Creating a Good Impression: PAS Statistics are not what they seem (2)

Continued from  Creating a Good Impression: PAS Statistics are not what they seem (1)

What happens to PAS statistics concerning the success of voluntary recording if we remove the records not made as a result of voluntary recording by artefact hunters? There are at least three groups of these extraneous data which are included alongside the figures (V) for voluntary reporting by artefact hunters:

A) chance finds made by members of the public who are not engaged in collection-driven exploitation of the archaeological record (my mum with a samian sherd from her rose garden, Bob the Builder who finds a Roman coin in upcast of a storm water drain trench on a building site). There is in fact no easy way to separate  these finds out from the rest except manually, as the current form of the advanced search of the database does not filter 'discovery method'.

B) Material coming from two archaeologist-compiled existing databases that contain information that does not all come from public finds. In late March 2010, the  Celtic Coin Index was amalgamated with the PAS database (and there is no filter in 'advanced search' to exclude them from the general records). The CCI database added some 37925 records to the PAS database, boosting overnight the number of Celtic coins recorded there from a few thousand to nearly 40000.

The Iron Age and Roman Coins of Wales Database was a one-year research project (2003-04) of the Cardiff School of History and Archaeology run by Dr Peter Guest (and Research Assistant Nick Wells). It gathered onto a database information on published and unpublished Welsh finds of Iron Age and Roman coins (excavated assemblages, hoards, casual single finds and indeterminate groups of coins, among them those recovered by metal detector or field walking and reported through the PAS). In the end, details of 52,838 coins (the vast majority dating to the Roman period) from 1,172 find spots were collected. The information was published as a detailed corpus. In March 2010, the dataset of this project was amalgamated with the Portable Antiquities Scheme database ‘significantly increasing the number of coins available for study’ and again boosting PAS record numbers. It is unknown here too whether there had been significant overlap between these two databases. Again there is no filter to remove these results from the general statistics.

To summarise, in 2010 the PAS database was stealthily increased by 39097 records of 90763 objects by adding these two databases - duplicating information available at the time elsewhere.

C) Treasure finds, reported by law.  Fortunately the 'advanced search' does contain the possibility of filtering out Treasure finds included in the PAS database. There is a button near the top of the search terms form. Pressing it gives some pretty surprising results. It turns out that there are nearly a quarter of a million additional finds on the database from this source (234,487 objects in 13599 records). There are 980 multiple-object hoards represented (one with 52504 coins in it) You can sort the results by the time they were added to the record. While a few Treasure items had been incidentally added to the database from June 1998, it seems there was a change in policy and a concerted effort to add them from August/September 2007. (Statistical analysis of the database for Saturday 1st August 1998 until Wednesday 1st August 2007 Total objects recorded: 285810 Total records: 187454).

To summarise, from August 2007 to now the PAS database has been openly increased by 13599 records of 234487 objects by adding treasure finds - duplicating information available at the time elsewhere in the Treasure reports prepared according to the Treasure Act art. 12.

So let's fix it for the PAS. Their blurb today proudly reads "1,438,864 objects within 923,127 records". But that result is A+B+C+V. if we want to know V, we have to extract B and C (but can't do much about A). The result is:

"([1,438,864 - 325,250] objects within [923,127 - 52,696] records") =

1,113,614 objects within 670,431 records.

How does that 'V' look in the twenty-year perspective and knowing that there are probably 27000 tekkies out there? Not very impressive. Those revised figures presented on a twenty-year scale come out at 43522 records annually, or 55681 objects. Far from the massive success claimed. It's actually less than two objects each (and remember, these figures are A+B).