Classical Numismatics Discussion
  Welcome Guest. Please login or register. 10% Off Store-Wide Sale Until 2 April!!! Explore Our Website And Find Joy In The History, Numismatics, Art, Mythology, And Geography Of Coins!!! Expert Authentication - Accurate Descriptions - Reasonable Prices - Coins From Under $10 To Museum Quality Rarities Welcome Guest. Please login or register. 10% Off Store-Wide Sale Until 2 April!!! Explore Our Website And Find Joy In The History, Numismatics, Art, Mythology, And Geography Of Coins!!! Support Our Efforts To Serve The Classical Numismatics Community - Shop At Forum Ancient Coins

New & Reduced


Author Topic: Woytek and weight of Trajan Coins.  (Read 6706 times)

0 Members and 1 Guest are viewing this topic.

Offline rasiel

  • Procurator Monetae
  • Praetorian
  • *****
  • Posts: 75
  • I love this forum!
Re: Woytek and weight of Trajan Coins.
« Reply #25 on: December 08, 2014, 09:03:27 pm »
This is probably the most sensible way of reconciling the findings. If you start your testing with a flawed premise then the results will necessarily be flawed as well. How deep the "surface" is disproportionately enriched (or conversely the core debased) was never determined. As outlined in my last post this is easily determined but has not yet to my knowledge actually been done. If and when I get the chance I will be happy to publish the results.

Ras

Offline imperialcoins

  • Consul
  • ***
  • Posts: 117
  • Agora Auctions
    • Agora Auctions, Inc.
Re: Woytek and weight of Trajan Coins.
« Reply #26 on: December 09, 2014, 05:48:37 am »
You should also take into consideration that the cleaning process for denarii involves some pretty harsh acid and/or base baths.

Jyrki Muona (sp?) worked on an analysis of coins of Otho and he literally took a core sample from the edges of coins so as to not destroy the coins.
-Alfred
Agora Auctions, Inc.
http://agoraauctions.com

Offline SC

  • Tribunus Plebis Perpetuus
  • Procurator Monetae
  • Caesar
  • *****
  • Posts: 6068
    • A Handbook of Late Roman Bronze Coin Types 324-395.
Re: Woytek and weight of Trajan Coins.
« Reply #27 on: December 09, 2014, 05:52:34 am »
I suspect, but can not prove, that the issues stem from the third option in your first scenario - "the equipment is crap".  I know that i may not make friends among the increasing number of people using these devices for coin studies but here is why I say this.

Hand held XRF devices are increasingly common and their sales literature implies that they are highly reliable.  This assertion is highly suspect.  Hand held devices were designed for initial survey of geological deposits and industrial scale items, for yielding rough estimates, not for proper analysis.

The Woyetk et al study utilized a micro-XRF device called COPRA, which stands for Compact Portable Roentgen Analyzer.  This "compact, portable" analyzer is the size of a large suitcase and also requires a computer and power source.  It also requires significant calibration after each set-up.  And that is the compact portable version.  In the nuclear field, which I know a bit about, any meaningful analysis  - where the results will be used for academic papers, annual reports, prosecutions, political action, etc - is done on machines the size of a large desk that have seven digit price tags.  

This is not a case where an old 1980s mainframe is now exceed by an iPhone.  It is more akin to the breathalyzer devices.  Even the breathalyzer devices used by police forces are not very accurate.  They do not provide any evidence.  Instead they provide a level of justification that allows a police officer to demand a proper test - usually a blood sample test - which does then provide admissible evidence.  No one can rely on a hand held breathalyzer device for evidence and no one should rely on a hand held XRF device for academically rigorous results.

I have looked for academic studies comparing the level of accuracy of the different types of XRF device - handheld, COPRA and full sized units - and surprisingly can't find any.  Just lots of glossy sales brochures for hand held devices touting their accuracy.  If any are out there they may prove me right or they may prove me wrong but until then I have to rely on speaking to people that use the full range of XRF devices in their professional life.

Furthermore, even the best XFR technology has been superceed for this purpose by other technologies.  As Ponting and Butcher note on the Archaeological Data Site when speaking of one of their studies:  "A more sensitive analytical technique was employed, inductively-coupled plasma atomic emission spectrometry (ICP-AES), that enabled much better quality data for important trace elements, such as arsenic, tin and bismuth, to be obtained. These data, together with scanning electron microscopy (SEM), micro-analysis (SEM-EDS), optical microscopy (metallography) and some limited lead isotope analysis by laser ablation multi-collector inductively-coupled plasma mass spectrometry (LA-MC-ICP-MS) form the basis of a series of publications in print".  The utility of technologies such as ICP-AES, SEM and SEM-EDS are not limited to the study of trace elements but are more reliable for basic fineness measurements too.

I know what you wrote about the device in your study.  From an academic point of view it is great and I have no qualms with your methodology.  I just have grave suspicions about the true accuracy of these devices for point samples.  As I noted above I may be proved wrong but wanted to share my concerns.

Shawn



SC
(Shawn Caza, Ottawa)

Offline jmuona

  • Procurator Caesaris
  • Caesar
  • ****
  • Posts: 402
  • I love this forum!
Re: Woytek and weight of Trajan Coins.
« Reply #28 on: December 09, 2014, 06:11:17 am »
Rasiel, if you look at the Forum for our article (named Denarii of Otho here), you will find more info of the methods used for the analysis of metal content in these studies. The key is to obtain a good sample of the metal from within the coin, like Alfred wrote. In our Otho paper there are also references to other works by Butcher and Ponting, in which the methods are explained in more detail. It is clear that surface analysis are not acceptable for anything below "Roman pure silver", which in itself contained led, gold and trace metals. The surface enriched silver layer does vary and after the copper was removed with acid, the blanks were also - at least at some stage - struck before the actual coins were struck, creating a hard "pure" silver surface.
These methods were used all over Rome as evidenced by the Otho tetradrachm from Alexandria shown on my Gallery here. The coin has a thick "silvery surface layer" and a massive copper-coloured core.
Jyrki Muona

Offline Paddy

  • Consul
  • ***
  • Posts: 372
  • Marcus Ulpius Traianus Maior
Re: Woytek and weight of Trajan Coins.
« Reply #29 on: December 11, 2014, 12:30:45 am »
1. My impression of this discussion so far is that most consider Woytek's method reliable. Is this correct?

2. Is a sample of 71 denarii large enough a sample?

3. Would melting the coins not give us the most accurate measure of the metal composition?

Offline SC

  • Tribunus Plebis Perpetuus
  • Procurator Monetae
  • Caesar
  • *****
  • Posts: 6068
    • A Handbook of Late Roman Bronze Coin Types 324-395.
Re: Woytek and weight of Trajan Coins.
« Reply #30 on: December 11, 2014, 02:36:25 am »
!) Yes.  Though that does not preclude that other methods are reliable too.

2) With sample the rule is generally "the more the merrier".  Given the cost, and possibly destruction of the coin, the numbers available are limited.  Thus 71 is quite a decent sample in this area.

3) Yes, though of course that is the most destructive - in fact total destruction.  And as you say it can give the most accurate - but is not perfect.  It measures everything that exists in the coin today which includes the effects of any pickling, aging in situ, corrosion, cleaning, etc. 

Shawn
SC
(Shawn Caza, Ottawa)

Offline rasiel

  • Procurator Monetae
  • Praetorian
  • *****
  • Posts: 75
  • I love this forum!
Re: Woytek and weight of Trajan Coins.
« Reply #31 on: December 11, 2014, 05:15:47 am »
Jyrki I searched for your paper but was only able to find your brochure on PMI (pretty slick I gotta say!). If you could please email it to me this would be very appreciated.

Shawn, the devil's advocate in me was skeptical from the get go but I had to give the analyzer a tip of the hat when it successfully identified some exotic "stumpers" I threw its way (zirconium, cobalt and tungsten). I did stump it with a pure sample of beryllium and another with the rare earth metal erbium but, to be fair, in both cases I knew these were out of spec beforehand. Also, if the machine was unreliable I would expect poor scores on repeatability which wasn't the case. All the same, I grilled the tech on handheld versus benchtop analyzers and it seems that the key part of the hardware, the vacuum tube and X-ray source, as well as the software, are the same and that only the form factor changes. He could've been misinformed or outright lying, of course, but it does make some sense that the manufacturers would streamline their parts bin to cut down on development and manufacturing costs. He explained that what you get as you go up in price, up into the six figures territory, is more resolution. More resolution is nice - that being the ability to discern even the smallest amounts of trace elements - but is not critical to the greater part of answering our most pressing questions.

And just to be thorough, in case anyone still reading this thread isn't bored to tears yet, there are actually three processes at play that prevent an easy understanding of the composition of ancient silver coins. First, as already discussed, the flans had their silver content artificially elevated on their surfaces via annealing (and possibly acid washing). Secondly, a heated silver-copper alloy when struck will have a higher amount of silver concentrate at the point of impact in a process I don't understand well at all but which is due to the limited solubility of copper and silver with each other while in a molten state. Whatever the chemistry at play, the bottom line is that it results in precipitation of microscopic blobs richer in silver and counterpart grains that are richer in copper to essentially form two alloy sub-groups. In the blink of an eye somehow these two stratify. The third and final process finds the surface copper slowly eroding through oxidation, molecular migration and mechanical stress. With the steady deterioration of these copper-rich grains comes a weakening of the overall matrix through internal fracturing. You see this popularly defined as "crystallization" and is the phenomenon of the brittle ancient silver coin that can shatter if mishandled. What is as yet undetermined, and this is the key point, is what a hypothetical section from surface to core would look like. Is it Ag-rich only in the first few microns at the surface with the rest fairly evenly distributed or is the core, like the surface, especially rich in copper and other base metals and only the middle layer relatively homogeneous?

Given the above, I remain very skeptical of Woytek's figures. The data sampling of 71 coins is great so long as you trust the methodology is sound... except this is precisely what is controversial! Just as it is inaccurate to measure the surface and extrapolate from there it is, or at least I believe it to be, equally inaccurate to measure the hyper-debased core and make the assumption that it's representative of the bulk of the material. I really have no personal investment in saying his results or mine are ultimately correct, I just want to know the facts. When I can again get the opportunity to use an analyzer I intend on running tests that I think will better determine the most authoritative measurements by taking a sample coin, testing as is, wearing it down and testing it again. I will repeat this a few times over and from that data generate a plot of the silver-copper gradient. Then, to be truly thorough, melt the coin (in an inert atmosphere!) and test it one final time. I suspect, but do not yet have proof, that the "sweet spot"; that is, the point where the two lines of each of these elements bisect on a graph, is below the surface but above the core. In other words, simply surface testing a worn coin should give a reasonably reliable analogue to the ideal scenario where you melt the coin and test it in its most homogeneous state. 

Even in the case where surface analysis is completely unreliable in determining overall composition (for silver coinage) it still provides very valuable data. Knowing whole-coin composition is important from a historical perspective but surface analysis potentially gives us a tool for determining age, modern and ancient counterfeiting and perhaps even a glimpse into manufacturing processes.

Ras
ps. Melting denarii in an argon-filled chamber, come to think of it, yeah, that's.... that's not going to be an easy task to pull off :-(


Offline PeterD

  • Procurator Caesaris
  • Caesar
  • ****
  • Posts: 1483
  • omnium curiositatum explorator
    • Historia
Re: Woytek and weight of Trajan Coins.
« Reply #32 on: December 11, 2014, 07:12:14 am »
This is a short section from 'The Silver Coinage of Septimius Severus and his family (193-211 AD) by Haim Gitler and Maththew Ponting, which helps to explain why different researchers have gotten different results for silver content.

"Metallographic sections of the Roman Imperial silver coinage of this [the third century] period reveal that silver-enriched surface layers of substantial thickness are present on the finished coins — almost without exception. These silver-rich layers have to be mechanically removed in order to obtain the most reliable analyses of the original coin alloy..." (Cope 1972, p.265).

Cope attributed this phenomenon to the hot-working of the flans prior to striking, which resulted in the oxidation of the copper-rich microstructure phase. This would have given the coins a black appearance and would therefore have needed to be removed. The black-oxidised copper-rich phase could be preferentially removed by blanching — cleaning in a dilute acid such as fruit acid or vinegar — which would have resulted in the coin blank having a bright yet porous silver-rich surface zone. This would then have been consolidated into a `silver coating' by the striking of the coin.

Cope goes on to say that the original consolidated silver-enriched surfaces can be distinguished metallographically from the more porous and deeper silver-enrichments created primarily by blanching at the mint, together with the burial environment and, the authors would like to add, aggressive chemical cleaning methods.

Unfortunately, this important discovery has been misinterpreted by most of the numismatic and archaeological science community, who confused Cope' s man-made silver `coating — essentially depletion silvering - with simple environmental surface enrich¬ment. Because of this and the desire for a `quick-fix' analytical method that is 'non-destructive', the technique developed to overcome the generally thin layer of environmental surface enrichment, has been applied generally to coins of all materials. Consequently, when applied to base-silver alloys, this methodology has proved unable to provide the reliable chemical data that it professed to do. Essentially the method involved the removal of the enriched metal in a small area on the coin' s edge, the idea of which was to create a `window' into the, hopefully, representative metal beneath. Unfortunately the correct identification of the `representative' metal proved problematic. One of the pioneers of this approach, which became fairly universal, was Giles Carter. Carter’s, technique was to air-abrade the area to be analysed, removing a total estimated 60 (micro)m of surface metal. This was the amount of metal removed in order to achieve a repeatable reading of the silver content (three repeats starting at 20(micro)m, ending at about 60 (micro)m), the assumption being that this final reading is representative of the alloy of the main bulk of the coin. However, other work has shown that this is misleading. For example, there is a notable discrepancy between the silver value for Severan denarii issued after 193 AD as reported by Condamin and Picon (45%) and as reported by Carter. Carter gives an average figure of 58.4%±0.5% (normalised) based on ten analyses. This is a discrepancy , over 13% that suggests the there is a serious problem with this approach. Yet it is this method (or a variant of it) which has been repeatedly used for numismatic research. Indeed, it was essentially this method that was used by Walker in his magnum opus, The Metrology of the Roman Silver Coinage, and gave a similarly high mean silver value of 57.6% for the same issues analysed by Carter.
Peter, London

Historia: A collection of coins with their historical context https://www.forumancientcoins.com/historia

Offline 77HK77

  • Caesar
  • ****
  • Posts: 432
Re: Woytek and weight of Trajan Coins.
« Reply #33 on: December 11, 2014, 08:07:49 am »
Rasiel,
in reading your paper I noticed the sample size was cited as 162 roman and Byzantine coins but I did not see a specific breakdown.

What was the sample sized you used for Trajan versus Woytek's 71 coins?

In both cases does sample group cover multiple years within an emperor's reign to lessen errors from outlying factors that might affect the mix in a single year, month day etc?

HK

Offline Andrew McCabe

  • Tribunus Plebis Perpetuus
  • Procurator Monetae
  • Caesar
  • *****
  • Posts: 4651
    • My website on Roman Republican Coins and Books, with 2000 coins arranged per Crawford
Re: Woytek and weight of Trajan Coins.
« Reply #34 on: December 11, 2014, 09:48:07 am »
71 is an absolutely enormous sample. That's not because of numismatics but because of statistics. Without having to know the exact numbers, one could look up a variety of reports of the fineness of slightly debased ancient silver coin series tested with any one standard method, in a range of publications, to see how debased silver "behaves". One quickly gets a sense that once sample sizes exceed perhaps a dozen examples, the uncertainty in the mean reduces to a very small amount, and if one adds another ten or twenty coins to the mix it's likely that neither the mean nor standard deviation will shift much. So the difference in results we see here can't be a statistical accident but must result from fundamentally different methods. Given the nature of the subject matter, a sample size of thirty would probably have been good enough to "lock in" the result for any given testing method, and a sample of fifty would be absolutely decisive. 71 is overkill.

In contrast to the comment "the more the merrier", there really does get a point where adding more samples isn't adding any extra accuracy at all, because the major sources of inaccuracy are no longer uncertainties within a single set of results, but between different testing methods. If, as seen in this discussion, there's about a ten percent difference in results using two different testing methods, it is of no value to add extra samples to one or other method, thereby refining that difference to 10.1% rather than 9.9%. None at all. The real issue is deciding which test method will give the technically better results. Once that's decided, the other set of results can be discarded entirely, thrown in the bin, as simply being wrong. Even a test based on a thousand samples doesn't improve accuracy of results if the test methodology is flawed.

Offline 77HK77

  • Caesar
  • ****
  • Posts: 432
Re: Woytek and weight of Trajan Coins.
« Reply #35 on: December 11, 2014, 10:07:17 am »
Andrew,
Your statement is to the point of my questions.
To accurately compare methods you need to match samples.

The closer you match Woytek's 71 in type, mint and date the more accurate the result. We can never have a exact comparison because the coins were destroyed.

If you only tested one coin you can't accurately make a comparison between testing methods. Alternatively you could to take the time to make your own mix of known materials and measure by each method. Short of that effort matching sample types is the next best alternative

Hk

Offline Andrew McCabe

  • Tribunus Plebis Perpetuus
  • Procurator Monetae
  • Caesar
  • *****
  • Posts: 4651
    • My website on Roman Republican Coins and Books, with 2000 coins arranged per Crawford
Re: Woytek and weight of Trajan Coins.
« Reply #36 on: December 11, 2014, 10:22:57 am »
Andrew,
Your statement is to the point of my questions.
To accurately compare methods you need to match samples.

The closer you match Woytek's 71 in type, mint and date the more accurate the result. We can never have a exact comparison because the coins were destroyed.

If you only tested one coin you can't accurately make a comparison between testing methods. Alternatively you could to take the time to make your own mix of known materials and measure by each method. Short of that effort matching sample types is the next best alternative

Hk

No, I don't agree. You only need to match the nature of the samples (era and mint of coin issues). You do not need to use the same actual coins. Once you get above 20 or so coins in the sample, and if they are all from the mint of Rome and of approximately the same era, then almost irrespective which 20 coins are selected, you'll end up with approximately the same mean and the same standard deviation for any given testing method. This is how statistical sampling works. You could repeat Woytek's method fifty different times using fifty different selections of coins (provided only the coins are of the same era) and I'd guarantee that you'll get almost exactly the same results each and every time, within one percent, so long as the testing method remains the same.

That's the power of statistics. It really means that once a sample size is large enough, no further benefit is obtained by using an even larger sample, and you are pretty much guaranteed to get the same results (approximately, within say one percent) from two different handfuls of coins (each melted down and destroyed).

The real uncertainty isn't in the sample. It's in the choice of testing method. Woytek's test can be replicated today by buying 50 worn denarii of ebay of the same era and carrying out his test method using the same equipment. The same results will be obtained (give or take one percent).

Offline 77HK77

  • Caesar
  • ****
  • Posts: 432
Re: Woytek and weight of Trajan Coins.
« Reply #37 on: December 11, 2014, 10:31:34 am »
We are agreeing on Woytek it is repeatable on a smaller scale

I'm questioning the comparison of rasiel's results. How many trajan samples were tested are they similar. It's not mentioned in the paper...for all I know he mistook a drachm for a denarius ( I know you wouldn't Rasiel- point for effect)


Offline Andrew McCabe

  • Tribunus Plebis Perpetuus
  • Procurator Monetae
  • Caesar
  • *****
  • Posts: 4651
    • My website on Roman Republican Coins and Books, with 2000 coins arranged per Crawford
Re: Woytek and weight of Trajan Coins.
« Reply #38 on: December 11, 2014, 11:23:30 am »
We are agreeing on Woytek it is repeatable on a smaller scale

I'm questioning the comparison of rasiel's results. How many trajan samples were tested are they similar. It's not mentioned in the paper...for all I know he mistook a drachm for a denarius ( I know you wouldn't Rasiel- point for effect)



OK, good on what we agree on. Also for Ras' sample, Ras is a very competent numismatist so I'm sure he didn't mistake a drachm for a denarius.

It is still possible that there was some form of systematic difference in the nature of the coin types chosen. But if this were the case then I'd say that was a stunning finding in itself, if, hypothetically, types with seated reverses differed from those with standing reverses. When one thinks of what realistic possibility there is that two different pools of coins were chosen, manufactured in different ways to different standards, I'd rate it as highly unlikely, excepting cases which themselves tell exceptional stories such as the fine silver Domitian types in the Snettisham hoard

It does seem much more likely that the differences can only be attributed to silver-content-measurement techniques. The choice of which or how many coins to test can have played almost no part in this. Instead technical judgements have to be made about the nature of the equipment used, and about the nature of how the sample was collected from the coins. Either or both could cause massive disparities in results, as has been already well commented on in this thread. This is of course a very long running controversy, so controversial that a lot of the first chapter in the Oxford Handbook of Greek and Roman Coins is given over to these issues.

Offline 77HK77

  • Caesar
  • ****
  • Posts: 432
Re: Woytek and weight of Trajan Coins.
« Reply #39 on: December 11, 2014, 12:48:38 pm »
I not referring to things like the specific types of designs but rather a comparison of sample variables.

In a perfect world I would make an alloy, strike coins, document my process and measure using both methods.
Ideally i hope to get the same readings..if i don't the variable lies in the equipment or process.

No one has the money or time

assumption one: both methods read the composition correctly
Assumption two: Proper procedure was used in both

which leads to sample variables
Hypothetically Woytek uses only horrible cleaned, terribly worn left to rot in acid soil samples for his test.
Rasiel on the other hand uses a single pristine, almost every micron of silver intact, sample
I expect to see a value of one, Woytek gives me .95 and Rasiel gives me 1.5
Individually understandable and not suprising but when compared one to other it appears as a large variance.

Understanding the sample pool helps judge whether assumptions one and two are adequate or suspect or if the variable in the sample pool explains the differing results.
It could also point to undiscovered variables.

on another tact
again hypothetically, the mint master is a little short of silver late in 112CE. As we compare samples and see a sudden difference from 111 to 112 and again to 113CE, we might not draw a conclusion, but would seek to understand the variance which could lead to additional testing and new information. I need to know data on the sample pool to to see the potential change and decide if I have an outlier or explainable varience

Offline Andrew McCabe

  • Tribunus Plebis Perpetuus
  • Procurator Monetae
  • Caesar
  • *****
  • Posts: 4651
    • My website on Roman Republican Coins and Books, with 2000 coins arranged per Crawford
Re: Woytek and weight of Trajan Coins.
« Reply #40 on: December 11, 2014, 01:04:56 pm »
The simplest explanation is that one or both testing methods are flawed.

One can look for way-out-of-ordinary bizarre sampling flukes such as you describe, but there's no need to look for a strange sampling bias when the fact that different test methods are used is just staring us in the face. I don't know the facts about the accuracy and repeatability of one or other method, but there are those who do have expert knowledge of the subject and can comment, and indeed are commenting, both in this thread and in publications.

There's no need to look for a complex and unlikely sampling bias explanations until after the differences in test methods have been ruled out as explaining the differences. Far from being ruled out, these methodological differences seem capable of providing large variances in results for reasons several people have commented on. If that's the case the issue of sample bias essentially becomes irrelevant.

Offline 77HK77

  • Caesar
  • ****
  • Posts: 432
Re: Woytek and weight of Trajan Coins.
« Reply #41 on: December 11, 2014, 01:27:04 pm »
we'll have to agree to disagree

I'm not looking for outliers or flukes- read my middle statement regarding the comparison of results from varied samples- but rather to understand if my method is even capable given the samples to work with

In that case it is important to understand potential variables as year of manufacture, wear, chemical alteration to decided if my method was adequate or even capable to the task.

if one uses a hammer and taste buds to analyze the composition of an Apple and the other uses a peeler and their nose to analyze an apple the only way I can begin to decipher the results and analyze the the effects of each method is to know both started with an apple. i should state both are equally valid but if asked to describe apple sweetness you would have a varied response because of the test method. Similar to the issue at hand

Offline rasiel

  • Procurator Monetae
  • Praetorian
  • *****
  • Posts: 75
  • I love this forum!
Re: Woytek and weight of Trajan Coins.
« Reply #42 on: December 11, 2014, 03:29:17 pm »
I only tested two coins of Trajan. It's all I had on hand at the moment I was called to come by and pick up the XRF gun. One was this pretty well preserved one http://www.tantaluscoins.com/coins/39186.php and the other worn to the nubbins. My results, for whatever good they are, are as follows:

39186   Ag 96.05   Cu 3.4   Au 0.072   Pb 0.482   
nubbins   Ag 96.76   Cu 2.59   Au 0.59   Fe 0.05

My assumption was rooted in Carter's assertion that silver enrichment is a phenomenon that affects a top layer approximately 5-10µ in thickness. I had no reason to question this and even if I did no way to prove otherwise. Take a hair and split it lengthwise ten times and one of those filaments would be about 10 microns. That's practically nothing. The delicate silvering of an Aurelian ant would be thicker. I can see that this premise is now questionable putting my figures for silver into doubt insofar as determining the amount of silver truly present in the coins. My point, again, is that if Woytek's model measured only cores, and this model is proven flawed, then it doesn't matter if he tested only two coins like I did or 71 or 7100 because the data is flawed and the study conclusions compromised.

His study and mine had different aims. He was focused on just Trajan while I wanted to get the big picture and see the decline from the time of Augustus through the mid-3rd century. Whatever the merits of surface testing (and in my case hardly any of the coins could be said to have had any of that top 5-10µ area intact) it shows a definite decline in silver at the surface so if we knew by what factor the surface is enriched compared to the bulk then we should be able to infer percentages reliably AND non-destructively. To give just three more examples:

Clodius Albinus  Ag 90.75   Cu 4.56   Au 0.46   Pb 0.55   Fe 3.3   Zn 0.065         Bi 0.31
Julia Domna      Ag 61.73   Cu 36.89   Au 0.33   Pb 0.394   Fe 0.07   Zn 0.13   Sn 0.47
Trajan Decius    Ag 75.04   Cu 22.23   Au 0.41   Pb 0.88   Fe 0.05   Zn 0.13   Sn 1.25

Even if the numbers are only reliable to within 5% isn't it interesting to find that however you slice it the silver content is more erratic than had been previously believed? The differences here are enormous and well outside of the curve explainable by surface Cu depletion. Note also that gold, unaffected by chemical degradation at the surface and playing no part in the silver-copper grain boundary tussles, is useful as a bellwether of silver content as it is intimately linked to the silver ores from which the silver coins are made. The model needs refinement but I have the feeling you should be able to determine Ag concentration in silver coins merely by analyzing the Au data!

Ras


Offline SC

  • Tribunus Plebis Perpetuus
  • Procurator Monetae
  • Caesar
  • *****
  • Posts: 6068
    • A Handbook of Late Roman Bronze Coin Types 324-395.
Re: Woytek and weight of Trajan Coins.
« Reply #43 on: December 11, 2014, 03:44:29 pm »
Full disclosure - I am hopped up on painkillers post minor and unimportant surgery.  So if I say something stupid just ignore me.

Andrew, thanks for calling me on the "more the merrier" comment.  You are right that it is not true and that 71 is overkill.  On the other hand 1, 2 or 3 is clearly not enough.

Ras, thanks for sharing all the details of your study.

I think that you are right that the way forward on this question needs to be to nail down the thickness of the area effected by surface enrichment/CU depletion.  I think that you could then use geometry to get some answers.  A very thin layer means that you can get a good "core" sample by delving down not too far.  It also means that surface readings should be highly variable as the thin layer is more or less worn away on existing coins.  However, a thicker layer and you need to revisit "core depths" of past experiments to see if they were really reached and you would also then need to do some geometric calculations to calculate overall fineness from surface and core sample and ultimately get to the key question (I think) of what went into the melting pot.

Shawn
SC
(Shawn Caza, Ottawa)

Offline 77HK77

  • Caesar
  • ****
  • Posts: 432
Re: Woytek and weight of Trajan Coins.
« Reply #44 on: December 11, 2014, 05:17:33 pm »
Ras,
thanks for providing the info on the Trajan coins

it would be an interesting project to look at a broader sample of the two coins.
There is a marked difference in Pb and Cu

It would be interesting to sample say three graduated "better" examples of the nubbins and similar "worse" versions of the pristine where we know they are produced at the same mint and within a measurable time frame one year etc. such information could lead to considerations of initial alloy content versus environmental affects. although still a small sample once you have an expected range, say Cu content changing with wear or widely swinging Cu content unrelated to wear, etc you can compare variables that might cause one method to report content different than the other




Offline PeterD

  • Procurator Caesaris
  • Caesar
  • ****
  • Posts: 1483
  • omnium curiositatum explorator
    • Historia
Re: Woytek and weight of Trajan Coins.
« Reply #45 on: December 15, 2014, 10:02:29 am »
Of interest is this video by Kevin Butcher and Matthew Ponting on the subject of surface enrichment (actually copper depletion) and the methods used to test the composition of coins.
http://www2.warwick.ac.uk/fac/arts/classics/research/outreach/podcast/vodcast/?podcastItem=depletion_silvering.mp4
They say that the thickness of the final silver surface of a coin is 100 microns and even more at the edges.
Peter, London

Historia: A collection of coins with their historical context https://www.forumancientcoins.com/historia

ras

  • Guest
Re: Woytek and weight of Trajan Coins.
« Reply #46 on: December 17, 2014, 02:39:51 pm »
Thank you for providing the video Peter. I don't hear the mention of 100 microns but assuming that that is their claim then I think they're making an argument against their own conclusions. 100 microns is the width of a hair. That's probably the wear on your average VF (as an unscientific guess). Notice that they're also making what appears to be a mistake when they show the Ag-Cu gradient and opt to base their conclusions by sampling only the core which is very poor in silver. This has been my point all along. You want your measurements to have, ideally, no bias one way or the other. If we can say that the diagram is representative of the gradient then the only accurate test is either to melt the coin down or dissolve it in acid (to get a homogeneous  solution) and make your conclusions from those tests. Logically, I just can't see how the core-centric testing holds up any better than surface-only. What am I missing?

Ras

Offline jmuona

  • Procurator Caesaris
  • Caesar
  • ****
  • Posts: 402
  • I love this forum!
Re: Woytek and weight of Trajan Coins.
« Reply #47 on: December 17, 2014, 04:49:36 pm »
Dear Rasiel,
first, the address is

https://www.forumancientcoins.com/numiswiki/view.asp?key=Denarii of Otho

The discussion here is approaching the situation where Medievals were discussing the number of teeth a cow has on the basis of Aristoteles rather than looking in the mouth of a fair number of cows. One should really read the articles cited in th paper on Forum I give the address for. Ponting and Butcher have done a lot of good work.
The video is not a substitute for the detailed explanation of the method. The method they use reaches MUCH deeper than 100 microns. There are no Roman denarii over 90% pure after Domitians debasement - TRP V - CENS POTES - being the cutting point. They are not making a mistake by sampling "only the core" - a more extreme example of the phenomenon is shown in my gallery for the Alexandrian TDs. Although weighing 3-4 tmes the weight of Roman denarii, they have about as much silver in them. The copper of a very thick surface layer has been removed with acid and the flan struck to solidify the surface and then the coin struck. This same thing was done in Rome, although in a less spectacular fashion.
The surface enriched denarius contains an x amount of silver (including trace metals etc) and the point here is that all methods looking at the surface layers give a "pure" silver result. Most of the metal in the coin is not "pure". With pure (Roman sense) silver coins this does not matter, but after the first Nero debasment (64?), it was never so except for a short period of Domitianus coinage.
Jyrki Muona

Offline SC

  • Tribunus Plebis Perpetuus
  • Procurator Monetae
  • Caesar
  • *****
  • Posts: 6068
    • A Handbook of Late Roman Bronze Coin Types 324-395.
Re: Woytek and weight of Trajan Coins.
« Reply #48 on: December 18, 2014, 02:01:43 am »
Ras,

Core-centric testing might be the best if you address some questions.  An important factor, as I noted above, is the true thickness of the enhanced/depleted zone.  Another important factor is whether the core is a generally homogenous zone.  If the second is the case then you should be able to do the math and figure out what percentage of the coin the core represents.  If high then core sampling would be a very valid way to get an idea of the original alloy.  Knowing the surface layer thickness and sampling the core and the surface should also allow you to get close to the results you could get with DA testing.

If the surface zone of a denarius of was 10 microns thick what percent would that be?  What if it was instead 100 microns?

Shawn
SC
(Shawn Caza, Ottawa)

Offline PeterD

  • Procurator Caesaris
  • Caesar
  • ****
  • Posts: 1483
  • omnium curiositatum explorator
    • Historia
Re: Woytek and weight of Trajan Coins.
« Reply #49 on: December 18, 2014, 06:46:03 am »
The only thing of interest is how much silver and copper or other impurities went into making the original flan. How much there is today is completely irrelevant. The original flan, before it was pickled, struck, buried and cleaned, was homegeneous (apart from the 2 phase microstructure). Subsequent pickling etc., eroded copper in the form of oxide from the surface and produced the typical surface enrichment. At no time did any silver move from the core to the surface, so the core of a coin represents exactly the ratio of silver and copper that went into the manufacture of the coin. Even if you melted a coin down, you would get a less accurate result than by sampling the core.

Incidentally, Butcher and Ponting have recently received a grant of 653K GBP to study the metallurgy of the silver coins from Severus to Gallienus, so they must be doing something right.
Peter, London

Historia: A collection of coins with their historical context https://www.forumancientcoins.com/historia

 

All coins are guaranteed for eternity