I've been working on writing a program to do the same things that your program does. One thing that I've found so far is (when manually checking the data-test results which I still have a ways to go on), in regards to your statement "My method also eliminates the problem caused by clustering.", that there are some quakes that occur separately from the clustered events. Its not very often, though, so it probably wouldn't have that big of an impact on the final result.

As an additional note, instead of using the CD-ROM of quakes that you're using, I've been using the PDE on my end which isn't quite as accurate. But then, nothing is perfect and really doesn't need to be perfect until somebody consistently stays beyond the statistically significant barrier.

Nice to see that you're still around.

Dennis