Pottery function and finewares in the Roman north

J eremy Evans

Introduction assemblage in use, to the archaeologically recovered assemblage (cf Schiffer 1976). In the case, however, The main aim of this paper is to attempt to sketch of a variation in breakage rates between different general trends in the functional composition of, and the types of site some general predictions might be level of fmewares in, pottery assemblages in northern made. Villas as the homes of wealthy individuals England particularly in the 3rd and 4th centuries. The might be expected to be more wasteful than lower majority of the data are based on the complete status rural sites. Also if differential breakage rates quantification of groups (and in some there is a were a major factor then the contents of 'destruction component of residual material). Most of the data are deposits' (though not necessarily of burials and derived from the author's research (1985), there still 'votive deposits') should differ markedly from being a general lack of quantified reports in the region normal sites. and especially a lack of reports which quantify the Archaeologically residual material can, with occurrences of all forms (including the sarnian and caution, be excluded from function figures and mortaria) as opposed merely to fabrics. grossly residual material, eg samian in fourth century Little work has been done on the functional contexts, has been excluded from most of the data composition of Romano-British pottery assemblages presented below. The effect of archaeologically although the earliest quantified studies did record this residual material on function figures will probably be (Newbold 1913), however, this W(lS as a method of to reduce apparent change in time and, if localised quantification per se. It is only really since the advent within a site, might distort spatial patterning. of the 'New Archaeology' of the 1960's that functional variation has been quantified for evidence of functional Definitions differences in the use of pottery within or between sites (Longacre 1970). Most work on Romano-British The methodology in quantifying vessel function has assemblages has been conducted by Millett (1979; been to follow definitions similar to those of Webster 1983; Millett and Graham 1984) who has performed (1969) and Millett (1979) quantifying the data by intra-site analyses on Roman material from Portchester, minimum numbers of rims per context. Cross-joins Verularnium and Colchester and the prehistoric Aegean have not been sought routinely as they tend only to site of Zagora, and he has also produced sequences for be successfully defined with unusual vessels and three southern Roman town sites (1979). rarely occurring fabrics, thus reducing the apparent In interpreting variations between different types proportions of these, whilst not doing so for the of site one possible explanation could lie in commoner types and fabrics. The functional groups differential breakage rates at different types of site into which all rimsherds were classified were jars, and possibly between different types of vessels. For bowls, dishes, flagons and pitchers, cups and example finewares might be treated with some care beakers, mortaria, amphorae, lids, and other. These whilst vessels used for cooking might become rapidly were defined as follows: foul and therefore be discarded or be broken in use on the hearth. If consistently different breakage rates Jars - vessels with a height of more than two thirds did occur and reflected vessel function then this of their rim diameter and less than twice their could provide a major transformation from the diameter.

95 96 Jeremy Evans

from those followed by Millett (1979) and his experiment of defining complete vessels was repeated

SCATTERGRAM OF VES SEL FUNCTIONS IN GILL AM 1970 using 266 vessels from Gillam's Types paper (1970). These were defined into functional groups using the 16 subjective criteria generally employed in this study o and the ratios of their radii to heights were plotted 14 on a scattergram (Fig. 1). This shows that flagons, x o • X jars and bowls are all well separated whilst there is 6 OD • 12 9 000 some overlap between dishes and bowls which do not x xo 0 00 • XX 0 break down very well into natural groups. Similarly O'~Cb cP • • 10 0x \3 00 • beakers and small jars intermingle and could not be RA01US lN CM X~fodo W ...... : ••• separated by this method. Mortaria, lids, amphorae XXl+l.x • • t •• 0 and'other' vessels were not plotted as their definition X tt6 • •• • • oo..J] 0 •• t.~" 0 relies on characteristics other than their general 00 LtJ • • ••• il '•• • x 0 0 • 6 •• t/r. • • shape. The lack of a clear distinction between dishes 00 • :... 6··· • i • and bowls has been noted in practice in the study, X !6 •• tF~6 I' +66 6. 00 particularly in relation to types such as Gillam 225 o ~ ~6 6 cfXOJ :6 66 6 6 0 @Bo8) and 313 which share an identical rim form and, tP·.~ 6 eO 0 whilst it is felt that there is some validity in ..6 6 @ -Cl 0 • 6 Cl!) <.r o subdividing them, account has been taken of this problem by also including the combined dish and

10 15 20 25 30 bowl percentages in the tables below. HE1GHT lN CM Most of the jars in the study fall into a height to

X DISHES rim diameter of around 1. 25: 1 and the two o BOWLS • JARS remaining groups are of some interest. Those items 6 BEAKERS with a height to rim diameter ratio of less than 1 are o FLAGONS the wide-mouthed jars/bowls of the Holme-on­ Spalding Moor and North Lincolnshire tradition. Fig, 1. Scattergram showing ratio 0/ heights and radii a/vessels in Gillam (1970) These have been ind uded in the jar group because all other bowls would be appropriate as tablewares, whereas these large vessels seem more probably used Bowls - vessels with a height of more than a quarter for storage like other jars and would seem out of and less than two thirds of their diameter. place in any group essentially composed of Dishes - vessels with a height of less than a quarter tablewares. They could equally well be regarded as of their diameter. an entirely separate functional group although their Flagons - vessels with a constricted neck in incidence is sufficiently low for this to be of dubious comparison with jars, too constricted to easily value. In regions where this sort of form is common, admit the hand, which would appear to have been for example amongst regions with high proportions used for liquid storage. Their height is generally of Severn Valley wares, this type may well be best greater than twice their diameter. treated as a separate class. Beakers - vessels, generally with a height greater The other interesting jar group are those vessels than their diameter and which appear to be of a which have a height to rim diameter ratio of around suitable size and shape for drinking from. 2: 1, a position where they are becoming intermediate Mortaria - dishes/bowls with a deliberately gritted between jars and flagons on the scattergram. These interior. vessels mainly comprise Crambeck types 3 and 3a Amphorae - large, two-handled containers varying in (Corder 1936) together with Norton and Holme shape and fabric but with rounded or pointed bases vessels of similar form to Crambeck type 3a (Hayes used for importing wine, oil, etc. and Whitley 1950, type 4; Hicks and Wilson 1975, Lids - shallow vessels similar to dishes , but with a nos. 6-8; Corder and Sheppard 1930, nos. 72, 80 handle on the exterior in the centre of the base. and 81 and 82, 84 and 85) all of which are handled. Other - any other type of vessel not in the above The common occurrence of Crambeck types 3 and 3a categories, eg candlesticks, cheese-presses etc. in well deposits, eg Langton (Corder and Kirk 1932), Benwell (Charlesworth 1960), Rudston (Rigby 1980) There are some differences in these definitions and Dalton Parlours (personal inspection; Sumpter Jeremy Evans 97

SEGONTIUM CATTERICK SITE 46 SITE 240 60- - 60- r- r-r- r- r- 40- r- r- 40- I- r- - r- I-- I-- I-- 20- I-- 20-~

Jars 0- o-WJllJ

40- 40-

Bowls

40-

Dishes

20-

Beakers

20- 20-

o cD [}-[[b--n Lids o-~ 2 3 4 5 5A 58 6 6A 7 7A 78 8 9 10 lOA II 3 4 5 6 6-7 7 3 =3-43-5 5 6 7 PHASE eC2 C2 eC3 C3 C4

Fig. 2. Proportions offunctional categories through time Fig. 3. Proportions offunctional categories through time at at Segontium (after Webster 1993) Catterick (Central Archaeology Service Sites 46 and 240)

1990), together with the evidence of cord attached to Within the scattergram (Fig. 1) it is useful that the one of the Dalton Parlours examples (Sumpter 1990, constricted necked jars (Hayes and Whitley 1950, 244) suggests that this group of jars was intended, at type 7, Corder and Sheppard 1930 nos. 95-6 and 98, least in part, for the purpose of drawing water. These Hicks and Wilson 1975, nos. 18-20, Corder 1928, jars have not been defmed as flagons/pitchers because no. 176) group centrally with the other vessels, this was not necessarily their sole purpose, their necks which tends to confirm their definition as pitchers, are wide enough to comfortably admit the hand, so that despite their lack of handles. As Millett notes (1979, they may have been used for storage, or perhaps, 37): though rather doubtfully, as cooking vessels. The case of the Huntcliff type calcite gritted jars also counsels "The marked overlap between beakers and jars suggests a caution on this point. There is clear evidence that this false distinction on these criteria. Nevertheless any Romano-British pottery specialist would be able to pick out type of jar was used for cooking but they occur in such a beaker; a fact which suggests that the criteria of style and numbers in the Rudston and Dalton Parlours well fabric are here being used instead of the height-rim deposits, that it is hard to escape the conclusion that diameter ratio" they were being used for drawing water. The consistent evidence of sooting and limescale deposits on the The storage jar class which Millett (1979) uses has Dalton Parlours Huntcliff type jars (Sumpter 1990) also not been employed here because many of the few demonstrates their use for both drawing and boiling vessels which might be put into this group share the water. The former could be simply achieved by same rim form as the smaller Huntcliff type jars and fastening a cord around the neck below the prominent with northern material it seems a rather subjective overhanging rim. category. 98 J eremy Evans

very large groups, rather than the restricted areas of Towns sites several of the groups in this study comprise. As le4 many of the groups in this study are from limited ee4 .,.y . areas of sites there is a fairly high probability that / these areas were being used for different functions in e1-3 , . . ~. different periods. This is especially so for those 11 ...... sequences which are from different locations within a site, so that it is unlikely that the single site Forts sequences will produce any more convincing results le4 ....• i. •• than all the groups together. There are some slight difficulties with some of the ee4 .. / . .. . data; the early 4th century Greta Bridge group has an e1-3 • • • • estimated beaker figure as colour-coated wares of : ••• -.%( 4 • I • this date would appear to have been lost since o la 20 30 40 50.. 60 70 80 90 JARS excavation . Fig. 4. Proportions ofjars on northern 1st-3rd,earlier The individual site sequences (Appendix I) show 4th and later 4th century town and fort groups. various developments in time but it would be difficult to claim that there are any clear trends applying to the data from the smaller groups. However, the more Functional changes in time substantial sequences from Segontium and the CAS Catterick sites, in particular, do show trends, with a There are two ways of examining the changes in the low level of jars and high proportions of tablewares functional breakdown of vessels with time in this in the early Roman period with the level of jars study. Firstly there are those sites with a sequence of rising in the 4th century to a peak in the later 4th deposits, especially Greta Bridge, Vindolanda, century (Figs. 2-3). The data from the overall range Shiptonthorpe, Catterick, Segontium and Brough-on­ of sites do show chronological trends more clearly Humber, and there are shorter sequences from (Fig. 4). There are no obvious differences between Beadlam, Castleford and Old Penrith (together with the 1st-2nd century figures, mainly derived from published sequences from Chester (Carrington 1980), published sources, and the 3rd century, but there is the Carrawburgh Mithraeum (Richmond and Gillam a trend in the early 4th century for jars to be rather 1951) and South Shields (Dore 1983). Secondly all more strongly represented than in the 3rd century. the sites from each period can be examined together There is then a marked change in the late 4th century to see if any general trends emerge. Millett has with jar totals higher on average than in preceding restricted his examination of chronological trends in periods, with consequently lower levels of dishes and function figures to sequences from south-eastern bowls and a marked weakening in the representation England. These work well, but they are derived from of beakers. The basic trend in the data is the direct opposite of that shown by Millett (1979) for three southern towns (Fig. 5), which must suggest that there are some regional patterns in the functional Mortaria 0 ~ = breakdown of pottery assemblages. The apparent lack Storage Jars ~ of much change in these northern groups from the later 2nd through to the early 4th century (Appendix I) may well be explained by the distinctly military emphasis of these northern sites. Whereas the use of Jars pottery in southern England developed from the pre­ Beakers Roman Iron Age tradition so that a consequent 'Romanization' in the use of pottery might be Flagons expected, over much of the North the Iron Age was essentially aceramic and many rural sites remained so BowlS throughout the Roman period. The northern forts,

Dishes and to some extent the towns, started with a full kit Neatham Chichester Verulamium of'Romanized' types for'Romanized' users without TIME any development from a pre-existing pattern of usage Fig. 5. Proportions offunctional categories through outside East Yorkshire and the Vale of York (cf time for three southern town sites (after Milieu 1979). Jeremy Evans 99

Evans forthcoming C). Significantly it is in East The 2nd century list comprises groups from seven Yorkshire, at Brough-on-Humber and Shiptonthorpe Hadrian's Wall turrets and milecastles, four sets of that high jar levels are found in these data. groups associated with forts, five civil settlements, a cemetery, and seven rural sites (one of which later Functional variations between broadly became a villa). In this group the turrets stand out contemporary groups because of their high jar levels and low percentages of disheslbowls (Fig. 6). In this the turret ceramic To examine functional variations between broadly assemblages might be compared with their small fmds contemporary assemblages the sites have been examined assemblages examined by Allason-Jones (1988) both in four groups; 2nd century, 3rd century, early-mid 4th showing a limited subset of the material range available century and later 4th-early 5th century. There seems to in the forts. Their jar levels, generally around or be a likelihood that variations in the functional exceeding 60 % of the groups, may be compared with breakdown of assemblages do reflect, at least in part, those from the rural sites listed here with which they the nature of the class of site. This is one of the are very similar (Fig. 7). The one exceptional rural implications of the study of the distribution of graffiti site, Bryn Eryr, Anglesey, consistently shows evidence on Romano-British vessels (Evans 1987, fig. 2; table A) of a high-status display orientated assemblage, which shows functional variations between the type of especially in its earlier occupation (Evans forthcoming vessels receiving graffiti from different classes of site A). The fort groups, from Segontium, the Vallum fill which cannot easily be explained except in terms of at Benwell and South Shields tend to have a similar consistent functional variations between different types range to those from the civil urban sites, excepting of site. Shiptonthorpe. This site, and perhaps Brough-on-

1st· 2nd CENTURY TURRETS & FORTS 2nd CENTURY RURAL & URBAN % 0/0 .. • ST86

80 80

• .ST85 • • • 60 A 60 • TDcemy • •• • • .PBmc Jars A Jars A • 40 A A 40 A • • BE A • A • • A .. • 20 20 • AA

20 40 60 20 40 60

Dishes & Bowls Dishes & Bowls

• Turrets • Rural

A Forts • Urban

Fig. 6. Proportions ofjars plotted against dishes and Fig. 7. Proportions ofjars plotted against dishes and bowls for 1st-2nd century turrets and forts bowls for 2nd century rural and urban sites (PBmc = Poltross Burn milecastle) (ST85, ST86 = Shiptonthorpe 1985 and 1986; TDcemy = Trentholme Drive cemetery; BE = Bryn 100 Jeremy Evans

Humber, show sequences more comparable with (Appendix I). These function figures tend not to reflect southern towns (Millett 1979), rather than with the the phenomenon because rimsherds are uncommon northern forts like the others listed here. The high jar relative to the large vessel body area, but the fabric figure from the group from layer 123 at South Shields figures also show that amphorae are also much scarcer may reflect, like those from the 3rd century Vindolanda on the rural sites (cf Collingwood and Myres 1941, groups, the location of this deposit in a marginal area 242; Carreras Montfort at Roman Finds Group 'Making of the fort, on the rampart-back, although layer 56 Waves' conference 1992). comes from a similar location (Fig. 8). (The Chester The 3rd century figures do not show many great figures are given as a range as although it is claimed variations according to type of site (Appendix I), that the material has been fully quantified the methods excepting the peculiar deposits from the Carrawburgh used in the report seem to have considerably over­ Mithraeum and Catterick Racecourse phase 3B, as all represented the samian content (Carrington 1980), the the other sites are forts vici or military associated towns upper figure, therefore, includes samian quantified by (Fig. 9). The military deposits from South Shields and minimum numbers of vessels and coarsewares by Bewcastle both show low jar levels with large quantities maximum numbers of vessels, whilst the lower one of dishes and bowls, although the Vindolanda sequence excludes the samian entirely. ) consistently opposes this, as does the small group from The rural sites generally have over 50% jars, with Segontium phase 7B. The enigmatic Castleford site, consequently lower levels of dishes and bowls than which is perhaps best viewed as a burgus is similar, as military and urban sites (Fig. 7). Beakers and flagons are the vici at Greta Bridge, Old Penrith and the are much less consistently and more poorly represented Chester mansio, whilst the Brough-on-Humber groups than on urban and military sites, but mortaria are not are more like those from Vindolanda. It is notable that

C1-3 FORTS C1-3 TOWNS

80 BO

SoH 5T Vin31455123 CAT 273 WELL 60 '_5eg7S 60 ·Vinll3 [SoH • Trentholme Drive Vin285 .-Vin91 11' Jars Jars SoH •• 40 • • 40 • .: . • • • • • .:. • CSM • •• • • ..:. 20 20 • • • • •

20 40 60 20 40 60

Beakers Beakers

Fig. 8. Proportions ofjars plotted against Fig. 9. Proportions ofjars plotted against beakers for 1st-3rd century fort sites beakers for 1st-3rd century towns (Vin = Vindolanda; ss = South Shields; (BoH = Brough-on-Humber; ST = Shiptonthorpe; Seg 7B = Segontium phase 7B) CAT 273 = Catterick, Central Archaeology Service site 273; CBM = Carrawburgh Mithraeum) Jeremy Evans 101 the odd groups, excluding the Carrawburgh Mithraeum the occasional and ritual use of the site. One interesting and Trentholme Drive for the moment, are from the point is the lack of flagons associated with the beakers, barracks and rampart-back at Vindolanda and two of and overall in the Appendix I there seems to be the Btough groups are from the town gate and rampart. remarkably little association between high beaker levels These groups which perhaps represent rather temporary and flagons. use are similar to the 2nd century turrets which were The Trentholme Drive cemetery figures, even probably used in a similar fashion. This does not without the sarnian, which is regrettably not fully explain the two groups from the interior of Brough published, form a curious group which does stand out which are both associated with buildings. The group from the other sites and illustrates the selection of from G 11 (6), (10), and (13), is a noticeable one vessels from the'normal' assemblage for grave goods. mainly composed of heavily gritted jars, and it may, The jars figure seems a little high for an urban site and perhaps, be that both these groups represent functional the dishlbowl total exceedingly low (the sarnian might specialisation within the site. The high jar level from have added to the latter but it also might have been an Catterick Racecourse phase 3B is dominated by material unusual group), the most striking features though are from the weathering cone from a well, which seems to the high levels of both beakers and flagons and the have been closely related to its use (Fig. 9) . presence of lamps. These features probably reflect The Carrawburgh Mithraeum provides two groups, the'ritual' nature of the cemetery deposit with vessels one very small, but the picture from both is consistent selected from the everyday assemblage but not and highly unusual compared with the others. Jar levels representative of it. The one rural site, Hawling Road, are a little low, but so are dishes/bowls, and 45-50% north Humberside, shows a reduction in jar levels from of both groups are beakers (Fig. 9) . This must reflect the very high ones shown there in the 1st and 2nd

EARLY 4th CENTURY FORTS & TOWNS LATE 4th CENTURY FORTS & TOWNS

80 eo P Burn

60 60 . ~~ Jars Jars ~ ~ ~ ~~. . ~ CBM OP • 40 • 40

20 20

20 40 60 20 '0 60 Beakers Beakers • Towns • Towns

~ Forts ~ Forts

Fig. la. Proportions ofjars plotted against beakers for Fig. 11. Proportions of jars plotted against beakers for early 4th century forts and towns. late 4th century forts and towns. (CBM = Carrawburgh Mithraeum) (PBurn = Poltross Burn mile castle; OP = Old Penrith) 102 Jeremy Evans centuries, but still retains a high level, in contrast with of the site had changed and that the'ritual' use of the most of the urban sites, as might be expected. site had either changed its form or ceased. The early 4th century figures are fairly similar to The two groups from the villa site at Beadlam fall those for the 3rd century, although jar levels are into much the same pattern as the towns but the higher on average (Fig. 4). Two of the fort groups, Rudston group is very different with only 3 % Segontium and South Shields, do show a rather dishes/bowls and over 50% flagons and pitchers. higher jar level than the 3rd century fort groups, and This would seem to be a result of its being a well the Vindolanda group has a high level of beakers, deposit containing material broken in use rather than with the Castleford group being rather similar. The refuse, or deliberate backfill, as the Skeldergate well urban groups show rather higher jar levels than in fill contains. The rural site at Apperley Dene is fairly the preceding period, except for Brough which now well differentiated from the others by a high looks more like the other towns. It is interesting, percentage of jars, but that at Hutton Ambo is not as once again, that the Brough group from the gate­ it is located on the edge of the Crambeck kiln area tower (Period VIII) shows a high jar level which where tablewares were the predominant products. parallels those from similar locations in earlier The late 4th century figures show a rather higher periods (Fig. 10). proportion of jars than in the preceding period (Fig. It is of note that the group from the Carrawburgh 4) and there is a slight fall in the proportion of Mithraeum, although too small to be very reliable, dishes and bowls, although the most marked decline has now settled down at similar levels to other forts is in beakers and flagons (Fig. 11 compared with and towns (Fig. 10). When compared with the 3rd Figs. 8-10). The Signal Stations and milecastle all century groups (Fig. 9) this clearly implies that use hang together as a coherent group with over 70%

LATE 4th CENTURY SIGNAL STATIONS & FonTS LATE 4th CENTURY RunAL & unBAN % 0/0 • R well • • 80 • 80 • • • • • LW • BED 60 60 .. 44 4 Jars 44 • HB 44 4 ••• 40 40

20 20

20 40 60 20 40 60

Dishes & Bowls Dishes & Bowls • Signal station • Rural

4 Forts • Urban

Fig. 12. Proportions ofjars plotted against dishes and Fig. 13. Proportions ofjars plotted against dishes and bowls for late 4th century signal stations and forts. bowls for late 4th century rural and urban sites. (Rwell = Rudston well; LW = Long Whins; Bed = Beadlam; HB = Healam Bridge) J eremy Evans 103 jars and generally less than 20% dishes/bowls and evidence of this from the data. The other and more virtually no beakers and flagons (Fig. 12) . The forts important prediction was that 'destruction deposits' in contrast have c 40-55 % jars and over 30% should differ markedly from normal sites. There is dishes/bowls whilst beakers and flagons are one 'destruction deposit' in this data, phase 2 at consistently represented, if at a rather lower level Greta Bridge, where the building was destroyed by than in earlier periods (Fig. 12). (The exception is a fire, almost certainly accidental, and a number of Old Penrith which is a small and rather unreliable complete vessels were recovered from this deposit. group reconstructed from unstratified material by The function figures, however, are almost identical excluding earlier fabrics) . The only northern town with the succeeding phase, and do not appear at all groups are those from Catterick (Gillam 1957; Bell exceptional. This could partly be due to residual and Evans forthcoming) and Shiptonthorpe (Evans in material from phase 2 in phase 4 and/or residual prep, a) and these are more similar to the forts (Fig. material mixed in with the phase 2 deposits. Whilst 13), although intermediate between the forts and the there is some residual material from phase 2 in phase Signal Stations. The two other town groups, from 4 it is not in that great a quantity and there is little Lincoln (Darling 1977) and Great Casterton (Corder sign of any serious quantity of residual material in 1961) are beyond the study area and might belong to phase 2. The index of rim equivalent divided by the a different regional pattern, but they seem to be minimum number of rims per context gives an consistent with that of northern England, having a far average rim equivalent which shows that the greater percentage of jars than is found on the 'destruction deposit' contains less fragmentary southern sites studied by Millett (1979). The towns vessels than the succeeding phases, a feature which seem comparable with the forts, a consistent result in tends to confirm the nature of the deposit - the this study regardless of period. figures being, phase 2, 15.1 %, phase 4, 7.4%, and The large villa group from Beadlam is phase 5, 9.2 % . It is clear that the 'destruction intermediate between the forts and towns and the deposit' is more disturbed than some other groups in Signal Stations as is the group from the civil site at this study, presumably because it has been levelled Long Whins near Filey (Hull 1932). The similarity for subsequent building and only a limited part of it of the Long Whins site to Beadlam does suggest that has been examined, whereas most of the other it could be rather more than a minor rural site (Fig. deposits come from undisturbed refuse or 13). The group from field walking the unknown site abandonment deposits. at Healam Bridge (Price and Evans 1991-2) contains Since there is little doubt that phase 2 contains a predominantly late 4th century material with little of major input of 'destruction deposit' material and an earlier date, but is actually more similar to the since it does not seem to effect its functional urban sites than the other rural ones. It also produces composition there may be less reason, in practice, to around 18 % of finewares, which, like its function believe that differential breakage rates are a major figures, suggests a site of some status, presumably a factor in functional variations between sites. This is villa. The late 4th century group from the Rudston also borne out by the Boudiccan groups examined by well (Rigby 1980) is as remarkable as its early 4th Millett (1983) which also do not appear to be very century predecessor with over 90% jars and less than different from their regional background. 8% dishes and bowls (Fig. 13) . This is clearly, once In conclusion it would seem that there are trends more, a selection of vessels used to draw water with of functional variation between different classes of very few pieces of ordinary domestic refuse site and the late 4th century groups in this study penetrating into the group (like the Dalton Parlours seem to reflect this best. The towns and forts seem deposits; Sumpter 1990) and as there tends to be a generally to have had very similar functional ranges, correlation between fabric and function in most of the turrets and Signal Stations also have similar the pottery in the north this has direct implications though not identical ranges. These latter would seem for the presence and absence of particular types (and to group with the normal rural site, as exemplified argues extreme caution when using absence in by Apperley Dene, though data on more sites of this arguments about dating; cf Evans 1985). class is badly needed. One possible explanation for the differences One constant part of almost all the assemblages between different types of site is differential under discussion are mortaria. They seem to show breakage rates. There is, however, no evidence of little variation between different types of site or this from the data available. It was predicted that chronologically. They have clearly become an villas might stand out, as the homes of wealthy essential part of the assemblage in all types of site individuals, as being more wasteful, there is no and this may have implications for the 'Romanization' 104 Jeremy Evans of food preparation, or, alternately, their use for some function for which non-ceramic vessels were previously employed, given that they were unknown in the pre­ PERCENTAGE OF SOOTING BY WARE CLASS Roman period and would not be found on minor sites % if they did not serve a practical purpose.

Evidence of the actual use of Roman ceramics 60

Thus far functional classes have been described in terms 40 of their shape, but little has been said of the purposes to which they were actually put. In many cases this is 20 fairly unknowable but some information is available and with the advent of residue studies much more may ~ ~ ~ ~ ~ ~ ~ ~ ~ Ic3 ec4 ec3 Ic3 mc4 Ic4 Ic4 Ic4 Ic4 become so with major implications for the study of 'Romanization' of diet (Heron et al 1991 ; Evershed et • Gritted w ares alI992). Historical sources on the whole make little, if o Other wares any, reference to pottery, and especially to its use. Most classical literature is the product of a self­ Fig. 14. Proportions of sooted rim and base sherds from conscious literary elite which would not consider various groups in heavily tempered fabrics, compared with all other fabrics. (CAS = Castleford; Vin = mentioning anything so sordid. Occasional historical Vindolanda; MAL = Malton; PBDG = Piercebridge; references are found in the New Testament, which, as B1N = Binchester; HUNT = Huntcliff) Reece points out (1981), is a useful source for plebeian life in Palestine in the fIrst century, and other christian authors who are writing for a mass audience. The wheaten fare, and insisted on making them dishes of beans and pulses and cooking the porridge of barley and millet according to biblical references (see endnote 1 for a list) suggest that the recipe in Ezekiel IV, 9. One of most charming of Paulinus' in Palestine it was the practice to clean ceramic vessels letters to Severus is that in which he describes with gentle and there is the distinct impression that tableware was humour the rigour with which Victor prosecuted his reforms at used communally rather than there being separate Nola and the horrible smells which permeated the entire vessels for each individual. No reference is made to establishment as a result of his ascetic cuisine. He tells us that cooking and whether or not ceramics were employed Victor's concoction was so horrible that the very pot in which it for this purpose. was boiled cracked - it was the mixture of things cooking together which was fighting . However, adds Paulinus lightly, it Further references to the use of ceramics are to be was wholly to the taste of the poor old countryman who had found in the letters of Paulinus of Nola, living in a joined them, partly because he had never known anything better monastic community in his Italian villa in the early 5th and partly because he had no teeth." century (Chadwick and Chadwick 1955, 72). Apart from the reference to a ceramic cooking pot this "The community of Paulinus and Therasia was austere rather passage also hints at the sort of diet which might be than ascetic. Paulinus and his family fasted until 3pm; but appropriate to the peasantry. their food and dress seem to have been simple rather than GraffIti provide some information on the terms used coarse. Wheaten bread was in use, but the platters and vessels were of boxwood and earthenware." for different types of vessels in Roman Britain;

It would appear from this that Paulinus had adopted the Cuppa for a dish, a variant of the classical latin tableware etc of a prosperous peasant. He exchanged cupa - tub or vat (Hassall and Tomlin 1978, no. simple gifts with his friends, a cap for St Jerome, 43). loaves for St Severus and also a boxwood platter OlIa on a colour-coated beaker (Hassall and Tomlin (Chadwick and Chadwick 1955, 79). It seems evidence 1980, no. 7) . that these gifts, including the platter, are trivial items of Both mortarium and pelveis for mortaria (Wright et everyday use. Paulinus also provides the only reference al 1976, nos. 54 and 66) . known to this 'author of ceramic cooking vessels Dolium on an amphora (Hassall and Tomlin 1979, (leaving aside Apicius's disposable pots for baking no. 11). bread);(Chadwick and Chadwick 1955, 72-3). Lagona on a storage jar (Wright and Hassall 1971, no. 50). "A certain disciple of St Martin, a monk Victor, who carried letters annually between Paulinus and his old friend SUlpicius Graffiti give very little information on the contents Severus in Aquitaine, professed himself scandalised at the of vessels, apart from amphorae, only three vessels Jeremy Evans 105

have graffiti recording their contents of all those by the material). The figures for sooted vessels from reported in Britannia since 1970, a storage jar from various sites are shown in Table 1. Apart from Middlewich containing amurca 'waste from brine', occasional vessels in other groups which may have and a flagon from Wallsend containing mulsum' been burnt or otherwise accidentally sooted after use honey sweetened wine' (Wright and Hassall 1970, the main sooted types are jars, together with a no. 37 and Wright et al 1976, no. 56). The vast significant proportion of dishes and bowls, which majority of graffiti on pottery simply give the considering their numbers are unlikely to have been capacity of the vessel in weight or volume, the most sooted accidentally. revealing example since 1970 being an example from Dishes seem to be sooted more often than bowls Southwark (Wright et aI, no 24) containing 'Mel and this is probably a real reflection of their use p[ondo} XXIIII' -'honey 24 pounds' . since they are more often found in coarsely tempered This type of graffito would rather seem to suggest fabrics than bowls. Fig. 14 shows that coarsely that the contents of the pot are for sale, and that the tempered fabrics were generally those used on an pot, like amphorae, is merely the packaging. open fire to judge by their being sooted at least twice Deposits on vessels can provide some evidence for as often as other types. The bases of several calcite the actual use of vessels as can the contexts of some gritted jars suggest how they were used for cooking; assemblages. The Rudston well deposit, and those they have reduced bases and about 10mm of the wall from Langton (Corder and Kirk 1932) and Dalton above the base, above which the wall has been Parlours (Sumpter 1990) clearly suggest that not only oxidised brown by heating on a fire . The lack of those vessels classed in this study as pitchers, but oxidisation on the base is probably because this has also Crambeck type 3 jars, handled jars of Huntcliff been set in the ashes, whilst oxygen is available to type form and Huntcliff type jars were all used in oxidise the walls above. It is apparent from the drawing water. It may be that this was the main above evidence that most of the vessels used for intended function of the Crambeck type 3 jars, like cooking, or at least heating on an open fire, were of the Throlam pitchers, with the spouted and handled heavily tempered fabrics. The tempering agent being vessels being reserved for use at the table. sand in the case of BB 1, large quartz in the case of Evidence of sooting has been consistently recorded most of the gritted wares, shell in the case of Dales in this study and enables those types of vessel used ware and calcite for most of the East Yorkshire for cooking and boiling water to be examined fabrics. Most of these vessels are derived from quantitatively. Some differences between sites may indigenous Iron Age traditions and this is part of the be found and these may partly be accounted for by reason for their use in this manner, as the bonfire enthusiastic cleaning of the excavated material (as firings to which they were subjected required that opposed to none at all on some other sites to judge they should be comparatively resistant to thermal

Site Dishes Bowls Jars Flagons Amphorae Beakers Mortaria Lids n.

Vindolanda eC3 0 0 58 .8 0 0 0 0 0 29 mC3 35 . 7 20.0 46 . 9 0 0 0 0 0 63 Ic3 0 14 .3 69.6 0 0 0 O· 0 36 Bewcastle C3·C4 0 3.8 54.2 0 0 0 0 0 69 Castleford 211 IC3 11. 1 0 23 . 1 0 0 0 0 0 38 260 eC4 0 16.7 27 . 3 0 0 0 0 0 26 Mal ton mC4 9.1 0 38 . 1 0 0 0 0 0 49 Binchester A2037 IC4 7.1 0 48 . 9 0 0 0 0 0 97 A2068 IC4 44 . 4 18.2 37.0 0 0 0 0 0 SS Huntcl i ff IC4 0 0 51.6 0 0 0 0 0 49 Piercebridge HS77EII IC4 0 9.1 58.1 0 0 0 0 0 SS Beadlam all IC4 9 0 30.1 21.4 0 6.7 10.3 22.2 956

Table 1. Percentages of sooted vessels by functional type 106 Jeremy Evans shock simply to survive firing as Woods (1983-4, found that in an oxidising atmosphere it was impossible 25) points out. to fIre one of the local clays with calcite tempering Most of the experimental work seems to have been above 675°C without the sample decaying badly (ibid, done on shell and calcite inclusions (Rye 1976; fIg 4). This probably accounts for the fIring of most Steponitis 1983). Steponitis demonstrated that of his heavily tempered calcite gritted fabrics and Dales ware fineware sherds, those with less shell tempering in a reducing atmosphere which would tend to inhibit this reaction and, although there is no evidence of it "would have a high initial strength, but would lose a very from this study, the use of salt to reduce this reaction great proportion of that strength if subject to thermal shock. A would seem to be a fairly strong probability (ibid, coarsely tempered vessel, on the other hand, would have less 131ft) if the clays do not contain reasonable quantities initial strength, but would retain most of that strength even of naturally soluble salts. after a severe thermal shock. This being so a coarsely tempered pot would probably have been more resistant and Rye (ibid, 118) suggests that quartz is a less longer lasting as a cooking vessel." (ibid, 44-5). suitable tempering agent because

"of its relatively high thermal expansion as well as its The popular choice of shell and calcite as a temper crystalline inversion at 573°C". would seem to be explained by Rye's point that it, like feldspars, has a thermal expansion similar to that However, it seems clear that BB 1 was used of typical low fired clay and therefore successfully for this purpose. It could be that the "would be the most suitable in avoiding stresses within the thermal expansion of the clays used for BB 1 were fabric during repeated heatings and coolings" (Rye 1976, greater than average. Bronitsky and Hamer (1986), 117). however, have suggested that fine temper of 0.5mm and less However, there is the difficulty that CaC03 may start to decompose at around 620°C ( ibid, 120) and Rye "is more resistant to thermal shock than coarse temper"

80 40

60 20 40 Jars 0

20

20 Jars 0 L-_'-_'-_'-_'-_'-_-'------' [[b Bowls oD

Bowls :0 d1b % % 2:0 Dishes & 40 20 cniJ 20 Dishes oD'----'-_-----' - 20 Beakers oD db Beakers 0Lr=J_-L ___-LLl=C1=J_-'-_-'-_-'-_--' 20

I 20 n Pitchers 0 I I I I I 10 Flagons o~ ~ r=J Mortaria 0 r=:JL--'-______C2-4 Cl0 15 C16 eCl7 IC17 C18 C19 2-4 8 11 12 13 14 Period Fig. J5. Proportions offunctional categories Fig. J 6. Proportions offunctional categories through time at Bath Orange Grove through time at Chester-le-Street J eremy Evans 107 for both sand and shell, although confirming that for 60 .--- shell greater quantities of the temper increased thermal shock resistance. Woods reports success in - 40 .--- practical experiments in cooking using pots - .----

"usually containing large amounts of grog and quartz sand, 20 - r--- though in one instance a fairly vesicular pot which had originally contained grass as an opening material was used. Jars Tinned baked beans, ravioli and spaghetti were heated and o reheated and no breakages occurred even when cold vessels were placed in direct contact with flames or glowing charcoal, or hot vessels were taken off the fire and placed on cold wet ground" (Woods 1983-4,27-8). 10 o __-====LD---->====:::::jCJ-----.J Bowls It seems that the popularity of coarsely tempered, handmade vessels, often derived from Iron Age traditions, was partly determined by functional % 10 considerations and that whereas more Romanized ~ Dishes vessels might make good tablewares and storage o vessels they may have been less suitable for cooking pots. This is not to imply that heavily gritted wares 80 were merely used for cooking, other uses have already been noted above, but their absence, or very low levels would tend to suggest that cooking was 60 being performed by other methods. The usefulness of this type of vessel in cooking is borne out by the 40 presence of calcite gritted wares on the Crambeck, Holme-on-Spalding Moor and possibly Norton kiln sites where all other ceramic needs could be provided 20 by kiln products. Flagons This being so it is interesting to examine the ceramic o history of the northern forts. Whereas rural sites in the north in the 1st and 2nd centuries seem to have had a considerable proportion of handmade, coarsely tempered fabrics (Evans forthcoming C), eg Rudston 60 (Rigby 1980), Beadlam (Evans in prep, b), Langton (Corder and Kirk 1932), Hawling Road (Evans - forthcoming B) and Crossgates (Rutter and Duke 1958), 40 the forts, apparently supplying much of their own needs in the Flavian-Trajanic period, do not. The Flavian­ 20 - Trajanic material from Binchester contains a very few handmade coarsely tempered vessels and when BB1 is S Cups 0 introduced it does not seem to make any great impact 2 3 4 5 6 8 11 13 15 17 until the 3rd century. The Hadrian's Wall turrets and mC13 ee15 eC16 mC16 IC16 milecastles as noted above are functionally dominated Fig. 17. Proportions offunctional categories through time by jars so that a high proportion of cooking vessels at Newcastle Castle Ditch (after Ellison 1981) might be expected, but BB1 does not amount to more than 30% at Poltross Burn even in Period 2 when 78% of the group are jars. rampart-back ovens seem to be commonest in 1st and The reason behind this may be the higher level of 2nd century contexts. Graffiti on millstones (Johnson organisation of the early army, its emphasis on 1983) and a few pots (Wright et al 1976, no. 66) communal cooking and probably the greater use of show that provisioning was organised on the basis of metal vessels. Most bronze skillets would appear to centuries and contubernia, probably with milling, and be associated with the military and are 1st to 2nd perhaps baking, at the century level, and cooking at century in date (Bennett and Young 1981) and the contubernia level since there are no structures Site % by sherd % by \Jt % by min nos % of n . Towns /vie; Ru ral and other nos of rims glass Forts Old Penri th Beadlam earl y di teh 1705 C3 24 22 612 s he rds m-IC2 6 181 s he rds Bri thdi r 23 65 497 s herds Greta Br i dge Carrawburgh Mi thraeum Segontium Phase 2 9 548 s herds Pers I + I I A eC3 50 14 rims Ps 2 Flav 22 11 Phase 4 12 288 s herds Pers 118 + IIC IC3 52 22 rims Ps 3 Flav 9 10 8rough-on- Humber Graeanog Ps 4 flav-Traj 17 Per VI eC3 11 48 rims C1-eC4 3 336 s herds Ps 5 Traj 11 Per VIII LC3 14 28 rims Ps SA Traj-Had I (13) IC3 11 27 r ims Bryn Eryr Ps 58 Had 14 G 11 (13) C3 o 1S rims C1-3 14 12 20 572 sherds Ps 6 Had-Ant 12 7 Ps 6A Had- Ant 8 2 Bus h Farm Ps 7 IC2-C3 10 Ches ter C1 - 3 7 6 13 409 s herds Ps 7A m/IC3 8 \le I I 11 (8) IC2 45 129 vessel s South Sh i elds \lell 11 (4-7) C3 30 227 vessel s Hawl i n9 Road 123 mC2 23 rims Catteri ek Con 500 C1st 301 sherds 156 mC2 30 30 rims Bainesse Farm - CAS Site 46 Con 4007 C1 s t 355 s herds 120 eC3 28 47 r i ms Ps 2 eC2 24 21 s herds ~ o n 547 Flav-Traj? 56 s herds Benwell above Ps 3 eC2 254 s he rds Con 4030 Had - eC3 63 s he rds Vallum fill Ps 4 mC2 12 4 0.7 273 s herds IC2/eC3 23 123 rims Ps 5lC2 11 0.6 981 s herds Catc ote Beweas t Le Ps 6 IC2 13 7 1.2 3316 s herds 01 2 C2-eC3 0.4 43459 C3-eC4 23 13 551 s herds Ps 6/ 7 IC2 16 8 1.2 1494 s herds Ca s tleford Ps 7A eC3 3 1 3 .0 33 s herds 211 IC3 22 9 328 s herds Ps 78 C3 15 12 17.6 34 s herds Vi ndolanda Ps 7C C3 20 18 44.7 219 s herds eC3 211 s herds Ps 7 C3 25 14 2.7 1391 sher ds mC3 8 8 211 s herds ??? IC3 6 5 595 s herds Catterick Racecou rse - CAS Site 273 Binehester Ps 1 eC2 13 2.9 87 rims A458 IC3 22 7 102 s herds Ps 2+ IC2 49 27 8 . 3 27 rims

Turrets Catteriek Bridge - CAS Site 240 Ps 3 eC3 15 7 0.6 31 rims Turret 26A Ps 3/4 C3 22 12 0.5 32 rims Per lA 73 sherds Ps 3/5 C3-eC4 19 13 0.9 46 rims Per IB 78 sherds Turret 35A Shiptonthorpe Per I 141 sherds 1985 Con 140 C2 4 291 sherds Turret 258 1986 Con 206 C2 7 9 84 sherds Per lA 452 s herds C1-3 7 6 13 409 sherds Per IB 139 s herds Turret 518 111 s herds HawL ing Road Turret 18B Con 500 C1st o 301 sherds Per lA 348 s herds Con 4007 C1st 355 s herds Per IB 154 s herds Con 547 Flav-Traj? 56 sherds Limestone Con 4030 Had- eC3 63 sherds Bank 39 rims PoLtross Burn Catcote Per I debris 48 rims 012 C2-eC3 0.4 43459

Table 2. Finewares in the 2nd and 3rd centuries J eremy Evans 109 suitable as messhalls in the average fort. Early forts have a very high incidence of graffiti compared with Si te % by sherd % by ~t % by min nos % of n. later forts and this probably reflects not only the nos of rims glassware Forts literacy of the garrison, but also the greater need for labelling and controlling provisions when they were South ShieLds layer 30 15 rims being used communally (Evans 1987). Thus the early VindoLanda 184 11 185 sherds northern military assemblages look quite different Cast leford 260 6 121 sherds Binchester from rural civil sites. phase 8b mC4 19 176 sherds Segont i urn Ps 7B lC3-eC4 13 Ps 8 lC3-eC4 15 Conclusions Ps 9 eC4 11 Ps 10 e-mC4 13

In the 3rd and 4th centuries there is a steady rise in TownsLv; Cl the use of heavily tempered fabrics through time and Old Penri th 7 274 sherds a rise in the proportion of jars in the function Skeldergate figures. The two sets of figures are interdependent ~ell mC4 13 159 sherds Mat ton mC4 13 11 299 sherds but they do not correlate exactly because of the use Brough -on-Humber of heavily tempered dishes and bowls as well. Period VIII eC4 0 11 rims The general trend towards cooking type wares, B II (4) eC4 7 29 rims F XII (3) mC4 7 68 rims shown especially in the rise of handmade fabrics, Catter i ck - CAS si te 240 points to the problem that ceramics do not form a Ps 5 eC4 13 0.7 76 rims Ps 6 mC4 16 0.9 161 rims stable segment of the human artefact kit throughout time. This is implicit in the existence of aceramic periods, but it tends to be overlooked when pottery Beadlam DXIISE is being used. Figs. 15-17 show the functional 'occupation' 7 6 189 sherds breakdown of longer time sequences for comparison. Hutton Ambo 156 sherds Apperl ey Dene 10 21 rims Fig. 15 derives from a sequence from Bath, at Carrawburgh 12 rims Orange Grove (Evans and Millett 1992 (written in MithraelJl1 Per III Hawl ing Road 1979», and though some changes may reflect C3-4 894 sherds differing uses of the site in time the overall trends are clear. Fig. 16 is the sequence from the Roman Table 3. Finewares in the early 4th century. and post-Roman occupation at Chester-le-Street, Co Durham (Evans 1991) and Fig. 17 is a late medieval to early post-medieval sequence from the Newcastle employed was subtly shifting within the Roman castle ditch derived from Ellison's (1981) report. period in the north, rather than it just being that These longer sequences show clearly that dishes certain functional types were becoming more and bowls, for example, were extremely scarce popular. It is notable that amongst the contents of the throughout the medieval period and it is not until the Huntcliff well Hornsby and Stanton (1912, 222) 15th to 18th centuries, depending upon the site, that report fragments of two wooden dishes and that they the diversity of a Roman style assemblage returns. were noted in Paulinus' letters (above) at this time as (They also suggest that post-medieval functional commonplace tableware amongst the less affluent. differences may reflect variations in site use and At the other end of the scale were glass and metal status (especially the Newcastle drinking vessel vessels which were probably used as tableware by figures». It seems apparent that the function of the higher social classes, which will be discussed dishes and bowls was not redundant throughout the further below. medieval period, but that it was simply reflected in some other, non-surviving material, in this case The concept of fmewares turned wooden vessels, treen (Dyer 1982) seen in the Young (1977) cast some doubt on the validity of the waterlogged well deposits from Southampton (Platt concept of finewares in antiquity pointing out that and Coleman-Smith 1975, II, 228ff). The long term Diocletian's price edict values pottery according to trends in technology and especially the late 4th to its capacity and not its quality. The logic of this must early 5th century trends in the function figures from be that the main market for ceramics in the edict was this study do rather seem to imply that the balance of seem as being for containers for other commodities functional types for which ceramics were being since it would be fairly meaningless to try to price 110 Jeremy Evans dishes and bowls by capacity. This is an important Site % by sherd % by wt % by min nos % of glass n. point since most finewares are tablewares and the nos of rims price edict is, therefore, irrelevant to their value. Forts

A survey of graffiti on Romano-British ceramics Binchester (Evans 1987) does produce evidence that samian, at A2068 14 9 313 sherds least, was regarded as a fineware in antiquity A1884 13 8 459 sherds A2037 12 14 618 sherds although the results were more ambivalent with Piercebridge regard to later Roman finewares. Given this the best HS77EW 12 8 316 sherds Old Penri th 9 18 44 sherds test of whether the concept of finewares is valid for the later Roman period is in the consistency, or Fortlets and otherwise, of the relationship between the proportion Signal Stations of finewares and the type of site. Marsh (1981) has Pol tross Burn 48 rims objected that it is invalid to expect the proportion of Huntcliff 9 7 353 sherds Scarborough (#) 7 353 rims finewares on a site to reflect its social status because Goldsborough (#) 655 rims of fluctuations in supply. This, however, is only true Filey; Carr Naze (#) 8 44 rims if sites of different periods are being compared, or if simplistic examination is being made of a single site Towns/vi ci sequence. The idea is quite valid provided that a site Catter i ck 18 250 rims is being compared with other contemporary sites and Catted ck CAS Si te 240 sequences can be examined providing they are ps 7 10 6 0.9 Lincoln 23 310 rims compared with contemporary sequences. Sh i ptonthorpe One of the main theoretical problems with 14 lc4 7 340 sherds examining the social status of a site by the percentage of finewares on them is the use of non­ Beadlam 7247 sherds ceramic tablewares on high status sites. Whereas Rudston well wood seems to have been the chief Roman alternative layers 266·7 >2 257 rims to coarse pottery there are many possible alternatives East HesLerton 4 28 rims Long IIhins (#) 7 61 rims to fine tablewares and high status sites probably Crossgates £8 routinely used much more metalwork than other # - In order to calculate these figures it has been assLI11ed that sites. Glass, gold, silver, bronze and pewter could all 50% of Hull's (1932) type 6 were in fineware for these sites. It provide elements of the tableware and the result is quite probable that this is an underestimate. could be that a high status site would have an apparently lower status pottery assemblage than Table 4. Finewares in the later 4th to early 5th centuries many other type of site eg the Boudiccan stone building at Verulamium examined by Millett (1983). much higher initial cost than earthenware it was Silver tableware would not seem to be exceptional more durable and for wealthy villa sites, remnants of several Romano­ British sets remain in hoards, of which the "pewter was actually better value because it retained its Mildenhall hoard is probably the most impressive monetary value, with old vessels valued at about two-thirds of the new cost, and they could be part-exchanged for (British Museum 1964). Pewter seems to have been new." becoming increasingly common in the later Roman period and it must have been used in at least some Chronological trends in fmewares East Yorkshire villas since a mould for pewter, or Tables 2-4 show the percentages of finewares from possibly sheet bronze, manufacture comes from the the groups which were fully quantified during the Langton villa (Goodall 1972) and another has been author's research (Evans 1985) by sherd number and found at Catterick (Blagg forthcoming). The use of sherd weight, and by minimum numbers of rims and bronze tableware is clearly exemplified by the sometimes RE for those groups which were lacking remains of a bowl found at the Beadlam villa site and in bodysherds or are taken from published sources. the same site has produced a fine collection of late Tables 2-4 do tend to show some long term trends in Roman vessel glass (Henderson J. pers comm). Some time, although there are considerable variation reasons for the growing attraction of pewter in the between sites of the same period which will be later Roman period may be provided by the 17th and considered below. Generally fineware values are 18th century evidence cited by Weatherill (1988, highest in the 2nd-3rd centuries and fall considerably 110-11) who points out that whilst pewter had a in the early 4th century, perhaps with a slight Jeremy Evans 111 recovery in the later 4th century. The individual site supply. This may be the reflection of the decline in sequences which reflect the early 4th century fall the use of beakers in the later 4th century in the area best are those from Old Penrith, Castleford and which may have forced the Lower Nene Valley Greta Bridge but there are regrettably no sites in the industry to attempt to diversify the type of product it study which have usable earlier and later 4th century was selling if it was to keep its northern market. groups. Overall the pattern of northern fineware usage is The earlier 4th century fall in finewares to some rather curious; with the end of samian supplies extent reflects the decision to accept all samian in producing a total lack of fine tablewares which was 3rd century groups but to exclude it from early 4th not filled on any scale until the late 4th century, century ones, especially at Old Penrith where there whereas fine drinking vessels continued to be used is quite a large quantity of residual samian in the up to that period after which their use declined early 4th century, but this would not seem to be the considerably. main reason, which is rather the cessation of samian supplies in the 3rd century. The contribution of Spatial patterning in fmewares Lower Nene valley finewares, although it is difficult An attempt will be made here to examine the to judge, would appear to have been at a fairly percentage of finewares in northern assemblages for constant level throughout the 3rd and earlier 4th spatial patterning and for variations according to the centuries. Therefore, it seems at first odd that Lower type of site. Table 2 shows the percentage of Nene Valley products should not have taken up some finewares from all 2nd and 3rd century sites in the of the position which samian held in the market, study and usable published sources. Overall the however, Lower Nene valley goods in the north were results do seem to bear out the possibility that the simply non-comparable in the early 4th century, the proportion of finewares would vary fairly vast bulk of the Lower Nene valley products being consistently between different classes of site. The beakers, whilst most samian ware vessels were dishes groups from the forts generally seem to have a fairly and bowls . Fineware dishes and bowls were simply high proportion of finewares apart from one of the not on the market in northern England in the early South Shields groups (layer 123) and the Vindolanda 4th century and no local kiln centre tried to fill the sequence; the Castleford group (from a site which is gap on anything more than a negligible scale. Cold perhaps best regarded as a burgus in this period) falls Cam, Crambeck, and the Crambeck copy industry at in well with the other forts. As noted above the Catterick, produced a few redware hemispherical barrack and rampart -back location of the Vindolanda flanged bowls imitating Dr 38, but these were always groups in a fairly marginal location within the fort scarce, in the case of Cold Cam and Catterick may influence the performance of this group. because they had neither the markets nor the scale of However this type of location is also found with the production required, in the case of the latter, South Shields group. It is notable that the Vindolanda perhaps, because rather unsuitable boulder clay was groups are intermediate between the forts and the being used producing a variable and often very turrets. second rate product (Evans 1989). In sharp contrast to the forts are the 2nd century It is strange that the tradition of red-slipped turret groups which have very low percentages finewares , which continued on through the Oxford, although dating to the mid-late 2nd century they New Forest and Much Hadham industries in the might be expected to have access to more plentiful south until the end of Roman Britain died almost samian supplies than the 3rd century groups. The entirely by the end of the 3rd century in the north interpretation of the percentages of fineware from the and no obvious explanation can be found for this in turrets, like that of the function figures for the terms of demand - possibly the answer laying in the turrets, would seem to be that they were purely inadequacies of supply. The slight increase in utilitarian groups, probably mainly for the finewares in the north in the later 4th century is due preparation of occasional meals and the storage of to the advent of Crambeck parchment ware. This food, which did not require the more sophisticated finally fills the void in the northern market for assemblages of the forts as they were only for short­ fineware dishes and bowls although the scale of use term use. The result being that fine tablewares would of finewares generally seems to be less than in the not have been appropriate on such sites (cf Allason­ 3rd century. It is also curious that it is in this period Jones 1988). that the Lower Nene Valley industry seems to have The groups from the vici and towns produce a diversified its northern market into dishes and bowls, fairly similar picture to the forts, but, perhaps with with beakers reduced to a more minor element of its a slightly lower average level, groups from only four 112 Jeremy Evans sites exceeding the 20% level, whereas groups from groups available. One site which is of note is the seven forts do, although they are well above the Carrawburgh Mithraeum, where the fineware trend turrets. One odd set of figures here is provided by reflects that of the function figures with the early 4th the published groups from the mansio at Chester century group being the first to look very similar to (Carrington 1980). These values are very high, even that from other sites. As noted above in relation to for a high status building in a town, and they are the function figures this change must reflect a change already calculated on a minimalist basis for the in the ritual in use on the site, or perhaps even a fineware. They would seem to demonstrate the change in the use of the site to something other than difficulties of this data noted above, and it seems that its previous religious purpose. the quantity of samian in particular has been over­ Table 4 presents all the fineware data for late 4th estimated in relation to the coarseware. The other century groups in the north which could be obtained. odd group is from phase 2 + at Catterick This data, like that for the late 2nd to 3rd centuries Racecourse, this is a gravelled area associated with shows very considerable regularities according to the a cemetery and the high fineware level, even if type of site. The best supplied site is Segontium, measured by weight, combined with a high followed by the poorly known rural site at Healam proportion of glassware relative to the quantity of Bridge and the Hilyard Catterick group (Gillam pottery suggests some form of ritual use. The other 1957), although the data from CAS site 240, which remaining odd site is the Carrawburgh M ithraeum , includes much more residual material but is from a where although the groups are small, the exceptional marginal location on the fringes of the town (Wilson result is unlikely to be a freak value. The fineware 1984), produces much lower levels. value here reflects the aberrant functional value, with These are followed by the fort groups from most of the vessels being fineware beakers, which Binchester, Piercebridge and the very small and were presumably used for the religious rituals dubious group put together from Old Penrith, which associated with the temple. run at a markedly higher level than the Poltross Burn The rural groups are the pre-villa ditch from milecastle and the Signal Stations which are all very Beadlam, Catcote (Millett unpublished), four north similar. The rural sites all seem to cover a very Welsh rural sites and the Iron Age to 2nd century similar range, excepting Healam Bridge, generally site at Hawling Road, north Humberside. These with a lower level of finewares than the forts and figures are at a level comparable with those of the urban centres and there appears to be no obvious turrets, except for Bryn Eryr, which stands out in difference between the villas and other rural sites. this data set as it does in comparison with other The very low level from the Rudston villa can be north Welsh sites. Here there is quite reasonable explained the deposit being from the well, discussed evidence, inter alia seen above in the function above, but the low level of finewares from the figures, for a rural site with a status display Beadlam collection cannot be explained away by orientated assemblage (Evans forthcoming A). variations within the site, it being a large collection Figures for the percentages of finewares on early covering all three villa ranges (Evans in prep, b). 4th century sites in the region are displayed in Table Once more forts and towns seem to have the best 3. The interpretation of this data would seem rather supplies of finewares, even at the end of the 4th more problematical than that of the preceding period, century. this would seem to be partly a consequence of the The position of the Signal Stations in the late 4th data themselves, but also because of there being less century is similar to that of the late 2nd century data than for the 2nd-3rd centuries, and perhaps also Hadrian's Wall turrets on first appearances, but they because of geographical and chronological variation do appear to be rather further up the hierarchy, being within the period. Thus whilst Lower Nene Valley clearly better supplied than rural sites, whereas the products might be found across the north, early turrets appear to be on much the same level. This is Crambeck parchment ware and redwares, which have in contrast to the picture produced by the function been regarded as finewares, were mainly restricted to figures where the Signal Stations appear to be quite the North-East and also were probably more analogous to the late 2nd century turrets. commonly available towards the mid 4th than in the Similar evidence to the general pattern outlined early 4th century outside East Yorkshire. Whatever above has come from other studies; the greater the cause there seem to be less clear general trends variety of fineware supply to towns and forts over in fineware distribution by class of site. However, villas and rural sites has been noted by Millett (1980) the urban and military sites, although varying, still in West Sussex. Similarly the distribution of produce generally higher levels than the few rural Romano-British graffiti cut after firing on ceramics Appendix 1. Second century junction figures

Site Jars Bouts Dishes Beakers Flagons ~orae lids Mortaria Other Dish+bo n.

Limestone Bank 70% 16% 5% 3% 3% 5% 21% 39 Pol tr~ss Burn mid C2 52% 15% 13% 8% 13% 27% 48 Turret 18b, period lA 88% 6% 6% 6% 17 Turret 51b, period IB 58% 8% 17% 8% 8% 8% 12 Turret 26a, period lA 67% 7% 7% 13% 7% 15 Turret 35a, period I 56% 6% 6% 19% 6% 6% 12% 16 Turret 25b, period lA 59% 6% 9% 22% 3% 6% 32 Benwell vallum fi 11, mC2 49% 5% 3% 1% 6% 41% 108 South Shields, layer 56, mC2 17% 53% 17% 7"10 35% 3% 70% 30 South Shields, layer 123, mC2 61% 22% 9% 9% 31% 23 Segontium phase 2 31% 15% 8% 23% 8% 8% 8% 23% 13 Flavian phase 3 40% 12% 20% 4% 16% 4% 4% 32% 25 Flavian phase 4 17% 12% 35% 17"% 9% 4% 47% 23 Flav-Trei phase 5 40% 13% 11% 16% 2% 15% 3% 24% 67 Traj-Had phase SA 35% 17% 22% 11% 3% 2% 8% 2% 2% 39X. 65 phase 5B 28% 16% 19% 16% 5% 15% 1% 35% 86 phase 6 24% 19% 26% 5% 17"10 10% 45% 40 phase 6A 47% 26% 13% 3% 1% 9% 1% 39% 93 Chester 27- 27- 12- 14- 10- 4- 3- 3- 39- 223- AD80-130 35% 18% 11% 9% 14% 5% 4% 4% 29% 171 Chester 22- 40- 7- 9- 7- 8- 7- 1- 47- 129 IC2nd 36% 12% 12% 4% 12% 13% 12% 1% 23% 78 Bri thdi r, F lavi an 19% 22% 22% 25% 6% 3% 3% 44% 32 Shiptonthorpe, IC2 70% 13% 9% o 2% o 6% o o 21% 47 Catterick Bainesse Farm, phase 3 27% 32% 8% 8% 5% 8% 8% 43% 37 phase 4 23% 23% 15% 15% 13% 3% 3% 8% 38% 40 phase 5 35% 22% 12% 13% 4% 1% 4% 4% 7% 35% 172 phase 6 31% 27% 12% 11% 5% 2% 5% 5% 41% 505 Catterick Racecourse phase 1 22% 15% 21% 7% 11% 1% 8% 5% 8% 38% 87 phase 2+ 22% 22% 30% 11% 4% 4% 7% 49% 27 Carlisle Blackfriars, P8b- j 23% 37% 15% 20% 3% 3% 13% 51% 91.9 vessels York Rougier Street,

P2 24% 27% 7% 10% 15% o 5% 6% 2% 38 98.4 vesse I 5 York Trentholme Drive# C2-3 58% 2% 1% 18% 16% .5% .5% 4% 4% 225 Beadlam ditch IC2 64% 18% 9% 9% 18% 11 L i ngcroft Farm C1-2 60% 20% 13% 7% 20% 15 Hawl i ng Road Cl 90% 3% 3% 2% 2% 7% 58 Hawl ing Road C2 91% 5% 3% 1% 8% 75 Graeanog Cl-eC4 56% 4% 28% 4% 8% 32% 25 Cefn Graeanog Cl-eC4 64% 11% 23% 3% 34% 66 vessels Bryn Eryr C1-3 33% 31% 24% 7% 2% 2% 55% 45 Bush Farm C1-3 44% 18% 27% 6% 6% 45% 55

# Samian missing from Trentholme Drive figures 114 Jeremy Evans

Appendix 1. Third century function figures

5ite Jars Bowls Dishes Beakers Flagons Amphorae Lids Mortaria Other Dish+bo n.

Vindolanda ec3 285 55X 15X 20% 10% 35% 20 314/317 65X 4X 19X 4% 4% 4X 23% 26 mC3 91 52Y. 13X 22X 3% 3% lY. 35% 60 IC3 113 59% 10% 5X 5% 5% 15% 15" 39 South Shields eC3 32X 53X 4% 11% SlY. 47 Bewcastle C3-eC4 34% 34X 18X 6X 3X 6" 52% 68 Castleford, IC3 34% 29% 24X 8% 3X 3X 53% 38 Segontiun 26% 30% 1lY. 9% 5X 2X 8X 3% 3% 4lY. 157 phase 7 phase 7A 36X 23% 9% 8X 6X 2X 12X 3% 1% 32X 121 phase 7B 61X llY. lY. llX 4X lY. 18% 28 phase 8 39X 31Y. 10X 5% 5% 8X 3X 41X 78 Chester C3, 25 - 26- 20- 11- 9- 2- 1- 5- 1- 46- 227 well 11 32X 15% 22X 10X 11% 3% lX 5X 2% 3lY. 178 Greta Bridge, mC3 39% 2lY. 1lY. 12X 2X 2% 44% 41 IC3 34X 25% 10X 10X 5X 2X 14% 36X 59 Old Penrith C3 25X 4lY. 15X 12X 2X 62% 147 Brough, Per VI e-mC3 55X 21X 7X 9X 2X 2% 5% 2lY. 44 G 11 (6)etc, C3 73X lY. 13X lY. 20X 15 Per VII, IC3 55% 14X 10% 7X 7X 7X 24% 29 B I (13) etc, C3 59% 15% 11% 4% 4X 4X 4X 26% 27 Catterick, Bainesse Farm phase 6-7 30X 21% 12X 14X 6X o 4X 6X 6" 36" 238 phase 7 23X 28% 15% 11% 5X lX lX lY. 8% 43% 263 Catterick Racecourse phase 3 28X 28% 15X llX 2X 4X 4" 49% 46 phase 3B 60X llX 16X 4X 4X 2% 2X 27X 45 Catterick ~ridge phase 3 32X 19% 19X 13X 3X lY. 3% 41% 31 e-mC3 phase 3-4 34% 22% 16% 6% 3% 9% 3" 44% 32 m-IC3 phase 3-5 30% 22% 13% 13% 4X 2X 15X 3S~ 46 m-IC3 Carrawburgh mithraeun I+IIA 29X 14X 50% 14% 14 IIB-IIC 22% 4% 17X 44% 13X 21X 22 Hawling Road c3-4 74X 11% 11% lX lX o o 3% o 22% 95

# - Samian missing fr~~ Trentholme Drive figures. J eremy Evans 115

Appendix 1. Early fourth century junction figures

Site Jars Bowls Dishes Beakers Flagons Amphorae Lids Mortaria Other Dish+bo n.

Vindolanda, eC4 (#) 50% 10% 17% 20% 3% ? 27% 30 South Shields, eC4 47"10 20% 20% 13% 40% 15 Chester well II (3) 18% 16% 14% 20% 14% 8% 8% ??? 49 Castleford, ditch 260 42% 23% 4% 12% 4% 8% 8% 27% 26 Skeldergate well, mC4 39% 23% 19% 4% 8% 4% 4% 42% 26 Mal ton, mC4 49% 23% 28% 5% 3% 3% 3% 51% 39 m-lC4 46% 7"10 39% 4% 4% 46% 28 Carrawburgh mithraeum,eC4 42% 25% 17% 8% 8% 42% 12 Greta Bridge 49% 18% 21% 6%? 6% 39% 33 Old Penrith 55% 20% 9% 7% 9% 29% 55 Brough'on- 62% 23% 8% 8% 23% 13 Humber, Per VIII B II (4) etc 38% 17% 24% 7% 14% 41% 29 F XII (3) etc, mC4 35% 30% 17% 6% 9% 3% 47"10 66 Beadlam, N of N wing B 48% 17% 22% 3% 1% 2% 8% 38% 116 D XI I SE mC4 43% 24% 17% 5% 2% 2% 7% 41% 41 Apperley Dene 67% 24% 5% 5% 29% 21 Rudston well eC4 43% 2% 1% 54% 1% 3% 148 Hutton Ambo 28% 39% 22% 6% 6% 61% 18 Segontium phase 9 44% 27% 11% 5% 1% 1% 5% 6% 38% 149 Catterick Bridge phase 5 e-mC4 48% 15% 4% 11% 1% 8% 8% 22% 76 phase 6 mC4 47% 24% 7% 7% 1% 1% 6% 4% 36% 161

# - Mortaria not included in Vindolanda figures

shows a pattern of the proportion of graffiti on transport costs, the problem of interpreting this fineware declining down the settlement hierarchy distribution must be addressed. Although the (Evans 1987, Fig. 2; Table A) from forts to rural proportion of finewares on a site would appear, a sites, a pattern which probably reflects the proportion priori, to be most closely connected with the social of finewares on the sites concerned since otherwise status of the site the results of both this study, poorer sites might be expected to take more care of Griffiths' (1989) Northamptonshire survey and their finewares than richer ones. Millett' s West Sussex survey (1980) do not really Given that there do appear to be regularities in bear this out. (A similar picture might also be fineware supply to sites determined by class of site expected in Warwickshire from the data published by rather than the pattern merely being determined by Booth (1991), although his figures are not directly 116 J eremy Evans

Appendix 1. Late fourth century junction figures

Site Jars Bowls Dishes Beakers Flagons Amphorae Lids Mortaria Other Dish+bo n.

Scarborough 79% 15% 4% 0.5% 0.5% 2% 19% 655 Fi ley 75% 14% 7% 5% 20% 44 Goldsborough 84% 8% 3% 5% 11% 94 Huntcl iff 71% 12% 17% 29% 49 Poltross Burn, Per III 77% 8% 6% 8% 15% 48 Catterick 59% 16% 12% 0.5% 1% 1% 5% 28% 250 town Catterick Bri dge* phase 7 44% 26% 10% 6% 1% 1% 8% 3% 38% 302 Lincoln 46% 19% 14% 9% 4% 2% 1% 33% 310 Shiptonthorpe 55 lC4 59% 12% 15% 2% 5% 5% 2% 27% 41 Great Casterton 45% 24% 12% 4% 5% 1% 5% 36% 350 Old Penrith 44% 50% 6% 56% 16 Binchester A2068 50% 20% 17% 6% 4% 4% 37% 54 A2037 48% 21% 17"10 3% 2% 4% 7% 35% 96 A1884 53% 15% 15% 5% 3% 5% 4% 30% 79 Piercebridge HS77EX 51% 29% 11% 1% 3% 4% 39% 76 HS77EW 56% 20% 15% 4% 4% 2% 35% 55 Segontium phase 10 47% 23% 9% 6% 1% 4% 11% 1% 32% 211 phase 10A 48% 24% 16% 4% 1% 1% 5% 40% 335 phase 11 57% 26% 7% 1% 1% 9% 33% 129 Rudston well, lC4 92% 6% 2% 8% 257 Beadlam, all lC4 groups 63% 19% 11% 1% 1% 1% 4% 30% 964 Long Whins 67% 21% 8% 3% 29% 61 E Heslerton 82% 7% 7% 4% 14% 28 Healam Bridge E.Field lC4 49% 14% 19% 8% 3% 8% 33% 37

*NB includes residual

comparable as they include amphorae, coarseware mass of common soldiery could hardly be so mortaria, and white-slipped flagons). Villas would described and social status could not possibly seem to be consistently lower in the hierarchy than describe the difference between the forts, turrets and might be expected and although towns might be Signal Stations. expected to house a collection of high status Given the general similarity of the 3rd and late 4th individuals, as well as lower status members engaged century groups it would seem that the proportion of in commerce, there seem to be few grounds at all for finewares in the assemblage is a product of the believing that high fineware values from forts reflect aggregate purchasing power of the site and probably the higher social status of their inhabitants. Whereas also its function as a redistributive centre, together commanders might well be high status individuals the with the function of the area or the site as a whole. Jeremy Evans 117

Weatherill (1988) has pointed to a similar Bibliography phenomenon in the hierarchy of sites recelvmg Allason-Jones, L. 1988: 'Small finds from turrets on 'modern' traded goods in the 17th-18th centuries, Hadrian's wall' in Coulston 1988, 197-233. where merchants and urban dwellers were more Bennett, J. and Young, R. 1981: 'Some new and some likely to obtain these in quantity than those in forgotten stamped skillets, and the date of P. Cipius Polybius', Britannia XII, 37-44. more'high status' rural residences (country houses). Bidwell, P. 1985: The Romanfort ofVindolanda, Hist Build The differences between the turrets and the forts (of Mon Comm England Rep 1, London. which the pottery from the turrets is most probably Blagg, T. forthcoming: 'The Catterick worked stone', in a sUb-set) must be determined on functional grounds. Wilson, PR, Excavations at Catterick 1956-88. The high fineware level in forts, and to an extent Booth, P. 1991: 'Inter-site comparisons between pottery towns, would generally reflect the concentration of assemblages in Roman Warwickshire: ceramic indicators a large number of consumers in a single place, a of site status', J Roman Pottery Stud 4, 1-10. Breeze, D. and Dobson, B. 1978: Hadrian's Wall, Middlesex. concentration which would make a fairly continuous British Museum 1964: The Mildenhall Treasure, a handbook, effort at supply worthwhile, whilst isolated London. consumers on rural sites would have such a low Bronitsky, G. and Hamer, R. 1986: 'Experiments in ceramic demand at each place, even if high status individuals, technology: the effects of various tempering materials on that it would not be possible to supply them directly, impact and thermal-shock resistance', American Antiq so that redistributive centres would be essential for 51(1), 89-101. finewares to be supplied to rural areas, whereas Budd, et a11989: P, Chapman, B, Jackson, C, Janaway, R, coarsewares might have been supplied directly. This and Ottaway, B, Archaeological Sciences 1989; Proceedings of a conference on the applications of is much in line with Peacock's observation that scientific techiques to archaeology, Bradford, September 'pottery shops' generally specialise in finewares and 1989, Oxbow monograph 9, Oxford. mortaria (and sometimes BB1), types for which there Carrington, P. 1980: 'The pottery', in Mason 1980, 00-00. is too low a demand on anything less than nucleated Chadwick, H. and Chadwick, N. 1955 Poetry and letters in centres for direct supply. If the towns were acting as early christian Gaul, London. marketing centres for finewares, if not necessarily Charlesworth, D. 1960: 'A Roman well at Benwell', Archaeol for coarsewares, then they might be expected to have Aeliana 4, XXXVIII, 233-5. Corder, P. and Kirk, J. L. 1932: A Roman villa at Langton, a higher level of supply than the forts because their near Malton, Roman MaIton District Res Rep 4, Leeds. demand was effectively not simply their own Coulston, J. C. (Ed) 1988: Military equipment and the population but also that of their market areas. identity of Roman Soldiers; proceedings of the fourth If this explanation of the fineware distribution is Roman military equipment conference, Brit Archaeol Rep reasonable then the only differences it is liable to Int Ser 394, Oxford. show by class of site which should be accounted for Darling, M. 1977: A group of late Roman pottery from mainly by social status are those between villas and Lincoln, Archaeol Lincoln, XVI-I, Counc Brit Archaeol, other rural sites, since both needed to acquire their London. Dore, J. N. 1983: 'The coarseware', in Miket 1983, 61-107. supplies through the same mechanism. Ellison, M. 1981: 'The pottery', in Harbottle and Ellison 1981,95-164. Acknowledgements Evans, J. 1985: 'Aspects of later Roman pottery assemblages The drawings are the work of Mr N Dodds. This in northern England', unpublished Univ Bradford PhD paper is a slightly revised chapter from my thesis thesis. (Evans 1985) supported by a grant from the SBAC of Evans, J. 1989: 'Crambeck; the development of a major SERC. northern pottery industry', in Wilson 1989, 43-90. Evans, J. 1991: 'The coarse pottery', in Evans et al 1991, Endnote 5-48. New Testament references to pots are: Evans, J. forthcoming (A): Bryn Eyre. Matthew 10:42 and Mark 9:41 Evans, J. forthcoming (B): 'The pottery from Hawling Road, Matthew 23:25-6, Mark 7:4-8 and Luke 11:39 north Humberside', in Creighton, J, Report on excavations Matthew 26:27 and Mark 14:23 at Hawling Road, North Humberside. Mark 14:3 and Luke 22:10 Evans, J. forthcoming (C): 'Reflections on later Iron Age Mark 14:20 and 'native' pottery in north-east England' in Vyner, B (Ed) Luke 8:16 The north Yorkshire Moors after the antiquaries. Luke 22:42 Evans, J. in prep (a): Shiptonthorpe. John 4:28 Evans, J. in prep (b): The Beadlam villa Roman pottery. John 13:5 Evans, J., Jones, R. F. J. and Turnbull, P. T, 1991: John 18:11 'Excavations at Chester-le-Street 1978-9', Durham Jeremy Evans. April 1994. Archaeol J 7,00-00. 118 Jeremy Evans

Evans, J. and Millett, M. 1992: 'Residuality revisited; an Millett, M. unpublished: The Catcote Roman pottery. examination of the behaviour of residual pottery ' , Oxford Newbold, P. 1913: 'The pottery and periods of occupation', J Archaeol 11. 2, 225-240. in Simpson 1913,297-397. Evershed, R. P., Heron, C., Charters, S. and Goad L. J. Richmond, I. A. and Gillam, J. P. 1951: 'The temple of 1992: 'The survival of food residues: new methods of Mithras at Carrawburgh', Archaeol Aeliana 4, XXIX, analysis, interpretation and application, Proc Brit Acad 77 , 1-92. 187-208. Rutter, J. G. and Duke, G? 1958: Excavations at Crossgates Gillam, J. P. 1957: 'The coarse pottery ' , in Hilyard 1957 , near Scarborough, 1947-56, Scarborough Dist Res Rep 1, 224-265. Scarborough. Gillam, J. P. 1970: Types of coarse pottery in northern Platt, C. and Coleman-Smith, R. 1975: Excavations in Britain, Newcastle-upon-Tyne. medieval Southampton, 1953-1969, Leicester. Gillam, J. P. 1968: 'The coarse pottery', in Wenham 1968, Perrin, J. R. 1990: Roman pottery from the colonia 2; 00-00. General Accident and Rougier Street, Archaeol York 16, Goodall, I. H. 1972: 'Industrial evidence from the villa at Fasc 4, London. Langton, East Yorkshire' , Yorkshire Archaeol J 44 , 32 . Rigby, V. 1980: 'The coarse pottery', in Stead 1980,45-94. Greene, K. 1978: 'ApperJey Dene Roman fortlet, a re­ Rye, O. S. 1976: 'Keeping your temper under control; examination, 1974-5, Archaeol Aeliana 5, VI,. 29-59. material and the manufacture of Papuan pottery ' , Archaeol Griffiths, K. 1989: 'The marketing of Roman pottery in Physical Anthropol Oceania 11(2),106-137. second century Northamptonshire and the Milton Keynes Schiffer, M. B. 1976: Behavioural Archaeol, xxx. area' , J Roman pottery Stud, 2, 66-76. Simpson, F. G. 1913 : 'Excavations on the line of the Roman Harbottle, B. and ElIison, M. 1981: 'An excavation in the wall in Cumberland during the years 1909-1912' , Trans castle ditch, Newcastle-upon-Tyne, 1974-6', Archaeol Cumberland WestmorlandAntiq Archaeol Soc 12, 297-397. Aeliana 5, IX, 75-250. Stead, I. M. 1980: Rudston Roman villa, Leeds. Hassall, M. and Tomlin, R. 1979: 'Roman Britain in 1978; Steponitis, V. P. 1983: Ceramics, chronology and community n inscriptions', Britannia X, 346-356. patterns; an archaeological survey at Moundville, New Heron, C., Evershed, R. P., Goad, L. J. and Denham, V. York. 1991?: 'New approaches to the analysis of organic residues Sumpter, A. B. 1990: 'Iron Age pottery' in Wrathmell et al from archaeological remains', in Budd et al1991 ? , 00-00. 1990, 128-30. Hilyard, E. J. W. 1957: 'Cateractonium, fort and town ' , Taylor, J. 1990: 'The pottery' , in McCarthy 1990, 197-300. Yorkshire Archaeol J 39, 224-265. Weatherill, L. 1988 : Consumer behaviour and material Hornsby, W. and Stanton, R. 1912: 'The Roman fort at culture in Britain 1660-1760, London. HuntcIiff, near Saltburn', J Roman Stud n, 215-232. Webster, G. 1969: Coarse pottery; a student's guide, Counc Hull, M. R. 1932: 'The pottery from the Roman signal Brit Archaeol Res Rep 6, London. stations on the Yorkshire coast', Archaeol J LXXXIX, Wenham, L. P. 1968: The Romano-British cemetery at 220-253. Trentholme Drive, York, Min Pub Build Works Archaeol Johnson, A. 1983 : Romanforts, London. Rep 5, London. Longacre, W. A. 1970: Archaeology as anthropology; a case Wilson, P. R. 1984: 'Recent work at Catterick' , in Wilson, study, Anthropological papers of the University of Arizona et al 1984, 75-82. 17, Tucson, Arizona. Wilson, P. R., Jones, R. F. J. and Evans, D. M. (Eds) 1984: McCarthy, M. R. 1990: A Roman, Anglian and medieval site Settlement and society in the Roman north, Bradford. at Blackfriars Street, Stroud, G1oucstershire. Wilson, P. R. (Ed) 1989: The Crambeck Roman pottery Mason, D. J. P. 1980: Excavations at Chester; 11-15 Castle industry, Yorkshire Archaeol Soc Roman Soc Mon, Leeds. Street, and neighbouring sites 1974-8, Grosvenor Mus Woods, A. 1983-4?: 'The old pot boiler' , Bull Experimental Excav Surv Rep 2, Chester. Firing Group 2, 25-31. Miket, R. 1983 : The Roman fort at South Shields, Newcastle­ Wrathmell, S. and Nicholson, A. 1990: Dalton Parlours, Iron upon-Tyne. Age settlement and Roman villa, Wakefield. Millett, M. 1979: 'An approach to the functional Wright, R. P. and Hassall, M. 1970: 'Roman Britain in 1969; interpretation of pottery, in Millett 1979 (Ed), Pottery and n inscriptions', Britannia 1, 305-315. the archaeologist, Inst Archaeol Occ Publ 4, 35-47, Wright, R. P. and Hassall, M. 1971: 'Roman Britain in 1970; London. n inscriptions', Britannia n, 292-304. Millett, M. 1983 : 'A comparative study of some Wright, R. P., Hassall, M. and Tomlin, R. S. O. 1976: contemporaneous pottery assemblages', unpublished Univ 'Roman Britain in 1975; n inscriptions' , Britannia VII, Oxford DPhil thesis . 382-392.