Monday, 5 August 2019

Big is Good, Big is Big Enough (We Checked it)?

Dead wood needs pruning
out of the UK heritage debate
Oh, my... In answer to my comments whether what is in the PAS database is representative of what the archaeological record of the sites exploited by collectors had contained before they stuck their spades in them, the hapless FLO from Salisbury is in complete denial:
PASWiltshire@PasWiltshire·8 minW odpowiedzi do @PortantIssues i @findsorguk
The shear (sic!) size of the database will average out any anomalies caused by local examples of selective reporting. That and the PAS database does not exist in isolation, so we can test it against other datasets. The database is representative.
Ouch.  That self-evidently is a load of the typical sort of codswallop that is doled out by the 'support the PAS, save-our-jobs, pat-a-detectorist-on-the-back' lot.

First of all, collecting is not archaeology. Neither are artefact (portable antiquities) collectors interested in creating a record of material culture.  What passes muster as Ancient Egyptian, Mesoamerican, Ancient Greek, or for that matter Roman 'antiquities', as sold by every single dealer and auction house and acquired by every single collector, bears scant resemblance to what is actually recovered by systematic fieldwork from the average site in Egypt, Mexico, Greece and Turkey, or for that matter any Roman site in Britain, Spain or France. There are other problems related to 'Bazaar archaeology', but it is clear (as the stamp collection analogy I used in the earlier post), collectors collections are not made with the aim of documenting archaeological reality. To pretend otherwise is just blind fanaticism.

So, looking at the PAS database we see that the boasted "1,428,744 objects within 918,768 records" right on the front page, can be matched with the information that 705,402 of those objects (in 442,405 records) are just coins (!). That's right, as many as half of the objects recorded are coins. That is not representative of the proportions between total artefacts deposited/ coin losses on any site I have worked on. Pottery however is present in enormous quantities on most of them, but the PAS database has only 161,955 sherds reported and recorded. And tile, tile is not really collectable, so just 18,878 fragments are  recorded there from the whole of England and Wales. Yet some of the sites exploited to produce the coins and other collectable metal 'partifacts' will have had zones that had been strewn with copious fragments of both. Likewise, when it comes to metal artefacts, every single timber structure, and the roofing (doors, shutters, fittings) of the masonry ones on those same sites will have needed nails to hold them together, and the PAS database contains .... 18,733 complete and fragmentary examples of them. Hardly 'representative' of anything when I've worked on single sites that have produced something like half that number. I could go on, but it would be labouring a point that should be perfectly obvious. I did a post about this 18 months ago that of course the Salisbury Museum guy will not be aware of. I have written a number of posts here on the reliability of the information in the PAS database, but the only response they elicited was abuse and an attempt (led by a PAS FLO) to get me thrown off the Rescue Facebook Group where a link to my texts was posted by the moderator - who then resigned as a result of the ruckus. I have written a text on the 'representativeness' of the "Viking" finds in the PAS database as a basis for study, published a while ago, nobody from PAS seems to have read it. It is easy to get the feeling that one is banging one's head against a brick wall when you are dealing with unreflexive UK jobsworths in denial.

These are not 'local' effects as the beleaguered and confused FLO intimates, these are biases that will occur in a record created as a result of the mechanisms used to attempt to use collections as an archaeological source, as the PAS do, and the archaeologists who swear by its use as 'evidence'.  Its like the teenager that relies on porn magazines to tell him 'what women are like'. Not at all reliable, if you think about it. 

But then are supporters thinking about it? We have a "never mind the quality, feel the width" argument above. The  ("WottaLottaFinds We Got") FLO reckons, "the sheer size of the database will will average out any anomalies". No it will not, since the anomalies are a factor of the manner of its compilation and this is a constant inherent in the nature of the record itself, the increasing size will just compound them. Bad data in, always bad data out, no matter how many of them there are. 
The PAS database does not exist in isolation, so we can test it against other datasets. The database is representative.
Databases of what, representing what? I tried to do this analysis a few years back, asking archaeologists on a professional forum gathering fieldworkers to kindly supply me with some links to online catalogues of fieldwork archives from excavations and fieldwalking of sites of different periods and in different parts of the country. Two units replied and sent some information, but it was not a big sample. The rest (archaeologists) said "*** off Mr Barford, you want to use that information against metal detectorists, don't you?". Well, I wanted to use that information to actually check some of the claims made about artefact hunting. It seems there were vested interests on the BAJR forum too in not having those claims subject to questioning.

No comments:

Creative Commons License
Ten utwór jest dostępny na licencji Creative Commons Uznanie autorstwa-Bez utworów zależnych 3.0 Unported.