\title{Baskerville Production --- the horror story}

\author[Robin Fairbairns]{Robin Fairbairns\\Computer
  Laboratory\\University of
  Cambridge\\\texttt{robin.fairbairns@cl.cam.ac.uk}}

\begin{Article}

\subsection*{or --- never mind the quality, feel the \protect\texttt{hsize}}
I've been part of the \BV\ production team for three issues now; you
may have noticed my name in Sebastian's `Colophon'.  What I want to
discuss in this article is the production business as I (and Jonathan
Fine) see it.  Sebastian's story is entirely another thing: he's the
one that does the \emph{real} work --- we simply proof-read and get
what he turns out onto paper.

Two things prompted me to produce this article: the first was an
actual technical production achievement, which I thought might be of
use as a `technical tip' (there's a standing action on all of the
\UKTUG{} Committee to produce such article for \BV).  The second
prompt came in a mail message from Allan Reese, who said ``[\ldots],
it seems to me that there has been a substantial reduction in the
physical quality [of
\BV{} production]''; I hope to shed some light on the effect that
Allan observed.

\subsection*{From proof to paper}

Sebastian works on \BV{} wherever he happens to be at the time; his
domestic arrangements are such that this often means `at home'.  When
an issue is nearing completion, he (e)mails the committee to say that
a proof \PS{} file is ready for inspection.  Those of us who have an
account on the CTAN archive \verb'ftp.tex.ac.uk' can then log in and
retrieve the file that he's placed in his private filespace there (one
of Sebastian's other hats is that of CTAN maintainer).

The \PS{} file is typically in the range 1.5--2MB long; pulling
several copies (as I have to do) is only really practical politics
because I have access to plenty of free disc space.  More
significantly, the file takes anything up to an hour to print (all the
printers I can use are connected by serial line); so I don't
\emph{print} more often than is absolutely necessary.  (Most of the
bulk of the file is made up of downloaded copies of the fonts: early
on in the present regime, Sebastian produced an issue as seven
four-page files, the largest of which was 1.7MB, and took nearly an
hour to print on its own!)

Once I've printed a copy, there's a rush on to proof-read.  This is
something I tend to be good at (I've a lot of experience of
proof-reading ISO standards, a peculiarly nit-picking way of spending
one's time).  I (and others) mail lists of perceived problems back to
Sebastian, and we iterate until all seems right.  Because everything
happens asynchronously, this iteration can be a long process.

Eventually, a week or two later, I have a revised file to print.  I
duly go ahead, using the `service' HP LaserJet~4 here.  While it's a
damned good printer, I have had unsatisfactory prints out of it~---
among other things, the toner seems always to run out just before I
need to do a production run!

At this stage, Jonathan Fine takes the results of my efforts for
copying.  My laboratory has a very respectable print room, with a
Xerox Docutech copier, but we don't use it because the cost is nearly
twice what we pay the copy shop (part of the reason is that the
price break at the copy shop is at a smaller number of copies than
that in my laboratory's print room).  To prepare the matter for
copying, Jonathan has to paste the A4 pages from my printer onto A3.
This game (`imposition') almost deserves an article in its own right.
It prepares \BV{} for presentation as you see it, stapled down the
middle of (typically) six sheets of A3 paper, and is something the
Laboratory's Docutech would do electronically, but the copy shop can't
do.

In fact (in my opinion), the weakest part of the
production chain is the copy shop; they are providing as cheap a
service as they can manage, the result of which is that we
occasionally suffer mis-stapled copies, copies with pages missing, and
so on.  (They are, at least, happy to put such problems right.)

All that remains to do is to stuff and then post the envelopes.  This
isn't a trivial matter, but its `technicalities' are hardly
significant to a group of people whose common interest is typesetting:
I'll say no more.

\subsection*{What could we do better\ldots?}

Baskerville's early (somewhat irregular) issues were produced at Aston
University on a phototypesetter, and then reproduced onto high-quality
paper.  This gave the impression of high quality\footnote{Though there
  are those who complained that those versions of \BV{} were
  `under-inked'}, but it bore a cost that we could probably not meet
on the punishing publication rate of \BV{} at present.

We could print onto better-quality paper: ``reprographic quality''
paper for use in laser printers is available: its use would presumably
improve the reproduction of what we've printed.

I could find a more lightly loaded (and hence more consistent quality)
LaserJet~4 (or other 600dpi or better printer); this may or may not be
possible~--- if anyone in the Cambridge area has such a machine to
offer at the right price\footnote{\relax\emph{Very} cheap!}, we'll
gratefully accept.

We could improve the imposition process and finally, we could go to a
better copy shop.

\subsection*{\ldots{} and why don't we?}

I'm always willing to consider a better printer; but since I know of
none, I can't use it.

A copy shop with tighter quality controls would cost us more, and
would probably (as a direct result of the controls) take longer.

The view of the committee is that \BV's quality is (just about)
adequate as it is.  We are producing the newsletter of a (not very
large) user group: the imperative is that we get the thing out as
often as possible~--- our members more often comment about the content
of \BV{} than about its presentation.

\subsection*{What \emph{will} we do?}

We are going to try reprographic quality paper: since the amount of
paper used for \emph{printing} is small, the incremental cost per
issue won't be serious.  On a similar basis, Jonathan has bought some
`professional' spray gum for use in the imposition process (it'll be
much quicker than using PrittSticks, and the results ought to be
better).

Unless (or in any case until) we find somewhere better, we will try to
improve our relationship with the copy shop.  When they make mistakes,
they lose money by having to redo work for us: we expect they'll be
happy to work with us to improve their service to us (thus,
incidentally, improving their profits).

\subsection*{The technical nugget}

Stuck on the wall above our LaserJet~4 are a couple of examples of
misprinting (one is a print of the University crest with a spike
sticking out several inches to the North-East).  Notes on the examples
suggest that the solution is ``to use another printer'' (it's evidence
of a firmware bug in the \PS{} interpreter).  In the last
edition of \BV{} (volume 4, issue 2), some text at the bottom of page
20 `jumped' from near the beginning of column one to just before the
beginning of column two; the effect was to obscure the line in both
columns by overprinting.

Changing printers wasn't actually an option for me, since I was
committed to using a LaserJet~4, and I know of none which don't have
the bug.  Some time ago there was a discussion on Usenet group
\verb'comp.text.tex' about just this problem, and I retained a
suggestion from Hans Visser (of the Technical University of Delft)
on how to solve the problem within \verb'dvips'.  But I don't run
\verb'dvips' for \BV{}~--- Sebastian does, and there's a delay
introduced by any interaction with Sebastian; so I decided to try and
solve the problem myself.

Visser's solution was to edit the definition of the abbreviated
command `\verb;w;' in \verb;texps.pro;, one of the files \verb'dvips'
inserts into the output of any file it produces.  The change he
suggested was to change the original \verb'/w{0 rmoveto}B' to
\verb'/w{10' \texttt{5 rmoveto 10 sub -5} \verb'rmoveto}B' 
% line split up to allow it to wrap while I'm previewing; may not be
% necessary when it's being printed in Baskerville...
(the command \verb'B' is defined earlier in the file to mean
\texttt{bind def}).

My problem was caused by \PS{} that said \texttt{(er)s(ville)} (part of
\BV{} as a font name).  Remembering Visser's instructions, I traced
the definition of \verb's' to mean \verb'/s{show 3 w}B'; wanting a
`quick fix', I decided simply to change the offending instance of the
\verb'w' command.  My third shot,
\texttt{(er)show 1.5 100 rmoveto 1.499 -100 rmoveto(ville)} finally
did the job.

I wouldn't recommend the game I played (locating stuff in \PS{}
generated by \verb'dvips', and then correcting it) to anyone; but it
\emph{did} work.  I would be interested to learn if anyone else has
encountered this same firmware bug in LaserJet~4s (I'm told it's been
eliminated from current production machines), and if there are
alternatives to Visser's solution.  In the meantime, I'm asking
Sebastian to change \emph{his} \verb'dvips.pro'\ldots

\end{Article}
% Local Variables: 
% mode: latex
% TeX-master: t
% End: