### What is in the PAS Database Part IV: Aoristic Analysis

Temporal uncertainty inherent in the Portable Antiquities Scheme database (PASd) poses challenges to representing and analysing the finds data. One powerful and straightforward technique for understanding the temporal distribution of the PAS finds is aoristic analysis. This technique has been used for example by criminologists for estimating peak offence times (e.g. variability in the frequency of thefts in a certain area over the course of a day), and it is suited for examining archaeological and historical datasets.

Briefly, a basic use of aoristic analysis in the context of the PAS data would be to calculate peaks and troughs in the chronology of estimated object deposition in a given area. A hypothetical example follows:

Let assume that four finds have been recorded in an area of a field, all dated between 1000 and 1400 CE. Find 1 has been dated 1000-1300, Find 2 dated 1100-1300, Find 3 dated 1100-1200 and Find 4 1000-1400.

Let us divide the period 1000-1400 into four segments or “bins” of equal temporal width, resulting in four bins that each cover a section of 100 years. Thus Bin 1 covers the period 1000-1100, Bin 2 covers 1100-1200, etc.

Now let us give each find a “weight” equal to 1, and then split each find between the bins according to how they fall on the temporal axis. So Find 1, dated 1000-1300, would be divided into three sections each weighing one third (0.33) and placed into the appropriate bin. Whereas Find 3, dated 1100-1200, falls completely into just one bin.

Having done this exercise for all finds, it remains to add up the total weights in each bin: this is the aoristic sum for each temporal segment and can now be displayed as, for example, a bar graph. This gives a model suggesting a temporal pattern in which object deposition intensified around the twelfth century.

Aoristic analysis therefore makes it possible to examine broad patterns in the distribution of PAS data across long periods. The below graphs gives, first, the straightforward aoristic sum all finds from the beginning of the Iron Age (800 BC) to the present day and, second, the same data with single finds of coins removed. The patterns indicate how both historical trends and modern biases operate in the PAS database.

In terms of historical trends, the second graph (omitting the coin data) could be said to capture information on the relative volume, growth and decline in production of metalwork objects, which in turn may tentatively inform us of historical demographic and economic developments. Here we see that finds pick up sharply in the early Roman period only to fall in the later. This is a topic of ongoing debate among Roman historians and archaeologists but it maps on periods of economic and demographic decline, or possibly simply on socio-economic shifts that resulted in certain types of metalwork objects becoming less common. There are, for instances, far fewer brooches from the later Roman period.

This trend continues to the Early Middle Ages, picking up during a period of substantial commercial and demographic growth in the Central Middle Ages. The impact of the Black Death (arriving in 1348) appears to leave a substantial trough in the finds data, followed by recovery in the Early Modern period.

This visualisation, however, also indicates some of the biases and limitations of the PAS data. Perhaps the most obvious is the sudden decline in finds dated post 1700, as it is the suggested chronological boundary for finds covered by the Treasure Act 1996 and the PAS is selective in recording the post-medieval material. Secondly, the predominance of certain object types, most notably the great peak of coin finds in the later Roman period, reflects what finders and recorders subjectively consider to be most interesting or worthy of recording. Recent interest in Roman coin finds was further enhanced by a drive by PAS officers in 2007 to encourage metal-detectorists to bring Roman coin finds (particularly ‘grots’ – poorly preserved coins) for recording, which resulted in a surge in this already considerable population in the PASd. A great deal of very valuable numismatic information was thereby recovered, but anyone examining the PAS must bear in mind that regional and temporal patterns in the database are not the result of only historical but also of modern activity.

Thirdly, these aoristic overviews underline the issue of temporal uncertainty in PAS dating. This can illustrated by zooming in on the medieval period.

The above graph gives the aoristic sums of non-numismatic PAS finds by decade. While the overall pattern is (very) roughly in accordance of what might expected based on independent sources of demographic and economic data, the sharp upticks and downticks highlight the course-grained nature of object dating.

The sudden drop in dated PAS finds from the fourteenth century to fifteenth suggests a substantial decline in the production and use of metalwork objects. Recent research by Carenza Lewis on pre-and post-Black Death ceramics evidence in eastern England has shown a similar decline in the volume of pottery around the fourteenth century, which, as she has noted, argues for a more maximalist interpretation of the impact of the Black Death on human activity. In the PASd, however, the whole pattern seems temporally shifted, with the big turndown taking place suddenly some fifty years after arrival of the plague. The continuing increase in the volume of finds through the fourteenth century also fits poorly with what is otherwise known of economic conditions and demographic development even before the Black Death arrived.

This sharp sea change at the year 1400 is likely to be an artifact of how information is entered into the PASd. As was shown in the previous post, 1400 is the most commonly entered todate in the medieval dataset with over 30,000 finds records keyed to it; its popularity may be due to it being conceptualised as a kind of an unofficial boundary date between the ‘Central’ and the ‘Later’ Middle Ages. A similar conceptual bias may account for the relatively low number of finds dated to the second half of the twelfth century, a period of otherwise significant economic expansion when the number of coins in circulation increased by almost tenfold.

The pattern is displayed in an almost exaggerated fashion in the temporal distribution of seal matrices, the seventh most common object type at 3% of the total medieval finds population. The vast majority is dated between 1200 and 1400. While the use of seal matrices, an object linked with literary and economic activity, may indeed have peaked between these dates, the aggregate picture is likely to be heavily skewed by the reliance on conventional boundary dates.

The archaeologically challenging nature of the finds evidence and the way data is structured in the PASd means that such bias is inherent in the database. Being aware of this, however, makes it much easier to work with and around the particular characteristics of the finds records.