4.1.3 Methodology

All the pottery has been fully archived, including sherd count and weight (4099 records) according to the recommendations of the Study Group for Roman Pottery. It was recommended that the waste from the fine ware production should be quantified by the addition of Estimated Rim Equivalents (EVEs), but funding was not available for this work. The relative merits of different methods of pottery quantification are fully discussed by Tyers (1996, 205). The samian has been reported by Brenda Dickinson, F.S.A., and that data is contained in a separate specialist samian database with details of potters and dates, and the archive fields of that database incorporated in the main archive database. The pottery from the 1991 evaluation Trench I has also been archived, and the database includes all the samian and mortaria from those excavations.


The dating of the pottery from stratified groups is examined using a technique developed to examine residuality (Darling forthcoming b), which results in a chart plotted by spreading the values using a program called Plotdate written by Paul Tyers (1994). This analysis is initiated by calculating a sherd count for all fabric and form combinations from the database for the relevant groups. This is then filtered through a ‘look-up' table, which assigns a numerical date-range to each fabric/form combination. The ‘look-up' table is the heart of the analysis, the date-ranges being based on the best available evidence for each fabric and form combination. These date-ranges are necessarily broad, the widest normally being 150 years, but due to the difficulties of dating the coarser Yorkshire ware, CASH, and in order to include as many as possible of the grey wares, the range has been extended to 200 years. The resulting data consists of a list of numerical date ranges (two fields giving the earliest and latest date) and a count field.

The final Plotdate program takes the resulting ‘value' (usually sherds) for each date-range and spreads it over its range either as the raw ‘value' or converted into percentages. The spread is often used for samian stamps, i.e. the value of the percentages for a date of 100-120 is spread over 20 years with one-half per decade, or a quarter per 5 years. For example, a value of 36 for 100-120 would be split into 18 for both 100-110 and 110-120, or 9 for each 5-year range.

To examine the main use of the site the local production waste and coarse OXSH fragments have been excluded. Approximately 49% of all sherds (nearly 5300 sherds) are plotted, the remainder being either undatable or having excessively wide date-ranges. Effectively this means that every sherd in the database to which a date can be applied is used to define the dated content of the site. The samian and mortaria included have used dates assigned by the individual specialists. The resulting chart (Fig.39), shows the plot of all pottery, and, to enable assessment of the relative contributions of samian and coarse wares, the plot of coarse wares alone.

figure 39
Fig.39: Plotdate of all pottery, and coarse wares only (plotdate values)

This shows a small quantity in the 1st century, rising to a relatively strong 2nd century content, and peaking in the latter part of 3rd century. It should be emphasised that the 4th century content is due almost entirely to the long date-ranges necessarily given to various wares and forms of broad 3rd to 4th century types, mostly Dales ware jars (range AD 250-350, including all bodysherds) and some NVCC vessels. The plotted 4th century content represents under 10% of the sherds, and is an acceptable ‘tail' given the quantity of Dales ware and the date-range necessarily applied. The absence of known 4th century coarse wares otherwise well represented at Brough (Wacher 1969, period VIII and later, figs 66-82), apart from a few in the plough-soil, makes it clear the main activity on the site had ceased by the end of the 3rd or very early 4th century.

Sample sizes

The value of groups of pottery is governed by the size of the samples, usually measured as a sherd count. The point at which samples become too small to be of value for analysis is very difficult to define, since it must depend upon the content of the group. A single smashed jar body, barely datable, could account for 100 sherds, while other contexts may consist solely of amphora sherds. Equally a context may contain only 10 sherds, five of which are diagnostically datable. As a rule of thumb, given a ‘normal' mix of fabrics and forms, any date assigned to a context containing fewer than 50 sherds is likely to be less reliable. Larger samples are necessary for analysis of the dated content of a group, preferably over 200 sherds, the larger the better.

The use of Lincoln as a comparison

The large fully integrated database covering all sites excavated in Lincoln from 1970 to 1987 (Vince and Jones 1990) has been used for comparanda. As a colonia and major urban centre in the east of England, Lincoln is particularly relevant to Brough-on-Humber where much of the pottery has a close resemblance to Lincolnshire ceramics. By the 3rd century, the central period for the Brough assemblage, Lincoln and Brough are likely to have had similar urban occupation and specific areas of Lincoln with similar date-ranges have been used comparatively.


© Internet Archaeology URL:
Last updated: Tue Nov 28 2000