Crit seems like it would be a Binomial Distribution either crits or doesn't, has n trials, and are independent (though I think this last one may not be true, thought I read somewhere that they tried to make crit and dodge less RNG). So we make p the probability of it critting and q = (1-p) or the chance to not crit. p in your case was .04 and q would then be .96 so the chance to crit once would be ((.04)^1)*((.96)^0). The chance for it critting twice in a row would be ((.04)^2)*((.96)^0) = .0016. Isn't math fun.
However I don't think they are independent and that the chance is even lower.
Yeah I remember them talking about making crits not as random and more distributed, but I don't know how that was implemented since they were very vague about it. Also, I like my way of getting .0016 better because it has less numbers...
Anyway I assume Coal didn't actually mean 2 consecutive crits but rather 2 crits during a brief encounter, but even if you took ten attacks it would still be a 1.6% chance of taking two crits.
Edited by The Hound, 20 January 2012 - 09:25 PM.
@Feanor: I was reacting to Trahas not to Perrycox. Read it ffs.