Thanks timka for posting the results and great photos.
n.igma's analysis is excellent.
We sometimes give the ancients too much credit for their technology and forget that it was more like a craft than an industry. The casting process was far from perfect. Not only were impurities present but the mixing and pouring process yielded highly varied results.
Below is some stuff I wrote that is more related to copper alloys but applies somewhat to any metal alloys:
"The
alloy components are subject to “segregation”. This can be due to: chemical factors – for example lead is relatively insoluble in copper; physical factors – lead is heavier and therefore subject to gravitational segregation sinking in crucibles and moulds (therefore the end of a pour has more lead than the beginning as it
comes from the bottom of the crucible); and technical factors – molten alloys were not mixed well and were often allowed to cool too fast (generally the faster a metal cools the more separation or segregation between the individual components of the
alloy). The result is that the
alloy mixture is usually quite uneven throughout a single crucible or “pour”.
A test done in the 1980s by T. Peters that attempted to replicate Constantinian-era leaded bronze revealed how the
alloy mixture within one
cast bar of bronze varied greatly from one end to the other. While the bronze mixture was made up to include 1.4% silver (4 scrupulae of silver
per pound of bronze) and 12.50% lead (36 scrupulae), the starting end of the bronze
bar had only 1.15% silver and 10.35% lead while the finishing end of the
bar had 1.57% silver and 14.62% lead. This shows a variance of 30% in the silver content, and 34% in the lead content, through one small
bar. Sampling one end of the
bar would yield a 3 scrupulae
fineness standard while sampling the other end would yield a 5 scrupulae
standard, when in fact the entire batch was made to a 4 scrupulae
standard. This level of deviation is likely representative of the variation to be found within any group of flans
cast together. To further complicate matters Cope has shown that the
fineness of two halves of the same late
Roman bronze coin can vary significantly, further complicating sampling methodology. In an example provided by Cope, the values in the two halves of a coin were 1.08% and 1.42%. Once again, this would point to different
fineness standards of 3 and 4 scrupulae
per libra respectively if taken in isolation."
While I was
writing about the problems associated with
fineness sampling the issues highlight how variable the contents of coins from the same issue could be. If we combine this with the differences in the timing and duration of the heating and cooling processes and we could expect significant variance in the structure of coins (or more accurately the flans) that came from one casting.
So, building on n.igma's diagnosis I would think that in any given batch some coins may suffer more from inter-granular embrittlement and therefore simply be more susceptible to this sort of damage. In other words the coin (
flan)
had a pre-existing condition.....
SC