So it appears "attached" by relevance... (despite thread being dormant over a year...)
I just fixed a 5216A with a peculiar, but ultimately similar, problem. At first I thought it was grossly out of cal, but the error was huge and changed depending on the sampling time/decade; whereas the well-aged internal crystal was dead nuts (<0.1 ppm error!).
Turned out the 5th digit appeared superficially to increment (even carried tens to the 6th digit), *except* it skipped 2, 3, 6 and 7. Somehow that digit's 1820-0093 decade counter (predecessor of the faulty 1820-0119 above) had a stuck bit, but still worked otherwise; it just divided by 6, instead of 10!
I surmised the 1820-0093 and -0119 IC pinouts were identical, based on schematic backdates. Indeed, a -0119, pulled from a later-model counter with other issues, works fine in place of the failed -0093. Which is lucky as the -0093 appears even more $$$ "unobtainium".
Also worth noting, these parts (and the unusual upside-down Burroughs B-5560 Nixies) are shared with the lesser 5221A and 5321A 'economy' counter models, which are more common, and less worth fixing.
Hope this info helps someone else down the line.