Tag Archives: science!

All that Glitters 7: the slight return

With due respect thus paid to events that have overtaken us, I return to my sort-of-scheduled programming here at A Corner of Tenth-Century Europe and also to the subject of Byzantine coinage, from which it seems I will probably never entirely escape. And why indeed would one want to? But I can bring one thing to a close with this post, which is my reports on the experiments that my collaborators and I did on the All That Glitters project analysing Byzantine gold coins by X-ray Fluorescence of which you have by now heard so much, at least until we actually publish properly on it. So here is the last post on the theme for the time being.1

Bruker S8 TIGER XRF analyser open for business

And therefore last chance to re-use this photograph of the S8TIGER WD-XRF machine in the Department of Chemistry at the University of Birmingham!

I’ve done so many of these posts now that summarising the experiments in any detail would be quite difficult as well as tedious, so for the purposes of this post, let’s just say, we’d gone into the project hoping to say something about changes in the trace elements present in the coins that might tell us something about changing metal sources and minting practices, and we’d found that for two reasons that wasn’t really possible, although we did still find some stuff out doing it. The first and more historical of those reasons why not was simply the nature of the Byzantine tax system, which persistently called in coins from across the Empire, melted them down and redistributed their metal centrally to the mints for striking, thus ineluctably mixing all the different mints’ practices together; the only place we could really see compositional difference was provinces that were falling off the Empire, and even then interaction was usually sufficient to keep things mixed up. But the other reason was that the detection machinery, be it never so sophisticated, couldn’t really tell us what we wanted to know, and that also for two reasons, one being because of invisible surface deposits from the soil that we couldn’t properly see through with the X-rays (and couldn’t safely remove very effectively), and the other being simple and frustrating variation in results.

At what had been supposed to be the end of the project, therefore, because of the various constraints and inefficiencies of getting the coins safely to the analysis machinery and back, we had unspent money left in our grant budget, and so I thought—I think this was me, but if not, I’m sorry to either Rebecca Darley or Maria Vrij for stealing their credit here—that one useful thing we could do with half a day was get some kind of baseline figure for how bad that variability was. So on 16th February 2016 we did a very simple experiment. We took one coin, put it in a sample cup and then without touching it, moving it or changing it in any way beyond what the automatic handling in the spectrometer put it through, ran it through exactly the same test five times, and then turned it over in the cup and did that again for the reverse side, giving us ten runs on the same object in which there was literally nothing more that we could have done to reduce variation.

Apparently I wasn’t taking the security photographs that day, so I cannot show you that coin in its sample cup, but here it is in shiny catalogue image; it is a gold-ish tremissis of Emperors Leo III and Constantine V probably struck in Sicily or Italy, as we demonstrated in the last one of these posts, between 717 and 741, Barber Institute of Fine Arts B4542

The results were not encouraging. Admittedly, in terms of gold content they were not too far apart, ranging from 85.26% to 87.21%, an error margin of only about 3%, but even that is 2% raw difference in apparent metal content. People have founded theories about currency alteration on the basis of disparities like that, so if you’re getting them between measurements of a single coin that’s a problem. But the less present elements had similar amounts of variation: silver 7.63% to 8.84%, copper 1.78% to 2.39%, iron 0.15% to 1.04%, aluminium 0.38% to 0.67%, magnesium 0.13% to 0.67%, and then a host of other elements that one didn’t even see in some or most tests. Again, these margins may not seem like a lot, less than a per cent in some cases, but those less-than-per-cent margins are in some cases more than the total percentage of a metal in question, which meant that the error margins we were seeing were mathematically huge, in the order of 300% to 500%. Basically, no respectable scientist would trust such figures, because they could have no confidence that they would be able to reproduce them, and fair enough, because here we were trying to do so and more or less failing.

So what did all this tell us? One gloomy conclusion, that for all we’d hoped to find differently, XRF probably still isn’t a workable way of analysing trace elements in coins, we had more or less already reached, but this let us actually put numbers on why not, which is worth something. I’ve since looked at quite a lot of papers using XRF analyses on coins, and I’ve found only one that used an average of several experiments as their working figure, and that was from 1983 (and was by none other than the late lamented Michael Metcalf, and he was dealing with variations of over 20% depending on what he’d done to Philip Grierson’s coins to get those results, which we know because he actually said so in his write-up).2 He wasn’t even a metallurgist! And he presumably also wasn’t paying for machine time, which is the basic reason that I guess people don’t otherwise do this. But it is, one might say, a little embarrassing for the subdiscipline. Still, I’m not sure that even an average figure from our tests would be very safe to use. How many tests would one need to run on each object to make safe a 300% error margin? What if one of those tests increased that margin? In general, I think that even the best XRF machinery we can get just can’t give accurate figures for small-percentage composition elements, even if it probably still has some application for the big-ticket components. It’s not the conclusion we’d aimed for but when we can get anyone to publish such a negative finding it may not be without value.3 And thus endeth, for now, the sequence.


1. Of course, we have actually published on the project a tiny bit, in the form of Rebecca Darley, “All that glitters…: the Byzantine gold solidus, c. 300-1092″, in Maria Caccamo Caltabiano (ed.), XV International Numismatic Congress, Taormina 2015: Proceedings (Rome 2017), II, pp. 982-985, but that was actually written and given before we’d done these final experiments.

2. That being D. M. Metcalf, “Interpreting the Alloy of the Merovingian Silver Coinage” in C. N. L. Brooke, B. H. I. H. Stewart, J. G. Pollard and T. R. Volk (eds), Studies in Numismatic Method Presented to Philip Grierson (Cambridge, 1983), pp. 113–126, out of 16 studies I know of from 1966 to 2019; I’m sure there are more, though, and if you feel I’ve missed an important one it would be really useful to know!

3. The main reason that we haven’t yet done more on publication of these experiments, even three years down the line, I should explain, is professional mobility. Even in the course of the project, every single project member changed jobs and only two of them even stayed within the same company/institution. Since then several more of us have moved again. Of course, our new employers all hired us for our own individual qualities and while some of them might like us to do this kind of collaborative inter-disciplinary research, they would prefer to have been part of it, so that only those of us who remain in Birmingham have any immediate professional interest in making this part of our workload. We will publish something on it, because we spent money on the assurance that we would, but it will be when one of us needs it more than whatever else we’re supposed to be working on, and the path to that isn’t yet clear.

All That Glitters, Experiment 6 and final

So, as just described, almost my first academic action of 2016 – for that is how far in the past we are for this post – was to head back to Birmingham, freshly remobilised, to pursue what was supposed to be the last run of experiments in the All That Glitters project of which I have now told you so much. Since the last one of those posts was only a short while ago, I’ll not reprise the project plan beyond saying it was to try and find out what was in Byzantine gold coins besides gold using X-ray fluorescence spectrometry and we were finding it difficult to get beyond what was on Byzantine gold coins. Now, read on!

Gold solidus of Emperor Constantine VI and Empress Eirini, Barber Institute of Fine Arts B4599

Gold solidus of Emperor Constantine VI and Empress Eirini, Barber Institute of Fine Arts B4599, in XRF analysis sample cup

Since we now more or less had a working method established, if it could be called that (since it didn’t really work), we had decided that our original research goal, of spotting changes in the trace elements in the centrally-minted gold coinage of the Byzantine Empire, was beyond the technology, and we needed to work out what else we could do with the remaining machine time. At first we’d thought we wouldn’t have enough, now we had more than we knew what to do with… But the most obvious thing seemed to be to broaden our sample as much as possible. So, we selected more of the Barber Institute’s coins, taken from imperial reigns we hadn’t covered, extra denominations from ones we had and sets from other mints than Constantinople that we could compare to coins of the same emperors there, and we took them all over to University of Birmingham’s School of Chemistry over a period of four days, where we were as usual excellently looked after as far as they could manage, and we subjected them to analysis. In all of this we were hampered by the fact that results were basically hard to reproduce; in fact, this became so frustrating that when it became clear that we still had a dribble of machine time budget left at the end of these experiments, we set up one more to address that problem specifically, and that will be the last of these posts when I get so far. But for this one I can basically give you only a very simplified set of findings, some of which might address real questions if only we could trust our results, and then gently suggest that even what we did get might justify some careful conclusions, though they might not really have justified the labour. So: some late antique numismatic questions, as answered by the S8TIGER in January 2016!

Bruker S8 TIGER XRF analyser open for business

Our tool of analysis, the S8TIGER WD-XRF machine, ready for action

Our first question in this set of tests was about fractional denominations. Though the primary imperial gold coin was the famous solidus, the “dollar of the Middle Ages”, there were also small numbers of halves (semisses) and thirds (tremisses) struck, with slightly different designs.1 Were these actually struck from the same metal as the solidi? Our results, shaky as they were, suggested that the answer was broadly ‘yes’, at least at Constantinople and, as far as we could test, Carthage. The only place where we picked up any reasonably substantial difference was Syracuse, in Sicily, but we’ll come back to that…

Gold solidus of Emperor Maurice struck at Ravenna 582-602, Barber Institute of Fine Arts B2390

Gold solidus of Emperor Maurice struck at Ravenna 582-602, Barber Institute of Fine Arts B2390

Gold tremissis of Emperor Maurice struck at Ravenna 582-602, Barber Institute of Fine Arts B2391

Gold tremissis of Emperor Maurice struck at Ravenna 582-602, Barber Institute of Fine Arts B2391; note the different design

The other thing we were hoping to establish in this set of tests was variations between mints. I admit that I was cynical about this; as I think I’ve said before, it had sort of become clear that almost all the elements were shared, and that this made sense in a world where imperial coin was being sucked into Constantinople in tax from right across the Mediterranean each year, melted down and then returned to the world as new coins; the recycling should have mixed everything together over time.2 So the only place we had a hope of seeing such variation was in places where that centralisation was breaking down, and in fact, from very early on it had become clear that late coins of Syracuse were gold-poorer than their Constantinpolitan contemporaries, to the extent where the one of us who hadn’t loaded a coin, so didn’t know what it was, could still tell if it was a Syracusan one from its results.

Graph of gold content over time for Byzantine mints of Constantinople and Syracuse

A very rough Excel-generated graph of coins’ gold content over time for the mints of Constantinople and Syracuse, by your humble author

Some of that impurity was visible by eye, indeed, but we could pick it up from before that. Indeed, there are one or two problem cases where mint attribution is uncertain for such coins, and for one of those at least, we were pretty sure we could now partly answer the question.3

Powerpoint slide showing three tremisses of Emperors Leo III and Constantine V from different mints

This is a slide I’ve grabbed from a presentation I will come to tell you more about in Princeton, and it’s the one on the right that’s the undecided case; but its metal content is much more Italian than Constantinopolitan, and we might get further yet

Why Syracuse was allowed to run its coins differently is a separate question, since as far as we know it was still paying tax to the centre and its coins must have been detectably poorer there too, but maybe what we’re seeing here is actually proof that it didn’t pay tax; its small change, too, seems to have been treated in such a way as to restrict its circulation, and Rebecca Darley (I can take no credit for this thought) wondered therefore if Sicily was persistent suffering a currency drain to the East that these measures were meant to stop by deprecating the exchangeability of Sicilian money.4 It might have helped!

Scatter plot of silver content versus copper content of Italian-attributed Middle Byzantine coins

Scatter plot of silver content versus copper content of Italian-attributed Middle Byzantine coins, which is probably Maria Vrij’s work, though I don’t remember; it was certainly her idea to do it

But as it turned out, we could get one step further with such distinctions. One of the other enigmas about coinage in Byzantine Italy is that we’re not totally sure which issues belong at which mints. Syracuse’s particular characteristics become distinctive after a while, but there are a rook of issues which are tentatively attributed to Ravenna, Rome or just ‘Italy’ that no-one’s really sure about.5 We haven’t solved this problem, but we may have spotted something that will help with it. I say ‘we’, but just as I owed the previous point to Rebecca Darley, this one was thought of by Maria Vrij; I sometimes think my sole intellectual contribution to this project was mainly defeatism. Maria noticed that whereas the Syracuse coins were debased with both silver and copper, and thus maintained a ruddy gold colour even once quite poor-quality, the elemental profile we were getting from supposed Ravenna issues included nothing like as much copper. Instead, the Ravenna issues seem to have turned ‘pale’, being adulerated only with silver. In that respect, they were following the trend of the post-Roman West at large, but it also makes sense in its own terms: Ravenna issued silver coin, which Syracuse didn’t, so when they had to cut corners with the solidi it makes sense that it was the refined silver from the local coinage that went into the pot, while Syracuse was presumably using less processed metal with accompanying copper content.6 So that’s something that belongs to Maria to write up properly, but hopefully it won’t be as many years before that happens as it has already been since we found it out… I make no promises there, as we all have other priorities, but nonetheless, we did find stuff in these tests that people might want to be able to refer to, and I hope this write-up at least gives some basis to believe that!


1. If you want the basics on these coinages, you can do no better even now than consult Philip Grierson, Byzantine Coins (London 1982), where pp. 50-56 will cover you for these purposes. The catchphrase, though, comes from Robert Sabatino Lopez, “The Dollar of the Middle Ages” in Journal of Economic History Vol. 11 (Cambridge 1951), pp. 209–234, online at http://www.jstor.org/stable/2113933.

2. My picture of this process comes pretty much direct from M. F. Hendy, “Aspects of Coin Production and Fiscal Administration in the Late Roman and Early Byzantine Period” in Numismatic Chronicle 7th Series Vol. 12 (London 1972), pp. 117–139, which is clearer than his later treatment in Michael Hendy, Studies in the Byzantine Monetary Economy, c.300-1450 (Cambridge 1985), pp. 257-303.

3. The standard reference for such matters, Philip Grierson, Catalogue of the Byzantine Coins in the Dumbarton Oaks Collection and in the Whittemore Collection, volume three: Leo III to Nicephorus III, 717—1081 (Washington DC 1973), Part I, where the coins in question are listed under Leo III 18a.1 (the Barber’s specimen online here), 48 (the Barber’s specimen online here) and, maybe, 12, 13 or 42 depending on what the Barber’s specimen (online here) actually is; the metallurgy makes type 42 seem likely though!

4. On the relevant Sicilian small change see for basics Grierson, Byzantine Coins, pp. 166-168, but for a different view of what was going on with its currency see Cécile Morrisson, “Nouvelles recherches sur l’histoire monétaire byzantine : évolution comparée de la monnaie d’or à Constantinople et dans les provinces d’Afrique et de Sicile” in Jahrbuch der österreichischen Byzantinistik Vol. 33 (Wien 1983), pp. 267-286, repr. in Morrisson, Monnaie et finances à Byzance : Analyses et techniques, Collected Studies 461 (Aldershot 1994), chapter X.

5. Grierson, Byzantine Coins, pp. 168-171.

6. Ravenna’s silver is discussed ibid., p. 140, but for the bigger picture see Mark Blackburn, “Money and Coinage” in Paul Fouracre (ed.), The New Cambridge Medieval History volume I c. 500‒c. 700 (Cambridge 2005), pp. 660–674.

All That Glitters, Experiment 5

Fittingly in some ways, given the distressing news of the last post, this post takes me back to Birmingham (which continues to happen, with a trip there on Wednesday coming that I will delight in telling you about before long if all goes to plan…). In fact, this is the last of the posts promised in my second Chronicle round-up, which means that we are now progressed in the story of my academic life to December 2015… It doesn’t look a lot like blogging progress, but let’s ignore that and instead tell the next part of the story of my project to zap Byzantine gold coins with X-rays, All That Glitters.

Bruker S8 TIGER XRF analyser open for business

The maw of the S8 TIGER XRF analysis machine in the Department of Chemistry, University of Birmingham, already much featured in these posts

It’s getting a little silly now to re-summarise the project every time I do one of these posts, however far apart they may be, so I’ll invite you to look here for the premise and just say where, by December 2015, the project had got up to. In brief, we had started from a belief that we might be able to find out about sources of metal for the Byzantine coinage and how those changed and maybe why by analysing them using a technique known as X-Ray Fluorescence (XRF). We got money to investigate this possibility in April 2015, and either before that or thereafter moved through the following developmental steps:

  1. finding out that the lightweight, energy-dispersive kit that we had hoped to use just wasn’t going to get the information we needed;
  2. finding out that the big, stationary, wavelength-dispersive kit we had to use instead (by kind courtesy of the Department of Chemistry, University of Birmingham) would get us the best results only on its longest cycle, cutting the number of things we could test in the time we could pay for down considerably;
  3. finding out that the results we were getting apparently included quite a lot of invisible surface deposits that seemed most likely to be leftover soil;
  4. discovering that, against all expectations, cleaning the coins in acetone actually made this problem worse, if anything;
  5. deciding, along the way, that we could not, as we had hoped, test different areas of coins for comparison of homogeneity either, because the results were just too darn variable to interpret;
  6. establishing that despite all these limitations, we could still distinguish between mint practices sometimes, but that only in the most difficult of cases was this telling us anything a competent numismatist couldn’t have seen by thmselves;
  7. and, although this was my colleague Dr Rebecca Darley, not myself, presenting these initial findings at the International Numismatic Congress in Taormina and at the Joint British Museum/Institute of Archaeology Seminar at University College London.1
  8. At the former of these presentations we got some pushback from the numismatists whose work we were implicitly questioning, which was understandable, but in the latter we got lots of pushback from one or two archaeometallurgists who felt that we were not people properly trained to do such work and that in fact it was pointless, which I saw as one of those ivory-tower problems; people are out there doing such work badly anyway, so would you rather just let them publish it and be accepted or shall we aim to do at least a bit better?2 Admittedly, we were having trouble doing much better, but that was what we now set about solving…

On 17th December 2015, therefore, three of us brought our test set of coins back to the Department of Chemistry, but this time with a difference. We’d already tried cleaning the coins in acetone, as said, so we had decided that we needed to try harder. But how hard should you try to clean a relatively soft precious-metal object of considerable value? Thankfully, this was a question that the team working on the Staffordshire Hoard had already faced, and since I’d been able to talk with one of them earlier in the year, we had a kind of answer, which was, berberis (or barberry) thorns: tough enough to shift surface dirt, soft enough not to scratch the metal!3 So before the test, Maria Vrij, by now in post succeeding me as Interim Curator of Coins at the Barber following my move to Leeds, had taken the coins and more acetone down to the Barber’s most suitable room for the purpose and, with the windows wide open, had laboriously worked over their surfaces with thorns under a magnifying glass.4 I can only say that this made me very glad to have moved jobs before this could have become my task, and I remain very grateful to Maria for doing it, but of course the real question was, what difference did it make? And the answer was, sadly, ‘a bit’: the levels of presumably-surface material that shouldn’t really be in the coins (calcium, silicon, potassium, aluminium) dropped, but were not gone.

A gold solidus of Emperor Heraclius struck at Constantinople in 613-616, Barber Institute of Fine Arts B2762, in a WD-XRF sample cup

A gold solidus of Emperor Heraclius struck at Constantinople in 613-616, Barber Institute of Fine Arts B2762, in its WD-XRF sample cup for testing

This was, in many ways, not the answer we wanted, as with so many of the findings thus far. We would much rather not have had to use the big, fixed machine to which the coins had to be brought, rather than one of the portable ones we could have taken to other collections; we would rather have been able to use a shorter test cycle and thus test more things in the time we had; we’d rather not have had to clean the coins at all; but if we had to clean the coins, we’d rather it had been possible just with a wash and a rub in acetone, not with hours of picking at them with thorns with your face close over a bath of solvent. If we had (and by we, I really mean Maria, sorry Maria), to do all that, however, we’d at least have liked it to produce good results. What it actually produced, however, was only measurably less bad results, which was not the exciting scientific conclusion for which we might have hoped. But it might be a bit more like actual science, and sadly, it’s a lot more like real life; messy, never quite sorted out, but still interesting…


1. The former of these papers is now published, in fact, as Rebecca Darley, “All that glitters…: the Byzantine gold solidus, c. 300-1092″, in Maria Caccamo Caltabiano (ed.), XV Internationa Numismatic Congress, Taormina 2015: Proceedings (Rome 2017), II, pp. 982-985. A cite for the latter would be Rebecca Darley, “What does the science mean? Interpreting metallurgic analysis of Byzantine gold coinage”, unpublished paper presented at the British Museum/Institute of Archaeology Joint Seminar, University College London, 15th December 2015.

2. It seems mean to point fingers, but once it’s being cited it is probably fair game and, on the basis of our experiments, I might raise questions about Rasiel Suarez, “A Metals Analysis of Silver Roman Imperial Coins using X-Ray Fluorescence Spectroscopy”, online here, whose precision just seems impossible with the equipment he used despite his checks (which were not carried out against a standard), and one would like at least to be able to ask more questions about the methods and reproducibility of the tests in Monica Baldassarri, Gildo de Holanda Cavalcanti, Marco Ferretti, Astrik Gorghinian, Emanuela Grifoni, Stefano Legnaioli, Giulia Lorenzetti, Stefano Pagnotta, Luciano Marras, Eleonora Violano, Marco Lezzerini and Vincenzo Palleschi, “X-Ray Fluorescence Analysis of XII–XIV Century Italian Gold Coins” in Journal of Archaeology (2014), pp. 1–6, online here. Note that we are not the only researchers wondering about things like this, by now: see also V. Orfanou and Th. Rehren, “A (not so) dangerous method: pXRF vs. EPMA-WDS analyses of copper-based artefacts” in Archaeological and Anthropological Sciences Vol. 7 (Basel 2015), pp. 387–397, DOI: 10.1007/s12520-014-0198-z, and E. S. Blakelock, “Never Judge A Gold Object by its Surface Analysis: A Study of Surface Phenomena in a Selection of Gold Objects from the Staffordshire Hoard” in Archaeometry Vol. 58 (Chichester 2016), pp. 912–929, DOI: 10.1111/arcm.12209.

3. See ibid.!

4. Of course, she is no longer Interim, but now actually properly Curator of Coins, and much better at it than ever I was, despite the acetone fumes!

Crusading and a Non-Deterministic Climate

The marking ebbs, and the ability to blog reappears… And for once it is clear what I should blog about, because I said I would pass over Conor Kostick‘s long-ago paper to the Digital Humanities Seminar in the Leeds Humanities Research Institute (which, as every sub-university-level academic organisation must every few years, has since changed its name), and then Dr Kostick himself cropped up in comments encouraging me not to, and so it seems rude to refuse. I admit that part of my initial reservation was that I might have to be rude, but now that I review my notes, even though the paper was called, “Digital Linguistics and Climate Change: a Revolution in the Digitisation of Sources since 2000”, which you can imagine annoying me in several ways I’m sure, I find less to be annoyed about than I remembered, but also less that one might call, well, conclusive.

Saul killing King Nahash and destroying the Ammonites, in the so-called Crusader Bible (c. 1250), New York City, NY, Morgan Library, MS M.638, fol. 23v

Saul killing King Nahash and destroying the Ammonites, in the so-called Crusader Bible (c. 1250), New York City, NY, Morgan Library, MS M.638, fol. 23v, image copyright not stated

Dr Kostick’s research at this time had arrived at the central theme of his paper from a circuitous direction. Starting with the study of the Crusades, he’d got into digital humanities as a lexicographical way of working out what medieval authors most probably meant by the words they used, which were of course changing as they used them. His example here, an interesting one, was that Archbishop William of Tyre, Chancellor of the Kingdom of Jerusalem already, may have been the first author to use the Latin word classis, classically meaning ‘fleet’, to mean ‘class’, as in first- and second-class, which are ways he divided up the nobility of Jerusalem in terms of tax liability. That wouldn’t have been clear without being able to find all the places he uses and all the places other people do and thus being sure that his is the usage that seems to begin it. This kind of technology lets us get further than the grand old lexicographers of old such as Charles Du Fresne Du Cange; as Dr Kostick put it, “we are standing on the shoulders of giants, with big binoculars”.

Charles Du Fresne Du Cange, from David d’Angers and Alfred Gudeman, Imagines philologorum (Berlin 1911), p. 19

Du Cange himself, from David d’Angers and Alfred Gudeman, Imagines philologorum (Berlin 1911), p. 19, public domain via Wikimedia Commons

From here, however, he had gone via an investigation of crusade preaching and had wound up at medieval climate data, not an obvious transition you might think, but several paths lead there. One is the kind of work that has been, let’s say examined, here before, attempting to correlate major political and social upheavals with climate events; another is the fact that at least one historian of the First Crusade, Ekkehard of Aurach, actually made the association for us, saying that the massive participation in that Crusade was at least partly down to a bad harvest, famine and ‘plague’ (perhaps ergotism, suggested Dr Kostick) in France that meant people with no other hope were willing to sign up with someone with a poorly-realised plan and take their ill-informed chances.1 The problem with many such analyses looking for other correlations, apart from the basic logical one of the difference between correlation and causation, has been poor focus of data, using, for example, tree growth in Greenland as a proxy for harvests in continental Europe, and this Dr Kostick avoided by taking as wide a range of sample evidence as possible. He started with chronicles, especially, using the same text-mining techniques as already mentioned, counting entries mentioning famine, plague and strange weather; added tree-ring data from a range of different areas (assembled by Francis Ludlow); and used ice-core data from Iceland and Denmark for finer dating. It’s a pretty good sample, as these things go, and this obviated many of the objections to such work I’d gone in with. So having done that, what do we then know? Well, the texts make it clear that both in 1095 and 1146, i. e. just before the First and Second Crusades, there were outbreaks of disease, which the tree-ring data suggests often coincided roughly with years of poor tree-growth, and the ice-core data sometimes allowed one to associate these and other such peaks with volcanic eruptions.

(I went looking for a climate data graph to put in here but the amount of short-sighted nature-blaming one quickly finds just made me angry so you’ll have to manage without an illustration between these paragraphs.2)

So case proven? Well, sometimes. It’s certainly possible, especially in the light of Ekkehard, to imagine how such a causal chain could fit together: a ‘year of no sun‘, poor crop yields, famine, destitution, desperate mobility, a convenient casus belli or particularly effective preacher, and suddenly what was meant to be a few thousand carefully-picked troops heading East, probably on the expectation of campaigning on an imperial salary for a few months, has become a horribly underplanned mass movement that winds up changing the world.3 The problem is that the chain doesn’t always work the same way. That works very well for the First Crusade, but in the Second Crusade, the popular participation was nothing like as large, though it was certainly large enough for Odo of Deuil to lament, I’ll admit; still, it was provoked by the fall of Crusader Edessa in 1144, and preparations were well underway by 1146 so I’d have thought that popular uptake is all that the bad year could have affected. Meanwhile, there was another significant peak between these two Crusades (not at 1101, at 1130 or so) which correlates with no such action, and there was no such peak before the Third or Fourth Crusades. Hey, maybe that’s why the Fourth Crusade couldn’t raise enough men, right? But the Third still presents problems.

A 15th-century image of the conquest of Constantinople by the Fourth Crusade, from  David Aubert, Livre traittant en brief des empereurs, II, fo. 205r

An unexpected result of a bad harvest? Probably not, eh? A 15th-century image of the conquest of Constantinople by the Fourth Crusade, from David Aubert, Livre traittant en brief des empereurs, II, fo. 205r, says Wikimedia Commons where this image is public-domain

Obviously, this paper was never meant to present a thesis as simple and obviously falsifiable as ‘volcanic eruptions caused the Crusades’, but without that, what do we learn from it? Our chroniclers already told us that plague and famine powered recruitment for some of the Crusades, and we didn’t need text-mining to see that. We might, now, understand better where that plague and famine had come from in these cases, but as with my earlier critique of Michael McCormick’s similar deductions about volcanoes, the problem lies in the volcanic eruptions that did not cause crusades, the famines and plagues that were not caused or strengthened by climate events, the crusades that did not correlate with bad weather or famines, and so on.4 No general rules could be extracted from this sort of causation, and neither was Dr Kostick out to present some, but without some such finding, it seemed like a very laborious way to conclude that a couple of our sources were maybe more right than we sometimes reckon. There seemed no question that Dr Kostick and his team had been more careful with data and correlations and even with causation than previous studies, but naturally enough perhaps, that had also limited what they could conclude.

That was my feeling as Dr Kostick wound up, anyway, but questions revealed other doubts and issues among the audience, many of which I thought he actually had good answers to. One of my colleagues argued that climate event references in chronicles are often wrong, to which Dr Kostick wisely observed that this was a good reason to correlate them with scientific data. Other questions focused more justly on causation: Graham Loud has in the past argued, apparently, that a famine which preceded the Third Crusade actually limited response from Germany, and here again Dr Kostick argued that while local responses to stimuli would obviously have varied, the bigger correlations still need explanation when they occur. True enough, but that seems to have been very rare… Well, I certainly don’t have better answers, and if Dr Kostick had been unwise enough to try and push his data further than it would go I imagine I’d have had bigger issues with that, but my feeling remains on this revisiting that his admirable caution robbed the paper of its potential power. The success of McCormick et al. suggest that, sadly, the route to publication of such work is not to care about such things but to push the deductive boat out well beyond sensible recovery, and maybe that’s why this one didn’t (yet?) achieve wider dissemination; it just wasn’t crazy enough!


1. F.-J. Schmale and I. Schmale-Ott (edd.), Frutolfi et Ekkehardi Chronica necnon Anonymi Chronica Imperatorum: Frutolfs und Ekkehards Chroniken und die Anonyme Kaiserchronik, Ausgewählte Quellen zur deutschen Geschichte des Mittelalters 15 (Darmstadt 1972), pp. 19-38 (commentary) & 123-309 (text), cap. 13/40, pp. 124-127, the relevant section transl. J. H. Robinson in Readings in European History Vol. I (Boston 1904), pp. 316-318, online ed. P. Halsall as “Medieval Sourcebook: Ekkehard of Aurach: On the Opening of the First Crusade”, online here.

2. I should clarify that the thing I think is stupidest in these arguments is neither that there is dispute over climate change at all, which I find explicable if dangerous, nor that there is argument over its causation, which is predictable really, but the conclusion that some people who believe climate change now is not anthropogenic reach that therefore we need do nothing about it because it’s natural. I imagine these people largely do not live in the areas most affected.

3. This interpretation of events largely rests on my old piece linked off this very blog, but is similar to that put forward in Peter Frankopan, The First Crusade: the call from the East (London 2012).

4. My target here is of course Michael McCormick, Paul Edward Dutton and P. A. Mayewski, “Volcanoes and the Climate Forcing of Carolingian Europe, A.D. 750–950”, Speculum, Vol. 84 (Cambrudge MA 2007), pp. 869–895, online here.

All That Glitters, Phase 4

The times continue strange in UK higher education, as you may have seen. Many of us are on strike for what is now the third week, more of us than ever now, and the employers’ representatives appear to be refusing to negotiate in person and then changing their mind by Twitter overnight. I don’t know what may happen in the next 48 hours and of course in case classes happen, they all have got to be got ready on the few days when we’re not on strike, in case something is resolved that means we go back to work. But, what this does mean is that my conscience is pretty clear about blogging. Having taken my first steps down a new road in the previous post, it thus behoves me to look around myself and say, ‘What was I doing in July to September 2015 that I haven’t already told you about?’, and the answer to that is not limited to but certainly includes, ‘zapping gold coins with X-rays some more’. So this is about our fourth set of tests.

Jonathan Jarrett and a gold solidus with XRF machinery in the Department of Chemistry, University of Birmingham

Posed, obviously; I may look intrepid, but you have no idea how tightly I was holding that coin. It rolling under the machine would have spoiled several people’s day quite badly…

If you remember, where we were with this is that having got money to evaluate techniques by which we might be able to use X-ray fluorescence spectroscopy to measure the metal content of Byantine coins, with an especial eye on trace elements that might betray metal sources, we had fairly quickly established that the kind of portable machinery which we could bring to the coins in their museum wouldn’t tell us what we needed to know. So the working set-up for these experiments was now that, after having checked our insurance quite carefully, as soon as I could get into the Barber’s coin room of a morning I would remove from it about 100 g of high-purity gold in the form of 20-odd Byzantine and other coins, then University security would turn up (in theory) and transport me to the School of Chemistry (in theory). We would then do as much zapping as could be done, with at least two people present where the coins were at all times, before Security turned up again (in theory) in time to get me and the coins back into the Barber before it closed. And this time we did this for four days running. I won’t tell you how many ways this process could go wrong, but I haven’t flagged them all. But Chemistry were lovely and very generous both with expertise and with biscuits, and though we never had quite the same team there two days together it was all quite a good group exercise anyway. So, what were we doing this time and how did it go? The answer is a long one, so I’ll put it behind a cut, but do read on! Continue reading

All That Glitters, Experiment 3

A problem with finally picking up this blog a bit is that some of the old stories in it have been sitting idle for really quite some time. Does anyone remember, for example, that while I was working at the Barber Institute we got a small amount of money to zap gold coins with x-rays? If you do, you may remember that I’d already reported on the first two experiments we did in that project when everything here ground to a halt in 2016. Well, now I resume, with a brief account of the third experiment we did, and before long who knows, we may be through the rest!

Bruker S8 TIGER XRF analyser open for business

The S8TIGER WD-XRF spectrometer in the Department of Chemistry, University of Birmingham, with five sample cups waiting and one under analysis

A brief recap may be in order first of all, though. The very short version of the project’s purpose was: we had all seen an increasing amount of work using X-ray fluorescence spectrometry to analyse the fineness of precious-metal coins and sometimes even assign origins to its metal based on the trace elements therein. We had our doubts about this, but the manufacturers of some of the best machinery involved wanted to quell our doubts about this and offered us quite a lot of help with it, so we came up with a plan and, somewhat to our surprise, got both permission from the Henry Barber Trust to muck around with their coins and funding from the Royal Numismatic Society to do it, and so set about it. We began this in December 2014, with two experiments. The first was designed to test the viability of hand-held, portable, energy-dispersive machinery that could be brought to the coins; the second did comparator analyses using a fixed-location wavelength-dispersive spectrometer to which we had to bring the coins. This showed us that to detect anything at all of trace elements we would need the big machine, but also that it was picking up a whole load of stuff that seemed very likely to be extraneous material on the surface of the coins, probably from the soil in which they had long ago been found, rather than the stuff of which they were actually made. And that is about where we come in!

Gold solidus of Emperor Constantine VI and Empress Eirini struck at Constantinople 785-797, Barber Institute of Fine Arts B4599

The coin whose results (and details) are given in the first table below, set up for the test in which it did so

So, the third experiment was very simple. Using the same set of coins as we had previously run, I got hold of some conservation-quality acetone and a toothbrush, took them and the coins down to a workroom in the depths of the Barber and gave them all a careful but thorough scrubbing and let them dry, all except one that we left uncleaned as a check. Then I packed them up again and we bore them off to the Department of Chemistry on 22nd July 2015. There we ran them again, on the shorter of the big machine’s two cycles, and this gave us some results both before and after cleaning.

Unfortunately, if anything, the post-cleaning results showed more extraneous grot. That ‘if anything’ will tell you, though, that consistency was hard to observe. The variation wasn’t vast amounts, you understand, between half and one-and-a-half per cent, but cumulatively certainly enough to push our figures for the actual gold content up or down by two or three per cent, which would for some people be enough on which to found a theory about debasement… Here are the top ten elemental results from just one side of one coin by way of example:

Coin 170 (obverse)1 Percentage observed first time Percentage observed second time
Gold 86.51% 84.49%
Silicon 3.44% 4.63%
Silver 1.61% 1.43%
Aluminium 1.27% 1.19%
Potassium 0.80% 0.63%
Sodium 0.74%
Copper 0.44% 0.34%
Calcium 0.32% 1.96%
Sulphur 0.28% 0.48%
Iron 0.27% 0.50%
Total percentage of coin observed 96.38% 96.70%

Down by two per cent gold, and that mostly gone to silicon and calcium, even though some other small elements had dropped in the readings as well. Also, what this arrangement doesn’t show, the second time small but possible readings showed up for chlorine, magnesium and, way down the scale, platinum, all absent the first time through. Surely cleaning with acetone couldn’t have added calcium to the coin? But each coin we ran had its own little tale of additional mess to tell, even (and this should have told us something) the uncleaned check. Here are its top-ten figures for the obverse:

Coin 21 (obverse)2 Percentage observed first time Percentage observed second time
Gold 82.57% 87.43%
Silicon 1.82% 1.61%
Aluminium 1.01% 0.81%
Silver 0.74% 1.10%
Sodium 0.65%
Potassium 0.47%
Chlorine 0.45%
Calcium 0.44% 1.78%
Iron 0.31% 0.31%
Copper 0.24% 0.26%
Total percentage of coin observed 89.12% 93.80%

Some things are beguilingly similar, yes—iron and copper levels stay about the same—but others are not. The first test picked up sodium, potassium and chlorine but none of those turned up second time even though we’d actually got better readings with more of the sample observed (possibly because we gave it a flatter bit to look at). Were we just looking at a different bit? If so, that bit apparently had less silicon in it, but a whole whack more calcium in it. What can you do with these sorts of data?

Gold solidus of Emperor Anastasius I struck at Constantinople 491-518, Barber Institute of Fine Arts B0006, in an WD-XRF sample cup

The uncleaned check likewise set up for the results above, though I have to say it’s not like it looks dirty

So this was somewhat daunting, especially as we had four solid days of testing booked in for the following week and now somewhat less certainty that it would produce anything. At the time, my best guess was that the acetone had removed only handling patina, thus exposing the actual surface of the coin and everything that had adhered to or reacted into it while it was in the ground. Actually, looking back, I think we were already seeing here the conclusion that the final experiment would necessarily lead us to. But that would be getting ahead of the story, and even though this story is so very far behind, I have hopes of telling it as I wanted to anyway. So, till next post but, er, two, if I have my plans right (and no-one else dies—did you see Mark E. Smith has left us for the bar of the great and final WMC since the last post?), I shall leave the question hanging…


1. We used randomly-allocated running numbers throughout these experiments, so that we couldn’t try and guess what the results should be according to where the coins fitted in Byzantine history (or the previous curves laid down by work like Cécile Morrisson, Jean-Nöel Barrandon and Jean Poirier, “La monnaie d’or byzantine à Constantinople : purification et modes d’altérations (491-1354)” in Morrisson, Claude Brenot, Barrandon, Poirier and Robert Halleux (edd.), L’or monnayé I : purification et altérations de Rome à Byzance, Cahiers Ernest Babelon 2 (Paris: C. N. R. S. 1985), pp. 113–187. This one was actually Birmingham, Barber Institute of Fine Arts B4599, a solidus of Emperor Constantine VI and Empress Eirini struck at Constantinople between 785 and 797, online here. I’m actually almost loath to identify the coins here, though, lest it be thought I’m actually publishing figures for their metal content. As is probably clear, though, that’s something I’m sure we weren’t reliably getting.

2. And this one was Barber Institute of Fine Arts B0006, a solidus of Emperor Anastasius I struck at Constantinople between 491 and 518, online here, with the same reservations as in the previous note.

All That Glitters, Experiment 2

Somehow my posts about me and my work—and that may not be what you’re here for but, you know, I like it—have got behind my seminar reports in such a way that they’re into February 2015 and I’m still in December 2014. Let me resolve some of that disparity by giving you a short report on the second day of experiments in the collaborative project I’m in for analysing Byzantine gold coinage by X-ray fluorescence, which was 14th December. (If you need background I announced this project ages ago here and dealt with some of our starting questions and the first day’s experiment here.)

Cover of J. O. Jeppson, The Second Experiment

Our results have so far not been this dramatic, but then, I’m guessing that our first experiment wasn’t quite as adventurous as this must have been

To recap, we had established that if our experiments were to tell us anything much about elements other than gold, silver, copper and maybe one or two other pre-determined elements, we were going to need not the energy-dispersive machinery we’d been using on the first day but the bigger, more expensive and, most importantly, immobile wavelength-dispersive machinery in the Department of Chemistry in the University of Birmingham, a machine called the S8 TIGER. I am only just able to describe the difference between these two analytic methods: in so far as I can, it’s to do with what is being used to pick up the energy given off by the things you’re bombarding with x-rays. The WD machinery includes crystal collimators that are sensitive to certain wavelenths of that energy, which therefore get picked up better, where the ED machines, which measure only in terms of intensity of signal, simply wouldn’t see such things among the massive gold return, as we had surmised. The WD machine also scans its samples in a vacuum, which eliminates interference from the air.

Bruker S8 TIGER XRF analyser open for business

The mouth of the TIGER yawning wide, with five sample cups waiting and one under analysis

On the other hand there are also problems with the WD machinery that don’t exist with the ED kit. For us the first of those was simply access; it’s nothing to do with the actual machinery except in so far as it’s immovable, but because we had to take the coins to the kit rather than vice versa, that meant arranging transport and insurance even on campus, and the transport repeatedly went wrong, which cut into our experimental time a lot. But, also, the ED kit works with narrow beams focussed on points; the WD machine scans its samples in masks such as the one below, of which the two sizes relevant to us were 5 mm and 8 mm, and those were therefore the only area sizes that we could analyse. Importantly, this also precluded examining coins at their edges or over piercings, because the sample has to fill the exposed area completely. This also highlights a problem with both ED and WD methods: non-homogeneity. If for some reason your coin had an odd tiny lump of platinum on its surface, say, the ED machinery would either miss it (in which case you’d never know) or find it and report a massive platinum signal (which would be misleading for the coin’s overall composition). The WD machinery, however, would factor it into the average, so that you wouldn’t necessarily realise that it was a coherent inclusion rather than a component of the main alloy. So there was plenty to worry about even if the machine worked perfectly.

Emperor Heraclius just visible on one of his solidi of Constantinople loaded behind an 8 mm mask for analysis in the Bruker S8 TIGER

Emperor Heraclius just visible on one of his solidi of Constantinople loaded behind an 8 mm mask for analysis

Anyway, we had our goals clear for this test. The first was to get our hands on the machinery and find out what the operational considerations in any further planning were, the results of which you sort of see in the musings above. Here I have to acknowledge the tremendous help and general goodwill of Dr Jackie Deans, official keeper of the TIGER, and Dr Adrian Wright, who had first let us involve the Department of Chemistry in the project and had helpful things to say whenever he dropped in. Our second priority was to run the same ten coins around which we’d built our first experiment on the S8 TIGER and see how the results differed from those on the ED kit. And as it turned out, our third one was to determine how we wanted to use the S8 TIGER, because as Jackie explained to us, it could analyse at three levels, a 2-minute cycle that would probably get us no more data than the ED machinery had, an 8-minute one which should do the job, and an 18-minute one which was the very most data it could gather. Adding 10 minutes to each analysis was obviously going to limit the number of coins we could actually analyse in any given timeframe, so we really rather needed to know whether or not it was worthwhile.

Gold solidus of Empress Eirini at Constantinople set up for analysis in a Bruker S8 TIGER XRF analyser

The rather different visage in gold of the Empress Eirini, likewise cruelly cut down to 8 mm of glaring royalty

And so what did we find? Well, this machine certainly had more to tell us. We were now getting returns in terms of many elements, at concentrations of down to parts per thousand or even less. This ineluctably meant a decrease in gold concentration reported, because there was now simply more data to fit into the percentages, but the overall picture of lots of gold, not much silver and less copper was still very apparent in the reported figures. What we hadn’t expected, and had now to deal with, was that copper wasn’t usually the third most detected element, and sometimes silver not the second: instead, we were seeing lots of calcium, silicon and sometimes aluminium beating them out. It seemed a priori unlikely that these were original metallic components of the coins in these quantities. That in turn implied that these elements had got into, or much more likely onto, the coins since striking, be that from use, preservation or anything else that might have happened to them. But, whatever they were, they also seemed to be more consistently detected on the long cycle than the medium-length one, meaning that we were going to need to use the long analysis to have any chance of consistent findings. So now we had two difficult questions to answer in setting up Experiment 3: firstly, what could we get done with less than half the scans that we might have hoped to do in any given day of experiments, but secondly, when we did, could we determine whether these results were merely contamination or do anything about that if they were? And these were things which we attempted to address in the New Year, so I’ll stop here for now.