i did some more work on gardner's book, "the advocate".

the (presumably incorrect) lines from the o.c.r. are in red,
while the lines from gardner's proofed copy are in blue...
(at least i hope i didn't screw up the colors like last time.)

here's a sample url, using page-by-page mode:
>   http://z-m-l.com/go/gardn/gardnp123.html

here's a single web-page containing the whole book:
>   http://z-m-l.com/go/gardn/gardn-hybrid6.html

and here's a web-page with _just_ the diffs on it:
>   http://z-m-l.com/go/gardn/gardn-159diffs.html

we've got 159 diffs.  that's about all the preprocessing
that can be done on this file.  159 diffs on ~4000 lines is
_great._  saves a ton of time on word-by-word proofing.

in this last round of preprocessing, i did a spellcheck...

i should explain that i do not use a regular spellchecker.

i copy all the text into a program (which i coded) that
spits out a list of all the words _not_ in its dictionary.
if you're interested, here's the dictionary that it uses:
>   http://z-m-l.com/go/regulardictionary.txt

then i go through the list and cull out the words that
-- for whatever reason -- look to me to be correct...
(most typically, they're names and compound words,
neither of which class happens to be in my dictionary;
doing this culling saves me a lot of time that would be
spent checking and then by-passing these valid words;
it's basically a quick way to create a custom dictionary.)

this leaves me with a list of words that i want to check,
which i then feed into my banana-cream app, so it will
show me all the occurrences of those words in the text,
with the relevant page-scan displayed at the same time,
so that i can check the text against its scan.  it's handy.

by the way, i can also feed banana-cream a list of lines,
which it will then search for and display automatically...
this is what i'll do to resolve those diffs displayed here:
>   http://z-m-l.com/go/gardn/gardn-159diffs.html

-bowerbird