Extracting meaning from ploughsoil as semblages

Edited by Ríccardo Francovich, Helen Patterson and Graeme Barker q.cv Oq,(a- tY)u/

Oxbow Books O *C 2000 -.s (/) +) 4. Ceramics and the site: is survey enough?

Vincent Gafrney

INTRODUCTION discussion with several observations on basic field survey The expansion oF the role of surface survey along with an methodology. Whilst the last 20 years saw a revolution in emphasis on the study of ploughsoil assemblages was a the development of survey methodology, it seems fair to notable characteristic ofEuropean archaeology of the late observe that, like excavation before, the basic techniques, 1970s and 80s. The reasons for these developments have offield survey are now well established and are unlikely to been discussed in a number of publications. Whilst these change significantly (Biddle, 1993). This, as in the case of . need not be repeated at length here, it is important to note excavation, does not mean that there will be no advances the role of conceptual developments related to the nature in scientiñc techniques which may be applied to surflace- of the site and an awareness of increasing threats to the survey data, or novel archaeological questions which may surviving archaeological record in these changes (Barker, be attempted through survey, Rather this suggests that the l99l; Gaffney and Tingle, 1984). The rapid development basic data collection techniques, such as gridded intensive of survey and the general optimism displayed during this and extensive collection strategies combined in a variety period rvas, perhaps, balanced by the observation made of sampling designs, are now likely to remain the mainstay by Haselgrove et al. in 1985 that'it was all still to do', of data capture. It is also true that alongside a robust set of' and certainly the impression of the period was that the field methods we also have a reasonable grasp of thel lack of detailed archaeological large parts processes information for taphonomic which may affect survey data. , of Europe could only be rectified by extensive survey Although such processes change with geographic, geo- rather than by any site-based methodology, logical and climatological situation, and specific re-qional In the interval following the publication of the Archae- circumstances will always vary, it is still largely correct to ology front the PLoughsoil volume (Haselgrove et a/., state that current survey theory and methodology can 1985), Europe has witnessed a floruit of surveys (Barker encompass such variation. and Lloyd, l99l ). The technical aids related to the capture One might also observe that even rvhen major techno- and analysis of survey data have improved dramatically, logical advances with, apparently, clear implications for particularly with the introduction of GIS technologies survey analysis emerge, these may not present immediate (Lock and Stanðió, 1995), and new publications have or significant advances for survey method or theory. GIS demonstrated the value of survey in a wide variety of is a case in point. Although the technolo_sy holds, perhaps, situations. From the survey of Mediterranean towns to the greater potential for analysis of survey data than for any study of the northwestern wetlands, survey has demon- other aspect of archaeology (Biswell e!. al., 1995), the strirted its value (Bintliff and Snodgrass, 1988; Carreté et current debate amongst archaeological GIS practitioners aL., 1995; Crowther et al.,1985). Equally significant have indicates considerable uncertainty about the overall context been the publications of archaeological syntheses, specifìc- of the technology and its role even within landscape analysis ally Sue Alcock's (1993) Graecia Capta, which demon- (Gaffney et al., 19951, Wheatley, 1993). There are grounds strated the potential of survey data for detailed synthetic to suggest that the real impact of GIS may emerge from its analysis. There can be little doubt that surface survey has integration within the general theoretical domain of archae- made a significant contribution to archaeology in general. ology, in its widest sense, rather than simply as a survey Horvever, nearly a clecade later it is perhaps worth re- specific tool (Gaffney and van Leusen, 1995), On the basis considering whrt has been achieved, rvhat is still left to do of the current situation it may be stated, lvith some and rvhat, ilany, are the limits of survey data for analysis. justification, that we have achieved nlost of the easy . In considering these questions it seems pertinent to start advances within the domain of survey methodology. Despite this, the relative stability and robustness of survey metho- dolo-eies over the last decade has allotved practitioners to establish survev as a legitimate pursuit in its own right, **r. rather than sinrply to act as a provider of sites to dig or a quarry for nice material (Richards, 1985). Horvever, the s¡atøs guo is still not entirely satisfactory. In those areas *'here survey is regarded as a natural precurser to most forms of archaeological f,reldwork, there may be grounds to believe that much *'ork is only lip service to survey and not particularly substantive or ellective. Elservhere, one feels that survey is sometimes seen as an alternative to other forms of archaeological activity, notabl¡'excavation. \\¡here this occurs, the decision to carry out survey may be taken purely on archaeological grounds but partl)' because there are difficulties in procuring excavation licenses or because excavation is considered too expensive for the diminishing resources of many academic institutions.In such instances one might consider u'hether, having argued for survey to be valued in its own right alongside excavation, there is norv a need to balance the important contributions of survey data with an appeal for integrated survey and excavation programmes (Bowden er al., l99l). There are tn,o basic contexts in which such an appeal may be necessary. The first involves those situations where excavation is essential to add to, and refine, the spatially Fig. 4.1 The Adriatic Island Project: tlrc study area in its extensive and relatively coarse database provided by regional conlext. survey. An example of such a situation is provided by rvork carried out over a number of years in by the N Adriatic Islands Project (see Gaffney el a/., this volume; Fig.4.1).

PHAROS,

The investigation of the relationship betu'een the assumed Greek and Roman urban settlement at Pharos/Pharia at Stari Grad on the island of Hvar drew upon survey and excavation carried out by a variety of institutions between 1987 and L992.The presence ofa Parian Greek colony on the' site is well attested both historically and archaeo- logically, although the extent of this settlement and its development over time has been the subject of much debate. The existence of a pre-Greek settlement has been discussed but never proved, whilst occasional finds of pottery pre- dating the historical founding of the colony hinted at the existence of a pre-colonial Greek settlement on the site of the colony (Mi-egoti, 1986; 1989). Perhaps the single important point of consensus concerning the urban topo- graphy of Pharos was the opinion that the remains of a subsrantial defensive drystone wall with a circuit of c. 1.3 A Excavated by J. Jclièié hectares the walls of the original Greek colony represented I Excavatcd by M. Zanovié (approximatc position) (Fi-e. 4.2). Despite this, reconstructions of the urban t.2.3 I 993 excavation topography i.-rsin-s the surviving remains have not carried Course ofancicnt wall, obsen ed and inferrcd anv form of consensus (Gabriðevió, 1966;Kirigin, l99l). The status of the town during the Roman period has also Fig. 4,2 Tlre putative deþnded area of Pharos with tench been a subject of intense debate followin-e the recent claim localions. Ceramics and the site: is survey enough? 3l that Pha¡os had achieved the status of municípiun during tables probably represents the average site density of the first century AD (Zaninovió, 1988). these sites - a mid-point average of c. 15 sherds per Beyond the urban area, Greek settlement has been collection unit (0.15 sherds per m2). Given that the limited to a handful of defensive sites, graves and one or smallest Roman activity areas (excluding one possible two possible Greek rural settlements all set in or around shrine) recognised with any confidence on the island the Stari Grad plain, an exceptionally large fertile area covered an area of 0.06 hectares (e.g. Maslinica SG0043, which carries traces of a Greek field system (Bintliff and Gaffney, 1992:276), areas with sherd densities of l5+ Gaffney, 1988; Gaffney, 1992). Roman data, however, is and with areas of 0,06 hectares or more have been ex- plentiful and the period is generally seen as a settlement tracted from this data to provide the extent of possible peak, There is an implicit assumption that the expansion sites in the area. A total of 31 separate polygons con- seen in the rural area is reflected in the urban zone (Fig. taining 7.31 hectares is derived from this. There are 4.3). clearly some problems with such an approach. Some of Quantitatively, the situation can be best illustrated by these areas clearly represent parts of known sites, whilst analysis of data collected as part of the Anglo-Croatian there are difficulties in comparing data from intensively survey of the Stari Grad Plain between 1987 and 1989 and extensively collected sites. Despite this, one suspects and the surface survey of all the available land within the that this figure represents a minimum directly utilized putative area of PharosiPharia carried out in 1991 and area and may suggest that actual numbers of lower order 1992 (Fie.4.Ð, activity areas on the Stari Grad plain afe greater than There are eleven recorded Roman rural sites and one had been appreciated. or two possible Greek rural sites within the 186 hectares Intensity of land use is also indicated by the general surveyed within the area of the Stari Grad plain. Six of scatter of ancient material beyond a¡eas defined as sites. the larger Roman sites were surveyed intensively as part Of 183.7 hectares surveyed c.64 per cent of the area of this exercise (Fig. a.5). Inspection of the site data provided evidence of ancient surface discard. Although suggests the top three quartiles of artefact frequency there has been some criticism of the manuring hypothesis

6384000 6387000 6390000 6393000 4791ffi- ++ + + -4791000

4788000- + -4788000

4785000- -4785000

4782cfn -4782cf0

4nw- 4nw

I I 6384000 6387000 6390000

Fig. 4.3 Roman sites in and around the Stari Grad plain, 6384000 6387000 6390000 6393000 6381000 I I I I

Lrþ ) 4788000 -t-1l,l + 4788000- )- ( \) l/ t, ,/ -l ì_) I \.-. I -4785000 4785000- + + "ì .'-r +

4782æ0 -4782æ0 )

4nw- -4næ00

I 6381000 ædooo æsiooo oagboo æeåooo Fig. 4.4 Extensive survey area in the Stari Grad plain.

townsfolk for such scatters, this still seems a reasonable explanation ment, and occasionally consternation, of the for the Hvar data (Alcock et al', 1994), As an aside, it of Stari Grad, teams surveyed every available backgarden c,3 hectares. should be noted that although much of the empirical basis and vegetable plot in the town - a total of for such practices is taken from northwest European The relatively low density of finds from Pharos relates agricultural literature (Gaffney and Tingle, 1984), there to taphonomic processes on the island which dictate that is frequently a regional historical literature in Medi- most material is found on clearance cairns rather than on (Gaffney et terranean areas to support the hypothesis' within the AIP the surface offields - an average ratio of4:1 survey area the rvorks of Don Milan Vusio (1887), par- at., l99I). Because clearance cairns are not significant ticularly Mladi Poljodelac (YoungFarmer), are extremely urban features, and the soils here have been intensively informative about agricultural practice on the island of worked, artefact densities are considerably lower in these Brað during the nineteenth century, and emphasize the areas. Although the survey represents a less than random prob- importance of manuring in what is actually a very im- sampling strategy and despite significant visibility six poverished agricultural environment. The need to garner lems, the data suggest that settlement extended over every source of fertiliser is displayed in Vusio's precise hectares at least. (Fig' a.6)' descriptions for the preparation of manures derived from The period survey data were very intriguing was indicated by a variety of sources including human excrement and road The existence of a pre-Greek settlement quafzite sweepings. the location of a very dense concentration of coarse walled area. The sequence within the rural hinterland of Pharos gritted pottery in the southeastern zone of the activity contrasts significantly with results from Pharos itself. Here The core of the scatter appeared to relate to building all available land was intensively collected within and which had disturbed deeply buried stratigraphy. The original its intensity around the area of the ancient seitlement. To the amuse- extent ofthe scatter is therefore uncertain, but Ceratnics and the site: is survey enough? 33

Lìlensive Survey Area

Extensive Survey Area

o Roman Sites

area' Fig. 4.5 Areas wirh site densiry pottery values, known sites and intensively surveyed

Greek colonial settlement. suggests the existence of a settlement which may have sent the limits of the original three trenches in relation to the walled beãn considerably larger. The urban scatter of Greek and The positions of all Greco-Roman material appeared to be surprisingly exten- area are given in Fig. 4.2. middle of the walled area and sive whilst Roman occupation was very restricted and c'60 Trench I lay in the this trench revealed ir- per cent and 94 per cent less dense than the Greek and measured 2x3m. Excavation of depth of 0.60m. Above this, Greco-Roman artefact densities. regular bedrock at an average had been completely disturbed by recent Despite the relatively poor quality of the dating evidence, ancient deposits (ritual?) pit in the rock the results suggests that Pharos was a medium-sized town cultivation, apart from a small BC pottery. Finds of similar during the early Greek period (fourth-third centuries BC), containing fourth century virtually no later with an 'industrial' zone associated with evidence for date predominated in the disturbed soil; here' pottery production of Greek character (tiles, amphorae, Greek or Roman material was encountered (2x3m) a short distance to the east, in coarse wares) along the southeast fringes of the scatter' Trench II, lay a of possible ancient origin These results can be contrasted with reconstructions by an open area, close to well by local residents with the previous researchers who have assumed that the surviving and near an area associated mosaic. Excavation revealed both wall of the colony provided evidence for the extent of the position of a Roman deposits. Remains of a largely robbed- settlement at this early date. Given the archaeological and Ro-un and Greek foundation, forming the limit of historical evidence, the relationship, and survival, ofsuch out north-south Roman present at a high level at the a small defensive circuit, even if linked to the original a floor to the west, were quantity of loose tesserae in colony, seemed unlikely' The apparent lack of a major west end of the trench' A layers may indicate that this room Roman settlement also seemed strange given the extent of the disturbed Roman Associated debris layers in- contemporary rural settlement. The available evidence sug- had a coarse mosaic floor. date this phase of c.AD 225-250' gested that further investigation, beyond the survey, seemed dicate a terminal for just below the features in the appropriate. The bedrock appeared and sloped down to the east' Consequently, in 1993, three trenches were excavated western part of the trench the may indicate the limit within the small rectangular walled area alleged to repre- where a äeep hollow in rock áltt0æ @

2Ç- ,qj /t.'J, e

A) Sample areas B) Prehistoric C) Greek D) Roman E) Greco-Roman

I

I L-

Fig.4.6 Survey datafrom Pharos (Stari Grad). Cerctntic's and the sil¿.' is sutt'ey enough? 35 of a scarp or gully beyond the trench limits. In the eastern half ol the trench a Greek rvall or loundation running north-south crossed the hollorv and retained a small terrace containing exclusively Greek material of c.350- 320 BC. The pre sence of layers u'ith Greek material both overlyin-e rnd below' this may indicate sttccessive ter- racing operations. These seem to have spanned a rela- tively brief period. Trench III (2x3m), dug a,eainst an exposed standing section of the north line of the supposed city wall, pro- vided the clearest sequence and the most surprising results (Fig. a.7). The excavation of Trench III, against the inner face of the defensive rvall provided both a lj construction history and associated dating material for the construction of the rvall. Here the structure proved to be erected directly upon the destruction material of Roman structures; it lacked any dug foundations. A circular brick from a Roman hypocaust immediately underlay the later rvall. Quantities of painted wall plaster were obtained from the debris. Ã terntinus posr quent of the earlyimid-third century AD was obtained for the defensive rvall from the pottery finds. The overlying layer against the lowest course of the wall produced a few sherds as late as the fifth century. The circuit rvall may therefore date from either of these periods, though the general lack of post-third century finds from within the walled area seems to point to the earlier date for its construction. The building destroyed prior to the erection of the circuit rvall may be part of a 0 lñ larger complex of second-third century date to which the mosaic floored remains in the vicinity of Trench II be- longed. Fig. 4.7. Pharos: Trench III, north section' Sealed belorv the third century AD debris, and separated from it by a layer of garden soil associated with the Roman occupation phase, rvas a debris layer of mid-Augustan date. Belorv this, a further earth layer (possibly of around town. However, with respect to the northern section of 200 BC) sealed a number of layers of the fourth and early the existing walls, whilst it is indisputable that the line third century BC. All tend to slope down at the northeast of the presumed early Greek town wall cannot have lain corner, suggesting a dip in the ancient landsurface or the within the area of the excavation, the ashlar blocks of presence of an early pit at this point. Though a depth of the existing wall do look Greek in character, and must c.2.2m was excavated in a narrow area at the north end, have have been moved from elsewhere, bedrock was not reached. The penultimate layer dug The dating provided from the excavation suggests that consisted of rubble and fragmentary roof tiles of Classical the Greek urban community was primarily active during Greek type (not paralleled elsewhere); the few associated the fourth and third centuries BC. Evidence for later sherds associated rvith this material may date from the Hellenistic occupation has not been forthcoming, and the mid-lourth century BC. site may have been abandoned or much reduced after c,200 The results of the Pharos Survey and tria[ excavations BC. Presumably this reflects the uncertainties and dangers prompts a radical reinterpretation of the history and of the later third and second centuries BC. If this is the development of Pharos during the Greek and Roman case, the associated field system on the Stari Grad plain periods. may plausibly be associated with the colony of the fourth Clearly, the Greek colony at Pharos was much larger and third centuries BC. The extent of the urban scatter at than originally believed. At a very early stage it may Pharos may explain the apparent lack of rural settlement have occupied an area of 6+ hectares. However, contrary rvithin the Stari Grad field system (see Gaffney et al.,this to carlicr thought, no upstanding sections of the original volume). period defended circuit of the colony are now known rvith cer- The revival of Pharos dates from the Augustan tainty. Such a circuit did exist and recent excavations rvith the incorporation of into the Ernpire. The have exposed a portion of these u'alls to the east of the excavated and survey evidence points to the existence of a major villa/adnlinistrative centre. There is, horvever, chronology. The lack of artefact data rrra¡' also be a culturill no evidence for a re-founded urban communit¡'. phenomenon, and as surveys are integrated into archaeo- Until the mid-third century AD, Roman settlernent in logical studies rr'ith rr'ider ternrs of reference this type of the Stari Grad plain rvas t¡'pit'ied by extensive villa develop- negative evidence achieves significance and desen'es a ment associated s'ith agricultural exploitation of the polje. rvider discussion. There is little evidence to suggest that the site of Pharos This may be demonstrated through reference to a rvas occupied b¡'an¡,thing other than a I illa or, possibly, an large-scale study of the twin problerns of urban-rural open vicus-t¡'pe settlenlent. There is no suggestion that a relationships and Romanization centred on a study ol torvn associated rvith salls existed on this site (contra the hinterland of the Roman town of Wroxeter (Viri- Zaninovié,1988). coniuttt Cornot,iontnr) in Shropshire (UK), Associated A fortilied coastal strong point \\'as constructed on the rvith this research goal is a major field programme in- site of Pharos follorving destruction of the villa./vicus cluding extensive sample landscape survey and explor- settlement in the mid-third century. This fortification, built atory excavation (Fig. 4.8). over the ruins of earlier buildings, probably contained a The attraction of Wroxeter lor such rvork lies in the core area of settlement, possibly rvithin the vicinity of Sv. amount of information we have on the town and its Ivan, a knorvn early Christian church. The defenses were remarkable preservation (Fig. a.9). The settlement at composed of blocks robbed from the derelict u'alls of the Wroxeter was, at 63 hectares, the fourth largest urban earlier Greek colony. The survivin-e 'Greek' rvalls of area in Roman Britain, and survives largely as a green Pharos are therefore the remains of a later third century field site. It has therefore attracted the regular attention Roman' refugiutu' enclosing c.1.3 hectares. of archaeologists, Consequently we norv have a consider- The lack of later building within the fortifications is able amount of information relating to the archaeolo-sy very suggestive. It is not insignificant that the early and history of the site. The work of Graham Webster Christian Church of Sv. Ivan lay within the area of the and Philip Barker, in particular, has provided invaluable rvalls and may have been built upon the remains of a detail on the development of the town from its origins as second-third century villa. Given the Iack of later building a fort for Legio XIV by AD 60 and the later foundation and the association of the church with an early bishop, it of the civilian settlement as the Civitas Contoviorunt is possible that the area of the fortress may have been during the '90s (Barker, 1990; Barker et al., forthcom- inherited by, or granted to, the church and functioned as ing). a precinct (Kovaðió, 1987). If this were the case it would The intensity of archaeological activity within the town explain why the site and its outline has been preserved at Wroxeter contrasts with the general lack of work in the as a relatively undeveloped area within medieval Stari settlement's hinterland where the principal source of Grad. information for land use has been aerial reconnaissance To some extent one might have concluded that much of (Baker, 1992; Whimster, 1989). Settlement in the hinter- this may have been derived from the survey results. Yet land of Wroxeter is largely represented by a variety of the relative precision and balance provided by excavation simple enclosures which have been studied and classified was of real importance in providing a fuller interpretation by Whimster (1989). Evidence from sources other than of the complex sequence of events and character of the aerial photography is slight and the overall economic and site which was not achievable through survey alone. social context of contemporary settlement remains unclear. Excavation actually removed several important archaeo- Perhaps the most notable characteristic of the area is the Iogical misconceptions related to the site which had been paucity of evidence for highly romanised rural settlement held, in the case of the identity of the 'Greek' walls, for types, especially'villas'. more than 100 years. The situation at Wroxeter, where we can contrast the development of a highly romanised urban core with the almost total lack of traditional evidence for Romanization rvithin its hinterland is very attractive to any study con- WROXETER, ENGLAND cerning Romanization. Consequently, the principal aca- A second situation which may prompt a call for an inte- demic objectives of the project may be divided into two grated excavation policy involves circumstances in rvhich broad areas: the nature of the material evidence is not necessarily amenable to survey. Here I am not talking about simple a) The investigation of the settlement history within the problems of survival, about rvhich there are numerous study area, concentrating on the iron age, Roman and references in the literature (Reid, 1984) or access to trade early medieval landscapes, with special reference to goods (Fentress and Perkins, 19891 Millett, l99l)' Ab- the Iron Age/Roman transition, the impact of Roman- sence of evidence is not alrvays the result of taphonomy ization, and the implications ol the decline and aban- or economy, although I suspecr that this impression donment of the Roruan to\\'n for the surrounding sometimes emerges largely because most surveys are cou n tryside. frequently concerned rvith settlement development and b) The development of novel approaches to the ânalysis Cerantics ancl the site: is survey ettouglr? 31

355000 360000 365000 370000 375000 325000

320000

315000 315000

310000 310000

305000 - -: ..- I i .t. 'ã.-..- l.J fFiTi.'I ¡l . . .¡J._+.1 . .. . ooo . +-J-..J r .t.. ì. 300000 1 l. J { o . o .¡ '11","u,lt,,.. .t.',1 ¡ r,:ìf. i r + *. i,-;i' -j- --. -295000

Fig. 4.8 The Wroxeter Hinterland Project survey area and sl.rvey transects.

Romanization in the of Romanization and its spatial correlates based upon To some extent the problem with incorporate everything GIS technolo-ey. literature is frequently that it can and nothing. In one area it may be infened via the Presence for literacy, in another Analyses related to urbanization and Romanization have, ofa house form or through evidence types of pot or even the of course, a lengthy research history in Britain at least the variable quantity of different the matter one (Haverfield, 1923). The reasons for this are not hard to shape of fields. Yet, if this was the sum of Romanization was not discern, especially in those Roman provinces, including might conclude that the study of emerge if we Britannia, where there was an apparent lack of urban such a challenge. Real problems start to as the combination of traditions or where pre-Roman trends towards urbanization begin to conceive of Romanization spatially variable were rveak and the development of towns and cities is a wide variety of cultural markers with of combinations interpreted as one variable within the longer process of signilicance, occurring in a constellation social and physical Romanization in a province (Millett, 1990). Integrating and differentially distributed across the analysis of Romanization with the study of town and Iandscapes. in Shropshire we countryside and therefore, perhaps, hinterland is attractive V/ithin the context of Romanization the town itself. Despite and many Romanists would see the essence of the civiras must start with the evidence for of the site has been investi- in the integration of town and country, probably within a previous work only a fraction under permanent unit derived lrom the late iron age tribal group (Millett, gated. Given that the current site is pasture, the town is currently the subject of a complete I 990). This has been planned to include Il the definition of a suitable research scale is funda- remote sensing survey. as well as air-based mental to the pursuit ol urban-rural relationships and a variety of grouncl-based techniques scanning. It is also intended Romanization, then so too is the choice of materials to be thematic and hyper spectral photographic data should be reassessed studied and the nature of fieldrvork to be carried out. These that existing aerial processing technologies. are not particularly nerv observations. they rvere admirably using image c.40 hectares of the site have put several years a-qo by Haselgrove (1990) in an article At the time of writing survey by the Ancient concerning the Romanization ol Belgic Gaul. One should been subject to magnetometry ancl Geophysical Surveys of Brad- rlso note, horveler, Phil Freeman's (1993) stated concern Monuments Laboratory been covered by rvith the Inck of a lturtpen rvithin all too nlany of otrr forcl, whilst c.75 per cent should have ground-based remote sensing anall'ses and our tendencv to repeatedly attelnpt to ans\\'er the encl ol the year. Further seisnlic and radar the sirnle old questions on Rotnaniz¿tion using the sanre surveys underrvay include resistivity, of this rvork are completely old data - villas. and Satllian rvare. survey. The initial results .N.-v

Fig. 4.9 Plan of Wro.reter derived from archaeological atd aeríal photographic data to 1990. Ceratttics ancl the sile; is survey enouglt? 39

356500 356600 356700 356800 356900

308600 308600

:C 308500 t,( -t_ 308500

308400 æ8400

356500 356600 356700 356800 356900

Fig. 4.10 Wroxeter: preliminary results oÍ magnetometer survey.

transforming our picture of Wroxeter (Figs. 4.9 and 4.10). by Whimster as clustering around Wroxeter, may be The very high quality response from the site has revealed associated with Romanized structures (M. Watson, pers. that the surveyed area contains significant new evidence comm.). There is also increasing evidence for the exis- for settlement and that areas previously believed to be tence of settlements like Meole Brace which have the blank are teeming with evidence for activity. Given this appearance of Romanized settlements, but which have response, integration of the varying types of air- and probably not been located in the past because they exist ground-based remote sensed data with enhanced aerial only as artefact scatters anjl4re not associated with photographic data within the GIS project is likely to enclosures identifiable fro¡í APs\(Ellis et al., 1994)' produce the nearest that we are ever likely to come to a Such points may, of còurseí be cleared up by the complete plan of the Roman town of Wroxeter. systematic survey planned to occur as part of this project. We can set the rural evidence against this new infor- However, whilst it is likely we will hnd a greater investment mation. The lack of Romanized structures, for instance, within the hinterland than previously believed, the apparent is frequently held to be a significant phenomenon associ- contrast between town and country around Wroxeter during ated with Wroxeter. The intensity and nature of urban the Roman period is likely to remain. What should we seek occupation may now go a considerable way to explaining to isolate from such an unprepossessing database for use the absence of some traditional indicants of Roman- within the study? The evidence for a considerable (undated) ization in the countryside around the town and in the density of settlement in the town itself is provocative and, immediate suburban zone. Howevet, recent work also in the absence of an alternative explanation, one must suggests that our knowledge of rural settlement can no believe that the pre-existing, rural, agricultural base largely longer be regarded as adequate. There is norv evidence supported the urban core. Given such a situation, we should that some of the univallate rectilinear enclosures, noted anticipate some structure in the archaeological record, Traditionall¡' one nli-qht expect strch evidence to enlerge landscape. In such a situation rve should also consider lrom anall,sis of the available rnaterial assenrblage. Arte- that the adequate description of settlement and society factual data should also be invaluable as a spatial guide to transcends the ability of surface survey alone. hierarchical and social ranking. .A'nalysis of cera¡rlic What then should we be doing in such situations? One assenrblages suggests that si-enificant differences do exist analytical path lies in the the study of the co-variation in the patterns olsupply to sorne rural sites (Evans, 1994). of rnaterial assemblages rvith field system and enclosure However, one of the most intriguin,e facts enterging froru morphologies, or the spatial patterning of status, defined the project's fieldii'ork is that extensive surface survey has through settlernent type, or site asse mblage, with eco- revealed a ver)' lorv level of artefacts outside the area of the nomic zonation. The linking of site assemblage (in- town at Wroxeter. Althou-gh sometimes suggested 1o reflect cluding structural data) with landscape features may rvell survival patterns, this was initially interpreted by project be the only way of attempting realistic analysis of social staff as primarily indicative of econornic activity, either or status relationships in the rural landscape of Roman relatively non-intensive mixed farming or as evidence ola Shropshire. There is, for instance, increasing evidence pastoral econorny, Hotvever, the continued lack of ceramic for the developrnent of Iandscape architecture in associ- evidence ri'hen surleying on-site is more difficult to explain. ation with non-Ronranized encìosures in the hinterland, On sonre enclosure sites and other low status settlements rvhilst the development of formal field systems in the such results may be explained by reference to status, access immediate suburban area of the town is very provocative. or even the incorrect dating of type sites. However, survey Excavation at Duncote Farm to the north of Wroxeter, and excavation of a villa building at Whitney near Shrews- has provided evidence for a regular patchwork of small bury as part of the project has added a new dimension to fields linked with the early phase of the site (Fig. a.l l). the problemÑ@n, Here extensive survey across Indeed, one of the authors of the excavation report noted the site and excavation of the bath suite and part ofthe villa that these fields conformed to a iugera measure (Ellis er itself provided a cerarnic vessel assemblage of less than 40 al., 1994). There is, therefore, the possibility for formal sherds, reorganisation of the Wroxeter landscape in association ' The evidence for Shropshire is beginning to suggest with the earlier phases of the settlement. Such a sug- that there are cultural factors behind the absence of gestion, however, demands corroboration and it is in- ceramics on Ro¡nan sites and hints at some degree of tended that further geophysical survey in association with cultural resistance to the use of ceramic objects and the reanalysis of the available aerial photographic evidence utilization of alternative, presumably organic, vessels. The should be attempted to clarify the situation sunounding - fact that this pattern is even represented in villa sites like these field systems. Whitley which have complex structural histories and Clearly, the situation at Wroxeter is one where infor- display most of the other trappings of Roman life, in- mation technology, specifically GIS, will have a major cluding bath suites and window glass, suggests that the role in storing, analysing and displaying data sets as diverse situation is extremely complex. as artefact distibutions at the intra and inter site level, There is, one suspects, a structure which is beginning digital environmental data and ground, air and satellite- to emerge within the Wroxeter area and which incor- based, remotely sensed data. It is also a situation where the porates the town, possibly with an elite antecedent, Iower full potential of such technologies may be explored. The level settlements including romanised structures associ- analysis of Romanization and its spatial conelates in an ated with rectangular enclosures, open settlements linked environment where common features of Romanization are with markets and below this an echelon of settlements absent is a challenge, and spatial linkages will frequently of native type. The emerging evidence for activity within be between landscape and material culture, elements that the town itself and the apparent lack of any material are not normally associated during archaeological investi- benefit to many of these farms suggests that the primary gations. The result of such an analysis should be the ability conduit for much material and social development was to produce secondary GIS data layers relating to artificial, through the urban population. If this group included the mappable 'values', including status, and an ability to use local elite, and there seems little reason to doubt this, such variables in association with other types of data. In their ability to nanipulate and invest within an urban such a project survey data can only be one data source structure presumably rests on preexisting social relations. from many. It could never achieve the primacy attained However, the strength of the continuing local substratum elsewhere. is evidenced in rvhat appears to be a resistance to some elernents of Roman material culture. In such a context, we consider that the absence of such data not only must CONCLUSION may be as important as its presence elsewhere, and that interpretation of the settlement evidence can only come In conclusion, one must acknowledge the role and potential through the integration of survey data rvith a wider variety of archaeological surface survey. However, there is the of cultural rnarkers u'itlr spatially variable significance fceling that a rubicon has bccn rcachcd in the development and differentially distributed across a social and physical of archaeological sun,ey, particularly in the Mediterranean. Cerantics and the sile: is survey enough? 4r

I Phase I Phase 2 '71 \

t39 I

I L

e)--S

I

I

I

-- -- a -- -- -= -- =- a' . -! =- -: = -- -- = 1 l- tt tl J P¡"'" ' tl $ rn..": o o r@ Þa! ænalala! . t È3tþtr o. Står.rctô

Fig. 4.1I Duncote Farm Phase PIan.

Hinterland Project is Following a period when there was a legitimate emphasis primary research for the Wroxeter and Mr Martijn van i simply on location, quantification and analysis of survey being canied out by Dr Roger White Leusen, BUFAU Leverhulme Research Fellows. Field- \ work and volunteer participation in the project is being funded by The Mercers Company of London, the Mary j Webb Trust, Barclays Bank, the Yapp Trust, The Roger and the Uni- scales and definition of different archaeological procedures, and Sarah Bancroft Clark Charitable Trust '- 'whilst on the other it will permit a fuller integration of versity of Birmingham. Ground-based remote sensing is different data types, including information from survey, being carried out by Geophysical Surveys of Bradford, (English into wider and more sophisticated archaeological studies. the Ancient Monuments Laboratory Heritage), This will not be as difficult as it may appear. Survey The Nara Cultural Properties Research Institute (Japan)' methodologies are now mature and the technical apparatus the University of Miami GAL (Japan), C.N.R.S.-Garchy (University and conceptual strategies are in place' All we need is the (France) and the Department of Earth Sciences has been money and resources! of Keele, UK). Airborne remote sensing data provided by National Environmental Research Council. Regular updates on the Adriatic Islands Project, the <-l Wroxeter Hinterland Project and other Birmingham Uni- \ ¡t ACKNOWLEDGEMENTS versity Field Archaeology Unit activities can be found on \ General enquiries to The Adriatic Islands Project: data from the 1987-9 survey www.bufau.bham.ac.uÙ -J was collected by the Anglo-Yugoslav team and processed [email protected] as part of the author's PhD. The AIP core team are B' Kirigin (Archaeological Museum, Split), N. Vujnovió (Hvar), S. F'orenbaher (Southern Baptist University' USA), REFERENCES Z. Stanðió (SAZU, ), S, ðaðe (University of ),J, Hayes and T. Kaiser (Royal Ontario Museum), Alcock, S. (I993) Graecia Capta' Cambtidge' Cambridge Uni- V. Gaffney and P. Leach (University of Birmingham).The versity Press. Alcock, S., Cherry, and Davis, (1994) J. J. Inrensive survey, Gaffney, V. and Tingle, M. (1984) The tyranny of the site: merhod agricultural pracrice and the classical landscape of Creece. In I. and theory in fiel

theory and the act of percept¡on. In J, Anderson, T. Madsen and Buried Landscape.Royal Commission on Historic Monumen¡s L Scollar (eds) Computing the Pasti 133-38. Aarhus, Aarhus of England. University Press, Zaninovié, M. (1988) Pharos - od polis do municipia. Zagreb, Whimster, R (1989) The Emerging Past: Air Plotography and the Arheàlolk¡ Radovi ì Rasprave l1: 3548.