Internet Archaeol. 17. Willis. The 'Long Goodbye' of samian ware

5.8 The 'Long Goodbye' of samian ware

5.8.1 The longevity of samian vessels in use

Dates ascribed to samian and other pottery types relate to (or should relate to) the periods when similar examples are most typically found in archaeological deposits. It is not uncommon, though, for fragments from samian vessels that would normally be ascribed a particular date to appear in deposits of later date, significantly when they do not appear to represent residual items but vessels in use close to the time of deposition. Samian vessels appear to have been looked after with care, and were often curated. The degree of this special treatment seems, as in other domains, to set samian apart from other pottery types.

Millett's suggestion that the frequency of a pottery type over time should usually follow a normal distribution curve is reasonable and is borne out by our knowledge of the biographies of pottery types of the Roman era, as found in archaeological deposits (1987a). This characterisation sees the frequency beginning with a moderate number of examples as production begins and a few early items are incorporated in the archaeological record; this frequency then rises to a peak as the type becomes popular, when sherds from broken examples, etc., are most commonly lost; and then the frequency tails off as styles and types change, production declines and halts, leaving a 'tail' of examples in the archaeological record as the last/old examples are broken, lost, etc. and incorporated into site deposits. A 'tail' will continue as residual pieces are found through site sequences. This 'usual pattern' lies behind perceptions of dates ascribed to pottery types in the Roman era. With samian, however, there is evidently a pattern of some difference. A large amount of evidence points to the prolonged life of samian vessels, through curation or careful use, resulting in a frequency curve that does not end in a rapid fall-off and a thin tail (as with a normal distribution curve), rather the decline was slower. This has implications for the use of samian as a dating tool, especially in the first half of the 2nd century and in the mid to late Roman period.

The longevity of samian has been observed from time to time. Booth, reporting on the pottery assemblage from the Roman settlement at Mansfield College, Oxford, notes: '...Characteristic Antonine forms are for the most part very clearly absent. The obvious exception to this, a few Central Gaulish samian sherds, occurred in Phase 3 and later contexts, and such material is indeed characteristic of late Roman rural assemblages, where it seems to have remained in use for very long periods after its manufacture, and need not necessarily indicate contemporaneous settlement' (2000, 311). In similar tone, in his consideration of the pottery from the rural/religious complex at Slonk Hill, north of Shoreham, West Sussex, Fulford had commented that in this case: 'To use samian is of only little help since, although providing a terminus post quem, its high survival rate may date groups too early, for lack of well-dated alternatives' (1978, 119).

5.8.2 The longevity of South Gaulish samian vessels

It was noted in the report on phase 1 of this project (Willis 1998a) that while the currency of South Gaulish, La Graufesenque, samian is conventionally considered to end by c. AD 100/110 (since it is from deposits dated to before this threshold that the majority of examples come), a significant proportion of vessels from this source in Britain evidently continued in use into the first half of the 2nd century. This was particularly noticeable with decorated bowls (Willis 1998a). This sustained currency occurred at a time when supplies from other sources (Les Martres and Lezoux) were comparatively low.

In the case, for instance, of the religious complex, presumably associated with the Roman army, at Orton's Pasture, Rocester, Bevan notes, with regard to pit groups F700, F701, and F703, that: '... the general impression is of contemporaneity between the features, although the dating of some illustrated coarsewares from F700 suggests a somewhat later date (Hadrianic to Antonine) than the Flavian to Flavian-Trajanic date recorded from the diagnostic samian ...' (2000, 27). Similarly, Clark, in reporting the pottery from Causeway Lane, Leicester, states of a group of site phase 2F (1762 sherds, 38.8kg) from the construction and use of 'Surface 3' that: 'The samian is typical of the late 1st to early 2nd century AD, predominantly Flavian in character, but with a high proportion ... of earlier material. Once again the absence of Trajanic and late Flavian South Gaulish material does not preclude use of the surface into the early 2nd century' (1999, 117); this phase group has an assigned ceramic date of c. AD 80-120 and a stratigraphic date of early 2nd century AD or later (Connor and Buckley 1999, 24), while the samian is clearly earlier in emphasis by its conventional dating. Other cases from this site of groups with only South Gaulish samian, but which are likely to be 2nd century in terms of activity and deposition, occur (e.g. Phases 2H and 2J: Clark 1999, 118; Connor and Buckley 1999, 25-6). In the West Bridge area of Leicester, Pollard had previously noted the same general trend, stating that: 'it is apparent that late 1st century samian continued to be used into the reign of Hadrian' (1994, 77). Clark concluded that the explanation for this apparent mismatch lay in: 'the curation and survival in use of earlier material' at a time when samian supplies were scarcer (Clark 1999, 118).

At Segontium, King and Millett (1993) looked at the proportions of decorated samian occurring through the site sequence; they found that there was an increase in the proportion of decorated South Gaulish ware through the sequence. They commented that: 'This perhaps suggests a higher rate of survival for decorated bowls, possibly resulting from differential use patterns ... If further evidence can be adduced in support of this hypothesis it may have wide-ranging implications for the use of samian for dating' (King and Millett 1993, table 16.6, 243). Phase 1 of the present project indeed showed that decorated samian bowls from South Gaul evidently did continue in currency into the 2nd century more frequently than plain ware (Willis 1998a, 103-4, compare illus. 10 with illus. 11). It was noted in the earlier report that this trend was one predicted by Marsh (1981, 219).