Figure 22: Factor analysis for Northern Gritty ware: plot of F1 against F2 scores
Excavations at the Upper Chapel, Sheffield, produced a collection of medieval pottery which included over-fired or burnt sherds, suggesting that the collection included production waste. Accordingly, a sample of five sherds was submitted for thin-section and chemical analysis, to establish the characteristics of the pottery and whether it can be distinguished from other similar vessels (see Table 10).
In the event, it appears that at least some of the samples are not of production waste but were produced elsewhere. However, three of the samples, all Northern Gritty ware, have a distinctive chemical composition which can be distinguished from samples from other medieval production sites (all located considerably to the north of Sheffield).
Thin-sections were produced by Steve Caldwell, University of Manchester, and a block was removed from each sample and the surfaces removed mechanically. This block was crushed to a fine powder and submitted to Dr J N Walsh, Royal Holloway College, London, for chemical analysis using Inductively-Coupled Plasma Spectroscopy.
A range of major elements was measured and the results expressed as per cent oxides (Table 11). A range of minor and trace elements was measured and the results expressed as parts per million (Table 12).
The ICPS data were normalised to aluminium to take account of the dilution effect of added quartz sand temper and were analysed using Factor Analysis (using the Factor Analysis option in WinSTAT, an add-on for Microsoft Excel).
Northern Gritty ware is a generic name for medieval pottery fabrics in an off-white to light brown clay tempered with angular quartzose gravel, with grains over 1.0mm across. Pottery of this general type was produced at numerous sites and those at Baildon, Winksley and Grantley, all in West Yorkshire, have been sampled and can be compared with the Upper Chapel samples. In addition, a large number of samples of Northern Gritty ware from the occupation site of Ingmanthorpe Manor were also analysed and these too could be compared with the Sheffield samples.
The three samples all have similar characteristics in thin-section. The following inclusion types were noted:
The groundmass consists of light grey to light brown optically isotropic baked clay minerals and sparse ill-sorted angular quartz grains up to 0.1mm across.
The quartzose inclusions are probably derived from a detrital sand derived from the weathering of Carboniferous sandstones, of at least two different textures and lithologies: a coarse-grained rock from which the quartz grains were derived, and a finer-grained red sandstone. Such sandstones mostly occur in the Coal Measures, which is consistent with the location of the Upper Chapel. The mudstone, some of which is organic, might be present in the parent clay or may also be detrital. It is almost certainly a Coal Measures shale.
The ICPS data were compared with that from Baildon, Knaresborough, Grantley and Winksley. The Grantley and Winksley samples include several different fabrics, and the Knaresborough samples are from a consumer site in the town, which is, however, known to have had a pottery industry.
Factor analysis was used to analyse the data and five factors were found. A plot of F1 against F2 scores (Fig. 22) indicates that two of the Sheffield samples are clearly distinguished from the comparanda, being closest in composition to the two Knaresborough samples, while the third sample is similar to the Winksley and Lumley samples (apart from Winksley Fabric 1, which is a distinctive white sandy fabric).
Figure 22: Factor analysis for Northern Gritty ware: plot of F1 against F2 scores
A plot of F3 against F4 scores separates one of the Sheffield samples from the remainder (Fig. 23) while a plot of F3 against F5 scores distinguishes the Sheffield samples from the others (Fig. 24).
The comparative data all come from vessels made with Coal Measures clays and tempered with sands derived from Coal Measures sandstones and therefore the similarity of the Sheffield samples might be expected.
Figure 23: Factor analysis for Northern Gritty ware: plot of F3 against F4 scores
Figure 24: Factor analysis for Northern Gritty ware: plot of F3 against F5 scores
Examination of the data, normalised to aluminium, indicates that the Sheffield samples have higher Potassium and Manganese values but that all other elements occur within the same range as the comparanda (Fig. 25).
Figure 25: Normalised data for Northern Gritty ware: plot of Potassium against Manganese values
A single sample from the Upper Chapel has the same characteristics as York Gritty ware, which is widely distributed in the Vale of York and less common in surrounding areas (Holdsworth 1978; 1995; Mainman 1990). Thin-section and chemical analysis has suggested a source for this ware at Potterton but it may well be that it is a generic fabric, produced at more than one centre but using similar raw materials and production methods (Vince and Young 2007; Vince 2004b). Pottery sherds found at Thorner in levels post-dating a 9th- to 11th-century kiln and published as Hillam ware were sampled and shown to be of this fabric (Cumberpatch and Roberts 1998-1999; Vince 2004a).
The thin-section has the following inclusion types:
The groundmass is optically anisotropic light brown baked clay minerals with no inclusions. Thin lenses of darker brown clay occur.
The ICPS data were compared with that from consumer sites at Doncaster, Knaresborough, Swillington, Thorner, Wetherby (Ingmanthorpe Manor) and York.
Factor analysis found three factors. A plot of F1 against F2 (Fig. 26) found that the F1 score of the Sheffield sample was within the range of the comparanda (as were those of the Sheffield Northern Gritty ware sherds). However, the F2 scores of both the Sheffield group were higher than those of the majority of the comparanda, the exception being one of the Doncaster samples.
A plot of the F2 against F3 scores (Fig. 27) for this dataset shows that the Sheffield sample has a lower F3 score than any of the comparanda (also lower than the Sheffield Northern Gritty ware).
Examination of the normalised data indicates that the Sheffield YG sample has a lower Calcium value than any of the comparanda, followed by the Sheffield NGR samples. The Scandium value separates the two Sheffield groups, although both lie within the range found in the comparanda. The Vanadium value is higher than in the comparanda, followed closely by the Sheffield NGR samples (Fig. 28). The ICPS data therefore suggest that the two Sheffield groups are more similar to each other than to other York Gritty ware samples, but that with the exceptions noted above, the range of values found is similar for both the Sheffield samples and the comparanda. The high Vanadium values are probably present in the dark brown to opaque coatings seen on the quartz grains, and in the siltstone.
Figure 26: Factor analysis for York Gritty ware: plot of F1 against F2 scores
Figure 27: Factor analysis for York Gritty ware: plot of F2 against F3 scores
Figure 28: Normalised data for York Gritty ware: plot of Vanadium against Calcium values
One sample could be distinguished by its higher iron content and the character of the sand inclusions. The following inclusions were noted in thin-section:
The groundmass consists of optically anisotropic baked clay minerals, moderate muscovite laths up to 0.05mm across and moderate angular quartz grains up to 0.1mm across.
The sand inclusions in this sample are probably derived from a Permo-Triassic sand (the rounded quartz, chert, siltstone and fine sandstone), together with some Carboniferous sandstone. Such a sand deposit should not occur in the Sheffield area and the closest source would be to the east of Sheffield, such as the Don Valley, or to the south, in the Nottingham area.
The normalised ICPS data indicate a higher iron content for this sample as well as a higher Calcium, Titanium, Phosphorus, Chromium, Copper, Vanadium, Zirconium, Lanthanum, Cerium and Lead content and a lower Potassium, Manganese, Lithium, and Strontium content. Some of these elements are easily affected by burial conditions, but in this case these are likely to be comparable for all five samples. Furthermore, the copper and lead values might be enhanced through contamination by the glaze. Even excluding these elements there are sufficient differences between this sample and the others to suggest a different source.
The normalised ICPS data were then compared with those from a series of samples of Nottingham glazed wares (NOTG), a sherd of possible late Saxon Nottingham ware from a site in Doncaster (NOTTS) and Doncaster glazed red earthenware (Doncaster A ware). Factor analysis revealed five factors and a plot of F1 against F2 (Fig. 29) indicates that the Sheffield sample is closest to the Nottingham glazed ware (and that the supposed Nottingham late Saxon sample is closer to the Doncaster glazed ware). However, examination of the normalised data indicates that the sample can be distinguished from the comparanda by higher Titanium and Chromium values (Fig. 30) and lower Lithium and Strontium values (Fig. 31).
Figure 29: Factor analysis for glazed red earthenware: plot of Fl against F2 scores
Figure 30: Normalised data for glazed red earthenware: plot of Titanium against Chromium values
Figure 31: Normalised data for glazed red earthenware: plot of Lithium against Strontium values
One of the five submitted samples is definitely not a local Sheffield product and can be distinguished both in thin-section and by its chemical composition. The remainder include one sample that is distinct in thin-section and, to a lesser extent, in its chemical composition. However, the raw materials to produce this fabric should be available in the Sheffield area and it is possible (though unlikely) that the 'York Gritty' ware sample is a Sheffield product made using similar raw materials and preparation techniques to the York Gritty ware found in the Vale of York and, more rarely, in neighbouring areas.
Three of the samples, however, were produced from similar raw materials and could well have been made in the Sheffield area. They are similar in thin-section and chemical composition to Northern Gritty wares produced at sites in West Yorkshire, but since these exploited the same Coal Measures outcrops this is unsurprising.
© Internet Archaeology/Author(s)
University of York legal statements | Terms and Conditions
| File last updated: Wed Jan 19 2011