That's for one run of 100 lines.

For much larger runs it is around 62.8%

FWIW I created a Huffman code for 50**2 pairs to improve the compression. (So called "blocking" )

It only saved something like 1% of the average code length, which seems to indicate that the code is already very close to the optimal entropy limit.

See German-WP for theoretical background.

Cheers Rolf
_{(addicted to the Perl Programming Language :)
Wikisyntax for the Monastery FootballPerl is like chess, only without the dice}