OK, I found the problem - a variable in the loop that was supposed to set the trial-type was counting the wrong way and hence the trial-type never got set and was instead set from random contents of RAM, which for me happened to be "0x00000001" so I got nothing-but-Mutate all weekend. Other people got out of bounds values and bad Mallards.
From Stephen over on the DPAD project forums.
get your fix