# Re: \substitutesilent{l}{m}

• To: fontinst@cogs.susx.ac.uk
• Subject: Re: \substitutesilent{l}{m}
• From: Ulrik Vieth <vieth@thphy.uni-duesseldorf.de>
• Date: Tue, 27 Oct 1998 10:38:13 +0100
• In-reply-to: <3634E08A.2CB9A4DE@cs.depaul.edu> (message from Alan Jeffrey onMon, 26 Oct 1998 14:50:18 -0600)

```
> A couple of gotchas... you've got to be careful about fontinst not
> generating cyclic substitutions foo -> bar -> foo.  This is particularly
> true in the case of m/n, since LaTeX will use those as `emergency fonts'
> to load if something goes drastically wrong...

I think the basic idea behind all those substitutions in fontinst is
to provide users with a reasonable setup, but no more than that, when
installing a font family, which doesn't exist in too many weights.

Most commercial typefaces come in families consisting of two shapes
(normal/italic) and at least two weights, but whatever these weights
are called may differ (e.g. regular/bold, light/demi, book/medium, etc.)
>From the fontinst point of view, we just try to install whatever we
have in the appropriate series and set up reasonable subustituions,
so as to provide something useful for \mdseries and \bfseries.

If a font family is available only in regular and bold, we use that
for the 'm' and 'b' series, but we don't particularly care about 'l',
because we don't have anything appropriate.  In the end, LaTeX's
default substitution to \seriesdefault will take care of that.

If a font family is available only in light and demi, we not only
install those series as `l' and `db', but we also set up explicit
substitutions for 'm' and 'b', so that users will still get something
reasonable without having to change \seriesdefault and \bfdefault.

For this reason, the substitutions presently included in fontinst
seem to be necessary, but adding the inverse substitutions isn't
really needed.

Cheers, Ulrik.

```