DeadProgrammer (a fine blog) has a post up about coin-counting machines wherein he mentions some of the datapoints we wondered about previously. (Of course, he seems to have just gotten his figures at the Mint, which is something we’re ashamed we didn’t do.)

(We note with amusement that his coin distribution is wildly different than ours; we suspect he cherry-picks the more usable change out before resorting to automated counts. We, on the other hand, just dump all our change into the jar every night, which should produce a more typical distribution. Or so we assume, anyway.)