When I say o-ring, thats pretty obvious, but "ring" refers to each course of the case itself. If laid on its side, they can deform out of round so the mating surfaces that capture the O-ring don't line up as precisely as intended. That then makes for the possibility of the o-ring not being captured properly on assembly or damage during assembly. I don't think one choice was determined over the other in this case.
BTW, if I remember the story right several rings were involved in a rail accident some time after the explosion. Rather than scrap them out of hand as possibly deformed (which would have really been the conservative right-way-to-do-it) they were actually measured for concentricity and certified to use. I bet the guy who headed that exercise couldn't pass anything for a week when they launched!
