Fraction of Surviving coins

From:  Ken Martins KenMartins@a...
Date:  Mon Feb 12, 2001 0:07am
Subject:  (was Raven Index) Percentage of Surviving coins?

<< The "Raven index" is a measure rarely used by name any more. It is,
for an issue that has been the subject of a scholarly die study, the
number of coins known divided by the number of dies known. For most
important Greek silver issues it is above 4 and sometimes above 10. >>

If I understand this correctly, we can surmise, roughly, the
percentage of surviving coins.  For example, Warren stated that the
typical range of "Raven numbers" is 4 to 10.  Let say 7 is the
average.  In the past we have discussed on this list that a die set
may produce somewhere between 500 and 10,000 coins before replacement
or recutting.  Once again, lets assume an average die set produced
2,000 coins.  Therefore, 7 of the 2,000 coins have survived and been
discovered, which is 0.35 percent.

Can the numbers be used this way?  What do you folks think?



From:  Robert Kokotailo, Calgary Coin and Antique Gallery 
Date:  2001 Feb 12, 1:07pm

At this time, trying to do any such numbers is pointless.  All
estimates as to the number of coins per die as simple guess work, as
are most of the estimates of the numbers of coins from each die that
currently exist.

The problem is that most such estimates are based on the specimens
that pass through major auctions with images, and are in the few
really major collections.  They do not take into account the specimens
that pass through the market without ever being imaged, which is the
bulk of what exists.

We basically end up with something like :

Original strikings from the die : 10,000 + or - 8,000

Surviving specimens specimens : somewhere between 5 and 50.

Average survivial % somewhere between 5 our of 18,000 = .002% and 50
our of 2,000 = 2.5%.

Basically it means little.

Only for the very rare and expensive coins, which tend to be sold by
public auction with images, or only reside in the major collections,
can any of the estimates of survivals be expected to be close.  But we
still have the problem that no one has any real idea how many of each
die would have been originally struck.



From:  "Chris Freeman" 
Date:  Mon Feb 12, 2001 3:14pm

Excellent points. I would add from working with early american copper
coinage, original mintage is sometimes entirely irrelevant with
regards to survival amounts. There is a die of an early american cent
that shattered very early, (known to us since the devices were very
sharp, indicating little wear), but is relatively common despite the
fact that probably only a thousand or less were minted, but there are
other dies that are extremely worn, (indicating heavy use and probably
many thousands minted), but are very rare with but a few examples
known. The areas where the coins were issued and distributed, both for
climate and coinage demand, as well as "luck", whether examples of the
coinage happened to be stashed away as savings, are probably much more
important to surviving specimens than initial mintage.

Given these facts from relatively "modern" coinage, I think we are
left to doing census' on desirable ancient coins, always remembering
that there is a hoard out there just ready to be discovered!



From:  Warren Esty 
Date:  Mon Feb 12, 2001 4:05pm

 The fraction of coins still surviving is known to scholars as the
"survival rate." Fifty years ago, some scholars guessed it was 1 in
5000.  Now, many more coins have been found, we have
more-sophisticated statistical methods, and far more issues have been
die-counted. We know more.

 But, we still don't know the number of coins struck per die, on
average.  Some say we can't know and would be best off not to even
guess. But, if you are willing to stick to *averages* and not push the
data too far, we can give a reasonable crude estimate of the survival
rate using the techniques you mentioned, Ken.

 The data for die-studied Hellenistic Greek coins suggest that
important silver Greek coin types worthy of being photographed in
major sale catalogs almost always exist in at least 3-10 recorded
coins per die. And, these numbers refer to *recorded* coins that are
in museums consulted, or coins that have been published in major sale
catalogs that the researcher used.  That probably leaves out many
examples. People (like Ted Buttrey on this list) who participate in
ongoing die studies find that, if they keep looking, year-after-year
the sample size increases. A thorough study can yield X coins, and ten
years later the sample has increased dramatically to 2X or maybe much
more!

 Imagine the difficulties. Write or visit twenty major museums. Pay
for photographs. Scour 10,000 sale catalogs (an omit another 10,000
that just might have an example). Photograph those photographs. And
realize you have missed almost all the smaller museums in the Greek
and Roman world, many coins in private collections in the west, and
almost all collections in the regions where the coins were made and
lost.

 Furthermore, many coins have been found since some of those
die-studies were done. Those usually don't affect the estimated number
of original dies (the estimators are designed to compensate for sample
size), but they do increase the number of coins per die.

 So ten coins per Hellenistic Greek die is probably a minimum
average. Now we need a guess at the number of coins struck, on
average, from dies.  Controversial subject!  But, the slim evidence we
have suggests it may have been 20,000 to 40,000 coins per die. Many
issues, particularly small issues, may have averaged substantially
less. But, big issues, like many Roman Republican issues, probably
averaged in that range. By the way, very extensive die studies of
Republican issues may find about 6 coins per die (Those of you on this
list doing die studies may wish to chime in with your numbers). But,
very many Republican coins are sold without a permanent record in the
form of a printed sale-catalog photograph. Surely there are far more
Republican coins extant than have been published.

 So, take 20,000 coins per die (if you are willing to play with crude
numbers and take averages) and divide by, say, 10 coins per die, and
you get about 1 in 2000 surviving.  For big issues in which the dies
were very heavily used (I'm thinking of mid-third century AD
radiates), the average number of coins per die may be larger, but the
number of coins in huge hoards probably increases the survival rate,
too. Of course, there are many caveats. Don't take "one in two
thousand" too seriously, but I think it is off by less than a factor
of ten.

 I would not be willing to assume the same numbers apply to AE or gold.