The Boudican destruction horizon identified in Colchester, London and Verulamium is very helpful in giving us insight into the nature of the built environment of those towns and cities before a.d. 60 and into their slow recovery, especially of Colchester and Verulamium, afterwards.Footnote ¹ Pre-Flavian urban development has been identified in other southern towns, too, notably Chichester (and Fishbourne) and Silchester,Footnote ² and Ernest Black sees the presence of ‘thin-walled box-tiles’ at the port site at Fingringhoe, Essex, and the ‘proto-palace’ at Fishbourne, as well as in Colchester and Canterbury as evidence of pre-Boudican bath building.Footnote ³ This is tantalising, but is it enough of a context to explain Dio Cassius’ comments that the causes of the Boudican revolt included the confiscation of money lent by Claudius to leading Britons which had to be repaid, coupled with the calling-in of a loan equivalent to 40 million sesterces by Seneca.Footnote ⁴ Who was borrowing and for what? Did it include buildings identified in the three towns that suffered at the hands of Boudica as well as in other towns where there is evidence of pre-Flavian building, but where it is difficult to distinguish between pre- and post-a.d. 60 contexts? To what extent was Roman-style building, especially of bath-houses, only confined to towns before a.d. 60? Of the several hundred legionaries and auxiliaries retiring each year with their savings and retirement gratuities, how many chose to settle in the province on retirement? How many in the towns and how many in the countryside? How did private projects relate to public building works? The purpose of this paper is to review the question of how far Roman-style building, particularly in masonry, had taken off across Britain, outside of a military context, whether before a.d. 60 or before the Civil War of a.d. 68–69 and the start of the Flavian dynasty.

In regard to public building works we should pause to consider how the road network was being developed and how precursors of the cursus publicus were supported before Boudica or during the Claudio-Neronian period as a whole. If we have a terminus post quem of a.d. 47/8 for the construction of the road leading west out of LondonFootnote ⁵ that subsequently divides to become the road to the north-west (Watling Street) and the road to Silchester and thence to south Wales and the south-west, it is likely that work on the other strategic roads leading in and out of London was equally well advanced to be capable of delivering supplies to the frontier garrisons. By the mid-/late 50s these were controlled by the legions based at Exeter, Usk, Wroxeter and Lincoln. Such a network could also support the movement of manpower between garrisons, including in and out of the province. As a working hypothesis we might envisage the building of road stations, mansiones and mutationes as an integral part of the support for those using the road network. An inscription dated a.d. 61/2 from near Philippopolis in Thrace,Footnote ⁶ a province from a.d. 46 and, like Britain, lacking roads and much urban infrastructure, records Nero's instruction to his procuratorial governor to build tabernas et praetoria along the military roads. Black interprets the praetorium as, in effect, a mansio, the tabernae as ‘shops’ and/or ‘inns’ sitting alongside the mansio in the same location.Footnote ⁷ Returning to Britain, such projects might not all have been completed by a.d. 60 or by a.d. 68/9, but it is hard not to imagine that several were, with others under construction, distributed along the major lines of communication. Equally, it is not difficult to suppose that much of the evidence for these hypothesised early projects has been destroyed by subsequent rebuilding. For example, Black makes a powerful case for Hadrian initiating a major reform of the cursus publicus associated with a programme of building mansiones across Britain.Footnote ⁸

Dating new buildings outside of a military context in pre-Flavian Britain is difficult and, as Black has shown with his examples of potential pre-Boudican bath buildings, considerable reliance has to be placed on proxy evidence. While destruction horizons that can be attributed with confidence to the Boudican revolt can be helpful, as in the cases of Colchester, London and Verulamium, in other urban situations the dating of the earliest buildings whose construction trenches typically cut virgin soil is difficult. Even if they happen to cut earlier ditches or rubbish pits, the character of the associated finds, which will be dominated by coarse pottery, is such as to give a pre-Flavian terminus post quem, inevitably pushing the likely date of build into the later first century. Only more closely dateable samian or a coin, unfortunately more often than not likely to be a Claudian copy with a broad date range, will allow the possibility of further refinement within the period c. a.d. 43 to c. a.d. 70. In a rural setting where dateable finds other than pottery are scarce, the continuity of Late Iron Age pottery forms and fabrics up to the late first century a.d. gives an even broader potential date range. Historically, all of this has inclined us towards seeing c. a.d. 70/the late first century as the period when Roman-style building took off in town and country in southern Britain. This of course chimes well with Tacitus’ often cited remark that Agricola, governor of the province in a.d. 78–84, ‘gave private encouragement and official assistance to the building of temples, fora and private mansions’.Footnote ⁹

Some examples of the difficulties of dating to and within the pre-Flavian period may be helpful. Two case studies, Fishbourne and Angmering from the Sussex group of early villas, illustrate this.Footnote ¹⁰ The Period 1C ‘proto-palace’ at Fishbourne is one of two masonry buildings that definitely pre-date the Flavian palace.Footnote ¹¹ Other than that it was sealed beneath the Flavian palace, there is no independent evidence with which to date the second masonry building. Likewise, it is its stratigraphical position beneath the Flavian palace that firmly puts the first masonry building, the ‘proto-palace’, into the pre-Flavian period. However, a more precise date of between a.d. 65 and 70 for the latter is proposed on the basis of the proxy evidence of the date of the builder's working yard a little way distant to the north.Footnote ¹² This contained waste and finished stone from a variety of sources, including imported marbles, which were dated by a few sherds supporting a late Neronian to early Flavian date. Because the deposit was sealed beneath make-up for the Period 2 palace, it is attributed to the first Period 1 masonry building, though it could quite possibly relate to an initial stage of building of the Period 2 palace, for example the West Wing, which was built first.Footnote ¹³ Even if the material from the builder's yard is to be associated with the first or any other of the three pre-Flavian masonry buildings, including Building 3 from the 1995–99 excavations,Footnote ¹⁴ the lack of dating evidence directly associated with either the first or the second masonry building does not preclude an earlier start date for one or both of them which could stretch back to the a.d. 50s. Turning to the dating of the Angmering bath-house, which is cited as a parallel for the bath-house of the ‘proto-palace’ at Fishbourne, we find that the only stratified material noted by the excavator consists of two sherds of samian, ‘one of Claudio-Neronian date and the other the base of an Early Flavian Form 18’ (not illustrated), found in the brick earth make-up for the floor of Room B.Footnote ¹⁵ The Form 18 is better regarded as pre- or early Flavian, rather than simply early Flavian. Notwithstanding further samian and coarseware of pre-Flavian or early Flavian date from the excavation, the excavator assigned an overall date of c. a.d. 70–160 for the occupational history of the site from pottery that was found in levelling material post-dating the abandonment and demolition.Footnote ¹⁶ On the basis of the stratified pottery, a pre-Flavian start date of the occupation cannot be excluded. Further, the ‘Ditches’ villa near Cirencester (Gloucestershire) provides a more recently excavated example; its first phase is dated to ‘around the third quarter of the first century AD’.Footnote ¹⁷ Other than a single, poorly dated brooch, which might post-date c. a.d. 70, the pottery evidence gives a terminus post quem of a.d. 50–70. While there is clearly a context for a wider re-examination of the dating of first-century Roman buildings in the countryside, these three examples at least show that the evidence can support pre-Flavian start dates or, in the case of the Fishbourne Period 1C buildings, the possibility of start dates earlier than Neronian to early Flavian.

Let us now turn to some of the materials which pre-Flavian bath-houses and other Roman-style buildings used, or could, potentially, have used in their construction. Two new and interlinked projects are shedding important light on the manufacture and consumption of ceramic building materials (CBM): first, research employing petrography and portable XRF has been carried out to characterise and, where possible, provenance the CBM from both antiquarian and modern excavations that were used in building the Roman town at Silchester;Footnote ¹⁸ second, excavation of the tilery, one and a half miles (c. 2.4 km) south of the Roman town at Little London, Pamber (Hampshire), where Lt. Col. Karslake reported the find of a Nero-stamped tile in 1925, has provided important insight into the date and range of its products and their consumption in Silchester and across southern Britain.Footnote ¹⁹

Considering first the research on the fabrics of the brick and tile used in Silchester, this has identified distinctive types which could be associated not only with likely local sources of clay, such as the Windlesham Beds and the more extensive London Clay Formation, but also with more distant production centres, notably Minety in Wiltshire and Eccles in Kent. All the known examples of Nero-stamped tiles could be linked by their distinctive fabric to waste material recovered from the plough soil at the presumed kiln site located by Karslake at Little London, Pamber.Footnote ²⁰ Following geophysical survey,Footnote ²¹ which indicated two potential major hotspots, the excavation of one of these at Little London in 2017 not only confirmed the existence of one large tile kiln, but also the inclusion among the waste material from it of two more Nero-stamped tiles. Among a wide range of types of brick and tile, including those necessary for building bath-houses, the excavation produced numerous examples of the distinctive, relief-patterned flue-tiles of die types 38, 39, 54 and 68, all variants of the ‘diamond and lattice’ design (fig. 1).Footnote ²² Relief-patterned tiles of these die types had previously been recorded from a wide range of sites across Berkshire, Gloucestershire, Hampshire, West Sussex and Wiltshire, and examination of the fabrics of those pieces that could be located has confirmed a Little London origin.Footnote ²³ Their distribution extends to the north-west to Wanborough (Wiltshire) and Cirencester (Gloucestershire) and south to Chichester (West Sussex) and Winchester (Hampshire) with an outlier at Rockbourne Roman villa in the south-west of Hampshire (see fig. 2). There is also a possible example of die 38, which could not be located, recorded from Canterbury (Kent).Footnote ²⁴

FIG. 1. Examples of relief-patterned tile (dies 38, 39, 54 and 68) from the 2017 excavation at Little London, Pamber, Hampshire.

FIG. 2. Distribution of relief-patterned tile from Little London, Hampshire, and Minety, Wiltshire.

Production of brick and tile at Little London took place alongside that of coarse pottery in a range of forms typical of the Claudio-Neronian period, their wasters associated with three small pottery kilns located in a second trench a few metres to the east of the excavated tile kiln.Footnote ²⁵ Although we cannot establish a certain start (indeed, did production begin before Nero's reign?) or end date, production of both pottery and building materials does not appear to have continued beyond the early Flavian period, when brick and tile in a distinctive ‘London Clay Formation’ fabric came to dominate the Silchester assemblage.Footnote ²⁶ Also, given that pottery made at Little London makes only a slight impression (<1 per cent) as a proportion of the Claudio-Neronian pottery assemblage from nearby Silchester,Footnote ²⁷ production at the site appears to have been brief, arguably confined to the reign of Nero. This is supported by the compact range of the radiocarbon dates from the excavation.Footnote ²⁸ Given that relief-patterned daub has been found in Boudican destruction deposits in Verulamium,Footnote ²⁹ it should not be a surprise that flue-tiles marked in this way can now be dated to the reign of Nero and, potentially, from before a.d. 60.Footnote ³⁰

Turning to the consumption of ceramic building materials within the nearby Roman town of Silchester in the Claudio-Neronian and early Flavian periods, a small collection retained from the excavation of Insula IX (Period 1) on the basis of distinctive typological or decorative features and the presence of animal footprints, rather than on a representative sample of the fabrics, has been shown to include six fabrics, of which three, probably four, are likely to be of local origin; two are from more remote sources.Footnote ³¹ One of the latter can be matched with examples excavated at Oaksey Park, Minety, and the other is probably attributable to a source exploiting clay-with-flints, but not matched to any known kiln. While 16 per cent by count are attributable to the London Clay Formation, at least half and possibly as much as 68 per cent of the retained sample can be matched with the fabrics represented at Little London. Of the non-local sources, 11 per cent can be attributed to MinetyFootnote ³² and 5 per cent to the source using clay-with-flints. Although the Nero-stamped tiles from Little London give a chronological terminus ante quem for this production, we cannot be certain when that at the other two workshops began and ended, but there is clearly some overlap with Little London. One important conclusion of the study of the Silchester material is that pre- and/or early Flavian building projects needed to draw on non-local sources, one of which – Minety – was at least 90 km distant from Silchester by road. Although a c. a.d. 100 date has previously been suggested for the start of production associated with the TPF-stamped tile at Minety,Footnote ³³ the unpublished report of excavations there by A.J. Scammell in the 1970s shows examples of associated pottery that are typical of the Claudio-Neronian period, and which compare well with the forms produced at Little London.Footnote ³⁴ Perhaps at Minety, too, coarse pottery was produced alongside ceramic building materials in the pre-Flavian period. There is also a record of three ‘diamond-and-lattice’ type, relief-patterned flue-tiles of die 56 (i.e. one similar to those found at Little London) from this site.Footnote ³⁵

In the same way that the relative proximity of Little London to Silchester suggests the latter was the principal consumer of its products, so the location of Minety, some 10 km to the south, suggests Cirencester was its main market. However, just as Little London lies close to the roads linking Silchester with Chichester and Winchester, Minety lies between the Fosse Way linking Cirencester with Bath and Ermine Street connecting Cirencester with Wanborough and Silchester; thus, the two kiln locations suggest that supplying the nearby towns was only part of their ambition. In addition to a supply that reached as far as Silchester, Minety relief-patterned flue-tiles of ‘diamond-and-lattice’ die 56 are attested in Cirencester and Bath, where the Minety fabric has also been identified by Sara Machin in situ in the voussoirs used in the vaulting of the baths and at Wanborough on the road towards Silchester (fig. 2).Footnote ³⁶ On architectural and art-historical grounds, a Neronian to early Flavian date for the initial construction of the baths and temple complex at Bath seems not to be in doubt, and it can now be linked, though not necessarily exclusively, to Minety.Footnote ³⁷

With contemporaneity of production at Little London and Minety, we can see the basis for envisaging several building projects across central-southern Britain that were initiated, if not completed, in the Claudian-Neronian period, if not specifically in Nero's reign. It is likely that the building of bath-houses was, as at Silchester, a priority, but in most places the remains of whatever was built at this time await discovery. However, we might venture to predict that, as they have been at Bath (but on a grand scale), the remains of pre-Flavian bath-houses and/or buildings with rooms heated with hypocausts will eventually be found in other southern towns such as Cirencester, Chichester, Winchester and Canterbury. We should also note the presence of both Little London (dies 39, 54 and 68) and Minety (die 56) patterned flue-tiles at Wanborough on Ermine Street,Footnote ³⁸ where aerial photography has revealed the remains of an, as yet, unexcavated (and probably later) mansio.Footnote ³⁹ At present we can link only Little London and Minety products that were actually built into two pre-Flavian structures, the baths at Bath and at Silchester, both substantial ‘infrastructure’ projects, but with none of the surviving examples of ‘diamond-and-lattice’ patterned flue-tile from those tileries attached to any particular building. To date >70 such pieces, the great majority comprising large and freshly broken fragments from a partly excavated late first-century dump, have been recovered from the >20 tonnes of CBM recovered from the 2018–20 seasons of the ongoing re-excavation of the public baths at Silchester.

Can we build further inferences from this evidence? As we have seen, the locations of both Little London and Minety suggest a degree of compromise between serving the needs of a project or projects in the towns to which they were closest and those of a wider network which could have included the building of mansiones and mutationes (or praetoria et tabernas), such as at Wanborough, along the nearby strategic roads between the larger settlements. While earlier research has highlighted Neronian building activity in general in towns, it is clear from the distribution of relief-patterned tiles from Little London that this also extended in some form to the countryside to what became villas, as at Littlecote (Wiltshire) and Rockbourne (Hampshire),Footnote ⁴⁰ and this is also true of Minety die 56, which has also been found at what later became villas at Lechlade (Gloucestershire), Shakenoak (Oxfordshire) and Truckle Hill, North Wraxall (Wiltshire) (fig. 2).Footnote ⁴¹

The coincidence of the manufacture of brick and tile alongside pottery at both Little London and Minety has certainly confirmed a Neronian date for the operation of these industries, but the very nature of the evidence does not exclude the possibility of a late Claudian commencement – the Nero tiles providing a terminus ante quem – nor, in the case of Minety, of later production, as of TPF-stamped tile in the second century.Footnote ⁴² We should also reflect that, without the close dating derived from the Nero tiles made at Little London and the presence of Minety brick in pre- and early Flavian Silchester, we could not make this case for Neronian/early Flavian public and private building projects across central-southern Britain. But is there also the possibility of more widespread pre-Flavian construction requiring ceramic building materials across southern and south-eastern Britain?

Since the great majority of excavated brick and tile is found loose, divorced from its original structural context, dating it is notoriously difficult, as, for example, is the case with relief-patterned tile in general. Even when found built into a structure, it is difficult to be certain of a primary rather than a secondary use, though the earlier the date of the structure, the more likely it is to be an example of primary use. As we will see below, the residual nature of CBM assemblages cannot be overemphasised. While none of the examples of relief-patterned flue-tile in the Little London fabric has been found built into a structure, the explicit Neronian association has provided an unequivocal terminus ante quem for this style of decoration, thereby raising the possibility that other relief-patterned tile production, particularly that of the ‘diamond-and-lattice’ design as found at Little London and Minety, is pre-Flavian in origin elsewhere in the province. Indeed, could all the workshops producing flue-tile marked with this design be associated with a single owner as part of a co-ordinated project to provide building materials for public and private projects across southern and south-eastern Britain? If building projects were to draw on more than one source for their ceramic building materials, as we have seen in the case of those using Little London and Minety products, consistency in the types produced and their dimensions would have been necessary. Tegulae, for example, would need to have been made to the same design and dimensions if the products of different workshops were to be used on the same roof.Footnote ⁴³ Thus the ability of projects such as the building of mansio bath-houses to draw on resources from different workshops implies some overarching private or public authority.

If we plot the distribution of all the finds of the commonest dies, those of ‘lattice-and-diamond’ type,Footnote ⁴⁴ we see it extends across southern Britain as far as Bath in the west, with finds in Hampshire, Oxfordshire and Wiltshire, but with the greatest numbers recorded from London and the south-eastern counties of Essex, Hertfordshire, Kent, Surrey and Sussex (fig. 3). There is a marked fall-off north of the Roman road that runs east–west from Colchester through Verulamium and Alchester to Cirencester, though there is a scatter of examples to the north as far as Nottinghamshire and to the north-west along Watling Street as far as Wall. This distribution is almost identical to that of all relief-patterned die types, but with slightly greater numbers north of the Colchester to Cirencester road.Footnote ⁴⁵ It includes a mix of urban, rural and roadside settlement sites. However, if we plot all the remaining relief-patterned designs (‘chevron’, ‘billet’, etc)Footnote ⁴⁶ other than ‘diamond and lattice’, the distribution is comparatively restricted, notably excluding the great majority of find sites from Hampshire, Sussex, south Oxfordshire and Wiltshire, the areas in which Little London, Minety and the products of a kiln suspected to be located between Chichester and Arundel in West Sussex have been found (fig. 4).Footnote ⁴⁷ Compared with ‘diamond and lattice’, the distribution is mainly to rural sites, rather than to a mix of urban, rural and roadside settlement sites. Chronology or the marketing strategies of different tileries may provide possible explanations for the difference (see further below). Interestingly, there are no recorded relief-patterned tiles of any type from Norfolk and the territory of the Iceni.

FIG. 3. Distribution of finds of relief-patterned tile using dies of ‘diamond-and-lattice’ type.

FIG. 4. Distribution of relief-patterned tile using dies other than those of ‘diamond-and-lattice’ type.

Prior to the confirmation of a Claudian-Neronian or Neronian date for the Little London products, it had been suggested that the production of relief-patterned tile did not begin until the Flavian period, c. a.d. 75, then continuing to c. a.d. 155–75.Footnote ⁴⁸ Such dates can only be regarded as provisional since they are overwhelmingly based on finds divorced from their primary structural context and therefore residual, as is the case with the great majority of examples listed by Ian Betts and colleagues in their 1997 Corpus. Even most of the examples from the seven sites listed where tiles had been found in a dated context or in situ are probably or certainly residual.Footnote ⁴⁹ The earliest potential examples (of ‘diamond-and-lattice’ dies 19 and 21) are from the bath building at Angmering, West Sussex, which, as we have seen above, the excavator conservatively dates to the early Flavian period. Our re-examination suggests that an earlier, Neronian or Neronian to early Flavian date is more likely, an important consideration given that the relief-patterned tile was found in situ at Angmering. This is supported by the evidence from Fishbourne where relief-patterned tile was found in the occupation and destruction levels of the Period 1C ‘proto-palace’.Footnote ⁵⁰ Although the ongoing re-excavation of the public baths at Silchester has so far produced a quantity of Little London relief-patterned flue-tiles, their presumed original context in the Neronian baths was destroyed by the latter's demolition and its replacement with a larger bath-house in the late first/early second century. This used a different source of brick, with comb-marked flue-tile overwhelmingly dominating (77 per cent by weight, 85 per cent by count) the total flue-tile assemblage of over 8,600 pieces (2.17 tonnes) from the site. None of the flue-tile of any period has been found built in its original position. After demolition, the Neronian relief-patterned tile was presumably either recycled into the later rebuild and its subsequent modifications into the fourth century or was made available for reuse elsewhere in the town to be used as hard core or incorporated in new build. Such a process of recycling through the generations may well account for the presence of relief-patterned flue-tile in buildings thought to have been constructed in the second, third or even the fourth century.

The Little London assemblage makes it clear that the manufacture of relief-patterned flue-tiles did not take place in isolation, and was only one part of a production that embraced a ‘complete’ repertoire, including bricks, roof-tile, floor-tile, flue-tile, with ‘thin’ scored tile as attested in pre-Boudican Colchester,Footnote ⁵¹ voussoirFootnote ⁵² and other specialist products required for building a bath-house of the period. Waste material was dominated by roofing material and bricks, with hypocaust tile, both plain and keyed, accounting for only 3.7 per cent by count of the assemblage. We should not assume therefore that it was only relief-patterned flue-tiles, themselves representing less than half of the keyed material (but typically all that survives of this production in museum archives), that were transported long distance, nor that such tiles were the products of specialised workshops. Similar conclusions regarding the production of such tile alongside other types and the potential for tileries to have supplied quite large areas were reached by Andrew Middleton and colleagues in their study of relief-patterned flue-tiles, employing petrography and neutron activation analysis.Footnote ⁵³ We should also note the widespread distribution of Betts et al.'s Fabric Group 1 (= Black's ‘London-Sussex’ tile group) which extends from Sussex to London and beyond.Footnote ⁵⁴ This grouping is based on inspection using a hand lens or a low-powered binocular microscope, but it is consistent with the extensive distribution demonstrated for the Little London products. Echoing the distribution of Little London and Minety fabrics, the influence of the road (Stane Street) in the distribution of the London-Sussex group is clear. Similar influences can be seen with other early, but not yet located tileries that supplied London, for example the proposed Eccles fabric group, with a distribution into London across north Kent along Watling Street and out of London to Colchester, and the MOLA fabric 3069, found along Watling Street between London and Verulamium.Footnote ⁵⁵ We should also note the proximity to Stane Street of the known production site of relief-patterned flue-tile at Ashtead Common, Surrey.Footnote ⁵⁶

The association of relief-patterned flue-tile with Claudian-Neronian or Nero-dated production at Little London has encouraged us to consider first the ‘diamond-and-lattice’ group and then the slightly larger entity of all relief-patterned tile. At Little London the life of the tilery may not have extended much, if at all, beyond Nero's reign; this is a short phase of production which is brought out by the comparative rarity of both its pottery and, indeed, its relief-patterned tile at Silchester. The brick and tile so far identified as having been used in the late first-/early second-century bath-house and dominant in the Flavian Forum-Basilica at Silchester came from a different local (London Clay Formation) source, which has also been identified in small quantities from earlier, pre- and early Flavian contexts from the Forum-Basilica and Insula IX.Footnote ⁵⁷

The situation at Little London and Silchester raises the question of how long production continued at other tileries making relief-patterned brick and tile. In this regard, we note that the overall distribution is largely localised to central-southern and south-eastern Britain, with all but two find sites (Kenchester, Wall) south and east of the Fosse Way. By far the largest number of finds comes from London, where Betts et al.'s 1997 Corpus lists examples from 103 sites, the majority located within the walled area, with c. 80 per cent producing no more than five such tiles.Footnote ⁵⁸ Only six sites are listed with more than 20 examples from within the walled area, with the Winchester Palace site in Southwark, south of the river, producing by far the greatest number (178). An absence of finds is noted from the site of the Cripplegate Fort dating from c. a.d. 120. Outside of London, only two major towns, Canterbury and Verulamium, are listed with as many as 20 dies, with Chichester, Cirencester, Colchester, Silchester and Winchester represented by between four and 11 die-types.Footnote ⁵⁹ While the majority (76 per cent) of the 165 find sites (other than London) are rural, with only one or two dies listed, there are a few, mainly bath-houses and villa sites, with numbers comparable with those from the major towns, for example Lullingstone, Kent (ten), Fishbourne, West Sussex (nine), Chatley Farm, Cobham, Surrey (nine). The same is true of sites with certain or probable praetoria or mansiones, such as Chelmsford, Essex (eight) and Lower Wanborough, Wiltshire (eight).Footnote ⁶⁰

Relief-patterned flue-tile has so far not been found in Dorchester (Dorset) (presumed civitas capital from c. a.d. 80), Exeter (presumed civitas capital from c. a.d. 80), Gloucester (colonia from a.d. 96–98) and Wroxeter (presumed civitas capital from the a.d. 90s), all of which have seen substantial excavations between A.W.G. Lowther's 1948 publicationFootnote ⁶¹ and the Corpus published in 1997. At Lincoln (colonia late in Domitian's reign, a.d. 84–96), also a city that has seen extensive modern excavations, only two pieces are recorded, both from post-Roman contexts (dies 27 and 102),Footnote ⁶² while at Leicester, the next most northerly town, there is a more substantial assemblage from modern excavations. Lowther published a total of 11 unstratified pieces from the Antonine Jewry Wall (Baths) site (dies 7, 9, 13, 30 and 67)Footnote ⁶³ (‘in place of the more usual [our italics] ‘combed’ type’),Footnote ⁶⁴ while further examples of dies 30 and 67 were found elsewhere in the town, including 14 small fragments of die 30 from Little Lane. A stratified example of die 67 came from a late second-/early third-century context.Footnote ⁶⁵ The range of dies and the numbers of examples compare favourably with some of the southern towns (other than London) listed above.

Nevertheless, the overall quantities are small, a characteristic that is also found with consumption sites in the south. For example, if we take finds from bath-houses, a sure destination for flue-tile, we find that at the Huggin Hill Baths in London, despite references to quantities of broken flue-tile (and other CBM) from the destruction phase, of the three published pieces of CBM from the 1960s excavation, two are of relief-patterned flue-tile (dies 42 and 85).Footnote ⁶⁶ The 1989 excavations brought the total number of pieces to 43, still with none found in situ, and a further two dies (5A and 27).Footnote ⁶⁷ Similarly, at the Cheapside Baths, the only CBM published by Peter Marsden consists of just three fragments of relief-patterned flue-tile (dies 18, 44 and 63).Footnote ⁶⁸ Two other fragments (dies 40 and 85) are also recorded from this site.Footnote ⁶⁹ It is reasonable to conclude that at both sites what was published represents all the pieces of relief-patterned tile recovered from the respective excavations and retained for study.Footnote ⁷⁰ Both baths underwent significant alteration after their initial construction, for which there was either no (Cheapside)Footnote ⁷¹ or insufficient evidence (Huggin Hill) of date,Footnote ⁷² before both were finally abandoned in the second half of the second century (Huggin Hill)Footnote ⁷³ or the late second/early third century (Cheapside).Footnote ⁷⁴ Although precise numbers have not been published, the excavation of the public baths in Tower Street, Chichester, also produced only a small number of pieces (>9) in four dies (4, 48, 55 and 81),Footnote ⁷⁵ including pieces which can now be attributed to the Little London kiln.Footnote ⁷⁶ Two pieces of Little London die 81 are from a context, probably a construction trench and therefore pre-dating the baths, where the latest pottery is of Flavian date, providing a terminus post quem for their initial construction.Footnote ⁷⁷

We can see repeatedly that the lack of even a basic quantification of CBM assemblages from published excavations is a major impediment to understanding the significance of what is reported. One of the larger collections of relief-patterned flue-tile from a single site outside of London is from the mansio and adjacent sites in Chelmsford. Although all the 114 fragments evidencing the use of seven dies (as well as a further 12 pieces from adjacent sites), mostly small and quite abraded,Footnote ⁷⁸ are carefully tabulated, it is clear from the report on the tile assemblage as a whole that, given the lack of numbers for any other category of CBM, only a small proportion of the original excavated material, but probably including all the relief-patterned flue-tile, was retained for archiving and publication.Footnote ⁷⁹ We cannot, therefore, assess the significance of that material relative to the rest of the CBM assemblage, including to other types of flue-tile. However, it is important to point out that, of the six fragments from stratified contexts dating to the first half of the second century, all are residual in secondary contexts; the earliest is from a beam-slot of the timber phase of the mansio and dated c. a.d. 120–25.Footnote ⁸⁰ The timber phase of the mansio thus gives us a terminus ante quem of c. a.d. 120 for the use of relief-patterned flue-tile in Chelmsford.

One site where there has been systematic quantification of the CBM is the tilery at Great Cansiron Farm, Hartfield, East Sussex, dated by the associated pottery to the late first/early second century with an archaeomagnetic date for the last firing of a.d. 100–30.Footnote ⁸¹ Here, a ‘mere’ 6.52 kg of relief-patterned flue-tile was recovered, representing 3.9 per cent of the 165.75 kg of three types of box-flue tile recorded, of which the combed material is by far the most dominant type (93 per cent).Footnote ⁸² This can be compared with the finds at Little London where, out of a sample of over 90 kg/655 pieces of keyed (combed, relief-patterned and scored) and plain flue-tile and parietalis, relief-patterned flue-tile accounts for 20 per cent of the assemblage by weight and count, and combed tile only 2 per cent.Footnote ⁸³ The dominance of combing recalls the situation at Flavian Silchester (see above), but the representation of relief-patterned tile by any measure at Great Cansiron Farm is so slight that it is doubtful whether it was a product of the kiln as it operated in the late first/early second century. Rather, it may be regarded as residual waste from an earlier phase of activity or from an earlier and, as yet, undiscovered kiln or debris from the demolition of an earlier structure. The situation is similar to that at Harrold, Bedfordshire, where late second-century production of relief-decorated tile has been claimed. The evidence is based on only five abraded, unstratified fragments, four of which come from the same tile.Footnote ⁸⁴ These pieces are clearly residual, as their number and condition indicate.

To conclude: the Neronian terminus ante quem at Little London, supported by the evidence from Angmering and Fishbourne, offers an earlier starting date for the production of relief-patterned material, associated with ‘thin’ scored flue-tile and parietalis. Of the few types of CBM that appear to have been selected consistently from site collections for deposition in museum archives, especially since Lowther's 1948 publication, the relief-patterned material is an almost unique resource for providing insight into early Roman building activity in southern Britain.Footnote ⁸⁵ Despite the incidence of finds in London and across southern Britain, the absolute quantities of relief-patterned tile remain small, and the combination of this and the earlier start date indicated by Little London lead us to question whether this tradition of keying flue-tile and parietalis continued until the mid-second century a.d., as has been proposed previously. Not only do the overall small numbers of finds argue against a long period of production, but the absence from towns in the south-west and West Midlands adjacent to the heartland of the distribution and where major building is either to be expected or can be demonstrated, as, for example, with the forum-basilica at Exeter,Footnote ⁸⁶ from the mid-Flavian period, also suggests an earlier termination. This is supported at Silchester, where comb-patterned material dominates the flue-tile assemblage, and, though mostly unstratified, it can be associated reasonably with the rebuilding of the public baths in the late first/early second century and subsequent modifications. In London, Betts and colleagues note the absence of relief-patterned tile from the site of the Cripplegate Fort, whose construction is dated to the Hadrianic period.Footnote ⁸⁷ We suggest that this negative evidence outweighs that for continuity to the mid- to late second century, where occurrences can be regarded as residual or, if in situ, evidence of recycling. Fishbourne once again provides a good example: as we have seen above, relief-patterned tile, which includes Little London material, is first present in the Period 1C Baths, but it is absent from the Third Period North Wing Baths, dating from the beginning of the second century. However, some reappears recycled later, presumably via reuse in the adjacent Period 2 Baths, into the Third Period East Baths.Footnote ⁸⁸ The lack, in our view, of compelling evidence for the continuity of the tradition beyond the late 60s or early 70s implies an abrupt cessation of consumption (and production) rather than a slow decline.Footnote ⁸⁹ Since we now realise that these tiles are the only survivors of a broad-based production and distribution of CBM, the implication is that the break in production in the manufacture of relief-patterned tile equates with that of all the other forms of CBM that were made at the relevant tileries. There are several possibilities to explain a widespread disruption in production: these include the withdrawal of loans prior to the Boudican rebellion as claimed by Dio, and discussed above, and the consequent financial insolvency of those funding the tileries, as well as major upheavals, such as the Boudican rebellion or the Civil War of a.d. 68–69, which themselves could well have triggered the withdrawal of credit. Insolvency could of course occur at any time and need not necessarily be linked to the Boudican revolt. Indeed, and paradoxically, all the evidence for this early development, where debt could have been incurred, is in the wider area affected by the Boudican rebellion, not in the territory of the Iceni. Whatever the reason for the tileries that made brick, including the relief-patterned flue-tile, stopping production, it is likely that the interval before tile-making began again was such that the memory of the old forms of marking tile had been lost, giving way to combing, already coming into fashion at Little London.Footnote ⁹⁰

Earlier we drew attention to the tendency to date the earliest masonry buildings in the countryside from the a.d. 70s onwards, even when the evidence allows the possibility of earlier start dates from the 50s or 60s, which is consistent with the earlier start date proposed here for relief-patterned tile.Footnote ⁹¹ While it is hard to interpret the significance of the many sites with single finds of relief-patterned style – so much depends on the date and scale of excavation – we should certainly entertain the possibility that the 40 or so sites with multiple finds (>2) outside of London included either bath-houses or buildings with rooms heated by hypocausts. In terms of Roman building styles and the introduction of bathing to Britain, this gives us an indication of a more developed province (in Roman terms) south and east of the Fosse Way than has previously been thought in the pre-Flavian or, more specifically, the pre-Boudican period. Alongside public works to provide urban bath-houses and praetoria et tabernas, the predecessors of the later mansiones along the road network, there is considerable evidence for investment in private housing in town and country and especially in London. There are notable blank areas in the overall distribution, particularly in East Anglia, corresponding with the territory of the Iceni, a client kingdom up to a.d. 60, and parts of the south-west east of the line of the Fosse Way (east Dorset, east Devon, including the legionary fortress at Exeter, and Somerset), still under military control until c. a.d. 70.Footnote ⁹² Alternatively, these absences may be explained by the practice of different traditions in the making of flue-tile, though they are also areas without evidence of first-century villas. We note above that, compared with the ‘diamond-and-lattice’ style, other patterns are extremely rare in the southern counties of Hampshire and West and East Sussex, those generally considered to have fallen within the client kingdom of Cogi-/Togidubnus. Perhaps the explanation for this is chronological, with the appearance of construction employing predominantly ‘diamond-and-lattice’ roller-stamped flue-tile slightly later and, in the south, linked to imperial support for the client kingdom as recognition of its loyalty during the Boudican rebellion (and in sharp contrast with the stark absence of relief-patterned flue-tile from the territory of the Icenian kingdom). Where the Nero-stamped tile, found so far (apart from Little London) only in Silchester, is concerned, the building may relate to those ‘civitates’ of the client kingdom ceded to that emperor. Perhaps the distribution of the Little London products correlates to some degree with those civitates?Footnote ⁹³

If chronology is indeed the explanation for the differential distributions, one implication is that Roman-style construction, including of bath-houses, was underway in London and the adjacent towns (Colchester and Verulamium) and countryside in the 50s, if not from the late 40s. This leads us to the question of who was paying for or borrowing to pay for this building in town and country? While members of British tribal elites may have been partly responsible, we might also consider whether veteran settlers could account for this phenomenon. At the beginning of this paper we questioned whether and where veterans with their disposable cash might have chosen to remain in Britain. While the veterans of legio XX might have settled near their colonia at Colchester, given that the next colonial foundations at Gloucester and Lincoln were not laid until the end of the first century, what happened to veterans from the other legions and units who settled in the province? Were they encouraged to acquire virgin land or existing farms from their British owners more widely across the province, but in the demilitarised zone south and east of the Fosse Way, which, with a very few exceptions, marks the limit of the distribution of relief-patterned tile?

Another important conclusion of this paper is that it can be securely demonstrated that ceramic building material travelled significant distances to the places where it was used in this early, pre-Flavian period. We are fortunate that relief-patterned tile has been retained in site archives because, in light of Little London, it can be used as a proxy for ceramic building material in general. Identification by petrographic analysis of products from Little London up to 100 km from the kilns strongly supports earlier research based on die evidence and inspection of fabrics with hand lens or binocular microscope that CBM of all types could travel considerable distances, mostly by road – a tribute to the capacity for the roads to take heavy traffic in the pre-Flavian period. By the same token, individual building projects were able to draw on more than one source for the brick and tile they required. If we are correct in our interpretation that relief-patterned tile-making was a phenomenon of the 50s and 60s, further work on the characterisation of fabrics will give valuable insight into the number of tileries operating and therefore of the scale of pre-Flavian civil construction in Britain.

The focus on relief-patterned flue-tile is a reflection of how it has been selected from site collections – no doubt influenced by, first, Lowther's study published in 1948 and, more recently, by Betts et al. Reference Betts, Black and Gower1997 – to be deposited in museum archives by virtue of its appearance and relative scarcity at the expense of the great majority of ceramic building material excavated from any one site, which has all too frequently been discarded without record. Historic collections of CBM retained in museum archives tend to include only stamped, decorated or otherwise unusually marked pieces, or tiles with animal-foot impressions, rather than representative samples of the range of forms and fabrics from each site. An important lesson for the future is that, in the knowledge that Roman building projects, both public and private, commonly drew on ceramic building material from more than one source and that such sources might be some distance from the point of consumption, we need to record the quantities of the CBM recovered from an excavation, not just by the various types represented, as is more commonly the case now, but by their fabrics as well. Representative samples of the different fabrics should be a required element of the deposited CBM archive of any site.Footnote ⁹⁴

In making a claim for a focus of early, i.e. pre-Flavian, construction projects in town and country across much of Britain south and east of the Fosse Way, this paper provides a context for determining whether similar kinds of development were undertaken subsequently more widely across the province, in the south-west, the midlands, the north and in Wales, which might include the coordinated provision of infrastructure to support the cursus publicus and movement more generally along the main arteries of the province. Fabric characterisation will also help us to determine how tilery markets changed over time.