Mostly correct, except for the sliding windows getting smaller at the end. The sampling would end when the window close of the last sample window hit the end of the catalog. We don't want any incredible shrinking windows, thats for sure. Sample window size must match the predicted window size, and must not change, else the whole thing might as well be thrown out the window (the real one).

As far as merit, it addresses both of my remaining gripes, and probably that of others that would eventually bring the issues up...

1. It solves the partial window problem.
2. Sample windows are now applied in a non-arbitrary manner, removing an additional layer of subjectivity.

The reason I suggested to apply this method down to the second is that most all EQ catalogs I've seen use this resolution to define EQ time. It would, again, be subjective of us to round up or round down earthquake times to the next nearest minute. Granted, this is seriously petty, but like I say, I'm looking for a method that is unsubjective as possible, and, the computer is doing all the work anyways, so why not. If we were doing this by hand, that would be a different story.....

Michael