Internet Archaeol. 17. Willis. Chronology and repair

11.3 Chronology and repair

Marsh pointed out that there was an orthodox view that samian was probably repaired more often from the end of the 2nd century when supplies in bulk to Britain were diminishing. This logical assumption though has not been verified by any assembly of data.

Monaghan observed in the case of York that vessels from Central and East Gaul were more commonly riveted and repaired than South Gaulish vessels (1998, 949). However, it is not clear whether this difference in the case of York is one of relative proportions of repaired items, or relates to absolute numbers of repaired vessels. The difference may simply mean that repair was not a phenomenon especially associated with 1st century samian, and may be site specific. Most of the samian from York dates from after the floruit of South Gaulish ware (e.g. Marsh 1981, fig. 11.8; Monaghan 1998, table 170) and so it may be that this trend simply reflects the predominance of Central Gaulish samian, supplemented by East Gaulish ware, at this centre.

As noted previously, there is no firm evidence to support the thesis that repair became more common in the mid or late Roman period as imported supplies dwindled and stopped (Willis 1998a). Study of the samian from Alcester (Evans 1996a, 89) found that the percentage of samian vessels repaired at this site remained more or less constant through time. Indeed, a growing amount of evidence points, surprisingly, to the reverse of the supposed trend of a rise in repairing as supply declined. At Catterick, Bell and Evans noted that the evidence indicated a higher rate of repair at the time when samian was actually most common (2002, 472); data from other sites is in agreement with this pattern. From various sites at Catterick, a consistent pattern has emerged, specifically that there is little indication of a greater tendency to rivet or otherwise repair samian from the late 2nd century (Bell and Evans 2002, 415, 457 and 472). A listing of 115 repaired samian vessels identified in the course of this project contains only four or five East Gaulish vessels (Table 72). Likewise, among the comparatively large collection of 78 East Gaulish samian vessels from the second fort at Dover (1970-7), for instance, none show repair, while there are five repaired Central Gaulish vessels among a total of 353 vessels from that region. This tendency for repaired vessels to be apparently most frequent at times when samian supply is comparatively good may arise because demand was greater than the rate of supply, during a period when there would have been a strong consciousness of samian and a relatively intense use of the ware: in other words the period/s when it was most frequently repaired, was the time when it was prominent in social life. This echoes Kevin Greene's observation that the periods when local and regional 'imitation' fineware pottery industries in Britain were productive was not during periods when the supplies of fineware (samian) from the continent to Britain were low, but conversely when they were at their height, creating an interest in and custom of fineware use, and a demand for greater quantities (Greene 1982).

Marsh compared the dates of the 137 decorated sherds with evidence of repair among the collections from London with the overall chronology of decorated samian sherds in the Museum of London archive, finding that, 'the pattern is similar and ... riveting was done because a certain percentage of vessels were capable of repair when broken and not because of a shortage in supply' (1981, 227). His graph (Marsh 1981, fig. 11.18) shows a correspondence between the frequency of samian over time and the frequency of repaired samian items; this chart does show a comparative increase in the repair of 2nd century samian, but this could represent more or less entirely early/mid 2nd century repairing, rather than being later in date (that is late 2nd or 3rd century). Similarly, at Piercebridge Ward notes that only a low proportion of the latest East Gaulish vessels show repair which might suggest that repairs were not sustained beyond the early 3rd century (Ward 1993, 20), and were not conducted in earnest. On the other hand Ward notes that, 'It is likely that some at least of the later 2nd century products were repaired in the 3rd century' (1993, 20), which would probably comprise mainly Lezoux items. In these cases, and elsewhere, though, what is really needed is collated data on the stratified occurrence of repaired samian so as to identify any chronological trends in the sequence of deposited samian and repaired items. The repair of, for instance, 1st century or early 2nd century vessels could well have taken place in the 3rd century (but this cannot be assumed). It is relevant, then, to note of the samian from Piercebridge (1969-81) that, 'In the Tofts Field excavations, most of the well-used and re-used pieces were found in later phases' (cf. Ward 1993, 20).

Turning to the issue of the chronology of repair types, Marsh postulated that the cleat or dovetail method was a later, 2nd century, method of repair. This may have become more common as it will have reduced the chances of sherds breaking, compared with the risks when drilling holes through samian (Marsh 1981, 227). Ward notes that in the case of the sizeable ensemble of repaired samian from the excavations at Piercebridge, all the instances of dovetail repair occur on vessels manufactured after c. AD 150 but that this is not necessarily indicative of the date of this method as most of the samian from the site dates from after c. AD 150 (1993, 19). Data collected for the present report include several examples of what might be early instances of dovetail method repair. From a total of c. 107 samian vessels from site phase D at the Baromix factory site, Alcester, covering the period c. AD 60-150, a repair to one vessel is described, probably with a dovetail setting; this is a La Graufesenque Drag. 15/17 or 15/17R which in Britain has a normal date-range in circulation of c. AD 40-100 (Ward 2001, 34; Willis 1998a; see database). At the Vintners site, Great Chesterford, a Drag. 29 occurs with a cleat repair, while at Rocester, Orton's Pasture, a Lezoux vessel dating, typologically to before c. AD 120 has been repaired via cleats (Table 72). Similarly, a Drag. 37 of Lezoux, dating to c. AD 120-150, from the fill of the Vallum at Birdoswald dating no later than c. AD 150/160 displays cleat holes (Appendix 11.1). It is possible, of course, that the cleats appearing on typologically early vessels actually represent repair undertaken at a date later than the normal circulation period of the majority of examples of these types, when they were already old. On the whole, as Table 72 clearly shows, the cleat method was far less frequently used than riveting and it is generally the case that while typologically early samian (c. AD 40-140) is occasionally found with cleat method repair, these cases are fairly rare (cf. Table 72). Judging from the data collected together in Table 72 the cleat method appears to become marginally more common with later/2nd century samian (essentially from Lezoux), but this has to be seen in light of the fact that there are more cases generally of repaired 2nd century samian (cf. Table 72).