 1988, 252; and by comparison with Canterbury
reference series in the British Museum). It has abundant clear to red-
122
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
brown quartz and occasional sub-rounded coarse shell on the surfaces.
This fabric type is dated to the late twelfth to early thirteenth century in
Canterbury (ibid.; N. MacPherson-Grant, pers. comm.). The only
examples of this fabric are represented by a base sherd and rim sherd of
a cooking bowl (No. 103) from topsoil, Context 301. Cooking bowl
rims of a similar type also occur in the sandy fabric, fabric type 1.
Fabric type 22, Coarse Shell-tempered Ware
A soft grey fabric with very coarse 2-3 mm. shell additions. The shell
appears to be roughly ground-down cockle shell. This fabric is
equivalent to Canterbury type MS4, a middle Saxon fabric type dated
to the late eighth to the mid ninth century (N. MacPherson-Grant, pers.
comm.). The only example of this fabric type is a poorly-preserved rim
sherd from Context 520, Trench 12.
Fabric type 23, South Coast Shelly-type Ware
A very lightweight fabric with sub-rounded shell and shell voids
apparent in a light grey to light pink clay with occasional grey flint and
moderate amounts of sub-angular clear quartz. One body sherd was
recovered from Context 363. This type has been dated to the eleventh
to twelfth century in Canterbury (N. MacPherson-Grant, pers. comm.).
Fabric type 24, North Kent Shelly Ware
This has dense moderate-sized (less than 1 mm.) shell inclusions with
coarse shell additions. The vessel type apparent in the assemblage is
similar to No. 99, a long everted rim cooking pot. The sagging base
appears to be added onto the vessel body. Two sherds are evident from
an unstratified layer from Trench 13.
Fabric type 25
A dark grey, gritty fabric with well-rounded moderate red-white quartz,
rounded voids indicating shell/limestone inclusions and very fine
mica/clear quartz on the surfaces. Only one base sherd from Context
541 was recovered in this fabric. It is dated to the ninth to tenth century
and possibly the eleventh century (N. MacPherson-Grant, pers. comm.).
Fabric type 26
A sandy fabric with moderate to sparse shell. A fine fabric with soapy
texture, clear fine common quartz and ill-sorted fine to 1 mm. sized
shell in a red-brown clay with black surfaces. Fine burnishing is
apparent on the exterior and haphazard burnishing/smoothing on the
interior. Only one sherd is evident from Context 363, Trench 11.
Canter bury fabric, MS5, dated to the late eighth century to the mid
ninth century (N. MacPherson-Grant, pers. comm.).
123
DAVID R.P. WILKINSON
Post-medieval Fabrics
Fabric type 100, Surrey Yellow-glazed White Wares
White, fine sandy fabric with clear yellow glaze. Only small-sized
examples were apparent from Contexts 841, 842, 312 and 315. The rim
sherd from 841 is similar to a cupped everted rim type from Dover
Castle from a pipkin, and a tripod foot from 312 indicates the presence
of tripod pipkin forms. At Dover Castle this type was dated to the
seventeenth century (Mynard 1969, 38).
Fabric type 101, Tudor Green type
Only very small sherds survive. The fabric is a fine white to pink with
fine pink stained quartz and a thick green glaze. Further material from
Dover may indicate the range of vessel forms.
Tin-glazed wares: Group 4
Fabric type 4
A buff-pink fabric with red iron ore stains. The only example of this
fabric is a semi-complete profile of a shallow bowl with a light blue band
around the rim and painted black lines with yellow infills from Context
301 (No. 109). This is a probable import from Liguria in north-west Italy,
and dates to the seventeenth century (Hurst et al. 1986, 26-9; Blake 1981,
114). Other examples have been recovered in Dover and in Canterbury,
and decorated examples with painted scenes are known from London,
Southampton and Exeter (J. Cotter, pers. comm.; Hurst et al. 1986, 26).
Fabric type 40
Blue and white English printed; tin-glazed, dated to the eighteenth
century. One example from Context 313.
Fabric type 41
Netherlands Maiolica? A light pink fabric with very fine black
inclusions. Yellow and blue painted pattern, from a backfill context.
Salt-glazed stoneware: Group 5
Fabric type 5
A light grey fabric with fine dark grey margins and dull, matt, light
brown glaze. The only example is a handle and body wall sherd from a
small drinking jug, the handle having an 's' shaped profile, from
Context 313 (No. 108). This may possibly be Beauvais stoneware,
124
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
similar to a small jug found at Norwich (Jennings 1981, 75-6). The
fineness of the sherd may date it to the fifteenth century.
All the other stoneware types consist of seventeenth- to eighteenthcentury
English stonewares mainly from Context 301.
Red Earthenwares: Group 6
This group forms the majority of the post-medieval wares. The fabric
types have been divided on colour, glaze types and any inclusions that
may be visible under a xlO eyeglass. The main type is fabric 65, dark
red with moderate fine sand and a dark orange green glaze on the interior
surface and a splashed metallic brown glaze on the exterior. This is
used in the manufacture of large bowls and pans. Grittier versions of this
fabric exist, such as fabric 600. Fabric type 68, a dark orange-red,
micaceous fabric with a yellow 'fem' type motif under a dark orange
glaze from Context 842 may be a seventeenth-century Metropolitan
slipware product but only two sherds exist so it is difficult to attribute at
present. A collection of lead-glazed earthenware from Dover Castle
(Mynard 1969, 42-5) indicates that the red earthenwares are dated from
the seventeenth century onwards, and the majority probably derive
from the Wrotham potteries in north Kent. The earthenwares were mostly
recovered from Contexts 312, 313, 315, 324, 328 and 301 in Trench 11
and 831, 832, 833, 837, 839 and 842 in Trench 13.
The factory-produced nineteenth- and twentieth-century wares,
Group 7, formed only a small proportion of the assemblage, mainly
from grave contexts in Trench 13. These have not been described in
detail, and the descriptions and sherd counts are in archive only.
Catalogue of Illustrated Examples
94. Decorated body sherd, fabric type 14, Ipswich ware. Context
546.
95. Rim sherd, simple rim with burnished flat top and burnished
exterior, fabric type 13. Context 902.
96. Rim sherd, slightly out-turned rim with punched dot decoration
in a single line on the shoulder, fabric type 16. Context 837.
97. Decorated body sherd, combed diagonal incised lines, fabric
type 16. Context 847.
98. Rim sherd, cupped rim, fabric type 20. Context 520.
99. Rim sherd, long everted cooking pot rim with rounded rim top,
fabric type 1. Context 363.
100. Rim sherd, everted bevelled cooking pot rim, fabric type 1.
Context 363.
101. Rim sherd, everted bevelled rim, fabric type 1. Context 363.
102. Rim sherd, everted bevelled rim, fabric type 1. Context 363.
125
DAVID R.P. WILKINSON
103. Rim and base sherds, flat topped rim and sagging base of
cooking bowl, fabric type 21. Context 301.
I 04. Decorated body sherd, rouletted double band decoration, fabric
type 118. Unstratified.
105. Decorated body sherd, rouletted single band and occasional
single roulette stamps on the rest of the main body of the sherd,
fabric type 118. Unstratified.
106. Decorated body sherd, combed oval patterns under a thick
yellow-green glaze. Partial orange-yellow glaze on the interior.
Fabric type 103. Context 315.
107. Hollow handle. Soft red sandy fabric, possibly north-west
French. No exact parallels have been noted. Context 503.
108. Handle, small drinking jug. 'S' shaped section to handle.
Stoneware with matt red-brown glaze, fabric type 5. Context 313.
109. Profile of shallow bowl. Tin glazed with light blue glaze and
painted yellow band within black-painted lines, blue and yellow
glaze on exterior. Ligurian, Genoa. Fabric type 4. Context 301.
ACKNOWLEDGEMENTS
I would very much like to express my gratitude to Beverley Nenk at the
British Museum for her help with the Medieval Pottery Reference
Collection, and to Nigel Macpherson-Grant and John Cotter of the
Canterbury Archaeological Trust for providing information on fabric
types recovered in Canterbury and elsewhere in Kent. Their help was
invaluable. Any mistakes or conclusions drawn are entirely the
responsibility of the author.
No. 110
THE WORKED BONE AND STONE
Catherine Underwood-Keevill
Top half of bone pin, with spherical head. Length: 34 mm. shaft.
Diameter: 1 mm., thickening at top to 2 mm., and top diameter
measuring 6 mm. Crummy (1979) Type 3, dated c. A.D. 200 - late
fourth century. Small Find No. 1. Location: Trench 1, Context 16.
No. 111
Incomplete bone pin. Tapering shaft with point and top missing.
Length: 71 mm. Diameter: at top - 3 mm., at bottom - 1.5 mm.
Location: Trench 1, Context 28.
126
EXC AVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
No. 112
Fragment of polished bone, possibly bird bone.
Length: 56 mm. Small Find No. 21. Location: Trench 1, Context 30.
No. 113
Bone fastener/toggle. Tapering bone pin with wedge-shaped cut in the
centre. Hand-finished rough surface, with striations from attempts to
smooth the surface. Flattened sides to the top half of the pin and
angular sawn top with three irregular surfaces. Length: 38 mm.
Diameter: at top 4 mm., at point 2 mm. Small Find No. 43. Location:
Trench 11, Context 333.
No. 114
Incomplete bone pin. Slightly tapering bone pin shaft with both top and
point broken off. Length: 51 mm. Diameter: at top 3 mm, at bottom
2 mm. Small Find No. 82. Location: Trench 11, Context 387.
No. 115
Bone pin. Probably re-used from a larger original. Flat sawn top and
blunt point. Length: 31 mm. Diameter: at top 3 mm., at point 3 mm.
Small Find No. 93. Location: Trench 11, Context 409.
No. 116
SF 95. Bone pin. Complete pin with three horizontal grooves beneath a
conical head. The point is made by slicing a small diagonal section
from one side of the bone. Length: 61 mm. Diameter: at top 3.5 mm., at
point 2 mm. Crummy (1979) Type 2, dated c. A.D. 50-200/250. Small
Find No. 95. Location: Trench 11, Context 413.
No. 117
SF 98. Bone pendant? Incomplete boars tooth with large hole drilled
through one end; larger diameter hole one side than the other. The
larger hole is poorly executed with stress fractures visible. Comparable
to bone tooth pendant from Fishbourne: Fig. 67, No. 13 (Cunliffe
1971). Small Find No. 98. Location: Trench 11, Context 409.
Stone Objects
No. 118
Conical shaped loom weight with large drilled hole and flat base. Made
from chalk with uneven pitted sides. Diameter: of hole 21 mm., of top
35 mm., of base 49 mm. Height: 31 mm.
Small Find No. 32. Location: Trench 2, Context 57.
127
DAVID R.P. WILKINSON
A BONE CHESS PCECE
David R.P. Wilkinson (with Kay Proctor)
No. 119 (Plate II)
Bone chess-piece (knight). Cylindrical and made hollow by removal of
cancellous tissue; probably from a horse metatarsus. Highly-polished
and decorated with three sets of five vertical incised Lines. The sets are
evenly distributed around the cylinder. A single inverted triangle
projects from the top and is asymmetrically placed between two sets of
Lines; it bears two ring-and-dot eyes to form a stylised head . The Dover
piece is unfortunately not from a stratified context, but the style is that
of the earliest (eleventh- to twelfth-century) chess-pieces known from
this country (MacGregor 1985, 137-8). Single-headed pieces are
identified as knights, and the centre of the piece may have originally
been plugged, as was the case with other more elaborate examples,
such as that from Helpstone, Northamptonshire (ibid., Fig. 73a).
Small Find No. 183. Location: unstratified.
(The authors are grateful to Arthur MacGregor for examining and
commenting on the above piece.)
PLATE II
= CD
-
-
= IO
Eleventh- to twelfth-century chess piece (knight) from the White Cliffs Experience site,
Dover (No. 119).
128
EXCAVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOV ER, 1988-91
THE ROMAN TILE
Leigh Allen
INTRODUCTION
All the Roman tile from the excavations at the White Cliffs Experience
site was recovered. In the final instance it was decided not to examine
fragments found in backfill from the 1970s excavations, as it was
unclear whether these had been previously analysed. The remaining tile
weighed a total of 69.094 kg., the bulk of which came from the tip layers
in the rampart of the Saxon Shore fort (particularly 835), and to a lesser
extent from contexts relating to the Classis Britannica fort itself. A very
small proportion (4.6 per cent) was from medieval contexts.
METHODOLOGY
Each fragment of tile was examined macroscopically with a x l 0
magnification hand lens to identify the fabric. The width, length and
thickness of these tiles were recorded where a complete dimension
existed; all the fragments without a measurable dimension were
grouped according to fabric under a miscellaneous heading. Individual
characteristics such as tegulae flanges, the curve on imbrices, box fluetile
keys and any graffiti, animal marks or signatures were recorded
separately. Twenty-four fragments of tile bore traces of Classis
Britannica stamps (including one complete example) and these are
discussed below.
TILE TYPES
Seven different types of tile were identified. The tegulae were
distinguishable either by a flange, a groove at the base of the flange or
a semi-circular signature at the top end of the tile; they made up
15.5 per cent of the assemblage. The average thickness of these tiles
was 20-26 mm. Tegulae flange heights varied from 46-55 mm. and
were of a standard form, the most common type being either squareheaded
or gently sloping on the inside edge. The tegulae were
manufactured predominantly in Fabric 1 - two examples of flanges in
F
_
abric 5 were noticeably taller and more slender, possibly denoting
either a separate production site or that Fabric 5 was an easier medium
with which to work.
129
DAVID R.P. WILKINSON
The imbrices were distinguishable by their curve and made up
12.1 per cent of the sample; their average thickness lay between 16 and
22 mm. The roof-tiles made up the second largest group in the
assemblage.
The box flue-tile or tubuli fragments were distinguishable by the
presence of a key for plaster or remains of perforated side panels
through which hot air could flow. These tiles made up only 5.5 per cent
of the total assemblage. They had an average thickness of 17-21 mm.
and were predominantly manufactured in Fabric 5. One example was
recovered with a complete width, measuring 156 x 19 mm.; it is similar
to the Type 3 box flue found at Dover on the 'Painted House' site
(Willson 1989, 10 1). The keys were all simple comb patterns being
either straight lines or curving designs made with combs with more
than five teeth.
The fourth group consists of all those fragments of tile that showed
none of the above characteristics, and was the largest group making up
36.4 per cent of the total assemblage. The range of thicknesses lies
between 15 and 39 mm. and this may include fragments of tiles from
other groups that could not be distinguished as such. Given that the
largest thickness measured so far is 26 mm., it is possible that all
fragments from this group with a greater thickness are floor or bonding
tiles. Unfortunately, only one fragment from this group had a
measurable dimension other than thickness - at 300 x 38 mm. it is
possibly an example of a lydion (average measurements being 403 x
280 x 4 1 mm.) often used in lacing- and bonding-courses. Brick
bonding-courses are known to exist in a bastion in the south wall of the
Saxon Shore fort (Philp 1984, 190).
The fifth group consists of large bricks with a thickness ranging from
40-48 mm. These tiles represent 9 per cent of the total sample (this is
in fact only ten fragments). Other than thickness there is an absence of
complete dimensions and it is, therefore, impossible to say more about
the tile types except that due to their sheer weight they would have
been floor- or bonding-tiles possibly lydia, pedales or sesquipedales.
The sixth group consists of three examples of mammae. These have
been placed in a separate group due to the lack of measurable thickness
for the tiles. The mammae are complete and they are all shallow and
circular with diameters of 37, 5 1 and 70 mm. Only one of the three
examples is actually attached to a fragment of tile. Mammae are either
used on floor-tiles to act as a key for the mortar, or to divide stacked
tile in the kiln (these are later chipped off). In this case the former
explanation is more plausible, as it would seem more likely that
mammae were removed on the production site.
Group 7 is the miscellaneous category and comprises 21 per cent of
the total assemblage.
130
EXCAVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DOVER, 1988-91
FABRICS
Examination of the tiles with a xlO magnification hand lens revealed
four distinct fabric types. Fabric 1 is predominant, comprising 88 per
cent of the total sample. In general, it is a reddish-pink fabric with a
variable degree of streaks and swirls of white clay. Inclusions consist of
abundant black- or red-brown iron ore, mostly less than 1 mm. across
but with occasional fragments up to 2 mm. across. Also present were
very fine quartz and mica inclusions and occasional voids. This fabric
corresponds in general to Peacock's Fabric 2 which is believed to
originate from the southern central Weald (Peacock 1977a).
Fabric 2 makes up 6 per cent of the sample. It is a red-orange fabric
characterised by very abundant tiny fragments of quartz and mica with
very occasional larger fragments of red-brown iron ore.
Fabric 3 makes up 5 per cent of the sample. It is red-orange in colour
and is characterised by very large inclusions including grog and flint
with some fragments measuring up to 5 mm. across. There is also a
very small quantity of iron ore, quartz and mica.
Fabric 4 makes up 1 per cent of the sample. It is a buff-coloured fabric
with common spherical lumps of red-brown iron ore measuring up to
1 mm. across and occasional larger fragments of quartz up to 2 mm. across.
STAMPED TILES
Twenty-four fragments of tile had stamps of the Classis Britannica type.
Rubbings were taken of each one in an attempt to find parallels.
One complete stamp was recovered, while 9 examples were too
fragmentary to identify. The stamps appear exclusively on imbrices and
floor-tiles. Twenty-three of the 24 examples were on tiles manufactured in
Fabric 1, the other being Fabric 5. No matching dies were found amongst
the sample. The stamps were compared with those from G. Brodribb's
(1980) work on stamped tiles of the Classis Britannica and W. Williams's
(1981) work on the stamped tiles from the Dover forts. Exact parallels
were found for all the stamps, and they have therefore not been illustrated
here. They fall into six distinct groups as follows:
Type 1 (one example only)
Brodribb Type 33; Williams Type K4. Complete examples contain the
letters CLBR in an oblong frame with squared corners. This example
includes the letters BR only. The B has serifs top and bottom. The R
has a big foot and serifs at top and bottom. The length is not known.
Width23 mm.
131
DAVID R.P. WILKINSON
Type 2 (five examples, all fragmentary)
Brodribb Type 2; Williams Type F l . Complete examples contain the
letters CLBR in an oblong frame with square corners. The letters are
crude and unequally spaced. The C is plain and the L is heeled. The B
and R have serifs at the top and the R leans back slightly. Width
27mm.
Type 3 (7 examples, all fragmentary)
Brodribb Type lC; Williams Type Hl. Complete examples contain the
letters CLBR set in an oblong frame with square corners. The letters
are large, simple and plain - without serifs. One feature of this type is
an irregularity in the letter B. In one case it has a bulge along its back.
Width 31 mm.
Type 4 (one complete example)
Brodribb no parallel; Williams Type G2.
Complete examples contain the letters CLAB in an octagonal frame.
The C and the L are attached at the top and bottom. The A is uncrossed
and the B is open at its top and bottom. Length 45 mm. Width 27 mm.
Type 5 ( one fragmentary example)
Brodribb Type 15; Williams Type E2. Complete examples contain the
letters CLBR is an oblong frame with rounded ends. The letters are
simple. The B and R are inverted. The R uses the straight back of the B
and touches the stamp frame at the top and bottom. Width 26 mm.
Type 6 ( one fragmentary example)
Brodribb Type 28; Williams Type E l . Complete examples contain the
letters CLABRI set in an oblong frame with rounded corners. The
letters are close together. In this example the L has a short foot, the A is
uncrossed and joins the back of the B. No measurement possible.
CONCLUSION
The evidence suggests that the bulk of the tiles from the White Cliffs
Experience site originated from the tile workshops of the southerncentral
Weald. All types of tile were being manufactured in this fabric.
To a lesser extent tiles were also being supplied by other smaller
sources. There is no evidence to suggest the use of French material.
The majority of the tiles are either roof- or floor-tiles with very few
fragments of box flue-tiles. This agrees generally with the evidence
from the 1970s excavations on the Classis Britannica fort (Williams
1981).
132
EXCAVATIONS ON THE WHITE CLIFFS EXPERIENCE SITE, DO VER, 1988-91
THE SEDIMENTOLOGY
Anthony J. Barham and Martin R. Bates
The O.A.U. excavation programme involved specialist support
provided by staff of the Geoarchaeological Service Facility (GSF),
University College, London. One important element in the joint
strategy was to identify sedimentological characteristics within the area
of the excavations which might be traced laterally as 'markers' into the
off-site environment. The other element was to try and elucidate further
the chronology and processes responsible for the abandonment and
destruction of structures on the site, and their relationship to an
overlying sand deposit, recorded previously in the area by other
workers (Mynott 1981) as the wind-blown sand.
The sites of both the Cl. Br. and Saxon Shore forts are located around
the 10 m. contour on the south-east seaward-facing margin of a low
promontory formed of an outcrop of the Middle Chalk, overlain by
Chalk Head (BGS 1977). This promontory extends out into the floor of
the Dour Valley and gives rise to a pronounced eastward curving
meander in the modern river course (see Fig. 1).
The western margin of the estuarine inlet also deferred around the
promontory in the Roman period (Rigold 1969) and would have placed
the southern perimeter of the forts in a relatively exposed position (see
Fig. 1), subject to wind fetch from easterly and south-easterly
directions. This estuarine inlet had, on the evidence recorded from a
deep roadworks shaft in York Street, infilled from -2.20 m. to +0.30 m.
O.D. by the late Roman period (Bates and Barham 1993). It is in the
area on the south side of the Shore Fort, currently between Queen
Street and Town Wall Street to the north and south, and bounded by
York Street to the west and Bench Street to the east, that extensive
spreads of well-sorted sands have been recorded (Mynatt 1981). The
sand unit stratigraphically overlies the infill deposits, and therefore the
onset of sand deposition must post-date harbour infilling at least in this
area of the estuarine inlet.
A NORTH-SOUTH BOREHOLE TRANSEC T AND THE LABORATORY
ANALYTICAL DATA
The stratigraphic relationship between the well-sorted sands and
underlying topography and deposits is shown as a transect running
north-north-west to south-south-east from the White Cliffs Experience
site (Borehole 01), via the Zion Chapel site (Mynott 1981) to
observations made in the roadworks shaft at York Street roundabout
133
DAVID R.P. WILKINSON
(see Figs. 1, 21). Figure 21 clearly illustrates the unconformable
relationship between the sand unit and the underlying deposits. The
sand unit thins landwards and upslope from> 3.50 m. thickness at the
York Street Shaft (where at +2.50 m. O.D. coarse gravel to cobble-size
well-rounded flint clasts are bedded within the sands), to c. 1.6 -
1.0 m. at the Zion Chapel site and Borehole 01. The unit forms a drape
over a pre-existing land surface sealing waterlain deposits at +0.30 m.
O.D. at York Street Shaft, the Roman metalled surface at +6 m. O.D. at
the Zion Chapel site, and deposits with reworked occupation debris in
Borehole 01 at +6.85 m. O.D. The basal contacts of the unit onto
underlying deposits are typically sharp and unmixed. This suggests the
onset of sand deposition was relatively sudden and subsequent
deposition rates were fast. Upper contacts of the sand unit are probably
locally truncated and therefore total thicknesses as logged for the unit
are minima as shown in Figure 21.
At the White Cliffs Experience site (Trench 12, Fig. 10), sand
deposits also interdigitate with (Context 919), and overlie (Contexts
903, 904, and 905) poorly-sorted fine matrix supported and clast-rich
(diamict) deposits (Contexts 902 and 911) within Trench 14 of the
O.A.U. excavations. These deposits are thought to represent the direct
collapse and subsequent downslope reworking of the fort wall
structures and associated debris.
The results from the O.A.U. excavation at the site were
supplemented by G.S.F. work. This included: stratigraphic recording,
monolith sampling and the manual augering of two boreholes.
Monoliths, and samples from the two boreholes were analysed by
G.S.F. using standard methods (see Appendix I). Only the detailed
analytical results from Borehole 01, located inside the eastern gate of
the Cl. Br. fort, and just outside the southern perimeter of the Saxon
Shore fort (see Fig. 2), are presented here (Fig. 22) and discussed in
detail.
Borehole 01 (BH 01) was manually augered to a depth of 3.50 m.
beneath present ground surface (+8.50 m. O.D.) using a Dormers light
alluvial drilling kit with 50 mm. diameter sampling barrel. A total of 35
samples of 30-40 gr. wet weight were retrieved at nominal 5 cm.
spacings down-profile. The stratigraphic profile is shown on Figure 11.
The sequence as logged (see Figs. 11, 21) comprised very wellsorted
clean sands from surface to 0.95 m. depth, continuing to 1.65 m.
depth with variable silt partings and occasional gravel inclusions. At
1.65 m. depth a sharp contact was logged, into more compact silts with
high frequencies of fine gravel sized clasts of ceramics, tile and flint.
This silt unit fined downwards to 2.75 m. depth where there was a
transition into cleaner well-sorted silts underlain at 3.05 m. depth by
archaeologically sterile clay-silts. The samples from BH 01 were then
134
......
u)
Ul
c:i
ci
9
7
5
4
3
2
0
-1
-2
-3
SSE
York Street site
1,..,. . ... 20
o