i've uploaded the archive.org o.c.r. for "the advocate"
-- the book that gardner suggested that i look at --
into the skeleton that was previously being used, at:
>   http://z-m-l.com/go/gardn/gardnp123.html

the last page of the book shows the changes i made:
>   http://z-m-l.com/go/gardn/gardnp126.html

the .zml that made these .html pages, as usual, is at:
>   http://z-m-l.com/go/gardn/gardn.zml

if you compare the .zml file to the original o.c.r.,
you can see that it is very similar.  it doesn't take
much work to massage typical o.c.r. output to .zml.

as no proofing has been done yet, the text is raw...

(although this book came from the internet archive,
it is a copy of a google book, which means the o.c.r.
is very shoddy, since google puts out low-res scans.
when archive.org does o.c.r. on its own scan-sets,
the o.c.r. is fairly good, since they're using abbyy.)

ordinarily at this point, in order to clean the o.c.r.,
i'd restore the linebreaks to gardner's p.g. e-text
by using the linebreaks from the o.c.r. as a guide...

however, gardner sent me a copy of the text as it
was _before_ he rewrapped the original linebreaks,
so i won't need to go through that boring exercise.

i decided to post this o.c.r. text anyway, just so you
could see what it would look like as it is "in process".
at this point, the structure of the thing is pretty solid,
in the sense that all of the paragraphing is correct,
and the chapter-heads are in place, and all of that,
it's just the scanning errors make the thing awful...

but if you look past those scanning errors, you can
see why this version is superior to the p.g. e-text:
it is obviously self-validating against the p-book
from which the text claims to have been generated,
since it is easily compared to scans of that p-book.
(or even against an actual hard-copy, if you prefer.)

even if the p.g. text _is_ accurate, it can't be _verified_
as accurate, not quickly and easily, like this text can...

-bowerbird