My mind is literally blown.
I’ve just run the counts and the stats of RRC 282/1b where the reverse dies have letters and numbers. Along the way I counted dies missing in the sequences. So where letter of the alphabet or on number in the order was not in Schaefer’s archive but would be predicted logically to exist. I found 35 die names that I would predict to exist that I’ve not seen (yet).
Esty’s formula would estimate with 95% accuracy that we are missing 34-23 dies.
This is pretty damn close and makes me exceptionally relieved.
But those of you in the know are saying What?! How can that be?! Crawford says that there are more than one die be controlmark?! This is true and untrue. Leaving aside 382/1a which you’ll just have to wait to read about in this article, on 382/1b repeats (i.e. multiple dies with same symbol) are unknown for alphabetic controlmarks and are exceptionally common for dies below xxxx in numerical control marks (80% of the numbers are known to have more than one die. Most have only two dies: there are only four numbers that have 3; none have more) BUT exceptionally rare for xxxxi-ccxxvi (only 2 instances or just over 1%).
I have never been so glad I spent an incredibly boring morning counting and checking through a batch of coins.