This method would repay further use since large numbers of samples can be processed quickly, as opposed to the quantitative methods of plasma-emission spectroscopy or neutron activation analysis.
Two methods of glaze analysis which help distinguish the glaze recipes are firstly, refiring of sample sherds under controled conditions and secondly, replication of recipes using the same clays and firing conditions as those suggested for the original vessels. Neither method has rigourously been applied in this study, although some glazes were refired as part of a series of refiring experiments on Hereford and Chepstow pottery. These suggested that the variations in surface finish (matt, crazed, glossy) were all due to firing rather than composition since most glazes on refiring produced a glossy, crazed, clear glaze whatever the original appearance.
Qualitative and semi-qualitative analysis by X-Ray Fluorescence does not distinguish the recipes since the final glaze composition is always a mixture of lead and the clay body. No attempt has therefore been made here to suggest which glaze recipes were used on any type.
Three methods of glaze application are known from documentary or from modern parallels; Splashing, painting and dipping.
Most of these glazes have noticable pitting, and in under-fired examples these pits still contain blobs of metal. This suggests that the pitting is the result of the glaze being applied as a relatively coarse powdered metal, rather than as a suspension of a lead compound.
Splash glazes have a long history in the London region, being first found in the mid-12th century and still being used in the early 14th century. Most splash glazes do not have colouring agents, although they are known. The technique has not been recognised on any locally made pottery in the study region.
Table 00 shows the results of XRF analysis of several locallly produced medieval and post-medieval glazes. The copper counts can be interpreted as three groups.
Group 1 has no copper whatsoever and consists predominantly of 12th century types, including both locally produced tripod pitchers and non-local jugs (Stamford and Andenne). The analysed samples of North Devon gravel-tempered ware and Wanstrow ware similarly have no copper.
Group 2 consists of glazes containing detectable copper (which probably means in the order of parts per hundred rather than parts per thousand, pers. comm. J. Bayley). Into this group can be put Ham Green ware, Bristol Redcliffe ware, Nash Hill clear glazed (although it is known that some vessels were deliberately coloured with copper) and Oxford AM together with two post-medieval samples; Ashton Keynes ware and ?Weston-super-Mare coarseware.
Group 3 consists of glazes deliberately coloured with copper. Into this group can be put Worcester-type jugs, Nash Hill green-glazed jugs, a sherd of a locally produced jug from Richards Castle, Hereford and Worcester, Nuneaton Ware, North French Monochrome, Saintonge polychrome and mottled wares, London-type ware Kingston ware and Developed Stamford ware. Of the post-medieval wares analysed only South Somerset slipware glaze had a high copper content, used in that case for paint rather than an overall colour.
These are wares for which scientific analysis is available. On the basis of visual comparison we can add Malvern Chase late medieval glazes and Newbury C green-glazed jugs.
From these results one can say that mottled green glazes on an oxidized or slip-covered body can be said with certainty to be coloured deliberately with copper. With mottled green glazes on a reduced high iron body it is not so easy to be certain, and until the analyses were carried out it was not known for certain that Worcester-type jug glazes were coloured with copper.
The identification of group 2 glazes is of interest. All are 13th century or later but have no other characteristics in common, except that they were probably all dipped rather than splashed or painted-on glazes.
Eraclius' treatise contains a recipe for making a green glaze:
"... However, if you want to obtain a green colour, take some copper, or better still some auri-calum [brass or bronze], and mix it with the lead as follows: take the lead and melt it in a pot; when it is molten stir it with your hands [!!!!] in the pot until a powder is produced, and mix this then with 6 parts of brass filings. When the poty has been dampened with water and flour sprinkle it immediately with lead, ie. with the filings mentioned above. if you want a yellow glaze sprinkle the pot with pure lead without brass filings. Then place this pot in a bigger pot and put it in the kiln so that it will become more brilliant and beautiful, but in a slow heat, not too much nor too little." (De Bouard, 1974, 69).
This recipe suggests that we should find a correlation between the presense of copper and that of either zinc or tin, at least in Northern French green glazed jugs. However, as can been seen in table 00 there is no correlation between tin or zinc and copper. Equally high concentrations are found in plain lead glazes. De Bouard reports on some experiments carried out at the Caen Centre of Medieval Archaeological Research to replicate the green glaze of Northern French green glazed wares, and on analyses carried out on archaeological samples. These showed that a good approximation to the medieval glaze could be obtained with a mixture of 96.72% lead to 3.28% copper. The quantity of copper found in the archaeological samples (which varied in date from the 9th or tenth century to the 13th or fourteenth century) varied from 0.3% to 4% (De Bouard, 1974, 75). De Bouard does not state whether the glazes with low copper, which are possibly comparable with group 2 above, were coloured green or, like group 2, clear.
Semi-quantitative X-Ray fluorescence analysis, carried out by Justine Bayley of the Ancient Monuments Laboratory reveals the same order of concentration of iron in all three wares (see table 00).
The distribution of values for iron in the XRF analyses has a single peak, between 200 and 400 counts. If there is any pattern at all to the results it is simply that those samples with a thin or patchy glaze cover show higher iron counts than those with a thick, overall glaze. Manganese was only present in six of the samples and in all but two cases the sample also revealed a high iron content (for the reasons outlined above). In the remaining two cases; Saintonge polychrome brown paint and Malvern Chase 'pink' ware with a brown slip, it is likely that the manganese is present as part of a manganese/iron slip or paint. There is thus no evidence from this series of analyses for the use of iron or manganese as an intentional glaze colourant. Biddle and Barclay record a series of quantitative analyses of the glaze on Winchester Ware (Biddle and Barclay, 1974, 140). These show negligable quantities of copper, nickel or cobalt but consistently 3-4% of iron. They did not, however, determine the origin of the iron.
There are some wares, however, where the use of iron or manganese to give a black, brown or purplish brown glaze is proven. The earliest of these is of late 13th to 14th century date and is obtained by adding iron scale to applied strips. On firing this produces a hideous bubbly glaze with a metallic gloss. The technique never had wide application in the region but is found rarely on Her A7b jugs, for example.
The colouring agent in the latter case was definitely introduced through the clay. In the next main use of iron/manganese colouring, Cistercian type ware, it is by no means certain whether the glaze was adulterated or whether the colour comes from the body. The latter is more likely since it is noticable that 'low-fired' examples of Cistercian ware have a clear, brown glaze and the distinctive coloured or opaque glaze is restricted to high-fired examples. In some thin-sections minute opaque crystals can be seen in trails leading from the clay surface into the glaze. These give many cistercian ware and later black glazes a distinctive glittering appearance under the binocular microscope.
There is in fact another type of blackish glaze found on cistercian ware; a mottled purple glaze. This has the same appearance as many kiln bricks, accidentally glazed by fuel ash. It is possible that some cups were intentionally given a fuel-ash/lead glaze by throwing ash into the kiln. This form of Cistercian ware, although noted rarely in the region, is probably not a local product.
Staffordshire and Surrey-Hants Border wares are sometimes covered with a mottled brown glaze over a light-coloured body, which could not have given rise to such as colour. The technique was definitely in use in the Surrey-Hants border by the begining of the 17th century but the first occurence of the technique in the Staffordshire potteries seems to be at the very end of the 17th or begining of the 18th century. As in copper-green glazes the mottling is probably a reflection of the distribution of iron/manganese in the glaze (whereas mottled brown salt-glaze is caused by the application of a uniform brown slip, the mottling is a reaction between the salt and the iron, causing the glaze to 'crawl').
This level of tin is insufficient to have any effect on opacity, and is too low to be the result of using pewter instead of lead. There is no correlation with the concentration of copper or zinc and it is thus inlikely that the tin was added as an impurity with these metals. The likelihood is that the tin was an impurity in the lead, in the same way that zinc and cadmium probably were.
The fact that the three highest tin counts obtained were for samples of Saintonge polychrome ware may be significant in relation to the problem of how such a clear glaze was obtained.
The earliest tin-glazed wares made in this country were of late 16th century date in Eastern England and mid-17th century in the West. The earliest documented tin-glazed pottery in the west was in Brislington and archaeological evidence confirms that tin-glazed wares were rare or absent before that date.
It is possible that they reflect a change in lead source. Wanstrow ware is known to have been glazed with Mendip lead (V.C.H. Som II, 1911, 429) and has a zinc count of 162.
Cadmium occurs as a trace in more than half of the glazes but there appears to be no pattern in its occurence.
ALKALINE GLAZE. One type found in the region has an alkaline glaze, Syrian, or Egyptian, Alkaline Glazed ware. This thick, sugary glaze is heavily crazed and usually has a slight blueish tinge. Alkaline glaze is translucent and usually covers blue and black painted decoration.
At temperatures above 200 degrees C. chemically combined water is driven out of the clay but if the vessel is taken out the kiln at this stage it will gradually re-adsorb water from the atmosphere.
At 550 degrees C. the quartz in the vessel will change from alpha to beta quartz. This causes the quartz to expand. If too much coarse quartz was present this might cause the vessel to crack or shatter. Vessels with a high quartz content therefore need to be taken slowly through this part of the firing. This might be one reason why coarse sandy fabrics replaced those with calcareous or organic inclusions.
Above 600 degrees C. the clay begins to undergo non- reversable changes. Gradually the clay begins to fuse together, forming an amorphous material distinguishable in thin-section by the absence of birefringence. The temperature at which the whole of the body becomes isotropic varies from clay to clay. In particular iron acts as a flux in this change and wares with a high iron content 'mature' at lower temperatures than those with a low iron content. Similarly the identity of the clay minerals affects the maturing point. Kaolinite requires the highest temperature.
Above the maturation point there is a range of temperatures over which no changes can be discerned followed by a point at which the clay starts to warp or bubble. This temperature too will vary from clay to clay. Calcite will decompose at c.850 degrees C. to form calcium oxide and carbon dioxide (although the exact temperature depends on the available oxygen and the pressure). On cooling this calcium oxide will take up water to form calcium hydroxide. This reaction causes the inclusion to expand. Isolated calcium carbonate inclusions will cause spalling whilst a heavily limestone-tempered fabric will completely shatter if fired above this crucial temperature.
For a given clay there is a certain range of temperatures over which it can be sucessfully fired. If this range is too narrow then the clay would not have been usable without considerable control over firing conditions.
The firing temperature is therefore an important feature of any vessel. Perhaps equally significant is the length of firing and the atmospheric conditions in the kiln. There are several methods available for determining firing temperature and conditions. Of these the simplest and least accurate is to refire samples of the ware at known temperatures and in known conditions and to compare the appearance of the resulting sherds with the original sherds. A number of samples were refired, principally from Hereford and Chepstow. These showed firstly and conclusively that the colour of greyware sherds was in the main due to oxygen starvation in the original firing and that with few exceptions the clays would fire red at 1000 degrees in an oxygen-rich kiln atmosphere.
Similarly a rough indication of maximum original firing temperature was given by comparing oxidized sherds with refired samples. In no case, until the 16th century, need the vessels have originally been fired above 800 to 900 degrees C. Even after this date many of the wares examined were probably relatively low-fired.
A more accurate determination of firing temperature can be obtained by differential thermal analysis, in which a sample is reheated whilst monitoring the energy taken to heat the sample. This can also reveal something about the compostion of the sample, since calcium carbonate will produce a distinctive change in heat output as will the different clay minerals. This method is still inaccurate and much more time-consuming than the refiring method.
The third method is to use scanning electron microscopy (SEM). Using this method it is possible to see the degree to which the clay minerals have been vitrified and also to compare the surface with the core (which should give an indication of the duration of firing). This method is prohibitively expensive.
The incomplete plan of a 12th to 13th Century kiln found at Penhow, Gwent, seems to discount the presence of two flues. Unlike other examples classified by Musty this kiln has a central spine without trace of a raised floor (Wrathmell, 1981, 4-7). Wrathmell quotes two parallels for the form of this kiln, one at Pottersbury, Northamptonshire, and the other at Nettleden, Hertfordshire. The products of the kiln, Chepstow HA (Penhow) ware, are also more similar to those of the home counties than they are to other local wares.
At a later date double-flued kilns are found. These too have been classified by Musty into those in which the pots were fired on the floor of the kiln (types 2a and 2b) and those in which there was a raised floor (type 2c). Three kilns of type 2 have been exacavated in the region. The earliest was probably at Ham Green, of early to mid-13th century date. This kiln had a central division (type 2b) which did not apparently support a floor (Barton, 1963). The other two sites are of later 13th to 14th century date, Laverstock and Nash Hill, both in Wiltshire (Musty et al., 1969, 88-90, Fig.4; Mc.Carthy, 1974). Double-flued kilns are known from the region in the post-medieval period in the North Herefordshire potteries (Marshall, 1948). All of these kilns were probably associated with very small scale pottery production.
To judge by the ground plans there was considerable overlap between the single and double-flued kilns in their capacity, although the simpler type 1a kilns were probably smaller that the remainder of type 1 and type 2.
Multi-flued kilns (Musty type 3) on the other hand have a much larger ground plan, hence probably the need for multiple flues. From the examples quoted by Musty (1974, 63-4) it appears that this type is essentially late and post-medieval in date and at present it is confined to the north-east of England and the Midlands. Examples are known from Chilvers Coton in Warwickshire and Sneyd Green in Staffordshire. These round, multi-flue kilns are similar to the earliest post-medieval kilns known in the Potteries, for example an example excavated at Hanley, Staffordshire (Celoria and Kelly, 1973).
The ground plans of kilns therefore show a gradual progression from small through medium to large kilns. As the size was increased there was a need to increase the number of flues from one, to two and then many (five or six being the most common number). It is idle to speculate on the relationship between the output of these kilns and the organisation of production in the region until more kilns have been excavated.
Saggars have also been found at several of the post- medieval Herefordshire kilns, mainly associated with the production of black-glazed tygs. In a few of the North Herefordshire kilns ring-saggars have been found. These are used to separate plates in firing. The Newent Glasshouse pottery also produced plates but there are no saggars from the site and the kiln scars found indicate that the plates and dishes were fired one inside the other, leaving blemishes on the rim and side of the vessels.
The Herefordshire kilns have also produced fragments of sandstone coated with glaze. These were probably used to separate coarsewares in the kiln. Many of the medieval glazed wares typically have rim scars on the base of the jugs, showing that these separators were not used. There is not enough kiln evidence to show when this technique was first used in the region.