A Little more on the live1 problem

It looks like ...
  1. Problem may only occur for bit 16 and bit 24
  2. Both live1 and live2 may share same pathology
This, however, is based on limited statistics and is not consistent with what Aram has seen.

Events printed from primex05128.dat.04 when the high byte (bits 24-31) changed. This seems to indicate the 24th bit of live1 always flips prematurely, but that it is counting in the right direction.

Events printed from primex05128.dat.04 when the third byte (bits 16-23) changed. This indicates that bit 16 doesn't always flip prematurely (see events 2584918 and 2585282).

Events printed from primex05128.dat.04 when the second byte (bits 8-15) changed. Scanning a few x 100,000 events, none were observed to have flipped bit 8 prematurely. This may indicate the problem only exists for bit 16 and bit 24 of live1.