KK>    That makes things a bit more fun.
 KK>    So then you are not only keeping duplicates, but keeping a
 KK>    record of how many collisions exist.
Er, no, I don't think so. I'm simply tabling, sorting and printing a "list" 
of how many times each unique char array occurs in the source data.
Problems so far:
1) With a linked list, the source data is big enough that I lose an awful lot 
of overhead searching the list to see if a given char array has been seen 
t;
2) With a hash table, I have the problem of sorting in order to rank the char 
arrays in descending order of frequency. I *could* change the hash table code 
slightly so as to not chain a collision off the existing table bucket, and 
therefore make the collision directly addressable within the table, but I 
start to get concerned about table size.
I'm starting to wonder if I can't get the best of both worlds by building a 
hash table (chaining collisions) and then using the hash table to build a 
linked list which I could then sort. I wouldn't have to search that linked 
list at all, because I'd have taken care of the "unique char array" issue via 
the hash table.
I suppose I'm going to have to start doing some timings.
--- GoldED 2.50 UNREG
---------------