Clay samples were taken as part of the research for this thesis. The majority come from two localities: the Malvern Chase and the Severn Valley, especially in the immediate environs of Gloucester. Samples from further afield were taken when the opportunity arose and are not a systematic survey of the clay sources of these areas.
Four or five basic clay types were recognised from this survey, although in most cases it is not the pure clay which was used in potting but a weathered or redeposited one. Analysis of parent clays by thin-section does not enable any pottery fabric to be adequately characterised, although it is often a contributory factor in the location or confirmation of a pottery source. Most of the information about the source and preparation of a pottery or tile fabric is derived from the study of the rock and mineral inclusions found within the fabric.
Geologically, the earliest clay is that from the Carboniferous Coal Measures. This is of two types; a light- firing clay and a red-firing clay. Both have very little naturally occurring quartz silt and are characteristically full of clay pellets. Coal Measure deposits outcrop in a number of areas within and just outside of the region but only three clay samples have been taken; one of red-firing Staffordshire Marl (M718), one of a lump of fired but unworked red-firing clay from the site of the 19th century Coleford Pottery in the Forest of Dean and one sample of white-firing clay from Pipe Aston in North Herefordshire (M717). The red-firing clay does not appear to have been widely sought after within the study region but was used extensively in Staffordshire and the north-west of England from the 15th/16th century to the 18th/19th century for black-glazed coarsewares. It was only used in the Forest of Dean coalfield in the 19th century.
The white-firing ware on the other hand was highly prized in the medieval period and was used in the Bristol area at Ham Green and in the suburbs of Bristol. Both wares have the characteristic white clay pellets but are tempered with different sands. Other industries using white-firing Coal Measure clay were at: Chilvers Coton, Nuneaton; probably Staffordshire (one kiln is known, at Sneyd Green, and there is a distinctive, partially glazed white ware found in South Staffordshire and West Midlands which has red-painted decoration). White-wares are also known in North Wales, at Ewloe (Davey, 1977). It is also possible that sites other than Bristol in the Bristol-Somerset Coalfield might have been using white-firing clay, there are, for example, references to Kingswood and Hanham potters (Price, 1979).
White-firing Coal Measure clays were used at Pipe Aston for clay pipe manufacture in the early to mid-17th century. They may well have been used in the Broseley industry, which rose to prominence in the late 17th century but it is more likely that imported North Devon clay was used, since there is documentary evidence from this period onwards for the transport of North Devon pipe clay up the Severn (S. Davies, pers. comm.).
Visually and in thin-section there appears to be little difference between the fabrics of wares made with these clays and the same range of inclusions is often present (for example, red sandstone fragments, clay pellets, iron ore fragments and quartz).
The Old Red Sandstone measures of the Welsh border include large tracts of clay. This clay has a high iron content, abundant quartz and white mica silt and occasional nodules of carbonate. Several samples of this clay have been taken but most are impure, containing sandstone fragments and some quartz (M530-2, M536, M724, M1053, M1116, 1120-1). These inclusions are probably present because of glacial reworking of the clays, since there was no overlying sand or gravel which might have contaminated them. The use of 'Devonian Marl' for potting dates back to the late 12th or early 13th century (for example Hereford A2 and Hereford A3) but was not extensive until the late 13th century. At this time numerous vessels, mainly glazed jugs are known from sites in Herefordshire, the Forest of Dean and Gwent. It is not known how many centres were involved and only one production site is proven, at Weobley in Hereford and Worcester. In the late 15th to 16th centuries little use was made of this clay type, mainly because of the predominance of Malvern Chase pottery in the region, but in the late 16th and 17th centuries numerous small industries existed using this clay. Their products are quite indistinguishable both visually and in thin-section.
Keuper Marl clays occur in a narrow north-south band leading through Worcestershire and Gloucestershire (west of the Severn) into South Wales. Separate outcrops are found in Avon surrounding the Mendip Hills and in isolated pockets in the Bristol-Somerset coalfield. A large number of clay samples have been taken from the Keuper Marl, principally around Malvern Chase but including Newnham-on-Severn and Woolaston (M804-6, M1051-2). In the study region only a small proportion of the total extent of Keuper Marl was workable potting clay. Much of the exposed clay was of a blocky texture with a high finely-divided carbonate content. Where samples could be prepared they were remarkably variable in texture and in the type and quantity of inclusions. It is likely that few of these inclusions were present in the parent clay and must have arrived by admixture with overlying sands and gravels. The fact that in places the clay samples were taken from some depth shows the extent to which clays can be contaminated by root activity, possibly animal burrows and by cracking.
General characteristics of the clay are an absence of quartz or mica silt, a variable iron content (but always lower than that of the Devonian Marl) and a variable quantity of clay pellets. In some wares made from Keuper Marl clay pellets are extremely common, for example Canynges-type and Gt. Malvern floor tiles. This was also a characteristic of a clay sample from the outskirts of Great Malvern (M181). Because of the absence of distinctive characteristics it is difficult to say how much use was made of Keuper Marl by medieval potters. It is the parent clay used for Malvern Chase pottery and may have been used for Worcester-type ware, although there is no proof. It is used for two groups of floor tiles; Gt. Malvern and Canynges-type tiles. The latter is unprovenanced but is most likely to have been made in Worcestershire.
Jurassic clays occupy most of the study region and are extremely homogenous. They occur in three main facies: fine estuarine clays, shelly marls and fine slightly micaceous clays. The former is by far the most widespread type and includes the whole outcrop of Lower Lias and much of the Oxford Clay. Numerous samples were taken of Lower Lias and Oxford clays but no regional variability was seen (Lower Lias M421-2, M424-5, M436-7, M588, M671, M719-20, M722-3, M1099, M1101, M1108-9, M1111, M1113-5, M1119; Oxford Clay M529, 750, 1107, 1118). The Jurassic estuarine clay was notable for the difficulty experienced in firing the samples, many of which blew up. This was not due to insufficient drying and is most likely a result of a high organic content. In thin-section, this clay when fired usually has a distinctive vesicular appearance and an extremely low quantity of quartz. Very little use was made of this clay by medieval or post-medieval potters, considering its vast extent. It was the basis for Minety ware, but in a heavily limestone-tempered and possibly weathered version and was used without much tempering in Gloucester TF79, a rare 15th century ware.
The second Jurassic clay type, shelly marl, occurs in the study region. A sample was obtained from a quarry at Hawley, where a thin band of shelly marl was sandwiched between massive Great Oolite beds (M1100). There is no evidence for its use for potting in the region, although just outside the region to the east, in the Oxford area, it was used between the late 8th and the 10th centuries. It is very noticeable in comparison with the untempered Jurassic clays that no problems occur in firing samples at low temperatures, although at temperatures in excess of 800 degrees C. the shell decomposes on cooling (due to the intake of water vapour combining with calcium oxide to form calcium hydroxide). This clay type is readily distinguished in thin- section from deliberately tempered shelly wares by the ill- sorted nature of the shell inclusions and by the presence of micro-fossils and bryozoa fragments alongside the shell. Quartz is usually extremely rare in these fabrics. It is much more difficult to distinguish different wares made from this type of clay. Three distinct wares are the St. Neots-type ware of the 9th to 11th centuries, the St. Neots Jugs of the 12th to 13th centuries and Oxford fabric B.
The last distinguishable type of Jurassic clay is typical of the Middle Lias and is variable in colour. Nodules of red iron ore are frequently found in the clay and are sometimes found in vessels made from this clay. White mica is commonly found but quartz is rare. This clay type is found in Gloucester TF41b and is represented by samples from Bath and Haresfield, in Gloucestershire (M423,M721, M1110, M1102, M1117). The clay occurs in a narrow band around the Cotswolds between the Upper Lias fine sands and the lower lias clays. A few sherds of the Chaff-tempered pottery from Andoversford, in Gloucestershire, have these characteristics, which also serve to distinguish the limestone-tempered Gloucester TF41b from the Gloucester-made limestone-tempered ware, Gloucester TF41a, and North Cotswolds I ware.
The Gault clay, of Cretaceous date occurs in a narrow band running north-east to south-west through western Wiltshire. It is characterised by a high quantity of ill- sorted quartz and in places is extremely micaceous. However, no clean samples of Gault clay have been examined in thin section and most of the inclusions found in wares made from the Gault were probably added, either intentionally or naturally, from overlying deposits derived from the greensand and chalk. Bath Fabric A, Crockerton wares and Nash Hill ware are all probably made from Gault clays. A sample of 'head' from the Crockerton area was fired and thin-sectioned (M734). It contains a very high quartz and chert content, so that it is extremely friable.
No other clays occur in deposits of Cretaceous date, although pockets of clay are found capping many areas of chalk downland. Samples of this 'clay with flints' have been examined visually but none examined in thin-section. A characteristic of this clay seems to be the variable quantity of quartz present. This must be due to the nature of the tertiary deposits whose erosion created the clay with flints. Most samples were not heavily calcareous and did contain large angular fragments of flint.
Tertiary clays occur within the study region only in the east and south-east of Wiltshire and in Berkshire. Two main series of clays occur within these Tertiary deposits, the London clay and the Reading Beds. The London clay in the London area is extremely fine-textured with a high organic content, comparable to the Lower Lias clay. It too sometimes contains microfossils of various types and has a low fluxing temperature, giving rise to a bubbly textured matrix at temperatures below 1000 degrees C.
Two types of clay are being formed in the study region at the present day, or have been formed in the recent past. The first is formed by alluvial deposition. In areas where rivers cut through clay deposits this can lead to the formation of a clay with mixed characteristics, such as a fine-textured matrix and a coarser sand fraction. Such clays have been sampled at Pill Harbour, where a silty mud is forming a clay suitiable for brick manufacture (M555), and Lassington, where they were formed by deposition from a small river draining into the Severn (M553). In this instance the clay contained quartz and sandstone fragments which probably originated in the Forest of Dean. The clays formed on the flood plain of the River Severn usually have a fine quartz and white mica silt temper often with a reddish colour which indicates their origin in the Tertiary deposits of Worcestershire (M535). Less heavily tempered clays are also found, formed by the redeposition of Lower Lias clay (M46-7, M549-51, M556). The silty Severn Valley alluvial clays have been exploited since the late 16th century for the manufacture of bricks, ruined brick kilns can still be seen protruding from the banks of the river. Many of these brick fabrics show 'salt surfacing' caused by the action of heat, calcium carbonate, iron and salt. A distinctive range of colours results, the most typical of which is a sickly yellow.
Clays deposited towards the mouth of the Severn or around the shores of the Bristol Channel have different characteristics. In general they are extremely sticky clays with few inclusions, although natural additions of beach sands do occur (M672-3).
The other type of clay formation is from the chemical weathering of rocks. In the south-west peninsula, chemical weathering has given rise to large deposits of kaolinite, the china clays. Within the study region no such large deposits are known but the clay-with-flints deposit found overlying the chalk downlands in Wiltshire was formed in this manner and is extremely extensive, even if not of great depth.
It is suspected that several of the limestone deposits of the study region have weathered to give rise to small pockets of clay. Excavations at Chedworth Roman Villa revealed that the Roman villa was situated not on the oolitic limestone which outcrops behind the villa but on a bed of sticky clay, a sample of which, M736, was fired and contained no limestone but had a very high iron content. One of the characteristics of some limestones is the presence of very small rounded pellets of iron ore, probably formed by algae. Some pottery fabrics have very fine clay matrixes which also contain these small pellets, such as Cheddar fabric E. The clays formed by the weathering of limestone will tend to contain very little quartz in the matrix, since quartz is rare in both the Carboniferous and Jurassic limestones of the study region. Bristol Fabric C contains angular fragments of limestone and flat fragments of red iron ore, both of which are most likely to have been formed by the weathering in situ of Carboniferous limestone.
The weathering of mudstones and siltstones, both of which have a substantial clay fraction, would give rise to a clay containing angular fragments of mudstone or siltstone while the weathering of sandstones would give rise to a deposit of sand and sandstone fragments.
The main outcrop of rocks of this type in the study region is in the Welsh borderland but the only pottery fabrics to have siltstone and mudstone inclusions are Hen Domen sandstone-tempered ware and Hereford A4. Neither of these fabrics contained angular inclusions and it is likely therefore that they both derived from deposits some distance from the outcrop of the rock fragments found within them. The source of the Hen Domen clay was confirmed by thin-sectioning of a series of clay samples, M914-7, taken from the site of Hen Domen itself (by Mrs. P. Irving). The Hen Domen clay contained the same range of inclusions as the pottery and in both the pottery and the clay the inclusions were rounded.
Sand and sandstone fragments occur in many pottery fabrics but there are no obvious characteristics to distinguish those of detrital origin from those that were formed by in situ weathering. Most medium-grained sandstones are too friable to become rounded by abrasion so that there is no way of using the roundness of inclusions as a way of separating the two types of deposit. Grain-size analysis may be used to distinguish ill-sorted deposits resulting from chemical weathering from well-sorted detrital deposits, although it is likely that the physical properties of the sandstone would also affect the grain size of weathering products. The other characteristic which can reliably differentiate between the two types of deposit is the presence of rock types in the same pottery fabric which do not outcrop together.
The majority of pottery and tile fabrics which do not include 'tempering' present in the geological deposit from which the clay was derived are 'tempered' with detrital sands or gravels. The distinction between natural 'temper' and that introduced by man does not appear to be testable by scientific means, since the mixture of sands or gravels with clays can happen naturally, as is demonstrated by the analysis of clay samples from Malvern Chase (M180-1, M524-8 and M540-8). Although every attempt was made to obtain 'pure' clay samples without contamination from overlying gravel deposits every sample contained rounded quartz sand and angular Malvernian rock fragments. The quantities of these two types of inclusion varied from sample to sample in a systematic fashion. Those samples obtained from the west of the chase, at the foot of the Malvern Hills, contained the highest quantity of Malvernian rock and the lowest quantity of rounded quartz sand. Conversely, samples from the east of the Chase, closest to the River Severn, contained the highest quantity of rounded quartz sand. The inclusions in these clays were derived from overlying gravel deposits, presumably by plant activity and by cracking of the clay. The relative frequency of the two types of inclusion is a reflection of the composition of the gravels. Those from the west of the Chase being formed mainly from talus from the Malvern Hills and those from the east of the Chase being deposited by the River Severn and ultimately deriving from the Triassic sandstones of Worcestershire and areas further to the north.
Similar evidence for local variations in the composition of sands and gravels comes from the Severn Valley around Gloucester. Some sand deposits in the Gloucester region contain mainly rounded quartz, identical to that found in the Malvern Chase clay samples, although usually with the addition of larger rounded limestone fragments (M533-4, M1112).
From the above description it will be clear that thin- section analysis of untempered pottery fabrics made from the main geological deposits of clay in the study region would not enable individual wares to be identified. Indeed, it would be difficult to positively identify the geological deposit from which the clay was obtained. However, the clays used in potting were rarely untempered and, even when they were, they were rarely pure geological clays. The natural weathering and mixing of clays quite often produced clays with a distinctive appearance and the addition by the potter of further temper also served to produce unique fabrics which, when found in the medieval pottery of the study region, enable the source of the vessel to be positively identified.