The Barth Type Caster

Type Height and Its Tolerances

Every letterpress printer knows that American type height is 0.918. It has become a sacred number. But it's an odd approximation.

The best history of American type height is in either edition of Richard L. Hopkins' book Origin of The American Point System for Printers' Type Measurement {Hopkins 1976}. In it he explains the first how the point system came to be adopted, but but then how its basic unit, the pica, was initially undefined. Two rival options were considered: defining a pica as 1/6 inch (an infinitely repeating fraction) and using the "Johnson Pica." This latter was the house standard of MacKellar, Smiths, and Jordan, and could be traced (it was claimed) back through a long history. The Johnson Pica prevailed.

The specification of the Johnson Pica as published by MSJ (in the Typographical Advertiser, Spring 1885; see {Hopkins 1976}, p. 60) defined it in terms of the centimeter, as 83 picas per 35 centimeters. This works out in to 0.016,018 inches. (As an aside, in addition to the several advantages argued for the Johnson Pica as reported in Hopkins, I would argue that it was the only sufficiently precise solution at the time. The definition of the inch has always been troubled (even today it has no legal basis), and in 1885/1886 (before the Mendenhall Order) there simply was no standard for an inch sufficiently precise for typefounding. The pica had to be metric.)

(But note that in Practical Typecasting Rehak reprints an ATF "Punch and Matrix Table" dated May 1902 which defines the pica as 0.166,044 ( {Rehak 1993}, p. 181). In practical terms, though (unless you're Lynn Boyd Benton), the difference of 0.000,026 inch doesn't matter.)

In the same 1885 published definition, MSJ related their type height (which became standard American type height) to the same 35 centimeter length: 15 type heights per 35 centimeters. This works out to 0.918,6 inches.

The problem, of course, is that 0.918,6 rounds to 0.919, not the familiar 0.918.

To complicate matters further, in 1902 ATF published a popular work claiming that standard type height was 0.918,5. {ATF 1902}, p. 18.

In Practical Typecasting, Rehak cites "the present accepted American Point Standard of 0.9180 for Type Height" and cites three ranges of allowable tolerances:

• +/- 0.000,5 for all sizes up to 72 point body at 60 point set
• +/- 0.000,8 for 72 point body over 60 point set [and I presume up to 96 point body at all widths]
• +/- 0.001 for 120 and 144 point bodies at all widths

Presumably these are ATF tolerances ( {Rehak 1993}, pp. 177-178).