[texhax] Cases for italic correction

Donald Arseneau asnd at triumf.ca
Sat Jul 21 11:02:31 CEST 2012

Barbara Beeton <bnb at ams.org> writes:

> On Fri, 20 Jul 2012, Philip TAYLOR wrote:
>
>     Barbara Beeton wrote:
>
>     > unlike all other letters in the cm math italic
>     > font, the d has no built-in "overhang" spacing.

Indeed the right side-bearing is minimal for d, and noticably
less than for l, but it is not zero.

>     > i've never actually asked, but i think the
>     > reason for this is that knuth uses d only for
>     > indicating derivatives (as in "dx"), not for
>     > variables, and for that use, he prefers it to
>     > be directly adjacent to the variable involved,
>     > not separated by a small space.

I think so too.

>     But what of a second derivative ?  That would require d^2y/dx^2,
>     which will surely generate almost exactly the same inteference
>     as Michael asserts obtains in d^3 (but which Lars and I cannot see).

The exponent is crowded, but does not collide with the d.
They might merge in a photocopy.

> however, i don't see the difference either;
> here's my test:
>
>   \documentclass[12pt]{article}
>   \begin{document}
>   \Large
>   $dx + d^2x + d{}^2x + d\/^2x$
>   \end{document}

Indeed those three spacings have to all be the same -- they
have the d followed by its italic correction. M Barr was right
to use a small space in his document, and wrong now to think
that \/ would have done the same.

An example that shows the (non-)differences is

\documentclass[12pt]{article}
\begin{document}
\setlength\fboxrule{.1pt} \setlength\fboxsep{0pt}
\Large
$dx + d^2x + d{}^2x + d\/^2x$ \fbox{\the\textfont1 d}

$lx + l^2x + l{}^2x + l\/^2x$ \fbox{\the\textfont1 l}

\end{document}

Donald Arseneau                          asnd at triumf.ca