X-VM-v5-Data: ([nil nil nil nil nil nil nil t nil]
	[nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil])
Date:         Wed, 18 Jul 90 13:47:17 CET
Reply-To: LaTeX-L Mailing list <LATEX-L%DHDURZ1@unido.bitnet>
From: bbeeton <BNB@MATH.AMS.COM>
Subject:      Re: two-column output
To: Rainer Schoepf <BK4@DHDURZ1.uucp>
Status: R
X-Status: 
X-Keywords:                  
X-UID: 200

rainer's suggestion that additional stretchable glue might be added
at the top of the right-hand column is one way of accomplishing the
goal of getting both columns flush to the bottom, but, truthfully,
you don't (usually) want to do that because the results look awful!
if there's more glue above one column than above the other, the top
baselines won't line up, and if the type size of the first line is the
same in both columns, the reader can get a slightly queasy feeling, or
at least will notice that something is peculiar.  (and the effect is
even worse when more than two columns are involved.  ragged bottoms
are much less disorienting, even if not desired as a matter of style.
listen to the voice of experience.)
by the way, it makes little difference whether the column involved is
the first or second (or ...), except when one is dealing with the very
last column, which is usually allowed to be shorter by a line or so,
even when columns are balanced.  ignoring that case (which should be
easy to detect), it's just as likely for a first column to be all text
and a later one to contain a section heading.
well, then, another idea:  make the skips before section headings
fixed -- no stretch.  (the skips after headings shouldn't have stretch
anyhow, if i've learned my lessons right from richard southall.)  and
make the space occupied by section headings a integer multiple of the
text baselineskip.  that way, all columns should come out the same, even
if both on one page are underfull by the same amount.  (still doesn't
help with a spread, but ...)
                                                -- bb