TRAC 1991:
THE HOARDING, DEPOSITION AND USE OF PEWTER IN ROMAN BRITAIN
By Rob Poulton and Eleanor Scott

The circumstances surrounding the deposition and hoarding of pewter in Roman Britain are examined – in particular material associations and broad archaeological contexts are analysed – and the conclusion is advanced that almost all such pewter material became buried as a result of ritual activity. It is further suggested that the deposition of pewter is but one element in a frequent and widespread use of seemingly everyday objects for non-rational purposes in Roman Britain, and that this reflects a development or re-invention of Celtic religious practices. Finally, the question of whether the character and distribution of the surviving pewter can be used to postulate a common domestic use for such vessels in late Roman Britain is considered. The arguments are advanced that much of the surviving pewter may have been specifically manufactured for religious purposes, and that the extent of domestic use cannot be directly inferred from the archaeological evidence.

INTRODUCTION
Pewter artefacts made in the Roman period have been described and discussed on a number of occasions (e.g. Wedlake 1958; Peal 1967; Beagrie 1989). Certain points have become generally agreed: while the circumstances and date of deposition of many of the known pieces of Roman pewter are obscure, the evidence from the remaining pieces suggests that the great majority were not manufactured until after circa AD 250 and that they were deliberately buried. In consequence, a common interpretation has arisen that the pewter represents the tableware of wealthy families, hoarded for temporary safekeeping in some time of crisis (e.g. Brown 1973, 201-4). Manning (1972, 248-9) suggested that religious activity lay behind the deposition of pewter hoards and Beagrie (1989, 179) agreed that the ‘theory of religious deposition would . .. seem to offer the best explanation for the context of the majority of Roman pewter vessels.’ It is the purpose of this paper to examine the interpretation of Roman pewter hoards in greater detail and consider how this affects our understanding of how pewter was used in Roman Britain.

CIRCUMSTANCES OF DEPOSITION
Hoarded material, by its very nature, is rarely recovered under controlled archaeological conditions. There is, therefore, very little that can be said with certainty about the circumstances of deposition of most Roman pewter, and even for the better known examples listed below the information is often inadequate.

Shepperton, Surrey (Poulton, in prep. a)
On 11 August 1987 a hoard of five pewter plates was brought up in the bucket of a mechanical excavator, which was removing the alluvial overburden in Shepperton Ranges gravel pit. The find was immediately reported and for this reason and also because the site was already under regular archaeological observation in view of the discovery of prehistoric finds of major importance (see, for example, Bird & Bird 1987, fig. 1) in the previous year, it was possible to determine the findspot with some accuracy and to assess the nature of its immediate context.

A bore-hole survey of the pit by the gravel company had indicated that there was considerable variation in the depth of alluvial deposits over the gravel, and site watching confirmed that this was due to the presence of a number of silted up river channels representing former courses of the Thames and tributary streams. These were evidently of various dates and it was from one such that the important prehistoric finds carne. The pewter plates, however, were from an area between the channels, in what must certainly have been boggy ground or under shallow water in antiquity. They were found at a depth of about two metres below modern ground level, but much of this is to be accounted for by post-Roman alluvial deposition. The plates were deposited as a group, probably in a container for which no evidence survived, and would therefore have sunk down into the mud. The digger driver reported no other finds, and the regular archaeological observation had established that the immediate area was free of occupation sites of any date.

The nearest Roman finds come from an area around 250 metres to the east though the nature of occupation there is unclear. About a kilometre north-west of the present finds a Roman pit was excavated (Frere 1943) and the 19th century finds of tesselated pavements near the Saxon cemetery at Upper West Field, Shepperton (Longley and Poulton 1982, 184) must be similarly located. About 1.5km to the north, there was probably another villa (Poulton, in prep. b), and 19th century records suggest the possibility of Roman occupation in the area of Shepperton village. The evidence is far from detailed, but the general impression is that this area of the Thames valley was well settled and reasonably prosperous in the Roman period. It might then be thought that one of the nearby Roman households had hoarded these items in a period of disorder, but, if so, they were rather dim witted. What sense could it have made to bury goods with intent to recover in boggy ground so far from the nearest settlement? Accidental loss is a bare possibility but one needs be highly imaginative to conjure up a satisfactory set of circumstances to account for its occurrence at this location. The explanation that best fits these facts is that the plates were votive offerings, for which a waterlogged site was deemed appropriate.

St Albans (Verulamium), Hertfordshire
A find of three pewter plates from Verulamium was from a very similar site context to that at Shepperton. They were stratified in bog mud beside the River Ver, and loosely associated with a large number of coins, which point to a date of AD 375-400 for deposition. The whole group of material has been suggested as votive in origin (Goodburn 1984,65), and, indeed, there seems no reasonable alternative explanation.

Bath, Somerset
A rich variety of pewter pieces have been recovered from the Spring in the temple of Sulis Minerva at Bath, a number with inscriptions to the presiding deity (Henig et al. 1988). These do not of course constitute a hoard in the strict sense, as deposition clearly occurred on a number of occasions. Nevertheless, this material is very relevant to the discussion here in view of the obvious association with both religion and water. Of particular interest is Cunliffe’s (1988, 361) suggestion that the pewter is temple plate, associated with official rituals, deposited at the end of its useful life or by special ceremony. The Romano-Celtic temple at Harlow has also produced a miniature pewter cup (Frere 1989, 303).

Stanwick, Northants

During excavations at the Stanwick villa between 1984 and 1988 a group of four pewter vessels was found. Its context is interesting (Neal 1989, 165):

“the [fourth century] building to the west [had] set into its rubble floor .. . a quem and two pottery vessels and on the north side was a stone-capped drain concealing a hoard of iron objects, including two unequal-armed plough shares and a large wheel hub for a cart. The plough shares are unworn. In a pit to the north was a hoard of four pewter vessels and offcuts of leather indicating the manufacture of shoes. The west wall of the hut was built over a well which was subsequently used as a rubbish and cess-pit. It contained coprolites of dogs.”

The association of the Stanwick pewter with a watery context – the well – and with metalwork (which has very obvious parallels with the ironwork hoards classified as votive by Manning (1972) and leather is significant and will be discussed below.

Figure 20. Brislington Villa, Avon (from Branigan).

Brislington Villa, Avon
The Pewter and Associated Objects. The Brislington villa, Avon (Fig. 20) was excavated in 1899. The last discovery made during the course of these investigations was of a well, and the contents were recovered by the resilient employees of the local Sanitary Committee. At 24 feet down was found ‘some tons of coarse building material, evidently the remains of the Villa’, and at 28 feet a large collection of faunal remains, mostly ox skulls, came to light, along with ‘other miscellaneous objects of the Roman period’ (Barker 1901, 18). In clearing out the section between 28 and 32 feet down, human skulls and skeletal remains were found. Below these (ibid., 19):

“came a graduated series of seven remarkable metal vessels, with curved and ornamented handles. Most of these were fairly perfect, except that they were more or less bruised and bent, but two had been broken to pieces by the fall of heavy material upon them … With these objects was found the only perfect specimen of pottery that came to hand. This was a small black pitcher or jug. Another good specimen, and larger, was slightly damaged … and it is worthy of note that a few tesserae of the destroyed pavements were among the miscellaneous objects.”

After these ‘finds’ came ‘another extensive deposit of rough building material … then … more bones and fragments of pottery … and various iron objects’ (ibid., 20). At the bottom of the well were found, amongst other things, bronze objects: an ornamented spoon, part of a fibula and a pin.

The Brislington Well Deposit, the Villa and the ‘Barbarian Conspiracy’ of AD 367: Branigan’s Theory Reconsidered. The Bodies In the Well are infamous. This well deposit has been championed by Professor Branigan as evidence of the Barbarian Conspiracy of AD 367 (1972; 1977, 95 fE.) and in this he follows Haverfield, Frere and Webster. A lengthy discussion of the issue is not possible here (see Scott 1988, 221-32). The following points, however, should be noted.

1) There is very little archaeological evidence for the ‘Barbarian Conspiracy’of 367, as reported by Ammianus Marcellinus. Only the north-western forts of Ravenglass and Bewcastle show signs of structural change at this date which may be attributable to destruction, though this is a moot point J.P. Gillam pers. comm.).

2) Ammianus Marcellinus may have intended to glorify Theodosius by inventing action for him to have skilfully ‘suppressed’ (Todd 1981, 232).

Figure 21. Brislington villa: a reconstruction of the well deposits found in 1899 (from Branigan).

3) The deposition/ dumping of material down the Brislington well cannot be firmly dated to 367. The coin of Constantine II (found seemingly trapped on a ledge above all the deposited material? See Branigan’s reconstruction: here (Fig. 21) is of limited value as a dating tool.

4) The material may have gone down gradually, not during one clearing up operation. The building material in the well may have been deposited as part of a ‘ritual of termination’ (Merrifield 1987, 49-50; and Scott 1991) during renovations to the villa house. The frequent use of foundation deposits on villa sites (Scott 1990; 1991) indicates that the Romano-British attached ritual importance to the building process as part of the fabric of their social existence.

5) The contents of the Brislington well comprise objects often found in other more easily recognisable ‘Celtic’ votive contexts: complete or nearly complete pots, metal vessels, metalwork, and faunal and human remains, especially skulls (below).

The excavation report concludes of this feature and its contents (Barker 1901, 20): ‘There seems every reason to conclude that this was originally the well of the Villa, but it was evidently made to serve as a rubbish pit when its use as a well was discontinued.’

Yet, if we examine the broad archaeological context, observing associations between material types, it might seem that on the contrary there are reasons to conclude that this well served a ritual, or non-rational, purpose (see below).

Appleford, Berkshire
The pewter hoard from Appleford is probably the best published of any such finds (Brown 1973). It was discovered during gravel extraction, and lay within a deep layer of soils which was recognised by the workmen as an intrusive and to them useless deposit which was thrown to one side in the pit. Some items of pewter were recovered from this and subsequently came to the attention of David Brown at the Ashmolean Museum. In consequence, the site was visited and in turn the deposit was carefully excavated. Despite the fact that it was a chance find this is therefore amongst the most informative of pewter hoards with regard to its circumstances of deposition.

The finds evidently came from the infill of a deep well or shaft cut through the gravel. Although a proportion may have been carted away with the surrounding gravel, it seems certain that a very high percentage survived for careful examination. The finds made consisted of:

1) A hoard of 24 pewter vessels, which seem to have been deposited in two piles, one of the small bowls and one of the plates.
2) A group of iron objects, consisting of an elaborate cauldron chain, a steelyard, part of a large scythe blade, a padlock, a shovel handle, a pan with folding handle, a chisel and sundry smaller items. Manning (1972, 235) has regarded this as a hoard, and in view of the parallels he quotes, and the character of the material, alternative explanations such as casual loss or rubbish disposal seem very unlikely.
3) Quern stone fragments.
4) Pottery, mostly of 4th century date, and including large portions of some pots.

5) Animal bones. A quantity were present but only a few (of Bos and Cervus) were collected, which is unfortunate in view of their potential importance (see above under Brislington).
6) Parts of a human skull.
7) Organic material including parts of a leather shoe and a plum or other fruit stone.

The immediate surrounding of the iron and pewter objects was a peaty soil full of rotting plant and twig remains. The human bones were all in a brown loamy earth (in parts very gravelly), with the remaining objects distributed between the two soil types. The peaty deposit had evidently formed in a waterlogged position, and it seems sensible to interpret it as a basal deposit with the loam above it. The hoarded items were therefore in a primary position in the well. This does not seem the most appropriate of hiding places if recovery was intended; on the other hand deposits of a votive character have been very commonly found in Roman well shafts (see below Wells on Villa Sites for references), and a votive explanation is favoured by Manning (1972) for this and the related group of ironwork hoards. Indeed, there must be some doubt whether shafts such as this ever functioned as wells; their original purpose might equally have been ritual.

The remaining finds may be compared closely with those from Brislington and are best understood by reference to the arguments advanced in that connection above. It would not be unreasonable to argue a case for regarding the pewter hoard as a votive deposit within a ritual shaft. Unlike Brislington, however, the shaft probably did not belong to a standard villa complex, as there was a complete absence of masonry in the immediate area, although there was plentiful evidence of nearby 3rd and 4th century occupation.

DISCUSSION
The above notes do not pretend to be an exhaustive review of the contexts in which pewter hoards have been discovered. They do, however, suggest that many pewter hoards should be regarded as having a ritual origin. Two different types of provenance seem to be involved; on the one hand deposition in riverine contexts, and on the other burial in ‘well’ shafts. In both cases the important factor is the association with water.

Romano-British ‘Celtic’ Religion and Ritual
That Celtic religion and its concomitant ritual burials and deposits continued into the Roman period in Britain has long been recognised (e.g. Ross 1968 passim; Macdonald 1977, 35-8; Laing 1979, 188), and Ross recognised that Celtic votive contexts were frequently ‘watery’ – wells, pits, shafts, etc. There even appears to have been an upsurgence or reinvention of the Celtic tradition in fourth century Britain judging by the number of new Romano-Celtic temples which appear in the south of England at this time (Painter 1971, 157). Certain Celtic ritual activity, such as deposition of ‘head objects’ re-emerged strongly in the fourth century in Roman Britain. Although Ross refers to this as a ‘cult’ of the head, it is probably best described as part of a general phenomerum, and not a ‘cult’ (Riddel 1990). Further, it appears to be a late ritual-religious phenomenon.

Much work on the survival of Celtic religion into the Roman period has concentrated on the evidence of epigraphy and iconography which reveals how particular Roman and Celtic gods were identified with each other (e.g. Macdonald 1977, 36; Salway 1982, 669). Far less attention is paid to the ritual human and animal burials, especially pit deposits, of Roman Britain which are clearly part of the Celtic ritual tradition known from pre-Roman Britain. Laing (1979, 118) simply notes in passing that: ‘Over a hundred ritual pits and wells, mostly Romano-British but some Iron Age, have been excavated in Britain.’

He adds that these pits and wells often contain the bones of dogs and ravens and ‘skulls’, but gives no further details. One must return to Ross’s original account for the only serious attempt to review this subject (Ross 1968). Her sample of pits, shafts and wells came from a wide variety of sites, such as Roman forts (mainly Newstead), settlements, hillforts and temples, but only one villa, Brislington. She does not include pewter in her list of objects which repeatedly appear in the watery features, but her list does include a great many artefacts and natural objects which have also been found accompanying deposited pewter: animal bones; buckets; burnt stone and flints; coins; cult objects; deer antlers; metal objects; organic matter; ox skulls and bones; pins of bronze and bone; potsherds; pottery vessels, whole or almost whole; quem stones; sandals and other pieces of leather; skulls, human and animal; smooth stones; twigs, leaves, acorns, nuts, stones and seeds.

Although it has been generally accepted that Celtic ‘pagan’ religion and its votive practices continued into Roman period, the expected cultural limits of such Celtic survival, however, have been made explicit (Macdonald 1977, 35; our emphasis): ‘Celtic beliefs are likely to have prevailed in country districts (though not in villas).’

The archaeological evidence, however, would appear to refute Macdonald’s hypothesis. Villa occupants should not be uncritically viewed as modern-rational. Different and culture-specific ‘cultural maps’ are to be found in other societies, both now and in the past. Neither should they be regarded as careless wastrels, dropping, losing and failing to recover an astonishing array of artefacts and livestock from pits, wells and shafts.

Wells on Villa Sites
Wells were frequently used for non-rational purposes in Roman Britain. For example, the sixteen dogs found down a Roman well in Staines, Surrey defy any credible ‘rational’ explanation (Chapman and Smith 1988). A number of Roman sites, including villas, have two wells, one of which appears to be functional – i.e. not full of ‘rubbish’ – and one which contains a large amount of objects known to have been used elsewhere as ritual offerings. This appears to have happened at the villa of Barton Court Farm (below), and at the Roman fort of South Shields, when the original well was replaced by a second well, the first was left open and upon excavation was found to contain a number of ox skulls (Paul Bidwell pers. comm.). This ritual process is evident on many villa sites and the votive assemblages deposited in the wells and pits contain recurring – thus significant – types of objects and sometimes assemblages. Given that some of these assemblages include pewter, these wells and pits are an important link in understanding the depositional associations and contexts of pewter.

Barton Court Farm
There were two wells discovered and excavated at Barton Court Farm, Oxfordshire. One deep well, feature 832, was in use in the fourth century and probably into the fifth (Miles 1986, 14-15, 46-7). This well contained in its lower levels the best collection of ironwork found on the whole site, a large number of leather shoes, several almost complete pots, an iron-bound wooden bucket and iron hook, and ‘large quantities of biological material’. It is clear that accidental loss cannot account for such a great proportion of the metalwork from the whole site having accumulated in the well. If this deposit at least was deliberate, so may other well deposits have been, particularly those whose contents tend to recur. The upper levels of this well contained ‘a deliberate dump of stone, animal bones, and well preserved vegetable matter’ (ibid., 15).

The second well on this site, feature 950, was situated nearer to the villa house and was housed inside a small rectangular masonry well-house. Only a small amount of ‘waterlogged material’ was found in this well (ibid., 32), and that this well differed in both style and its contents may indicate that the wells, though contemporary, had different uses. 950 may have been purely utilitarian, whereas 832 appears to have had a votive function.

Rudstone
At Rudstone, Humberside, the enormous well contained four stratigraphical groups of animal bones, including skulls and whole bodies, as well as pottery, the remains of buckets and chains, coins, an ‘enigmatic well deposit’ (Buckland 1980, 164), antlers, stonework, tesserae and wall plaster. Among the deepest stones was a block carved with the figure of a deity or genius. The pottery collection interestingly comprised vessels not represented elsewhere on the site (Stead 1980, 29-30, 36, 149-50). The building debris and some of the nearly whole animal skeletons are reminiscent of the Brislington and North Wraxhall well deposits which have been considered evidence of the effects on villas of the ‘Picts War’ of 367. However, the deposit at Rudston was dated by Stead as being not earlier than the early fifth century (1980, 29-30).

Denton
The villa at Denton, Lincolnshire, also had a well which was found to contain building debris, and, nearer the bottom of the well, cattle bones, moss, and items of leather, including shoes. Pottery was of circa 350-400 (Greenfield 1971, 47-8, 53).

Rockbourne
There were two wells at Rockbourne, Hampshire. The ‘main well’, situated by the bath suite, had deposits of pottery, coins of the second to third centuries, moss, parts of a leather sandal, including its hobnails, and – reminiscent of Appleford – various fruit stones and hazel nuts. The second well appears to have replaced the first, containing fourth century ‘infill of the usual debris’; and we learn that the ‘quantity of animal bones was unusual'(Morley Hewitt 1971, 18, 15-16).

Rivers and Ritual
There has been much recent discussion of how ‘prestige’ metalwork and other items covering a chronological span from the Neolithic to the late Saxon period (as it does at Shepperton (Fig. 22) came to be in the rivers from which they have been recovered (e.g. Needham and Burgess 1980, 442-9; Bradley and Gordon 1988). A consensus view seems to be developing that the normal explanation must lie in ritual/religious activity, and it is interesting to note that types of objects found in rivers here are often found with burials elsewhere in Europe (Torbrugge 1972) in view of the contrast between the frequency of pewter in burials on the continent and its rarity in Britain (Beagrie 1989, 174-81).

Figure 22. Finds Jrom Shepperton Ranges gravel pit. The deep alluvium indicates Former courses of the Thames and tributary streams.

Finds from riverine contexts are perhaps less spectacular from the Roman than from the prehistoric period, but finds, for example, of whole pots are not uncommon from the Thames through Surrey (Surrey County Council, Sites and Monuments Record). These must surely have found their way into the river as a result of ritual activity. Apart from the pewter hoards at Shepperton and Verulamium, several pewter hoards have been recovered from extinct river channels in Cambridgeshire (Lethbridge and O’Reilly 1933). In explaining their deposition, Lethbridge and O’Reilly (1933, 166) invoked the idea of concealed plate in the Picts War and suggested that ‘It is absurd to suppose that all these hoards could have been lost by accident … it is possible that these pewter services were placed in chests and sunk in the rivers with a small buoy to mark the place, much as fishermen now keep lobsters alive in the sea.’ This is an extreme example of reluctance to consider explaining the archaeology of Roman Britain in terms of non-rational or ritual activity.

A Context for Pewter Deposits
The above analysis of associations of deposits reveals that perhaps a more proper context for the study of a deposit of archaeological material is other deposits of archaeological material, and not an ‘historically documented fact’. A useful and tenable late Roman context, independent of the nonexistent ‘Picts War’, has been demonstrated for the pewter from Brislington, Appleford and Stanwick. For instance, the use of leather shoes, skulls and other animal bones, pottery, fruit stones, coins and metal objects as votive deposits on villas is clearly important in understanding pewter deposits. The deposition of leather shoes may relate to the Romano-Celtic practice of burying hobnail boots at sites such as Bradley Hill, Catsgore and Avebury (e.g. Salway 1982, 704-5); a burial rite which was a feature in the south, west and midlands in the fourth century. Another noteworthy hoard of Roman pewter associated with shoe leather, as well as ironwork (Manning 1972, 235), pottery fragments, a coin of Constans and a spindlewhorl, was found at the base of a stone-lined well 61ft Sin deep at Thatcham, Berkshire (Collingwood 1931).

There is a great deal of material from villas which waits to be interpreted as the by-product of the non-random ritual human behaviour that it surely is. Neither is this behaviour culturally intrusive. It belongs to the Celtic tradition, and Macdonald’s premise that Celtic beliefs did not prevail in the villas of Roman Britain is no longer tenable.

The pewter hoards from riverine positions lack direct association with contemporary material. Their location, however, makes the notion of hoarding with intent to recover highly unlikely, and, in a wider sense, their context is a sacred one: it is evident, for example, that certain stretches of the Thames, such as that at Shepperton, were foci of ritual activity over a very long period.

THE USE OF PEWTER
It has been a common assumption that pewter ware was a normal accoutrement of the dinner tables of villa owners in the Late Roman period. As Brown (1970, 108 – quoted with approval by Salway 1982, 636) expressed it, ‘At this time it must have been usual for every reasonably stocked household to have a dresser stacked with six or a dozen or even twenty or more, assorted plates, bowls and jugs.’ Such a belief is obviously closely related to the interpretation of pewter hoards as buried for safety with intent to recover. If the hoards are correctly understood as originating in ritual activity, this in itself makes the ‘villa dinner tableware’ notion difficult; after all, if its final usage was votive, how can we tell what its earlier function may have been? The question may, perhaps more usefully, be expressed another way: are the items in pewter hoards to be understood, like those in the comparable ironwork hoards, as everyday objects given a new function? In order to answer this it will be useful to look at some of the characteristics of the pewter ware itself.

The first point to be made is that none of the hoards recovered seem to represent anything like what we should regard as a ‘dinner service’. Instead, as Brown (1973, 201) pointed out, the material in the Appleford and other hoards has comparatively few matching pieces and instead seems to represent piecemeal acquisition. This might be better explained if one supposes that the hoard has its origins in material collected at a shrine or temple. Indeed, one piece at Appleford has a graffito – ‘Lovernianus presented the things he had bought’ (Brown 1973, 193, no. 24) – which implies a religious donation. Graffiti on other pewter pieces seem mostly to be personal names, but these too would seem more sensibly explained in the context of votive offerings. This in turn relates to a number of otherwise puzzling features of pewter ware. Could the dinner table of a prosperous household ever have been graced by objects with their rims splitting off due to defective manufacture (e.g. Brown 1973, no. 14, described as ‘very little worn as though the plate had not been used much’, and no. 20); or with casting marks and other irregularities still visible (e.g. Shepperton no. E2 (Poulton in prep. a) or Appleford nos 1, 18 & 21 (Brown 1973))? Finally, it would seem that much of the pewter we have has very little sign of wear in use (Peal 1967, 28; Poulton in prep. a), something which is remarkable when one considers the ease with which this soft metal can be marked.

In sum, then, it may be suggested that some pieces seem unsuitable for the dinner table, some seem to have religious connotations (and those with Chi Rho symbols (Peal 1967, 28) should be included here), and others seem to have been little used. The nature of the hoards themselves is ritual and the great majority lack even proximity to known villas (Brown (1973, 204) was driven to suggest that the Appleford hoard had been buried for safety by occupants of the Dropshot villa, a mile distant). It could be concluded that the overwhelming majority of the pewter we have represents only items collected or used in temples or shrines, or acquired for specifically votive purposes. There is no certain evidence that any of it was used for domestic purposes, and the question must be raised whether, speaking more generally, pewter was ever so used.

In considering this question, two other aspects of Roman pewter which have been much studied need to be taken into account. The most recent discussion of metallurgical analyses of Roman pewter (Beagrie 1989, 171-5) has made a number of suggestions to account for the variable composition of Roman pewter, which is almost always a tin-lead alloy, with the tin content varying from less than 40% to greater than 95%. He rightly points to the comparative cheapness of lead as a factor: this was also true in the medieval period, when it is clear that lead was added only to hollow wares in small amounts (10-20%) (Homer 1985, 151-2) and in very small quantities (typically less than 0.5%) to flatwares (Brownsword and Pitt 1984), presumably because greater amounts would have produced items which were unsatisfactory for daily use. The point is sharply emphasised by the discovery that sepulchral chalices and patens have up to 75% lead (Brownsword and Pitt 1985: in the 19th century such discoveries were described as lead chalices and patens, as for example one found at Chertsey Abbey (Poulton 1988,47». Might it not be the case that the composition of the Roman pewter we have is largely determined by the use to which it was put, and that the high lead pewter was manufactured solely for ritual purposes? This in turn may explain the apparent contrast between early Roman pewter which is normally high in tin and Late Roman pewter which is predominantly high in lead, the latter reflecting the increased use of pewter for ritual purposes in the third and fourth centuries.

The second aspect to be considered is the distribution of finds of pewter and evidence for pewter manufacture. The most recent distribution map (Beagrie 1989, fig. 3) shows a concentration of finds south of the Fosse way beyond Cirencester and to the north of a line drawn between the southern shore of the Bristol Channel and Canterbury. Finds elsewhere are more thinly spread (except, perhaps, for a secondary concentration in Cornwall) and almost completely absent from large areas of Wessex, South East England, Wales and the North. If this distribution is meaningful (and the fact that it is little altered by the plethora of more recent finds from that published by Wedlake (1958) may encourage such a view) then it should, on the arguments advanced above, reflect regional patterns of ritual activity rather than any domestic use of pewter. That such was occurring may be indicated by the wider spread of pewter small finds (Beagrie 1989, fig. 3), which are likely to represent casual losses of domestic material, and by the slightly broader distribution of evidence for pewter manufacture (ibid.), which might reflect use of pewter for non-ritual purposes. Finally, it should be emphasised that its indestructability and the ease with which it can be recycled mean that pewter is highly unlikely to be discarded as domestic rubbish.

CONCLUSIONS
Wherever there is adequate evidence for the circumstances of deposition of Roman pewter an association with religious activity is indicated. It is therefore a reasonable assumption that the vast majority of the extant pieces are the result of such activity. This metal seems to have found favour in the third and fourth centuries for a variety of religious purposes, most probably because it was a relatively cheap material which could nevertheless offer the illusion of the splendour of silver plate, and as such represented both an affordable and appropriate offering to the gods. The extent to which pewter was used for domestic purposes in the Roman period is unclear, but the character of that which survives may suggest that much of it was manufactured for specifically ritual purposes or had never been used for other purposes.

Acknowledgement. We should like to thank Phil Jones for much stimulating discussion.

Bibliography

Barker, W. R. 1901 Account of Remains of a Roman Villa Discovered at Brislington, Bristol December 1899. Bristol.

Barrett, J. C. and R. Bradley (eds) 1980 Settlement & Society in the British Later Bronze Age. Oxford; British Archaeological Reports 83.

Beagrie, N. 1989 ‘The Romano-British pewter industry’ Britannia 20, 169- 9l.

Bird, J. and D. G. Bird (eds) 1987 The Archaeology of Surrey to 1540. Guildford, Surrey.

Bradley, R. and K. Gordon 1988 ‘Human skulls from the River Thames and their significance’ Antiquity 62, 503-509.

Branigan, K. 1972 ‘The Romano-British villa at Brislington’ Proceedings of the Somerset Archaeological and Natural History Society 116, 78-85.

Branigan, K. 1977 The Roman Villa in South-West England.

Brown, P. D. C. 1970 ‘A Roman pewter mould from St Justin Penwith Cornwall’ Cornish Archaeolog;y 9, 107-10.

Brown, D. 1973 ‘A Roman pewter hoard from Appleford, Berkshire’Oxoniensia 38, 184- 206.

Brownsword, R. and E. E. H. Pitt 1984 ‘X-ray fluorescence analysis of English 13th-16th century pewter flatware’ Archaeometry, 237-44.

Brownsword, R. and E. E. H. Pitt 1985 ‘Some examples of medieval domestic pewter flatware’ Medieval Archaeolog;y 24, 152-55.

Buckland, P. C. 1980 ‘Insect remains from the well’ in I. M. Stead 1980.

Chapman, J. and S. Smith 1988 ‘Finds from a Roman well in Staines’ London Archaeol. 6, 3-6.

Coles, J. M. and D. D. A. Simpson (eds) 1968 Studies in Ancient Europe. Essays presented to Stuart Piggot.

Collingwood, R. G. 1931 ‘Roman objects from Stanwix and Thatcham’Antiquaries Journal 11, 37-45.

Cunliffe, B. (ed.) 1988 The Temple of Sulis Minerva at Bath: Vol 2. The Finds from the Sacred Spring Oxford; Oxford University Committee for Archaeology.

Frere, S. 1943 ‘Romano-British finds at Littleton’ Transactions of the London and Middlesex Archaeological Society 9, 203-4.

Frere, S. 1984 Verulamium excavations, 3 Oxford; Oxford University Committee for Archaeology .

Frere, S. 1989 ‘Roman Britain in 1988’ Britannia 20,303.

Garwood, P. et al. 1991 (eds) Sacred and Profane Oxford; Oxford University Committee for Archaeology.

Goodburn, R. 1984 ‘Objects of pewter’ in S. Frere, 65-7.

Greenfield, E. 1971 ‘The Roman villa at Denton, Lincolnshire. Part II’Lincolnshire Hist Archaeol. 1.

Henig, M. 1988 D. Brown N. Sunter L. AllasonJones and D. Baatz ‘The small objects’ in B. Cunliffe (ed.).

Homer, R. F .1985 ‘The medieval pewterers of London c. 1190-1457’ Transactions of the London and Middlesex Archaeological Society 36, 137-63.

Laing, L. 1979 Celtic Britain. London; Methuen.

Lethbridge, T. C. and M. M. O’Reilly 1933 in ‘Archaeological notes’ Proc. Cambridge Antiq. Soc. 33, 164-7.

Longley, D. and R. Poulton 1982 ‘The Saxon cemetery at Upper West Field, Shepperton’ Transactions of the London and Middlesex Archaeological Society 33, 177-85.

Macdonald, J. 1977 ‘Pagan religions and burial practices in Roman Britain’in R. Reece (ed.).

Manning, W. H. 1972 ‘Ironwork hoards in Iron Age and Roman Britain’Britannia 3, 224-50.

Meates, G. W. 1979 The Roman villa at Lullingstone, Kent.

Merrifield, R. 1987 The Archaeology of Ritual and Magic.

Miles, D. 1986 Archaeology at Barton Court Farm, Abingdon, Oxon. Council for British Archaeology Research Report 50.

Morley Hewitt, A. T. 1971 Roman Villa, West Park, Rockbourne, Nr Fordingbridge, Hants. Revised Edition.

Neal, D. S. 1989 ‘The Stanwick villa, Northants: interim report of excavations 1984-88’ Britannia 20, 149-68.

Needham, S. P. and C. B. Burgess 1980 ‘The later Bronze Age in the Lower Thames valley: the metalwork evidence’ in J. C. Barrett and R. Bradley (eds).

Painter, K. S. 1971 ‘Villas and Christianity in Roman Britain’ British Museum Quarterly (Prehistoric and Roman Studies) 35.

Peal, C. A. 1967 ‘Romano-British pewter plates and dishes’ Proc. Cambridge Antiq. Soc. 60,19-37.

Poulton, R. 1988 Archaeological Investigations on the site of Chertsey Abbey. Surrey Archaeological Society Research 11.

Poulton, R. in prep. a Archaeological discoveries made in a gravel pit at Shepperton, Surrey.

Poulton, R . in prep. b Excavations and geophysical survey on an earlier and later Saxon site at Shepperton Green, Surrey.

Reece, R. (ed.) 1977 Burial in the Roman World Council for British Archaeology Research Report 22.

Riddel, F. 1990 Stone Heads of the North – Some Observations. Unpublished BA Dissertation, University of Newcastle upon Tyne.

Ross, A. 1968 ‘Shafts, pits, wells – sanctuaries of the Belgic Britons?’ in J. M. Coles and D. D. A. Simpson (eds).

Salway, P. 1982 Roman Britain 2nd Edition.

Scott, E. 1988 Aspects of the Roman Villa as a Form of British Settlement. PhD Thesis University of Newcastle upon Tyne.

Scott, E. 1990 ‘A critical review of the interpretation of infant burials in Roman Britain, with particular reference to villas’ Journal of Theoretical Archaeology 1. Oxford; Oxbow.

Scott, E. 1991 ‘Animal and infant burials in Romano-British villas: a revitalisation movement’ in P. Garwood et al. (eds).

Stead, I. M. 1980 Rudston Roman Villa.

Todd, M. 1981 Roman Britain 55 BC-AD 400.

Torbrugge, W. 1972 ‘Vor-und friihgeschichtliche Fliissunde zur Ordnung und Bestimmung einer DenkmaJergruppe’ Bericht Romisch-German. Komm. 51-2, 1-146.

Wedlake, W. J. 1958 Excavations at Camerton, Somerset.