Subscribe now for full access and no adverts
Tiles and bricks, collectively known as ‘ceramic building material’ (CBM for short), are ubiquitous finds on many Roman sites. Indeed, it is precisely because of this abundance that many 20th-century archaeologists took little interest in CBM. Excavators frequently discarded the bulk of the seemingly uninteresting fragments with only cursory examination at best. An exception to this rule, however, was tiles that had been impressed with stamped letters before they were fired. These have long been of interest to antiquarians and modern archaeologists alike, as they can provide illuminating clues about the industry that produced them. For instance, tiles found in Roman forts are sometimes stamped with the name of the army unit that made them, while civilian tile stamps give the names or codes used by certain municipal or private tileries.
Civilian stamps were never common in comparison to military ones. They were used on the products of municipal tileries in London and Gloucester, but virtually nowhere else; as for privately operated kilns (which served both urban and rural projects), fewer than 500 examples of their stamps are known from the whole of Britain. Two-thirds of all stamps issued by private tileries have been found in Gloucestershire and north Wiltshire, so here, uniquely in Roman Britain, we have a better chance than anywhere else of gaining insights into the organisation of the civilian tile industry during this period.
Around ten or so different letter groups can be distinguished in the Cotswold region, and sometimes a series of different dies (objects made of wood or metal in which lettering has been engraved so as to produce a stamp when the die is impressed in the wet clay of the unfired tile) produced the same basic letter stamp. Only a very small proportion of the output was stamped, perhaps just a few tiles per firing, and the two most common stamps that we see use the letters LHS and TPF (together with derivations of the latter: TPFA, TPFB, TPFC, and TPFP). TPF could stand for T(egularia) P(ublica) F(ecerunt) or T(egulariorum) P(ublicorum) F(iglinae), both of which can be translated as meaning ‘made by/product of the public tileworks’. Alternatively, TPF could just be the initials of the tilery’s owner, as is presumably the case with LHS.
The vast majority of LHS and TPF tiles (96 out of 150 known examples) have been found in Cirencester, but where were they made? For many years, the source was thought to be a site called Oaksey Park, near Minety in north Wiltshire, which lies just six miles south of Cirencester. This was largely hearsay, though, based on the supposed findings of an excavation of two kilns and associated features by amateur archaeologist Anthony Scammell in the early 1970s – and which contradicted what Scammell himself had actually written about the site. The results of the excavations were never published, but Scammell did lodge a report with Devizes museum in which he is quite clear that he did not believe the stamped tiles had been made in the kilns he had investigated. This has now been confirmed by the results of a community archaeology project led by Cotswold Archaeology at another Roman tile kiln in Minety at Brandier Farm, just over a mile from that at Oaksey Park.
The Brandier site had attracted little previous attention, but when photographer and ceramicist Peter Lavery and his family bought the farm in 1992, Peter noticed a distinct mound in one of his fields, and around it he found fragments of heavily fired Roman tile. Peter was keen to see the mound investigated in more detail, and in 2022 Cotswold Archaeology instigated a project to learn more about it. A preliminary geophysical survey showed evidence of intense burning within the mound and surrounding area, and this led to two seasons of excavation, with a third and final season this summer.
When our project began, our initial hopes were to establish whether there was indeed a Roman tilery at Brandier, as opposed to some other type of industrial activity such as a medieval pottery kiln (Minety was the centre of an important ceramics industry during that later period, so it was possible). And if it was a Roman tile kiln, how did it relate to the Oaksey Park site? Was it contemporary, indicating a more-dispersed industry in the Minety area than hitherto had been realised, or did Brandier replace its neighbour? We were also excited by the chance to investigate a tile kiln under modern conditions, and to apply scientific techniques to understand more about the fuels used to fire it and the environment in which it had stood.
A new Roman tile kiln
We began by targeting Trench 1 on the area of high magnetic response at the northern end of the mound. Within a couple of hours on the first day of the dig, it became apparent that we did indeed have a Roman tile kiln – and we were also greeted by our first stamped tile, which, appropriately, could be read as ‘HI’ (in reality, it was a fragment of a rather more conventional LHS stamp). The kiln itself (termed Kiln 1), was made from brick and tile; its central flue, stoke hole, and the southern part of the firing chamber were well preserved, while the northern part was much less so, although enough survived to show that the original chamber measured c.5.4m by 4.8m. The flue had been kept dry by a drain formed from curved tiles, and we could also see that Kiln 1 had been repaired on several occasions. The wall of the firing chamber had been rebuilt with clay-bonded bricks that survived up to eight courses high, and the central flue had also been relined and made narrower, perhaps because the original design had not functioned overly well. A line of post-holes just beyond the back of the stoke-hole either formed a standalone structure such as a drying shed, or supported a canopy roof to protect the kiln from the elements.
We did not recover any dating evidence associated with Kiln 1’s construction: the little pottery from the demolition layers is no earlier than the middle of the 2nd century AD – but there are strong suggestions that this was not the first kiln to operate on the site. Beneath the southern chamber wall, two other tile walls were found at a lower level with vitrified internal faces. They may have formed the flue of an earlier kiln (Kiln 2), and we will investigate this area further in this summer’s dig – so watch this space for further updates on that later in the year. We identified a large clay-extraction pit, too: 3m deep, close to kiln 1. Its lowest fills were waterlogged and contained preserved timber planks; above this, the pit was backfilled with kiln rake-out, waste tiles, and fragments of vitrified kiln wall.
Interpreting tiles
As might be expected, we recovered a lot of CBM – some 20,000 fragments, weighing 5.6 tonnes in all. What we did not anticipate was the number of stamped tiles that this would contain – but to-date we have recovered 136, an extraordinary number, almost doubling the number of known stamps, as the table on p.29 demonstrates.
There was a plethora of LHS stamps in the upper layers of debris associated with Kiln 1 and, with two exceptions, they all came from the area in front of the stoke-hole – it is possible that they represent the collapse of a tiled roof supported on the row of post-holes mentioned above. The rakings pit was particularly productive of stamped tiles, all of which (with the exception of a single LHS stamp found in the level immediately beneath the topsoil) are from the TPF series. TPF and TPFA stamps were present at all levels in the pit backfill, but TPFC was only present in the higher layers and TPFP only in the highest. From these results, we might interpret TPF and TPFA as the first stamps used in the kiln, followed later by TPFC, and then TPFP. TPFB must presumably have been produced at some point in this sequence as well, although it was not present in the rakings pit.
The effective absence of LHS stamps from the rakings pit strongly implies that there must be a second, as yet undiscovered, pit where debris from the production of these tiles had been discarded. Stamp distribution patterns also lend further strength to suggestions of an earlier Kiln 2, as TPF-series tiles had been built into the structure of Kiln 1. While it is possible that these had been made in a first phase of Kiln 1’s life, and subsequently rebuilt into a second phase structure, the drain formed from curved tiles (stamped TPFP) is probably associated with the initial use of Kiln 1, suggesting the stamps had been made in an earlier kiln – perhaps represented by the remains that emerged right at the end of the 2023 season.
We can draw several conclusions from the stamps. First, the suffixes A, B, C, and P on the TPF tiles evidently do not represent the output of different tileries, as they are all present at Brandier. The sequence identified in the rakings pit suggests that the plain TPF stamps came first, followed alphabetically (and logically) by TPFA, then TPFB, TPFC, and TPFP. Given the number of stamps, we can be reasonably confident that there weren’t any intervening TPFD, -E, -F, etc stamps, and so the choice of P after A, B, and C is curious and not readily explicable. Chronologically, the LHS stamps, assumed to be the product of Kiln 1, would follow on afterwards.
What sort of tiles were the kilns making? The main form recovered by us was tegula (a flat roof tile with a raised flange at each side), with lesser amounts of imbrex (a tile of semi-circular profile used mostly to cover the flanges of two adjacent tegulae) and tiles used for cavity-wall heating associated with a hypocaust. Not too much should be read into the proportions, however, as they may simply reflect that imbrices and flue tiles were not used in the construction of the Kiln 1, other than some imbrices that formed the drain.
Of the tile forms associated with heating wall spaces, three different types were manufactured at Brandier: parietales (a flat tile fixed vertically to a wall and held in position by cramps), box-flue tiles (a hollow tile about the size of an open-ended shoe box), and half box-flue tiles (they resemble a normal box tile cut down the middle). Traditionally these have been perceived as different stages of wall heating technology, with parietales and half box-flue tiles usually dated to the period up to the middle of the 2nd century. All three of these tile forms were found with TPF stamps, and occasionally some box-flue tiles were stamped TPFA. In addition to wall-heating, we also found three different types of tiles (known as voussoirs) used in forming arched vaults: so-called armchair voussoirs, hollow voussoirs, and solid voussoirs. The armchair voussoir is a rare form of vaulting normally only used by the military and occasionally for public buildings. Stamps were only found on the solid voussoirs, so it cannot be demonstrated whether the three forms were produced together, although, like the flue tiles, this is a distinct possibility.
Travelling tiles
Where were the tiles going, and how important was the Minety industry? Naturally enough, the prevailing view has been that the Minety kilns were established to service the construction of the major public and private buildings of nearby Cirencester, the second-largest town in Roman Britain after London. This is suggested by analysis of tiles found in Cirencester – which demonstrates that they were made from Oxford Clay, the Minety clay source – and by the numbers of LHS- and TPF-series stamps found there.
Pottery wasters found at Oaksey Common reflect types of vessels present in the earliest deposits in Cirencester, which date to the AD 50s and 60s, suggesting that the Minety industry was established within a decade or so of the Roman invasion of AD 43. When production started at Brandier is less clear, and the 2nd-century date suggested rests on the assumed date of the stamps, for which there is currently little solid evidence. A simple explanation would therefore see pottery and tile production starting at Oaksey Common in the 1st century, with the kiln at Brandier a 2nd-century offshoot. Early tile production was seemingly not solely directed to Cirencester, however, as 1st-century deposits at Bath (23 miles to the south-west) and Silchester (42 miles to the south-east) also appear to contain Minety products. If correct, this indicates that the industry supplied other major public building projects in western Britain.
We can now be confident that the source of the stamped tiles was solely Brandier, not Oaksey Park, and one possibility is that the Brandier works were owned by the Cirencester town council and the TPF series of stamped tiles was manufactured for the construction or repair of public buildings in the settlement. If we accept TPF as meaning ‘product of the public tileworks’, this would have identified the tiles as public property with the A, B, C, and P suffixes marking either different contractors who ran the kiln on behalf of the town, or possibly different contracts. LHS may have labelled the private production of the contractor for commercial sale.
The TPF series tiles are not restricted to Cirencester, however, as they also occur at some smaller settlements and villas on or close to Ermin Street (the Roman road linking Cirencester and Gloucester, not to be confused with Ermine Street, which ran between London and York) in a zone between the small town at Wanborough, near Swindon, to just outside Gloucester. But the focus is clearly on Cirencester. The LHS tiles have a somewhat different distribution. Some of the individual dies are restricted to Minety, Cirencester, and Wanborough, while certain other dies have a much wider distribution, from Silchester in the east, to Kenchester near Hereford in the west, and Old Sarum near Salisbury in the south. Curiously, neither stamp has been found in Bath.
Minety was therefore a regional supplier of tiles, not just the tileworks for Cirencester, but sending its products to several other urban centres, as well as a variety of smaller settlements. Tiles were heavy, bulky materials and it may be surprising to modern eyes that they were transported over such distances, rather than major towns setting up local kilns in their immediate environs. But the evidence shows that tiles did move long distances, an example of what to us looks like the irrationality of the Roman economy. Presumably this is because the economy was not driven by simple market forces, but instead by state contracts, local monopolies, and vested interests.
Source:
• Neil Holbrook is Chief Executive of Cotswold Archaeology, one of the largest archaeological companies in Britain. He has a special interest in the archaeology of Roman Britain.
• Peter Warry began researching Roman ceramic building material alongside a career in industry, and was awarded a PhD from the University of Reading in 2005. He continues to report on sites and publish research articles.
All images: Cotswold Archaeology
You must be logged in to post a comment.