Internet Archaeol. 17. Willis. Verifying the Date of Samian Form Types

5.4 Verifying the date of samian form types

The database enables the conventional chronology of the individual samian form types, established through piecemeal observation, to be 'tested' for the first time against a large body of stratified examples. 'Master curves' in the shape of graphs for the frequency over time of each samian type represented can be generated from the project database using the summary data in Table 81 (an Excel spreadsheet). Some examples of main samian types are presented and discussed here. These curves can be compared with the existing 'conventional' ascribed date-ranges of types, and in this way, conventional dates may be assessed, and either 'verified' or found, perhaps, to be in need of qualification.

5.4.1 How ascribed dates for samian types have been deduced

Presently the dates applied by specialists to the individual samian form types are the result of a cumulative observation of the occurrence and association of these forms. This, ultimately, is traceable to reliably dated horizons linked to coin evidence and/or historically attested events, such as the establishment of military installations of the Limes, including the northern frontier in Britain and 'campaign forts' (cf. Hartley 1972b; dating the debut of the Drag. 31R bowl to post c. AD 160 is an example; cf. Oswald and Pryce 1920). Information from military sites has been particularly instructive as many such sites, not least in Britain, had short lives or distinct phases, and their foundation, abandonment and perhaps re-occupation can often be more or less securely tied to historically attested events (cf. Breeze and Dobson 1985). The result is what may be termed 'conventional date-ranges' for forms (cf. Webster 1996).

It is significant that the scheme of samian dating has in genesis and development been very closely linked to military horizons. This has been advantageous in terms of, for instance, association with historically identifiable episodes, with evidently fresh supplies being used at newly founded installations, minimal residuality at sites occupied for a short duration, etc. On the other hand this gives rise to the probability that, as discussed below (this section), these conventional datings are less accurate for non-military sites. This is a potential problem that, previously, has never been properly realised.

Those reporting upon pottery types, including samian, often refer to the date that they ascribe to pieces as 'the date of manufacture', meaning the period during which a type was being made. This is not so; we know a little about when types were produced from details of production sites. What we know much more about from the archaeological record is an approximation as to when types entered the archaeological record. In fact this ascribed date is really an aggregate date of deposition for the type, based upon the incidence of other known examples; it is not a production date (cf. Millett 1987a). In other words the dates ascribed to pottery vessels, sherds and stamps actually represents the date range of deposits with which like pieces are most normally associated. (The emphasis of this date range will normally be somewhat later than the start date of production, since there will usually be a time lag between production and the deposition of sherds in site deposits (Millett 1987b)).

To take up the point outlined above (this Section), an important factor at work with the dating of samian ware relates to its dating via reference to its appearance at military sites. Units of the Roman army evidently received fresh consignments of samian on a regular basis, and newly established forts often appear to have a strong element of new supplies (pers. comm. Felicity Wild; Section 6.3). That military sites and horizons are often closely dateable by other, non-samian, means such as coins, texts and known campaigns, makes these contexts especially useful for dating artefact types, including samian. Hence military sites can be a more sensitive indicator of changes in samian typology and supply than are other types of site; from military sites and horizons we can gauge and date the appearance of new types and note the absence (demise) of other types, as Hartley has demonstrated (eg. Hartley 1972b). Hence attention to pottery on the Limes, has been central in developing ceramic chronologies (as is so plainly apparent from specialist reports upon samian assemblages). However, the likelihood is that given a regular turnover of samian at military sites, and the consumption of large amounts of samian at these sites (cf. Section 7) associated with discrete phases, pottery groups and assemblages from military sites are often more closely dateable than are those at other types of site. At non-military sites, including major civil centres, where there is perhaps slower turnover of samian, groups of stratified samian may appear older than contemporary groups at military sites; thus there may be something of a bi-modal element in dating based upon samian which has not been acknowledged in the literature to date (cf. Section 6.3).

5.4.2 Examining ascribed dates: questions and potentials

In truth the date-ranges ascribed to pottery types, including samian, have not usually been arrived at by a systematic analysis, nor have they been rigorously tested against the evidence of their stratified incidence (see Tyers 1996, 36-42). This is not to say that these conventional dates are inaccurate, rather that it is appropriate that they be tested in some way. The database has the potential to systematically 'test' these conventional dates in a relatively independent manner for the first time.

The database can make a further contribution to the refinement of samian dating. Since it is founded upon the evidence of stratified instances of samian items, it can provide a comparatively independent check upon the dating of types.

It might be contended that there exists an element of circularity in this method, in so far as the conventional date-range of a samian type in a group will have been a contributory piece of evidence in the general dating of that group. This will be so in degree. However, dating of site deposits and of artefact assemblages in the Roman period is not determined these days solely by samian ware; indeed, far from it. There are now other chronologically diagnostic categories which have proved themselves to be robust dating indicators. In large part dating depends on stratified sequences, and since the introduction of the Harris matrix as a routine procedure site sequencing has become systematic. Moreover, samian is usually only a moderate to small constituent of stratified pottery groups, with other types of pottery present not only more numerous but also being useful for dating in varying degrees, and often contributing key chronological information. Going (1992a) has shown that the view that site horizons are dated by their samian is rather out-moded as our knowledge of the coarse ware pottery and non-samian finewares of the Roman era has become sufficiently sophisticated to yield a robust independent chronological index for some while. This chronological information has been fed into the dating of sites accordingly over the past 25 years or more (Going 1992a, 96). The importance of coarse ware and non-samian fineware in dating assemblages can be seen in Going's Chelmsford report (1987) and in other reports (eg. Martin 2000a; 2000b). The best outcome is when it is possible to combine chronological information from both samian, other pottery types, and other sources. Hence the dating of site horizons is in practice not dependent upon samian alone: far from so.

In addition, a number of the groups used for the database have associated coins, or other dateable artefacts associated, such as brooches. These aspects are significant as they show that there are elements in the ascription of dates to groups and phases that are independent of samian and hence the potential impact of circularity is diminished (cf. Section 5.2.2).