"History" written by Coins and Metal Detectors, a Cautionary Example from Wiltshire Reviewed

 "evidence thus produced can contribute
enormously to the knowledge of a site"

Richard Henry @richardhenryflo · 23 maj
I recently published an article with one of my former @findsorguk volunteers @SalisburyMuseum where they tested metal detecting methodologies and also rewrote our understanding of the Roman small town of Sorviodunum (Salisbury)

So these karaoke recorder volunteers of the PAStexplorers project (as suspected but never explicitly admitted by the PAS) included artefact hunters? A bit of a conflict of interest, no?
Then there is a reference to a piece of work on the outskirts of Salisbury:

Alix Smith and Richard Henry  2020 'A controlled metal-detecting survey: Revising the Roman numismatic perspective of Sorviodunum' Wiltshire Archaeological and Natural History Magazine, vol. 113 (2020), pp. 190–201 (Henry is the local FLO, Smith lives on the site and she is presumably one of the two detectorists mentioned - and the acknowledgements tell us that the other one was Jonathan Brooks who did not contribute).
Abstract

In 2017, an opportunity arose to carry out a systematic metal-detecting survey in the environs of the Roman town of Sorviodunum. Whilst controlled excavations and geophysical investigations have previously taken place both inside the hillfort of Old Sarum and the area immediately surrounding it, this is the first time, to the authors’ knowledge,that any metal-detecting survey in the vicinity has been carried out. This study also provided an opportunity to test the methodological approach of a 10 per cent sample versus a 100 per cent metal-detecting survey. Discussion of the coins recovered from Sorviodunum provides for (sic) new insights into the town from the perspective of numismatic evidence. In addition to Roman coins, metal finds of medieval and other periods were also recorded, and these are argued to contribute to our understanding of Roman Sorviodunum, as well as later occupation at Stratford sub Castle.

Note: 'an opportunity arose' and 'to the authors' knowledge'.

The first thing one notes is that the actual area of the site examined is not in any way explicitly indicated or properly described in the publication, it provides (not 'for' !)  loosely 'floating data'. It looks like the field in question is that centred on 51° 5'17.16"N   1°48'31.91"W. The administrative district of the site is not given, neither are the dates of the survey (needed for searching for information on the PAS database). The scales of the maps are idiotic and seem deliberately designed to obscure information rather than inform.

One of the figures - two dot distribution maps, of course

The second is that the editors of Wiltshire Archaeological and Natural History Magazine seem to have no idea about how to edit. The first three pages of this text is a solid block of narrativisation (called 'background') that could have been presented in four paragraphs with references to the literature. This is just padding and 'bibliography bulking'.  After the waffle we get (p. 192) to the research aims. This too is equally waffly. The searchers wanted to "determine the paucity or otherwise of artefacts within the site", in other words find out "how productive" the area next to a scheduled site was, a typically detectorist concern. Anyway, they say:

When practised responsibly following the Code of Practice (2017), metal-detecting outside the areas of the scheduled ancient monument [...] can cover a large area reasonably quickly and produce noteworthy results. If carried out with care and precision, evidence thus produced can contribute enormously to the knowledge of a site. To put this into context, before this study the total number of coins discovered at Sorviodunum was 91; by contrast, two detectorists undertaking a survey over a total period of 16 days, produced 133 coins.

Super eh? How could it be otherwise? But merely "recovering more artefacts" is not a valid research aim. That's just an extension of the  "wottalotta finds we got" PAS-mantra. They also reckon along with the coineys that "moneta uber alles" and merely getting 224 coins to put into "Reece periods" and make a histogram is 'doing archaeology' and  "can contribute enormously to the knowledge of a site".

By the way, note that 133 coins alone in 16 days is the equivalent of 130 recordable items each in a month - as opposed to Heritage Action's 30.25 per year.

To return to Brooks and Smith's "research aims":

The final aim was to test field methods for conducting metal-detecting surveys [...] Assessment of these results would, in turn, help demonstrate the value of metal-detecting as a complementary component of any archaeological examination of a site, adding substantially to the knowledge and evidence obtained from other elements such as non-invasive surveys and excavation. 

So rediscovering the wheel. Except they did not do any of those other elements, merely collected as many COINS as they could from an area of about 160 x 255 metres ( so that's 4 ha off the edge of a site that has an area of 17ha). Hmmm.

Methodology (pp 192-4) is more waffle (a lot of this should have been in the 'background') They followed the 'Code' (but make no mention of 'Our Portable Past') and say they had 'photographic evidence from the owners which showed the site had been ploughed in the 1970s'. They could have used Google Earth, where you can see it was ploughed in 2017. Padding.

Shockingly:

Due to previous ploughing and agricultural use, the soil contained a large quantity of ferrous objects within the detectable depth, giving a high level of ‘background’ noise across the entire survey area [...] both detectors were nevertheless set to discriminate against ferrous objects, as is common in most metal-detecting surveys [refs]

Note that here the notion of 'background noise' is "not-coin noise" - they are aware that they are leaving evidence of Roman use of the site in the soil, but all they want are the goodies for their coin loss histograms. On p. 293:

finds were retained and recorded by being placed in individually numbered finds-bags, with an accurate GPS position of the finds recorded on the bag, using a Garmin Etrex 20 which is accurate to ±2m. All recovered finds were shown to the Finds Liaison Officer and all finds meeting the remit of the PAS (i.e. dating prior to AD 1700) were recorded onto the PAS database (www.finds. org.uk/database, search ‘Sorviodunum survey’).

Go on, search. That information is hidden away under 'Other reference' in the PAS records, as shown by the anonymous record ('created 2 years ago, updated about one year ago') of the sole find that is (for some reason, its a bog-standard two standards Gloria Exercitus) presented in the article WILT-9D9B89. But that data-field seems not to be publicly searchable by the search engine provided. Padding.

Note the glaring fact that no mention is made in the article of where the objects recovered during this project were deposited. But then, the PAS record states 'Subsequent action after recording: Returned to finder'. This is important in the light of the declaration that an upcoming course of the so-called Institute of Detectorists on doing precisely such surveys "is about how to use the metal detector as an archaeological tool to contribute knowledge for public benefit - nothing to do with any "collection-driven exploitation". Here it looks like the detectorists pocketed the finds from this "survey". In other words, this article refers to a bout of collection-driven exploitation of a known site, right next to a scheduled site that is masquerading as an archaeological survey.

As for what it 'tells' us:

The distribution of Roman coins can also be compared against that of medieval and postmedieval finds (Figure 5). Roman coins have a clear focus towards the south of the field with only 17 coins located in the northern half, whilst the medieval and post-medieval material distribution is greatest to the north and is more evenly spread. The distribution of Roman coins potentially indicates that the extents (sic) of the settlement are greater than previously thought

Since along the west side of the field there is a modern hamlet, the later finds will more likely relate to the earlier stages of its development, and in particular to material carried out as manure from it . But the authors are more interested in making a histogram of the coins (pp. 195-198). They try to assess (classify!) the site, ignoring any archaeological evidence that was not a coin. There was pottery, probably slag, ferrous objects such as mounts, fittings and tools in that ploughsoil - some of it (if we believe detectorist lore) being eroded by being in that topsoil. Yet the opportunity to study it was ignored in favour of the COINS because they are (a) collectable and (b) have pictures and writing on them so they don't take a lot of effort to analyse and understand, unlike the rest of the archaeological evidence from a site where an understanding of context of deposition and discovery are needed more than an x-marks-the-spot-accurate-to-two-metres.

I'd have more sympathy for their notion that they have "discovered" that the extent of the settlement was greater at this point than previously thought if it was based on similar gridded fieldwalking of pottery, tile, nails and other archaeological evidence, and not based on the selective pocketing of a single class of evidence, ignoring its relationship with others. All we have is a pattern of dots.

As for the alteration of the coin histogram from this site that seems to so excite these writers, the histograms that 'show' this are not really very convincing. Both the peak of coins of Reece's period 13 (260-275) of the pre-survey sample and those of period 17 (330-348) of the survey results as well as that 'Flavian peak' (p. 192) into which much was read, could merely be due to localised effects. For example in the infield to the north of the settlement may have been a field ditch or largish quarry pit with a shallow upper fill that had been levelled by the dumping of domestic rubbish  in the 330s and 340s that was later ploughed out and that would explain why in this modern field, the coin assemblage differs from other areas of the site. In order to interpret these 'data', more information is needed about both context of deposition of these items and the context of discovery. It seems to me that apart from these peaks, the rest are within the range that could be explained as random sampling effects.

I think most of the authors' conclusions (pp 198-9) can be challenged, but this post is too long anyway and I can't be bothered. What is not coin-based is trite stuff (about waterways and all that) that is not particularly innovative, while other bits just go into text-driven speculation. The authors seem to think their (selective) 'evidence' from a single field, instead of reflecting what was happening in  a small part of the area just adjacent to part of a site apparently a kilometre long, can be a pars pro toto representation of the site's history that they then go even further and extrapolate into the regional context, but without any discussion (or even apparent awareness) of whether or not that is justifiable. Bonkers. 

The conclusions of this article are bulked out by including bits that should be in the 'recommendations for future metal-detecting surveys' but then both needed editing to remove the repetition. I'd add that any detectorist intending to attempt such a survey first needs to learn how to properly write up the results to avoid a train wreck like this paper - even if an archaeologist is involved in its drafting.

Two tekkies searched a field next to a Roman road, without saying which field, and without saying when, without saying what they eventually did with the material. They tell us about the coins, but nothing about the other finds (which are difficult to identify - if they are there at all - in the current form of the PAS database). They draw some 'conclusions' that are impossible to verify due to inaccessibility of the details of the 'survey'  or the material recovered. And then somebody writes, with the aid of the FLO, a paper 11 pages long that probably could have covered the same material in five if the excess narrativisation and padding were cut out. Yes the metal detector can be a useful archaeological tool, but only when used as such, and not a means to provide loose items to collect and witter about.