Almost the last academic thing I did last year before breaking for Christmas was the first two sessions of a project that is now nearly finished, All That Glitters, announced here so long ago. It was so long ago that it might be worth reintroducing. Basically, by great happenstance one of my predecessors in taking care of the coin collection at the Barber Institute had run into two people at a conference who were presenting about the ability of the X-ray fluorescence (XRF) equipment their company, Bruker Industries Ltd., produced to do metal analyses. That predecessor, Rebecca Darley, also knew Robert Bracey at the British Museum, who had done work with XRF on the coinage of the Kushan Empire, and now I was in charge of the Barber’s coins. None of us are archæometallurgists and our knowledge of archæology and numismatics didn’t necessarily combine with the metals analysis expertise upon which we could call so as to somehow make a composite archæometallurgist out of us all, but Robert could borrow the British Museum’s handheld XRF scanner, Bruker had a new machine they wanted to test against real problems, the Barber is on the same campus as several much larger and more expensive machines also supplied by Bruker some of whose host departments proved happy to let us use them and to tell us about them; a project seemed obviously to exist in potentio and our task was to work out how to make it tell us something useful.
It’s too early to say whether we achieved that, pending actual publication of our results and conclusions which we are even now working on writing up, but right from the beginning it was clear to us that before we could have any historical conclusions we would have to have methodological ones. This is because there are so many potential problems with XRF analysis, problems which not all publications using it consider, and when they do, consider most often for silver alloys rather than the high-purity gold we would be testing, that the first priority had to be not to find things out, but rather to find out what we could find out.1 Accordingly, our pilot experiments were designed to evaluate the machinery more than the coinage, and we started on the 3rd December 2014 when Robert brought the British Museum’s handheld scanner up from the British Museum, along with its calibration standard, and Mike Dobby and Colin Slater from Bruker brought in their new M1ORA energy-dispersive scanner and we put some coins through their x-rays.
We had chosen a test set of ten coins running from Anastasius I (491-518) to Constantine VI (785-797), with two coins for each ruler from the same mint and if possible, the same workshop of the mint (if that’s what officinae were), hoping that whatever results we might obtain would thus be internally secure against outlier coins from a bad day at the mint or similar.2 All of those got zapped on both sides with both handheld and the microwave-like M1ORA. Both of these use a spot beam, but with the M1ORA it’s possible to target it precisely, so we aimed for flat surfaces wherever possible.
The initial results of this made the handheld device look like much the poorer sibling, as its readings were extremely variable. Robert worked out, however, that this was related to how much of its expected sample it had been able to observe—this in turn probably down to surface relief but already we were into unknowns—and when the figures were all normalised to a notional 100% of sample they came out much more like what the M1ORA was seeing. The M1ORA was able to dump its readings straight into a laptop equipped with suitable software, and did this levelling-up in that software, so made things immediately clearer for us by automating that step, solely an issue of configuration but we still needed to be aware of it. I keep stressing variables and difficulties because I don’t want to imply that we were getting actual true results, but that said, once they were both talking to us in the same framework the message of the machines was pretty consistent: all these coins were being analysed as very high-purity gold, 97% or more, which is astonishingly high for any pre-modern metallurgy. However, other than silver and copper, which occurred in about the proportions one would expect (i. e. not very much and even less), the only other element that was consistently detectable was iron. That was in part a factor of what we had asked Mike and Colin to make the machinery look for but if you have to do that at all, there is obviously a restriction inherent in your question-setting…
The very high fineness was sort of what we might have expected, anyway, as it is roughly what earlier analyses of Byzantine gold coins by a number of methods have suggested.3 It still made me uncomfortable for its lack of variety, however, and so since we had run-time left in the day I started hauling other things out of the cabinets. These included two Persian gold dinars, of Shahs Shapur II (309-379) and Varhran IV (388-399), and the above piece which is pretending to be something like the former, as well as an Arab-Byzantine solidus from Carthage and some more Byzantine pieces. Mainly I just wanted to be sure that the machinery actually would report lower gold finenesses, and so it duly did, with the Persian pieces both lower (but not by the same amount) and the Sindh piece even less fine (which, to be honest, was already apparent in its colour, but that was why I’d chosen it). The Byzantine stuff, and the coins imitating that, remained high in the machinery’s estimation.
This bit was fairly unsystematic sampling, but it did give us some reason to believe that the machinery was observing something consistent with our expectations, and which therefore fitted into existing understandings of the early Byzantine gold coinage. That is circular, though, obviously! This and the likely effect on the readings of differences between the surface of the coins and their cores, because of both manufacturing factors and subsequent environmental exposure, meant that we weren’t willing (and still aren’t) to say that these figures are actually how fine those coins were. We also weren’t seeing a range of trace elements which we had expected on the basis of older work, and which might have suggested things about changes in metal supply and treatment that would potentially be historical evidence.4 So, while these machines might serve other people’s purposes, we ourselves were going to need some bigger kit. And that would be Experiment 2, about which I shall write in a couple of posts’ time. In the meantime, however, here is some shiny metallic blogging for the Christmas season and I wish you all a happy holiday!
1. Part of our problem was that so much of the literature about these problems was old enough to relate, potentially, only to a much more primitive incarnation of the technique. Nonetheless, Michael F. Hendy & J. A. Charles, “The Production Techniques, Silver Content and Circulation History of the Twelfth-Century Byzantine Silver Trachy” in Archaeometry Vol. 12 (Oxford 1970), pp. 13-21, William A. Oddy, “The Analysis of Gold Coins—A Comparison of Results Obtained by Non-Destructive Methods”, ibid. Vol. 14 (1972), pp. 109-117 and J. Tate, “Some Problems in Analysing Museum Material by Nondestructive Surface Sensitive Techniques” in Nuclear Instruments and Methods in Physics Research Part B Vol. 14 (Amsterdam 1986), pp. 20-23, all suggest that differences should be observable between surfaces and cores of coins and between methods that measure only the surface and those that measure total composition, and L. Beck, S. Bosonnet, S. Réveillon, D. Eliot & F. Pilon, “Silver surface enrichment of silver–copper alloys: a limitation for the analysis of ancient silver coins by surface techniques”, ibid. Part B Vol. 226 (2004), pp. 153-162, DOI: 10.1016/j.nimb.2004.06.044 and Vasiliki Kantarelou, Francisco José Ager, Despoina Eugenidou, Francisca Chaves, Alexandros Andreou, Elena Kontou, Niki Katsikosta, Miguel Angel Respaldiza, Patrizia Serafin, Dimosthenis Sokaras, Charalambos Zarkadas, Kyriaki Polikreti & Andreas Germanos Karydas, “X-ray Fluorescence analytical criteria to assess the fineness of ancient silver coins: application on Ptolemaic coinage” in Spectrochimica Acta Part B Vol. 66 (Amsterdam 2011), pp. 681-690, DOI: 10.1016/j.sab.2011.08.001, give some explanations of why that should be so. (I have to thank Dr Eleanor Blakelock for some of these and several other useful references.) All of these except Oddy and Tate were working with silver alloyed with base metal, however, and so another of the problems we have is in knowing how far the same applies to gold and if it does, whether if alloyed with base metals only or also with noble metals such as we expected to see. And the mess only gets worse from there…
2. Barber Institute of Fine Arts B0005 & B0006 (Anastasius I, Constantinople, officinae Eta and Iota), B2761 & B2762 (Heraclius, Constantinople, both officina Eta), B4384 & B4385 (first reign of Justinian II, Constantinople, former marked Theta), B4464 & B4465 (second reign of Justinian II, Constantinople, no control marks), B4598 & B4599 (Constantine VI and Eirini, Constantinople, no control marks).
3. Those earlier analyses being principally those gathered and conducted in Cécile Morrisson, Jean-Noël Barrandon & Jacques Poirier, “La monnaie d’or byzantine à Constantinople : purification et modes d’altérations (491-1354)” in Morrisson, Claude Brenot, Jean-Pierre Callu, Barrandon, Poirier & Robert Halleux, L’or monnayé I : purification et altérations de Rome à Byzance, Cahiers Ernest Babelon 2 (Paris 1985), pp. 113-187.
4. The kind of conclusions, indeed, that were coming out of ibid. and another study there, Jean-Pierre Callu, Claude Brenot, Jean-Noël Barrandon and Jacques Poirier, ‘”Aureus obryziacus”‘, ibid. pp. 81-111, albeit with a rather more variable sample of evidence!