Re: Widths in tables

Wed, 26 Apr 95 21:35:11 EDT

> This should be documented as meaning "big points" (Adobe "points") of
> exactly 72 to the inch rather than standard Anglo-American printers
> points of 72.27 to the inch (or indeed continental European Didot
> points).

That's what I proposed in the first place:

pi = 12 pt
in = 6 pi = 2.54 cm (exactly).

That other pica (the traditional U.S. pica, formally defined as
0.16604 in.) was a compromise arranged by the big U.S. type foundries
somewhere in the late 19th century and later adopted by the British;
before then, each foundry used a slightly different value. The odd
U.S. point/inch relationship is thus a relatively recent kludge that
runs counter to the basic thinking that used to be common to virtually
all of the European systems of measurement, which divided the inch
unit -- whatever it was called -- into 72 subdivisions, usually called
points, and the point into 12 subdivisions, usually called lines.
This wasn't just for printers, it was for everybody. The whole-number
relationship between points and inches that people seem to ascribe to
Adobe is a return to the basic idea rather than something new.

> Better yet, how about "pica"? I mean, really, four characters
> is not such an onerous burden.

But then it wouldn't be HTML. A central design principle of the
language is keeping performance as high as possible (and typing as
easy as possible) by using the shortest possible names for everything.
Besides, pi and pc are very common abbreviations for "pica" in nearly
all desktop publishing applications. In some, the two abbreviations
pi and pc are interchangeable; perhaps we should allow that here.

> point/points/pt/pts (pt is OK for Point, so is plain p if you must)
> pica/picas,
> inch/inches
> mm
> cm/cms
> m (metres, very optional, could also be miles :-))
> em (a square of the current type size, e.g. 12 points when using a 12pt font)
> en (0.5 em exactly)
> diderot, dt (point diderot)
> cicero (point cicero)

Yes to mm, no to m; meters are found in typographical specifications
about as often as parsecs. Ens are not generally used in typography
as a unit for specifying widths wider than a single en, as I pointed
out before, and are completely unnecessary in this scheme. The Didot
system is simply an analog of the Anglo-American one based on the foot
standard given to Charlemagne by Haroun-al-Rashid instead of the foot
standard of England; it doesn't need to be specified here because
we're not establishing a real system of length measures but rather a
set of aliases for unstated ratios to pixels. The same applies to
ciceros, which are just an attempt to get picas back into the metric
system. Points and picas are part of the necessary conceptual
apparatus of typography, but we don't need more than one of each in
the virtual system of the screen.

Aside from the pixel itself, the only real unit on the screen is the
point, which has to be determined by the designer of the interface in
order to display type. Picas and inches are merely handy aliases for
whole numbers of points. All the rest are aliases for various strange
and (in my opinion) not very helpful non-whole-number relationships to
points. Outside of making it easier to import legacy data, the only
reason for allowing mm and cm is so that printouts work as expected,
but I suspect that a case could be made for getting rid of them, too.


Jon Bosak, Novell Corporate Publishing Services
2180 Fortune Drive, San Jose, CA 95131 Fax: 408 577 5020
A sponsor of the Davenport Group (
The Library is a sphere whose consummate center is any hexagon, and
whose circumference is inaccessible. -- Jorge Luis Borges