Improved Selectric Composer Catalog Available

John Savard

On the Internet Archive, at

https://archive.org/details/ibm-selectric-composer-type-style-portfolio/

the author of a forthcoming book on the typewriter has uploaded an excellent quality scan (it is a large file, at 900 megabytes, though, I must warn you) of a fairly late Selectric Composer type catalog, thus giving most, if not all, of the typefaces for it.

John Savard

Looking through this book, I noticed that for both Bodoni Book and Century, the lower-case italic x was of the conventional style, rather than of the style often seen in the italics for Scotch Roman.

I went through the 1923 ATF catalog for comparison, and saw that in the case of Bodoni Book, this was not a mistake, but in the case of Century, it was a slip; Century Expanded definitely does use the other style of lower-case x, the one Donald Knuth liked, in the italic.

John Savard

Ah, but Century Schoolbook used the conventional x, and perhaps that is what Century for the Selectric Composer derived from.

Mark Simonson

Definitely. Century Schoolbook has always been more common than Century Expanded, and thus more likely to be adapted for something like this.

Jens Kutilek

Thank you for posting! I just love the composer fonts.

I have a physical copy of a German catalogue, from which I made a digitisation of Century. It’s included in this collection:

https://github.com/jenskutilek/quarantine-fonts

I have wondered about the design of Composer Century, some glyphs like a and g are very un-century-like.

Mark Simonson

If you look on pages 4 and 5 in the catalog, it says that there were only three alphabet widths, coded blue, yellow, and red, spread over seven different point sizes. Not only that, but if you look carefully, the relative proportions of the characters is the same for all fonts: The width of the lowercase alphabet is always a little bit wider than A through S in the caps.

So, basically, they had to design all the different faces to match the same set of widths. This would account for the strange proportions in the a and g of Composer Century. It's a bit like the kind of distortions you get with fixed-width faces. I would guess that the widths they chose were based on what they expected to be the most popular styles, like Press Roman and Univers.

Florian Hardwig

IBM’s engineers had come up with a system of nine units for all letter widths, which proved to be a challenge for Adrian Frutiger when he was asked to adapt his Univers to this coarse system:

[…] the biggest problem was that each letter of the alphabet was given a fixed unit, no matter which typeface was used. They had used Times as the basis, and texts that were typed using a classic typeface didn’t look like they were done on a typewriter – they printed well. With Egyptienne and grotesque typefaces, i.e. with all other styles, however, there was a problem. Let’s take the s for instance: in Times it’s relatively narrow, but in a grotesque like Univers it’s wide. And that’s exactly where it started to get difficult. The Univers s should have had five instead of the allocated four units. The g is too narrow as well. The classic g has a narrow form but the grotesque g has a wide one, just like a d or q. The crippled g is a typical characteristic of the Composer-Univers. But it was worse with the F and T. You can clearly see the big gaps.

From Heidrun Osterer, Philipp Stamm (ed.): Adrian Frutiger – Typefaces. The Complete Works. 2014.

John Savard

Mark Simonson said:

I would guess that the widths they chose were based on what they expected to be the most popular styles, like Press Roman and Univers.

In the January 1968 issue of the IBM Journal of Research and Development, in one of several articles describing the development of the IBM Selectric Composer, there was a table showing how IBM averaged the widths of a number of different typefaces to come up with the ones they used. The typefaces averaged were Baskerville, Bodoni, Garamond, Caslon, Janson, Meridien, and Univers on a photocomposer, and Bembo, Centaur, Rockwell, Gill, Perpetua, Clarendon, and Times Roman on a mechanical composer; both used the 18 unit system. This article was "The IBM Selectric Composer: The Evolution of Composition Technology", with Adrian Frutiger as author. So I'm surprised that he wrote that the unit system was based on Times Roman alone, as he would have known better.

However, the result they obtained does happen to scale very well to the widths used by Monotype for Times New Roman, and so Press Roman did get just about the best possible fit.

Although even in the case of Press Roman, the letter widths of the Selectric Composer were imperfect in one respect: the widest letters, m, M, and W got 9 units when 10 or 11 would have been appropriate. This resulted from limitations in how wide a character could be on the Selectric golfball element on the one hand, and the desire to support type sizes as large as 11 or 12 points on the other.

On my own web site, on the page

http://www.quadibloc.com/comp/pro01.htm

I have a chart where a number of unit systems are compared; from it, one can see that the unit system of the 7-unit ATF Typesetter was likely to have been based on Baskerville, one of the first typefaces used with it, and that of Quick-Set Roman was likely to have been designed around a Scotch Roman or around Century Expanded.

The later Electronic Composer allowed one to enter a two-digit code to select a set of widths to be used; this allowed, for example, the use of elements with the Cyrillic alphabet for Russian-language typesetting. Even on the mechanical Selectric Composer, one could remove and replace the coder - due to the inconvenience of doing so, and the expense of purchasing the additional coder, this was used only in a very limited fashion for certain types of specialized elements. There were "Presswire" versions of some typefaces where the digits were narrower, and there was a Copperplate Gothic element where the lower-case was all capitals as well but in a smaller size.

No doubt, if it was even considered at all, using the potential additional versatility of the Electronic Composer (and presumably both the Mag Card and Mag Tape versions of the Composer as well) to allow changing to different widths optimized for individual typefaces was rejected as it would have created a confusing incompatibility with the original mechanical Selectric Composer.

Stephen Coles

Thank you so much for posting this, John! I recently acquired and posted a 1966 Composer type specimen that may be the first one IBM published, but it’s in very poor shape.

BTW, said author is Marcin Wichary and the upcoming book is Shift Happens.

John Savard

The one you posted is interesting too, an early one that was a smaller-size booklet. I see you also posted some pages from a later one; I think I've seen those before on Flickr, and I was hoping to see more pages from that specimen. Oh, I see: they're not yours, they're from the Letterform Archive.

I figured anyone who followed my link would see Marcin Wichary's name (in the collection name on the Internet Archive), but I really should have mentioned it. The book is about typewriters and keyboards rather than typefaces, also an interest of mine.

Jens Kutilek

In case anyone is interested, I wrote an article about my experience digitizing some IBM Selectric Composer fonts: Reviving the IBM Selectric Composer Fonts (and especially about the math involved…)

Mark Simonson

I remember so many publications set with the Selectric Composer back in the seventies, notably various editions of The Whole Earth Catalog, all set in Selectric UN.

Mark Simonson

... and, I never expected anyone to revive the IBM Composer fonts digitally. Nice work, @Jens Kutilek !

Jens Kutilek

Thank you!

konrad ritter

No, thank you for digitizing the versions of Century you've done so far! I'm a big fan of them, especially Century 8. I can't wait to see your Century 10 -- I plan to use it a lot, when it comes out. I hope you'll let us know then it launches!