Estimation of the number of original dies 

To calculate the original number of dies that have been used to produce this series, we will use the statistical procedures published by several authors, which we find perfectly detailed in the work of Leandre Villaronga (1).

The estimates are based on the proportion between the number of known coins of a certain type and the number of dies used to produce this type; we collect this information in the database. For example: we have registered 200 coins with the obverse of a specific emperor and in this set we have observed a total of 40 different dies; this would give us an average proportion of 5 coins for each die.

These two parameters are sufficient to apply the Guilbaud and Carter methods, while the Goods method requires additional information: the number of singleton dies, which are the dies of which we know only a single coin.

All of these statistics are automatically calculated in the database every time a new coin is added.

The precision of the estimate improves as the number of coins registered in the database increases and the coin/die ratio increases.

According to Villaronga (Op.Cit, p.98-102), a proportion between 2 or 3 coins per die is considered quite reliable in order to know the original number of dies for this type; if the ratio is greater than 4, we have around a 95% chance of knowing all the original dies for that type.

The following tables show the current status of the estimates.

 

Emperor Coins
Registered
Diferent
Dies
Known
Rate
Coins/
Known Dies
Number of Original Dies
Estimated by Goods' Method
Dies Stand.Dev. Singletons %Known
Augusto 318 69 4,61 72 2 12 96,2%
Vespasiano 316 76 4,16 80 2 15 95,3%
Tito 250 50 5,00 51 1 7 97,2%
Nervae 135 19 7,11 19 1 3 97,8%
Traiano 408 91 4,48 95 2 16 96,1%
Hadriano 90 24 3,75 28 2 12 86,7%
Antonino 541 94 5,76 96 2 14 97,4%
Marco 98 18 5,44 18 1 1 99,0%
Commodo 231 47 4,91 49 1 8 96,5%
Severo 107 19 5,63 20 1 4 96,3%
Alexandro 304 51 5,96 53 1 10 96,7%
Total 2798 558 5,01 579 5 102 96,4%

(*) In Total cells the global estimation is reflected, not the sum of partial estimations.
 
Original Reverse Dies Number Estimated
Type Known coins Known Reverses  Coins/ Dies
Rate
Estimated Reverses Stand.Dev.
+/-
Singletons % Known Reverses
Aquila 1072 206 5,20 217 3 52 95,1%
Pira 1726 361 4,78 383 4 100 94,2%
Total 2798 567   600 5 152 95%


An example - Nerva

Let's take as an example the restitution coins for Nerva which, although not the most abundant in the database, have a high proportion of known coins by die (7.11).

Therefore, an estimate of the original number of Nerva dies is considered to be more reliable than in the case of Antoninus Pius which, despite having a greater volume of documented coins, has a smaller ratio of known coins per die (5.76), due to the large number of different dies identified so far.

Statistical methods predict that these 19 known dies could be 20 originally with a reliability of 96.4%.


Minting considerations

If we consider the entire set of registered DiviSeries coins, we have a sample close to 3000 coins, exceeding the global proportion of 5 coins per die; being higher than that value in most subtypes.

A sample of this size, and with this proportion of coins/dies, allows us to identify practically all of the original dies, with a reliability greater than 95%. In fact, rarely when incorporating new coins into the database do we find a new die.

According to our current data, we estimate that around 600 obverse dies were produced and a few more for the reverse, distributed among the different types.

At the end of the chapter on mint attribution and issuing authority we noted a certain constant regarding the participation of DiviSeries coins within the hoards closed after the reign of Trajan Decius, we estimated it between 2% and 3%. Now that we also have an estimate of the total number of original dies for Divi, we can think about making a rough estimate of the total volume for antoninians that were produced during the reign of Decius. In this estimate we will not differentiate between the coins that Decius minted in his name and those that he minted in the name of the empress or her children, in the same way that we will not differentiate by mint; since when studying the volumes of coins within the hoards we have not differentiated by person represented on the obverse or mint of origin.

This opens up the possibility of, by extrapolation, estimating the volumes of emissions for other reigns. We understand that the error in the estimates can increase if we work exclusively with the presence of coins in hoards. That is why we would have to carry out studies of original dies for certain series in order to limit the error, which the calculation only through hoards can imply and which is much greater than the error that occurs when applying statistical methods based on the proportions of coins/dies.

From what we have seen so far, if the proportion in hoards of coins between the DiviSeries and the issues in the name of Decius and his family is 2.5% (1 to 40), we can make a first estimate of 24,000 (600 x 40) as a number of original dies for the issues ordered by Trajan Decius.

Given the years that I have spent calculating the original dies of the DiviSeries, the proposed new work for Decius' antoninians is a task that exceeds the effort of one person. The proposal remains made as a line of work for future generations.



(1) Villaronga i Garriga, L., Statistics applied to Numismatics, Spanish Numismatic Association, Barcelona, 1985, pp 98-102.