I seem to remember an issue like that (one piece being particularly difficult to find) back during Ep3 testing. The reason, as explained to me by ET, and if I’m remembering the correct issue (which I quite likely may not be), is that the randomization algorithm is partially based on the number of active players at the time of your pwning. Since there were very few active players in the test environment, it was biasing towards some numbers. ET tweaked the code to not be as dependent on it, but if you’re doing most of your pwning at the same time, mterek, try shifting that time by several hours and see if your drop numbers change.