Problem with nroff/troff

[Rick Zaccone]

I've been chasing down a problem with nroff/troff, and after several
days of trying, I just don't have any more energy.  Can anyone help me
with the following problem.

I've set up a test file that produces the rotten output.  It looks like
this:

Some text.
.bp
.EQ I
a rather complicated matrix
.EN
.bp
.EQ I
another complicated matrix
.EN

When I run the file this way, nroff gets confused with the second
matrix and jumbles where certain columns are (some things come out left
justified that should be on the right).  I've also made the following
observations:

- Eliminating the initial text and the first .bp causes everything to
come out properly.

- If run by itself, the second matrix comes out correctly.  In
addition, the code generated by neqn is the same for this matrix
whether it is run off by itself or with the rest of the test file.
This seems to indicate an nroff problem.

- The problem occurs with both the -me and -ms macro packages.

- This is part of a larger paper for which I am using troff.  When
troff gets to this section of the paper it quietly dies.

- Making the second matrix a little smaller causes the problem to go
away.

Does anyone have any ideas.  All suggestions are encouraged.
-- 
Rick Zaccone                    Penn State University
UUCP: {akgua,allegra,burdvax,ihnp4}!psuvax!zaccone
CSNET: zaccone at penn-state       BITNET: zaccone at psuvax1