5.3 Exploring the date of Samian groups

5.3.1 Dating groups of samian

A method of dating samian which focuses upon forms present, especially plain forms is advantageous at a number of levels:

• The comparative ease with which plain samian forms may be identified
  Firstly, samian forms were highly standardised and recognising a specific form from a sherd 'in the hand' is usually straightforward and may be readily accomplished by the non-specialist using a guide (e.g. Webster 1996). It is surprising that such identifications have traditionally been left to specialists, given that the form typology for samian is, on the whole, less complicated than for other common wares of the Roman era found in Britain and north-west Europe; as in other fields confidence in identifications is bolstered by experience.
  
  
• The dating potential of plain samian forms
  Secondly, plain forms are, in virtually every instance, more common and often markedly more common, among recovered samian groups than are decorated items. Given a samian group is of some size, there is a reasonable chance of establishing some ratios between certain chronologically specific form types present, whether decorated or plain (but especially between plain forms; cf. below for method). This approach can establish a date for a group; previous work has shown that this approach has good potential.
  
  
• The infrequency of stamps among archaeologically recovered samian groups
  Makers' stamps appearing on samian forms (cf. Webster 1996, 7-9) have been one of our main props in dating sites of the Roman period, but they are often infrequent site finds and examples are often best evaluated as an amalgam from a site, rather than being particularly useful for dating site layers and phases.
  As a matter of course, recovered samian stamps are despatched for identification and dating by reference to the Leeds Corpus of Samian Stamps established and accrued by Brian Hartley and Brenda Dickinson, who have collated dating evidence for each maker's dies. This scholarly resource has been of invaluable assistance to the archaeology of Roman Britain and beyond, as Hartley and Dickinson have been able to furnish dates for innumerable stamps submitted over many years by reference to this corpus.
  It is important to recall, however, that not all samian forms were stamped, while some forms were variably stamped. In certain periods and certain industries, stamping was less frequent than at other times, as for instance with the East Gaulish industries, prominent in the later 2nd to mid 3rd century AD (cf. Bird 1993, 2-3).
  To gauge how infrequent stamps are among samian groups, it is worth recalling that there will be many instances of site phases with which no stamps are associated, for instance, as with the earliest phase at Ribchester 1980/1989-90 (Dickinson 2000a, 202). Table 2 lists the incidence of stamps amongst samian groups showing the size of each group in terms of number of sherds. These data suggest that on average stamped items occur less than twice per 100 samian sherds recovered from fieldwork (though of course this ratio will be related to how fragmented each group has become). Of these stamps, some will be illegible, some fragmentary or otherwise unassignable, and thus undated. Table 3 shows the frequency of stamps amongst groups of samian where the number of vessels represented has been established; again the ratio of stamps to vessels represented is striking, being on average 1 stamp per 26 vessels represented. In sum, the percentage of stamps recovered from usefully stratified contexts is very low indeed. In other words, in the pursuit of dates for Roman contexts there has been a reliance on a very small sub-fraction of the samian assemblage that may be infrequently present. Moreover, while individual stamps can furnish comparatively short date-ranges it is uncommon for groups to yield more than one or two examples at best, and hence the potential for verification of dates or the refinement of the date of a stratified group is circumscribed; stamped items present in any context could also be residual.
  
  
• The incidence of decorated items among archaeologically recovered samian groups
  Similarly while decorated sherds are more common than stamps, decorated forms are less frequent finds than are plain types by some margin: with the exception of a few Roman military sites and vici, decorated items typically, at best, account for no more than c. 35% of the vessels within samian groups (cf. Table 42). At major civil sites the average figure is around 26%, while at 'Small Towns', religious foci, roadside settlements and rural sites, that is to say, by far the majority of sites of the Roman era in Britain, this percentage is normally lower still at c. 17-20% (cf. Section 7.3; Table 42). Moreover, in a rather high proportion of cases, sherds with decoration, or otherwise from decorated forms, cannot, anyway, be dated more tightly beyond the date indicated by the fabric of the sherd, its form, or the quality of its surface gloss. The chances that a sherd from a decorated vessel might be 'tightly' dated depends upon several factors: the sherd has to be from or include part of the decorated zone of the vessel, the decoration must be reasonably well preserved, the decoration must be diagnostic, etc. Often sherds are too small for the decoration represented to be chronologically diagnostic and in some cases the condition of the samian is poor due to taphonomic processes (as at Birdoswald Spur: Willis forthcoming f).
  
  
• Summary
  Consequently, a means of dating samian groups via the plain ware forms present reduces reliance upon the comparatively infrequent chronologically diagnostic decorated pieces and stamps. Building a method that considers proportions of plain forms present, therefore, may mean that as the number of archaeological field interventions expands, the few samian specialists are less hard pressed to provide dates and identifications for all reports.

Table 2: The representation of stamps among samian assemblages by number of samian sherds

References: Birdoswald (Willis 2000c; forthcoming f); Great Chesterford (Miller 1995; Willis 2003); Great Dunmow (Rodwell 1988); Haddon (Willis forthcoming h); Lancaster (Willis 2000f); Rocester (Willis forthcoming e).

References: Alcester (Booth and Evans 2001; Ward 2001); Alchester (Booth, Evans and Hiller 2001; Dickinson 2001d); Bar Hill (Robertson et al. 1975; Docherty and Wild 1975); Brockworth (Rawes 1981); Carlisle (McCarthy 1990; Taylor 1990; Dickinson 1990); Catterick (Wilson 2002; Bell and Evans 2002; Dickinson 2002a); London, forum (Philp 1977; Dannell 1977); Maxey (Pryor et al. 1985; Wild 1985b); Metchley (Jones 2001; Pengelly et al. 2001); Milton Keynes, Woughton (Marney 1989; Pengelly 1989); Orton Hall Farm (Wild 1996); Southwark (SLAEC 1978); Towcester (Brown and Woodfield 1983; Bird and Pengelly 1983); Verulamium (Frere 1972; Hartley 1972a).

© Internet Archaeology URL: http://intarch.ac.uk/journal/issue17/1/5.3.1.html
Last updated: Mon Mar 7 2005