Overall, the single most contributing factor I found to probe counts for LISCH and EICH was the key generation algorithm. I spent a great deal of time fiddling with this to try and get the best distribution of hashed keys, thereby reducing primary hashing. To look at the results with LISCH it appears I did it very well, as it is quite a bit less than the mean given in the book.
Originally, I had tried using the entire title concatenated with the artist as the alpha key, and converted the ascii characters to integers. This led to alot of primary clustering. I tried various combinations of letters from the title and artist, as well as adding the length of each field to form the final numeric key that was used for hashing.
In reflection, the hardest part of this program was the Binary Insertion with Two Buckets. It took alot of thought to figure out how to handle the extra right children and especially mapping the variant n-ary tree into an array. The end result is pleasing to me, as I did the work so that I could have used any number of buckets, and didn't constrain myself to just solving the two-bucket problem. This allowed me to reuse the binary insertion code and just alter the number of slots and buckets between the two sections.
The following is a table which summarizes the results of each of the previous sections.