Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!tut.cis.ohio-state.edu!snorkelwacker.mit.edu!ai-lab!karl@ai.mit.edu
From: karl@ai.mit.edu (Karl Berry)
Newsgroups: comp.fonts
Subject: Control word for font design.
Message-ID: <12592@life.ai.mit.edu>
Date: 2 Jan 91 22:57:10 GMT
Sender: news@ai.mit.edu
Organization: MIT Artificial Intelligence Laboratory
Lines: 29

The word is, variously, `hamburgerfonts', `hamburgefonsiv',
`hamburgefonstiv', etc.

It's true that a single word is not going to help the average user of a
font, as Anders points out.

But there is some reasoning behind this word; it's not just random.
Namely, the letters in that word cover most of the elemental shapes in a
font (the [ill-defined] set of letters that covers all of them is called
the ``control letters'').  That is not to say that a `c' is an `e' with
the crossbar taken out, or that `d' is `b' flipped; of course they
aren't.  But nevertheless, to establish a (usually desirable) sense of
unity and coherence throughout the font, the shapes of `e' and `c' must
relate visually, in some way that I can't describe in words.

When designing a typeface, the letters `h' (or `n'), `a', and `e' are
often done first, since they contain many repeated elements.  Other
letters, like `f' and `g' and `t' are oddballs; there are no other
letters in the modern Latin alphabet like them.  

Finally, I suspect the ordering of the letters in `hamburgefonstiv'
tests common letter pairs; the word `ghbfsntviaeu' (or whatever)
wouldn't be nearly as useful.  No hard data on this, though.

You certainly do need other tests.  The `testfont.tex' file that Knuth
wrote to help him develop Computer Modern shows some of the
possibilities.

karl@cs.umb.edu