Samian ware was a much prized form of table ware. As a valuable commodity, it would have been carefully preserved and lasted long in use, particularly after the end of the second century a.d., when export from the Central Gaulish potteries at Lezoux ceased and it became more difficult to replace breakages. Hence, broken vessels were commonly riveted together. The mend may not have been water-tight, but could have been made so by the use of adhesive. Sherds with rivet-holes and grooves are a common find on sites.Footnote ¹¹⁴ Others have the break coated in a black tarry substance, probably some form of adhesive.

A measure of the quality of samian ware and its prestige is also shown by the way in which vessels were reworked to form other, smaller objects. Geoff Marsh provides a useful summary of these procedures.Footnote ¹¹⁵ Broken rims could be filed down to make smaller vessels. Bases could be trimmed to produce small dishes, lids, or palettes. The ‘kicked’ base of form 31 could have been used as a spinning top. Smaller sherds could be converted into counters or spindle whorls.

In her study of the samian ware from the late second century onwards at Piercebridge, Margaret Ward notes 74 vessels showing repairs, about 40 trimmed bases with evidence for reuse, and a large group of counters and spindle whorls.Footnote ¹¹⁶ On the basis of this evidence, she suggests the presence of a workshop in the vicus, responsible not just for repairs, but also for the production of reworked items for sale. That repaired and reused samian should have been on sale is a not unreasonable supposition, particularly in less affluent settlements and in the later Roman period. Wroxeter, however, where Atkinson mentions repaired items along with a deposit of samian interpreted as a stall-holder's stock,Footnote ¹¹⁷ does not seem to offer an altogether convincing example of the regular sale of second-hand products. The material interpreted by Atkinson as reserve stock, as no doubt it was, was found in the fire-damaged East rooms of the forum rather than in the gutter, into which the contents of the stalls had fallen. In discussion, intended primarily to establish the date of destruction, he notes the presence of repaired samian and also of coarse ware in these rooms in addition to the piles of samian. Whether these, too, were for sale is a matter of interpretation, particularly in the case of Room 3, the findspot of the single repair illustrated,Footnote ¹¹⁸ where later disturbance is mentioned. The repaired pieces were not openly displayed on the stalls, but with the reserve, and could equally be interpreted as specially commissioned repairs. While it is not impossible that the odd repair was on sale as a ‘second’, in a major town and tribal capital in the mid-second century the inhabitants were able to obtain, and could no doubt afford, new products.

The most convincing evidence for a workshop for the repair and reworking of samian ware, by commission or for sale, comes from Kempston Church End, a village or roadside settlement in a rural area, on the bank of the River Great Ouse outside modern Bedford. Recent excavations revealed a group of pits,Footnote ¹¹⁹ one of which contained a large group of samian ware, almost all of which showed evidence for repair or reworking. No buildings were found in direct association with the pits, but these may have been ephemeral structures which left little trace in the ground. It seems unlikely that a weight of broken samian ware would have been transported over a distance merely to dispose of it. The potters’ stamps suggest that the workshop was in operation at the end of the second century or, allowing for a period of use before breakage, the beginning of the third. With the exception of the two stamps of the potter Littera on form 18/31R, mentioned below, all the stamps dated to the second half of the second century a.d. and were on later second-century forms.Footnote ¹²⁰

The pit contained substantial parts of up to 19 vessels.Footnote ¹²¹ All were plain forms. With the exception of sherds from a South Gaulish Curle 11 and an East Gaulish form 18/31, all were Central Gaulish and from Lezoux. Three showed evidence of repair. Two dishes (online fig. 1), both of form 18/31R and possibly belonging to the same owner, showed an initial break across the centre of the dish which had been repaired with dove-tailed grooves containing lead fillings, some of which remained in situ. Both showed stamps of the same potter, Littera i (c. a.d. 120–50), and were among the earlier pieces in the group. Broken into 18 and 9 sherds respectively, the dishes may have been returned for further mending, to be rejected as irreparable. The third piece, of form 36 (online fig. 2), showed two circular rivet-holes, both broken through. Marsh suggests that the use of dove-tailed grooves, less likely to cause breakage than drilling, was the later method, becoming more popular in the second century,Footnote ¹²² a fact disputed by Ward on evidence from the Middlewich vicus,Footnote ¹²³ where the two methods appeared equally on both South and Central Gaulish products. At Kempston, too, both methods appear to have been in use concurrently, with the grooved repairs potentially the earlier.

At Kempston, however, the craftsman's speciality appeared to have been the trimming of bases, of which there were six examples ( fig. 10), one of a cup (form 33), the other five of dishes (form 18/31–31). These had been trimmed by splitting off the wall at the junction with the base. The break was then rubbed down or polished to provide a smooth edge. Not all the trimming was complete; some points and uneven breaks remained to be removed. In one case the trimming was still under way. Part of the wall remained attached to the base, while two loose, joining, wall sherds had been removed. Also in the pit were groups of rim sherds, many joining, which had been split from the bases, though the secondary trimming and rubbing down of the edges of the base meant that it was impossible to tell from which of the bases they may have come. The groups of joining rim sherds, amounting in one case to 80 per cent of the rim, would appear to contradict Marsh's suggestion that reuse of this kind took place when the vessel was so badly broken that only the base remained intact.Footnote ¹²⁴ None of the bases showed the bands of wear inside the footring noted by Ward as indicative of their reuse as lids,Footnote ¹²⁵ nor were there any other signs of wear likely to have been caused by secondary, as opposed to primary, use. This, together with the presence of the rim sherds, provides strong evidence that reworking was taking place on or near the spot where they were found and that they had not yet been sold or distributed for subsequent use.

FIG. 10. Kempston Church End, Beds.: Trimmed samian bases and accompanying wall sherds from the pit. (Photo: Priscilla Wild)

Other types of reworking are less certain. A bowl of form 38 (online fig. 3), broken across the middle, was almost complete apart from the removal of two pieces of rim on opposite sides of the bowl, possibly an attempt to make pouring-spoutsFootnote ¹²⁶ when the bowl broke. The flange of the South Gaulish form Curle 11, mentioned above, had two lumps broken off it, of roughly equal size. Whether this was deliberate and if so, for what purpose, is unclear. Six joining sherds of form 79 showed no evidence for repair or reworking.

Four of the vessels showed owner's marks, two in the form of notches on the footstand, one an X beneath the base, and on the outer wall of the form 38, the letters CAR, presumably the start of the owner's name. The marks, implying different owners, may provide evidence that the craftsman was working to serve the wider community rather than simply the needs of his own household. A further, and more general, implication is that at least one of the owners was literate. The other three, in all probability, were not.

The pit contained no decorated bowls. In general, even in rural areas, decorated bowls, more highly prized and expensive, tend to be repaired more frequently than plain vessels. Ward cites examples from military sites and their vici in the North-West where repairs to decorated bowls are in the majority.Footnote ¹²⁷ In a rural area such as Kempston, where decorated ware was less frequent,Footnote ¹²⁸ proportions of this sort could not be expected, particularly in the later second century, when decorated bowls were clearly becoming more difficult to come by. The proportion of decorated ware recovered in the recent excavations from the site as a whole, however, is 13 per cent,Footnote ¹²⁹ so it seems strange that none should have ended up at the workshop for repair.

There were no counters or spindle whorls in the pit. These do not appear to have been in the craftsman's repertoire, nor were any found elsewhere on site during the recent excavations. It is also worth recording that there were no examples of the reworking of other types of pottery in the pit.Footnote ¹³⁰

A final question is why all this material was jettisoned. Some pieces, such as the riveted dishes, were clearly irreparable. In other cases, it is likely that breakage occurred while the reworking was in progress, making continuation pointless. Some of the neatly-trimmed bases, however, would surely have been useable, or saleable, with only a little further polishing. It is hard to see why these items should have been regarded as failures.

Of the extent to which products of the workshop were in use elsewhere on the site we can only guess. Two other examples of trimmed bases were found, one the base of a small, unstamped, East Gaulish cup (form 33 or 46), with heavily worn footring. The original inner surface showed abrasion marks suggesting that it may have been used in its recycled form for rubbing or polishing. The other was half of a trimmed form 33 base, broken across the edge of the stamp. A further case of possible reuse was a rectangular piece of mortarium wall (form 45) centring on the ‘bat-headed’ spout. The form of the spout suggests manufacture at Trier in the first half of the third century.Footnote ¹³¹ There appear to be traces of a rivet-groove on one break. The spout is heavily worn, though whether through primary or secondary use is not clear.

Repair and recycling of materials took place in the ancient world as a matter of course.Footnote ¹³² The method of recycling varied according to the nature of the material. Some, such as glass and metal, could be melted down and used to make new objects, as witnessed by evidence at many sites for the collection of broken glass as culletFootnote ¹³³ and of hoards consisting largely of scrap metal.Footnote ¹³⁴ Building materials could be reused in their original form. In the post-Roman period, the nearest source of good building stone was likely to have been in disused Roman buildings, which were happily plundered for their masonry to build churches, as at York.Footnote ¹³⁵ Pottery was more difficult to recycle, though, as we have seen, it could be mended or cut down to form smaller objects. While the evidence is greater for samian ware, other types of pottery could also be recycled. Instances of repairs to mortaria and black burnished ware have been noted.Footnote ¹³⁶ A group of trimmed pot bases, possibly for lids, was recently found in a well or tank at Bretton, Peterborough.Footnote ¹³⁷ Amphorae could finally be used for land reclamation.Footnote ¹³⁸ Like pottery, clothing and textiles were repaired, by darning and patching, or cut down and made into smaller items.Footnote ¹³⁹ When the rags became irreparable, they could be used for general purposes such as padding.Footnote ¹⁴⁰ The Edict of Diocletian mentions refurbishers of different kinds of textiles.Footnote ¹⁴¹ Perhaps we have an example of their ceramic equivalent at Kempston.