I think taking different limiting processes corresponds to using different partitions.

]]>Couldn’t this problem (and possibly the more general problem of conditioning on events of probability zero) be dealt with merely by a passage to the limit ? i.e define a sequence of events that tend to the desired event (in this case, one could imagine ever-thinner bands around the equator), and define the conditional probability at the limit as the limit of the conditional probabilities ?

I imagine there are problems with this approach, and I was curious what they might be.

thanks !

]]>