> I can't reproduce this either.

Strange. I'm attaching a (heavily compressed) screen shot that
illustrates the problem. (Please let me know if this is a breach of
list etiquette...)

This is what I get when I do "C-u C-x =" on one of the characters in
the problem area:

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
  character: 0 (48, #o60, #x30, U+0030)
    charset: ascii (ASCII (ISO646 IRV))
 code point: #x30
     syntax: w 	which means: word
   category: a:ASCII l:Latin
buffer code: #x30
  file code: #x30 (encoded by coding system undecided-unix)
    display: by this font (glyph code)
     -apple-monaco-medium-r-normal--14-140-72-72-m-140-iso10646-1 (#x30)

There are text properties here:
  face                 (:strike-through t org-table)
  font-lock-multiline  t
  fontified            t
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Does it have something to do with the font-lock-multiline variable?

> However did you consider using the symbol  " ± " ? ;-))

Funnily enough, I had thought of that. This is OK so long as as you
are not using formulae, but calc will not interpret a ± b correctly as
an error form.

Slightly off-topic, but another problem I had with using unicode math
characters in tables is that the special characters (e.g., √〈α²〉) are
variable width, even though I use what is supposedly a fixed-width
font. I imagine that this is very system specific (I'm using aquamacs
on os x), but I wonder if anyone on the list has a solution to this
(on any system...).

Cheers

Will

-- 

  Dr William Henney, Centro de Radioastronomía y Astrofísica,
  Universidad Nacional Autónoma de México, Campus Morelia