jon said:
>   By cross-checking I mean
>   using #84 to polish #41445,
>   not the other way round;

i'd love to see you document that work-process.

because the image that springs to mind is that
you're washing a clean window with a dirty rag.

the clean window, of course, is the new d.p. text.
it's just come out of 6-7 rounds of close analysis.
i'm sure that it has errors in it, but not too many.

#84, on the other hand, would be the dirty rag...

and it's not that #84 is chock full of errors, either.
the reason it's such a rotten e-text, though, is that
its provenance is not just unknown, it is _murky_...
(can't remember why, at this point, but there were
_reasons_ noring/passey made it a stalking-horse.)

a few quick looks indicate you'd be looking at about
2500 diffs, and i'd guess 2450 of them will be clean.

the rest will be version-differences, or #84 errors...

that's a lot of work to locate -- maybe -- 50 errors.

or 25.  or 10.

(and to be honest, the number might be as low as 3;
and by then, you are doing more edits for the cause
of intra-book-consistency than "error-fixes" per se.)

and i know it, for sure, since i just did it, on "huck"...

the window/rag metaphor is not perfect, of course,
since you _will_ end up with a little cleaner window.

but the small benefit probably will not be worth the
high cost that you will have to pay in order to get it.
(if i could do it over on "huck", i would _not_ do it.)

that's just my opinion, though, and even if i'm right,
your time is _your_ time, to waste however you like,
so i welcome you to take up the challenge and do it.

and if you do, please document it, so we'll all learn...

***

robert said:
>   The page images are from a "photo-reprint"
>   of the 1818 Lacking, Hughes, et al. edition, which,
>   judging by the scans, is code for microfiche/microfilm.

oh geez, i wish i woulda known that before i put in
_any_ work on these.  do _not_ use reprints, folks!

if one of the _original_ editions isn't scanned yet,
the just wait until it happens, eventually, because
these reprints will never serve as canonical copies.

anywho, baby-que, here's a version i threw up:
>   http://zenmagiclove.com/frank/wisdom.py

since the so-called "page-scans" aren't real scans,
and the text has errors introduced by jon's process,
i ain't gonna use any more time to make this better.

for reference, wikipedia says that the 1818 edition
had 280 pages, but we've got 311 scans somehow.

of course, who knows if wikipedia is wrong or right?

i am drowning in a sea of undependable data being
pushed at me by people who don't care if it's wrong.
so probably time to get out of this ridiculous racket.

-bowerbird