METAL ANALYSIS OF TWENTY TETRARCHY ARGENTEUS and TWENTY FOLLIS
( A very long post)
I recently had some Tetrarchy coins analysed for their metal content. Two groups of coin were tested. The first group was 20 GENIO POPULI ROMANI post reform billon follis, with assorted levels of surface silvering still visible, from seven mints (Treveri x 5, Heraclea, Alexandria x 8, Londinium, Cyzicus x 2, Thessalonika and Antioch x 2), all struck between 295 and 306 ce. The second was 20 argenteus from nine mints (Treveri x 3, Rome x 7, Alexandria, Siscia x 2, Cyzicus, Thessalonika, Antioch x 3, Aquilea x 1and Nicomedia x 1), all struck between 294 and 302 ce.
I needed the analysis to be non-destructive ( they were my coins), and couldn't get access to the most sophisticated methods that can penetrate right through the metal, so I settled for a spectroscopic analysis which penetrates about 1.3 mm into the surface. The analysis was done professionally, using a properly calibrated DELTA Premium Handheld XRF Analyser, which is primarily used in the field for mine ore assay work, and for the technically minded uses a 4W Rh, Au, or Ta anode X-ray tube as it's excitation source. The machine claims a penetration of up to 2 millimetres into the metal but the technicians running the test said that in reality about 1.4 millimetres is more realistic.
The current literature says that average silver content of a post reform tetrarchy follis between 293 and 303 ce should be about 3%, and that the argenteus had a silver content of over 90%, averaging at about 93 %. The raw data from the testing, and a detailed description of each coin tested is in the appendix. The sample size is less than I would have wished for, but it seems worth posting the results and my tentative analysis, as there doesn't seem to be much around recently on this topic, and the relative values of Tetrarchy Aureus, Argenteus, Follis and Aurelianii is a hotly debated subject .
1) The 20 billon follis (nummus?) GENIO POPULI ROMANI coins.
All the coins tested were confirmed as copper/silver alloy with trace tin and lead, the copper content ranging from a high of 87.25% to a low of 64%, and the silver from 16.82% down to 3.4%. here was also quite a high lead content, from a low of 1.14% to a high of 17.18%.
The mean silver content was 7.94% per coin , and copper 80.22% per coin. There was also a mean of 5.6% lead and 2.49% tin in each coin.
I knew that the results would be of limited use in determining overall silver content for the billon coins because of the different composition due to silvering near the surface, and I am aware of the statistical issues with such a small sample but I was interested to see if the least well preserved coins, that had no remaining visible silver and had been worn down so that the spectroscopic analysis was of metal closer to their core, showed a silver content that approximated to that predicted in the literature. I was also interested in differences between the coins and if I could find any pattern, and expected the follis in general to show comparative silver contents roughly proportionate to the amount of visible silvering. Of the twenty coins tested, five had full silvering and the features on the coin were in EF condition so they had lost little if any of their original content, six had what I have called half silvering left , four had some traces of silvering but not many, and five had no visible silvering at all and were well worn down.
The first hypothesis I tested was that the coins with most or all of their silvering visible would show a higher silver content with this type of analysis than those with less visible silvering. I was surprised to discover that at least with my sample, they did not. The five coins that had the best visible silvering and were in the best condition had silver contents of 4.47%, 3.59%, 13.25%, 9.48%, and 6.12% respectively for an mean average of 7.38%. The six coins that had some but not all their visible surface silvering intact ( the ones I called 'half-silvered') had silver contents of 8.68%, 16.82%, 10.49%, 11.25%, 9.61% and 8% for an average of 10.81%. The four coins with bare traces of silvering left had silver contents of 8.73%, 3.64%, 6.76% and 8.91% for an average of 7.01%. The five 'worst' coins, with no silvering left at all, well-worn and VF at best, had silver contents of 7.07%, 6.49%, 4.65%, 7.44% and 3.4%, for an average of 5.81%.
Some coins with full silvering left on them were showing silver contents half as high as some completely denuded of all visible surface silver. The highest mean average silver content was shown in coins with half their surface visible silver denuded. The mean average silver content was almost the same for fully silvered and barely silvered coins. On balance, there is not much discernable co-relation between the visible silvering on the surface of the coins and their silver content. That content differed widely, but not directly proportionate to visible silvering. The amount of silver in a tetrarchy coin is not related to how 'silvery' it looks, so whatever method was being used to give the surface silvering, it was not a layer, however thin, of pure silver.
Next I looked at the data by date of striking. Some of the coins are difficult to tie down to a specific year, so I have divided them into three broad categories. Seven coins were struck after 294 ce ( obviously, because that is when the reform happened and they started to be produced) but before 300 ce. Seven coins were definitely struck in 300 or 301 ce, and six coins were struck between 300 ce and 307ce ( it is possible some of the last group were struck in 300 or 301 ce). The proportions of fully silvered, half silvered, barely silvered and no longer silvered coins in each of these groups are well distributed (See chart 1), making it unlikely that the results are a reflection of visible or coin quality.
296-299 ce 300-301 ce 300-306 ce
Fully silvered 1 3 1
Half silvered 2 1 3
Trace silvering 2 1 1
No visible silvering 2 2 1
The average silver content in the early group (296-299ce) is 8.42%, with a high of 16.82% and a low of 4.47%. Average copper content is 80.216%. The mid group (300-301ce) has an average silver content of 8.99% with a high of 13.25% and a low of 6.12%. Average copper is 82.49. The late group ( 300-306ce) has an average silver content of 6.15% with a high of 9.61% and a low of 3.4%. Average copper for this group was 79.92.
The copper content is consistent over all the groups, with the mean for all three groups within 3% of each other in copper content at around 80%. The silver content varied by much more. The coins struck in the late period had on average 30% less silver in them than the earlier groups.
There does seem to have been a general debasement of the silver content in the coins struck after 300 ce. Six of the seven coins in the middle group ( 300-301ce) come from one mint, Alexandria, which as I discuss below, may be a more important factor than date of striking, but the debasement applies equally when compared to the earliest group, which had a good spread of mints.
The last arrangement of the data I made was to look at metal content by mint. Many of the mints are represented by just a single sample, but it seemed worthwhile to at least try and separate by geography as well as condition and date of striking. Two mints are well represented, with eight coins from Alexandria and five from Treveri. There are two coins each from Cyzicus and Antioch and one from Londinium, Heraclea and Thessalonika.
Two of the mints ( Treveri and Londinium) are in the Roman west, Maximian's territory rather than Diocletian's. Six of the coins were struck in those two mints, the remaining 14 at mints in the Eastern Empire governed by Diocletian. The six western coins average silver content is 4.74%, the eastern 14 coins average is 9.31% , more than double. The difference cannot be explained by coin quality, amount of visible surface silver, or the dates the coins were struck, as there is a good spread of quality and striking dates in both Western and Eastern mints. ( see charts 2 and 3)
Western mints Eastern mints
Fully silvered 2 3
Half silvered 6
Trace silvering 2 2
No visible silvering 2 3
There is also a good spread of dates in both groups, ( see chart 3)
Western mints Eastern mints
296-299 ce 3 4
300-301 ce 7
300-306 ce 3 3
The five coins with the lowest silver content of the entire sample are all from the Western mints. Two of those are fully silvered EF coins, so Maximian's mints were striking inferior coins to Diocletian's, even though they looked identical. Not all coins had mintmarks on them and there must have been practical problems in having identical looking coins with such large differences in intrinsic value. The literature predicted a lower silver content in Western minted coins, but the difference is startling and occurred from the very beginning of the minting of the new reformed coinage.
The Western mints also seem to be using more tin in their alloys. The mean tin content per coin from Western mints is 3.4%, while in the eastern mints it is only 2.09%. The five coins with the highest tin content were all from Western mints, perhaps lending some indirect evidence for the importance of the Cornish tin mines, which would have been much more accessible to a Western mint than an Eastern one. It would be really interesting to test the tin content of various coins during the earlier ten years of Carausius and Allectus Gallo-British revolt.
The lead content of the coins varies more than most of the other metals. There is a mean content of 5.6% lead per coin, but it varies from a low of 0.97% to a high of 17.18%, and there is no apparent correlation based on coin quality, date of striking, or the part of the Empire the coin was struck in. There is one very clear pattern. The Alexandria mint used much less lead in its alloys than any other mint, Western or Eastern. The mean lead content from the eight Alexandrian coins is 2.11%, while for the other 12 coins from six different mints the mean is 7.97%, close to four times the lead content. No coin from any other mint has a lead content of less than 3.6%, and the seven coins with the lowest lead content ( all below 2.7%) are all in the Alexandrian sample.
Alexandrian coins also had a comparatively high silver content, at 9.14%, well above the mean of the coins from all the other mints of 7.14%. It appears from the data that Alexandria was using a completely different recipe for their billon alloy than anywhere else. They had only just stopped minting their own currency based on a billon tetradrachma and started using Roman denominations. It would be interesting to analyse some Egyptian Tetradrachmas from the late third century ce to see if they also had a very low lead content.
The sole coin from Thessalonika, with about half it's surface silver still visible, has a decent silver percentage at 8.68%, but a completely different alloy mixture with only 66% copper and a lead content of over 17%. Unfortunately a sample of one has no statistical validity but it would be worthwhile to test a few more Thessalonika Follis if I can get my hands on them.
2) The 20 Argenteus coins.
These coins do not have a surface coating, and are supposed to be silver all the way down, so in theory spectroscopy should give a more accurate picture of the metal content than for the follis. The coins analysed range in date from 294 ce to 302 ce. Nine mints are represented. Rome has seven coins, Treveri and Antioch have three each, and Siscia has two. Alexandria, Cyzicus, Thessalonika, Aquilea and Nicomedia are represented by one coin each. Unlike the 20 alloy coins, all the Argenteus show very similar characteristics. The mean silver content is 96.65% which is higher than the literature predicted. There is very little deviation from that norm, whether the coin was struck in 294 or 302 ce, and whatever mint it was truck in. The coin with the least silver content ( struck in Treveri between 295 and 297 ce) has 95.33% silver, and the coin with the highest ( struck in 302 ce at Thessalonika) has 98.69%. Fifteen of the coins fall within one percentage point of the mean. The one Western mint, Treveri still has the two coins with the lowest silver content, and three of the lowest five, but the debasement is far less, at barely 1%.
The other metals in the coins, at least from this surface analysis, are comparatively insignificant, with no other metal in any of the coins at more than 1.6%. There are trace amounts of Tungsten, Iron and copper in all of the coins, and trace amounts of lead, Alumina and sulphur in most of them. For most intents and purposes, the Argenteus was a pure silver bullion coin, minted at a carefully controlled weight of 96 to the Roman pound advertised (at first) on the coins reverse.
If anyone looking at this has any suggestions, criticisms or corrections I would be delighted to hear them. I am very much feeling my way into this. My email is email@example.com
Howard Posner. August 2014.