The Iceni hypothesis — part 25

Illustration from The Seven Wishes by Alfred Smedberg

While the recording of find spots for British Celtic coins for distribution maps has been encouraged for well over a hundred years, in recent decades their reliability has been questioned. You can read about Robert Van Arsdell's and Warwick Rodwell's thoughts on the matter by following those links.

The original card-file Celtic Coin Index at Oxford contains reliability ratings for the recorded find spots but Carrie and I did not include them on the online version because each keeper had their own system for this and I am not enamored by this sort of system (fuzzy logic) which can have only subjective criteria. For expert systems, I find a binary system to be far more workable. An unreported find spot only reduces the amount of data, but a false find spot corrupts the data-set.

Iceni hoards of silver coins were apparently buried quite a long time after the coins were minted and could even be secondary hoards (a hoard that usually contains the contents of smaller hoards that have been combined in a new hoard, or any hoard that was moved at some point). These hoards have yet to give any information about the original distribution patterns of the three main issues because of circulation mixing over time.

The Iceni hypothesis looks for multiple deposits (a number of coins found within a small area that are not a hoard) that have certain similarities to each other through die-links on fairly unworn coins at a ratio higher than would be expected as a random effect; that are accompanied by other items of value such as metal pellets, ingots, coin blanks, a large number of brooches, etc. that cannot be explained easily as a founders hoard (a stash of metal for use by a smith); where the area might show bone fragment evidence of a large feast, or where there could have been some natural feature thought to be sacred such as a spring or a bog. One of the bonuses about including such discrimination is that they would be highly unlikely to include coins with false find spots. Only if the entire multiple deposit was given a false find spot would the data become corrupted. That has happened when coins have been found at a listed archaeological site and the finder has lied about the location in order to be allowed to keep or sell the coins. One notable case was the reported find of some Cunobeline bronze coins in unworn condition by a person who had done some grounds work at Harlow Temple (where unworn coins are plentiful). When archaeologists examined the reported find spot, there was no evidence of any Iron Age activity in the immediate area. Legally, of course, this is just circumstantial evidence.

No dot on a map will reveal any of these things, nor will databases or spreadsheets that are not set up to allow for very specific queries. The term we are looking for is associated finds. These include entire assemblages of archaeological sites where the target coin was found or other objects and coins found in the immediate vicinity of a stray find and of the same period and not necessarily part of any other archaeological site. I suggest that any appropriate find within half a kilometer would be a range to start with, and that could then be adjusted to whatever radius yields the best results in experiments for that coin location. The density of other finds around one coin find could be quite different from that of another coin find, so there is no universal formula. Thus the database query must be able to handle any radius fed to it. An enhancement would be where its GUI could present targeted terms such as "ingot", "bone" or "excavation report" (wherever a database has such data) highlighted, or in a different colour to the rest of the items listed in the results. A further enhancement would be simple pattern recognition such as having an untargeted object flagged in the lists where more than X number of the same occur in the given radius. For XRF data, "greater than" and "less than" capabilities are essential in order to target alloys within a defined range. Needless to say, the Iceni hypothesis relies not only on closely linked dies within a small geographical area, but die studies to reveal where they are in the chronology of the series. It plots combined spacial/temporal data.

What is done is done. the information we now have is in whatever form it is in and it would be far too expensive to rebuild. It is not impossible to work with, just difficult. We are seeing these day, though, an almost promiscuous amount of reporting of finds swept along by an "every sherd is sacred" fervor. For some categories of common types of finds, this is a huge waste of money from all the way from initial recording to ultimate bandwidth costs. My hoard charts on the left are an almost text book example of how little information is required to see a pattern. The horizontal hoards show essentially the same patterns to each other in the XYZ bar chart on the right of each chart, especially when compared to the vertical hoard differences. It did not matter that much that the hoards had 86, 89, 502 and 1,756 coins and the Fates arranged these hoards especially well by each zone having both a small hoard and a larger one.

So such a project could be undertaken by a group of "elite metal detectorists" or perhaps a club in Norfolk for Iceni coin finds. They would take GPS coordinates for every Iceni coin, note its die combinations, weigh it, photograph it, and upload everything to their database. Finding statistically valid patterns might not take as many finds as they might think.

Tomorrow, making do with what you have.