I can’t tell if I’m procrastinating writing or doing do diligence but I realized I couldn’t say really what I wanted to say in this article without engaging more with Buttrey’s Crepusius data. Buttrey was ‘missing’ 171 reverse die numbers. Schaefer’s archive is only missing 114. A big improvement!
BUT Esty’s formula thinks that estimate is way too high:
By contrast I also ran RRC 360 (also one die per control mark and the controlmarks used are numbers). Here Esty’s model seems nearly perfectly right on for our predicted number of missing dies:
I’m so uncertain what to make of this.