I am especially grateful to Greg Beroza for allowing me to use his cleaned version of the IDA datasets, for providing the utility software, and for answering my many questions about the data. He and his students spent several man-months doing the manual cleaning of the datasets. I also owe thanks to Martin Karrenbach and Carlos Cunha Filho for helping me with my Connection Machine codes. Everyone at the SEP has been patient when I needed large amounts of disk storage, when both the tape drives were tied up with the large datasets, when the network became clogged with my data, and when I used the Connection Machine for hours at a time. I am most grateful to Jon Claerbout for allowing the work on this somewhat unconventional project.