Analysis of the pottery fabrics was undertaken by Helen Ashworth in the 1980s (Ashworth 2010, 261), developing and modifying the classification of Val Rigby (Stead and Rigby 1986, 260-7). Individual sherds were examined at 10× magnification to identify inclusions, which was the basis on which fabrics were initially distinguished. Dates of likely use at Baldock were then established based on other dating evidence, including coins and stratigraphic position. In this way, a small group of fabrics that were found only in the very late sequences could be assigned a 5th-century date. These were initially recognised in the upper fills of a large doline, which sealed contexts dated by coins of the House of Theodosius to the end of the 4th century (Fitzpatrick-Matthews 2010, 134). It was then possible to recognise the same fabric types in other 'late' features that were not elements within stratified sequences.
The fabrics isolated in this way resembled Iron Age types (including one type that is similar to a local Late Bronze/Early Iron Age fabric) and without the stratigraphic data, they would have been seen as residual in their contexts. Comparison with types known to be of prehistoric date, though, highlighted important differences, both technological and stylistic. In most cases, the vessels were handmade but finished on a turntable, although one type (perhaps the latest) lacked the wheel finish. One had a burnished variant. All were more highly fired than the prehistoric types. Where forms can be recognised, they are late Roman or, in several cases, early medieval ('Pagan Saxon') in character.
Two broad groups of these sub-Roman fabrics have been identified, both basically sandy (Figure 2). One consists of a dense ware (Ashworth Fabric 31), with a heavily quartz-tempered variant (Fabric 50), a similar ware with additional flint and chalk (Ashworth Fabric 56) and a mixed tempered ware containing quartz, grog, crushed shell and chalk (Ashworth Fabric 51). Fabric 31 has a fine, hard matrix, tempered with abundant small quartz grains (<0.15mm diameter) and a rough, sand-papery feel. The interior surface is orangey-pink, the core dark grey to black and the external surface dark grey to black. At least one sherd shows external decoration. Fabric 50 is heavily tempered with quartz grits (<1.0mm in diameter), producing a slightly pimply surface. The colour is dark grey to grey throughout. The interior surface is wiped; the one sherd identified has a worn external surface that may also have been wiped. Fabric 51 has a sandy micaceous matrix with sparse to common small to medium quartz, grog, crushed shell and chalk. The core has a narrow grey central band with brownish-red outer layers; the surfaces are dark grey to black and there are voids on the outside surface where shell has burnt out. Fabric 56 has a coarse sandy matrix, with common to abundant small quartz grits, sparse medium flint and sparse medium to large chalk. Some of the tempering material stands proud of the surface. It is a yellowish-grey throughout, handmade and of very poor workmanship.
A sandy fabric with vegetable temper, one containing additional quartz (Ashworth Fabric 52) and one containing grog and crushed shell (Fabric 54), forms the other group. Fabric 52 is the finest of these wares, with a sandy matrix tempered with common small quartz and very small sparse organic material. The interior is dark grey but the surfaces are oxidised to a uniform brownish-orange; one sherd is decorated with a burnished band. Fabric 54 has a micaceous fine-grained matrix with sparse mixed tempering of small grog, crushed shell and organic material. The interior is mid- to dark grey and the surfaces dark grey.
A third type occurs in only a few sherds (Ashworth Fabric 53), which is tempered with abundant quartz grains (0.1-0.2mm diameter) and organic material. It is poorly fired and corky in texture. The core and interior surface are grey, while the exterior surface is a dark reddish-brown; large parts of the surface have spalled away. It is identical with Bedford Fabric A1 (Baker and Hassall 1979, 152), where it is dated to the 7th century AD on the basis of vessel form. This is almost certainly too late for Baldock, where a 6th-century date may be proposed.
What is striking about this material is the apparent continuity of potting technology with earlier, mass-produced traditions: Fabric 31 resembles Verulamium region sandy ware (Ashworth Fabric 16) in its inclusions and colour range, while Fabric 54 resembles Much Hadham wares (Ashworth Fabrics 11, 12 and 49), although, unlike them, it contains shell, which is a characteristic of wares from Harrold and elsewhere (Ashworth Fabric 4). In most senses, apart from likely mode of production (below), this material is Roman pottery.
The recognition of the forms produced in these fabrics is difficult, as the material exists for the most part only in small sherds, generally from the bodies of vessels. One sherd of Fabric 52 shows a burnished vertical neck but too little remains to enable a confident reconstruction of the vessel; it may be from a necked jar in the same tradition as a very late Much Hadham Oxidised Ware jar published by Val Rigby (Stead and Rigby 1986, 378, no. 838).
However, there are complete (or near-complete) vessels in fabrics associated with the major late Roman industries found in Baldock (Shelly Wares (LRSH), Oxfordshire Colour Coated Wares (including OXRS) and Hadham Wares (including HARS) that are associated with these post-400 fabrics or occur in other stratigraphically late contexts (Figure 3). For instance, a 5th-century grave  in the California late Roman cemetery contained two vessels in Oxfordshire Red-slipped Ware (Ashworth Fabric CC8, OXRS) of apparently unparalleled form: one is a pedestalled handled cup, the other a small handled jug, which appear to form a matching pair. Another grave contained a miniature carinated jar in Hadham Oxidised Ware (Ashworth Fabric 49A) that copies an earlier common form, a third contained a black-slipped Hadham Ware miniature bowl (Ashworth Fabric 11C), a fourth a probably Hadham copy of an Oxfordshire Red-slipped beaker (Ashworth Fabric CC12, HARS). The remaining vessels are of forms that have parallels in late 4th-century contexts but appear in graves with the sub-Roman fabrics. They are frequently poorly made and have very worn surfaces.
As all these vessels appear to be the products of manufacturers that supplied pots to local markets before 400, do they constitute evidence for the continuation of the industries after that date or are they better explained as rare goods curated for some time after they had ceased to be manufactured? Several of the vessels are certainly technically less accomplished than 4th-century examples. The unusual forms may indicate that they are products of an industry that was working on a smaller scale than previously, without following standardised patterns and perhaps operated by less skilled potters.