2. Later Romano-British Pottery at Mucking

While as a whole there is little later Romano-British pottery at Mucking (certainly when compared to the substantial earlier Roman assemblages), there does appear to be a significant, although small, assemblage of later 3rd- to later 4th-century wares. Possibly of this date were 394 sherds of Nene Valley colour coats, 590 sherds of Much Hadham wares, 57 sherds of Parchment wares and 24 sherds of Late Shell-tempered wares, while definitely 4th century in date were 28 sherds of À l'Éponge, 195 sherds of Portchester D Ware, three sherds of Alice Holt wares, seven sherds of Mayen wares, 18 sherds of Rettendon Ware, and 501 sherds of Red Oxfordshire Colour coats (Table 1). Going (1987) has argued that these wares, when found in Essex, are characteristic of the latest 4th-century activity, even when they are found in earlier contexts elsewhere.

At other Romano-British settlement sites, later phases of activity, particularly 4th-century activity, is often characterised by the build-up of surface deposits in a form of middening; where surfaces are stripped, these are often 'caught' in the top of negative features such as large ditches and hollows. This is true, for example, of the sites at Langdale Hale and Campground within Earith parish in Cambridgeshire (Evans et al. 2013), at Kilverstone, Norfolk (Garrow et al. 2006) and at Vicar's Farm, Cambridge (Evans and Lucas forthcoming). However, this middening is invariably associated with other firm signs of occupation, such as the laying out of new enclosure systems (often curvilinear in form), the building of structures and craft and agricultural production. This is reflected in the material culture of these sites, for example in the date-profile of coinage, or of pottery.

As noted previously by Going (1993b, 72), 44 of the 133 Roman coins from the excavations at Mucking were recovered from Anglo-Saxon Grubenhäuser or other Anglo-Saxon features, including 22 from a single structure, GH57. Comparison of the adjusted coin profiles for coins from Roman features against those from Anglo-Saxon features reveals an extremely divergent pattern. Of the Roman-feature coins (Figure 2), the profile shows a sharp fall after AD 238, suggesting a dramatic decline in activity at this time (something that would also be supported by the low numbers of fineware pottery after this date).

The bulk of the later coinage from the site was recovered from non-Roman features (Figure 3): of 53 coins firmly dated to after AD 238, only 14 (26%) were recovered from Roman-period features, while the remainder were found in Anglo-Saxon contexts. Of those 14, all but six were found in contexts (mainly upper fills of ditches) that also contained Anglo-Saxon pottery, and can be assumed to have been redeposited. The six exceptions were found within one area of the site, within the upper fills of the 2nd-century enclosure ditch (D15) that lay to the north of RBI, and associated with internal features. Various gullies and posthole alignments suggest that there may have been a structure within this enclosure, possibly dating to the 4th century, but there are also four Anglo-Saxon Grubenhäuser and a post-built structure, along with postholes, pits and a gully associated with Anglo-Saxon material. It is possible that all the coins deposited in this area too were part of a post-Roman phase of activity.

The main potentially later Roman enclosure ditch discussed above (D15) is worth examining in more detail for the depositional patterns it reveals. A total of 1422 sherds of fine and coarseware pottery were catalogued from the ditch fill (Table 2), as well as two sherds of amphorae (depths unrecorded), and six sherds of mortaria and twelve sherds of samian. Examining the fabrics by phase, it became clear that the later material resided solely in the upper fills of the ditch: this was clearly a 2nd-century ditch that gradually became infilled in the later 2nd- or 3rd century, with elements of later depositional activity in the resulting hollow. One such activity might have been deposition of a small coin hoard: as well as the three 4th-century coins found in the vicinity of 550N/325E, a further three undated coins were recovered from the same location, also in the upper ditch levels.

Pottery of the later 3rd and 4th centuries was only found in the uppermost spit (this is also true of Anglo-Saxon pottery). This pattern of ditch infilling is also seen in the other main enclosure systems (Tables 3–4); those spits with later 3rd- or 4th-century pottery invariably see the deposition of Anglo-Saxon pottery alongside.

It thus appears that these specifically later Roman pottery fabrics may have formed a part of the Anglo-Saxon domestic assemblages that were being deposited on the site, presumably in the 5th century and later. There are two possible explanations for this: contemporary usage or curation. On other Anglo-Saxon settlement sites, the occurrence of late Roman pottery in early Anglo-Saxon features has occasionally been noted. At both Heybridge, Essex, and West Stow, Suffolk, analysis revealed that this Roman pottery was of an unusual composition, with higher than expected proportions of colour-coated and other 'significant' sherds (Going 1993a, 71–2; Drury and Wickenden 1982). The assemblage from West Stow is interesting, as Plouviez (1985, 84) has compared its composition to later Roman assemblages from Icklingham. She concluded that the West Stow assemblage differed from a 4th-century settlement site assemblage in its bias against grey and later shell-gritted fabrics, and in its imbalance in parts of pots represented. She concluded that a 'normal' assemblage would display numerous body sherds, and that rim sherds would outnumber base sherds, and that the West Stow assemblage represented deliberate selection.

Plouviez (1985, 84) thus argued that late Romano-British non-samian pottery found in SFBs at West Stow had a breakage profile (i.e. the proportion of rims vs bases vs body sherds) that represented deliberate curation or collection (that Anglo-Saxon settlement being located close to a former Romano-British settlement). There were as many of these late red sherds in later SFBs as there were in earlier ones. The pattern of 19% rims, 44% bases and 37% body sherds at West Stow was contrasted with the Romano-British site at Icklingham, which saw 22% rims, 8% bases and 70% body sherds (leading to the conclusion that at the former site, rims and especially bases were preferentially collected). A similar pattern to Icklingham is reflected for Romano-British Mucking as a whole: 29% rims, 11% bases and 59% body sherds, and it can be suggested that this is what a 'normal' Romano-British broken pottery assemblage looks like. It is important to emphasise that this in itself is different from when whole vessels are destroyed and deposited en masse; the Well 4 assemblage at Mucking (which represents the clearance of a destruction event) gives us a typical pattern for such an event: 12% rims, 44% bases and 44% body sherds.

When we look at the breakage patterns for the later Roman pottery at Mucking, it conforms more to the Icklingham pattern than the West Stow one (Table 1). Looking at the Oxfordshire Red Colour coats, for example, we see 29% rims, 12% bases and 59% body sherds; this suggests more 'normal' deposition patterns rather than deliberate collection or curation. Similarly, Portchester D Ware sees 16% rims, 9% bases and 75% body sherds; Much Hadham has 12% rims, 7% bases and 81% body sherds. The pattern for all the other later imported wares is similar, with a proportion of body sherds of around 70% or higher, except for those with very small numbers of sherds represented. It is of interest, however, that it is the red wares that see proportionally more rim sherds and bases represented than the non-red wares, and this may represent a later collection effect. To investigate this in more detail we need to examine the contexts in which these sherds were found.

One way to do this is to examine the assemblages of Roman pottery found in Grubenhäuser at Mucking, to see whether a 'normal' or a 'curated' assemblage appears to be present. The picture is slightly complicated, however, in that this Roman pottery appears to fall into two distinct camps: that from structures cut directly into Roman features, and from those that were not. Looking at the pottery assemblage from the two structure groups (Table 5) suggests that the former group displays a 'normal' settlement assemblage, with a predominance of greywares (perhaps a result of pottery from the underlying ditches becoming later incorporated in the structure fill, or the fills not being adequately distinguished during excavation). The latter group, however, looks much like the West Stow assemblage in terms of fabric proportions, with high numbers of colour-coat sherds, including samian.

It does look different, however, in terms of the parts of vessels represented (Table 6). Both in the 'Roman' Grubenhäuser and in the other group, these Roman sherds have a 'normal' breakage pattern, with at least 66% body sherds, and higher proportions of rims than bases. Either the usual selection practices for late Roman pottery were not being followed, or these late Roman vessels were being used and broken in everyday ways. Just 7% of these later Roman sherds were recorded as being abraded (86 of 1213); this compares to a site average of just under 2%, and 3.5% of the specifically 4th-century sherds. However, this appears to be restricted to certain fabrics: 19% of all the Oxfordshire Red Colour coat was abraded (97 of 501 sherds; 14% of which were recovered from Grubenhäuser), compared to 3% of the Much Hadham products (18 of 590), 2% of Nene Valley products (9 of 394 sherds) and no abraded sherds at all in Portchester fabric D. Similarly, 7% of the samian from the site was recorded as abraded (197 of 2691), but 23% of that in the Grubenhäuser (34 of 145 sherds).

It appears as if bright red wares were being targeted for selection during the Anglo-Saxon occupation of the site (and being more likely to be abraded), but that other fabrics formed part of the domestic repertoire, with less abrasion, and seeing more normal breakage patterns. What we do not appear to be seeing, in the 'non-Roman' feature Grubenhäuser is any redeposition of Roman midden material back into the fills (as might be argued for the structures dug through Roman features); the comparative lack of any greywares demonstrates this conclusively.

Interestingly, the pattern for samian in Grubenhäuser is different: 26% rims, 23% bases and 52% body sherds – could there be a combination of redeposition and curation involved here? Several Grubenhäuser in the northern part of the site have occasional sherds of samian and/or Red Oxfordshire Colour coats, but all of the Grubenhäuser with four or more sherds of later 4th-century pottery are situated in the southern part of the site (where Hamerow 1993 argued the earliest Anglo-Saxon phases of settlement were to be found).

Such structures include GH57, with its distinctive 5th-century pottery assemblage of mainly sandy fabrics and carinated vessels with a high proportion of Schlikung and burnished vessels, later 4th- to 5th-century belt attachment, Romano-British shale bracelet fragments and a scattered Romano-British coin hoard dating to the later 4th century (22 coins, of which the latest was AD 367–75); there was also an elongated triangular comb of probably 5th-century date. The later Roman pottery also recovered from this feature comprised five sherds of À l'Éponge, nine sherds of Portchester D Ware, six sherds of Red Oxfordshire Colour coat, three of Nene Valley colour coats and one of Much Hadham ware. Hamerow actually assigned GH57 to the 6th century, on the basis of the solitary sherd of a probably Frankish vessel (possibly a bottle); this was a large fresh sherd with a raised rouletted cordon, for which no parallels from England were known (Hamerow 1993, 22). Hamerow suggested that this could be related to the mid- to late 6th-century products of the Artois region, but acknowledged that exact parallels were hard to find, and that the closest parallel (which was still stylistically different) was the biconical vessel from Gr212 at Krefeld-Gellep, which Pirling (1966, taf. 19, 133) dates to the 6th century. Setting aside this rather tenuous parallel, GH57 would sit more comfortably in a very early 5th-century setting, alongside some of the earliest inhumation graves at Mucking (see below).

Another possibly very early Grubenhäuser is the nearby GH55, which contained a late Roman penannular brooch and steelyard in its fill, along with fifteen sherds of Portchester D Ware, five of Red Oxfordshire Colour coat and one of Nene Valley ware; its Anglo-Saxon pottery comprised carinated and simple grooved wares suggestive of an early date. Neither of these structures cut through Romano-British features.

There would thus seem to be fairly strong evidence to suggest that part of the later Roman pottery assemblage at Mucking was in use contemporary with the earliest Anglo-Saxon structures and pottery at the site, presumably in the early 5th century. The distribution of these sherds is largely confined to the fills of Anglo-Saxon Grubenhäuser in the southern part of the site, and upper ditch fills in the same area, particularly those of RBI, RBII, the 'F' ditches and the rectilinear ditch D15.

A strikingly similar pottery assemblage has been reported from Ship Lane, Aveley (Foreman and Maynard 2002, 144–6). This comprised small amounts of late shell-tempered ware, Portchester D, Oxfordshire Red Colour-coated ware and Alice Holt greyware, but with no obviously locally made vessels; there were no Nene Valley or Much Hadham wares at this site. Foreman and Maynard argue that this assemblage may represent an early 5th-century one, with the fabrics represented being those whose production is argued to carry on well past AD 400. A similar argument has been put forward for a variety of very late 4th- to 5th-century assemblages in an unpublished paper by T.S. Martin, which suggests that these regional imports continue being produced and used several decades after the demise of any local greyware industries. This would accord well with the lack of greyware in the 'non-Roman' Grubenhäuser at Mucking.