[texhax] Something funny about \mathop (fwd)

[Michael Barr]

I have answered my own question below.  The answer is found in point 13 of 
Appendix G of The TeXBook bottom of page 443 and top of 444 in my 
(the first) edition.  Although I didn't follow the technical details it is 
clear that TeX will lower a single character \mathop to center it along 
the horizontal axis.  It is not clear why an empty pair of braces defeats 
this mechanism but it does.  It is clear that this was intended for things 
like integral signs and not single alphabetic characters, but it doesn't 
distinguish between them.  Curiously, this is not indexed (not under 
\mathop, in any case) in the TeXBook.

Michael Barr

---------- Forwarded message ----------
Date: Mon, 26 Jul 2010 06:26:25 -0400 (EDT)
From: Michael Barr <barr at math.mcgill.ca>
To: texhax at tug.org
Subject: Something funny about \mathop

This file illustrates the "something funny"
\documentclass[12pt]{article}
\begin{document}

\def\I{\mathop{\rm I}}
\def\II{\mathop{\rm II}}
  This is an example $\I\II\I\II$
\def\I{\mathop{\rm{}I}}
   This is an example $\I\II\I\II$

\end{document}

The two definitions of \I ought to give the same results, but don't.  The 
lowering doesn't occur with \II.

Michael Barr