I think the problem solution explanation for the SORT5 problem isnt clear enough. I came up with a solution which entails 8 uses of SORT5 and am not sure if this is the most optimised solution. I believe in a future version of the book you should explain the ideal solution a bit more clearly.
I believe the same applies for the Boyer Moore explanation in the Iterative Refinement of a Brute-Force section of the same chapter
My clearer explanation of the SORT5 solution for other people below : -
Use SORT5 to extract the highest (h1), second highest (h2) and third highest (h3) numbers for subsets S1 to S5 and then do individual SORT5 for the H1, H2 and H3 subsets of the highest, second highest and third highest. If there is a better solution please explain below.