The Call of the Wide

Jim Felici

I’m in one of my bossy moods today, so this column is about what not to do when it comes to type. This week’s no-no is making visible changes to the character widths of the type you set. The key word here is visible.
I was provoked to write on this subject by an election flyer handed to me in the local market ahead of the recent French elections for regional governments. When I saw the text, I had one of those “What’s wrong with this picture?” moments. The type looked weird, but why?

The answer is that the widths of the characters in this text vary wildly from line to line, giving the text a slightly seasick quality. Type detectives will realize that this is an indication that it was set using (or rather, misusing) Adobe InDesign, which is the only program that can vary character widths as a part of the hyphenation and justification process.
In the following view of that paragraph, take a look at the lowercase es in the highlighted section, and you’ll see gross variations in their shapes, from squished to almost round.

The idea of altering character widths to help with good type composition was hatched by type-design doyen Hermann Zapf (Palatino, Optima…) and adopted by Adobe some years back. (You can find controls for this, called Glyph Scaling, in InDesign’s Justification dialog box, accessed via the Paragraph palette menu.)
When these alterations are kept within a range of ±1% or so, they can have a surprisingly beneficial effect on consistent type spacing, because it allows the program to find more and better hyphenation points and line-ending opportunities. And within such a constrained range, the variations in character width are undetectable. But if you get too slack when defining that range, you can make a major mess, as in the sample above.
Some type purists believe that tinkering with character widths should be simply verboten, but I’m with Herr Zapf on this one. The key is visibility. It’s like the tree falling unheard in the forest: If you alter a character widths and no one notices, did it really happen?
In general practice, character widths get altered for two main reasons. One, as above, is for mechanical purposes, such as enhancing composition or fitting type into spaces where it doesn’t fit. As an example of the latter case, on a newspaper page you could probably get away shoehorning various headlines into place by altering the character widths by more than a few percent because the distance between them hides the contrast in their appearances. But squeezing one line of a two-deck headline and not the other is a formula for ugly. If you can see the differences in the widths, you’ve gone too far.
The other, much more troublesome reason for altering character widths is in an effort to create a new typeface, or rather a new typeface family member. This is a big mistake, and the result is never pretty.
Trying to create a new typeface by simply altering widths only creates a more or less distorted version of the original. If the changes are subtle, and the face’s original design is amenable (some faces can take a surprising amount of abuse), you can get away with some degree of alteration before the results start to look bad. But your room for leeway is slight, as the following examples show.
The samples on the right (expanded by 15%) don’t look wider exactly, more like stepped-on. Capital letters look disproportionately wide, and round lowercase letters look like they’re on a high-gravity planet. Seriffed characters look bulgy rather than curvaceous.

As you can see, seriffed faces, which already have much more contrast between thick and thin strokes than sans serif ones, fare particularly badly when their widths are altered. It happens because varying character width has an inordinate impact on the vertically oriented parts of a character. When making characters wider, these parts get thicker, while horizontally oriented parts retain their original weight. The increased contrast throws the shape of the character out of whack. Instead of Garamond Extended, you end with Garamond Distended.
The same principle applies when reducing widths, and for the same reason: You’re fouling up the balance of the vertical and horizontal forms.
There’s only one commercially popular face that I know of that’s the result of simple electronic width alteration: Helvetica Narrow. This face was a quick fix in the early days of desktop publishing to address the need for a condensed typeface, and it soon became a standard part of the PostScript and Macintosh core font sets. The following illustration shows that it’s a simple electronic variation on Helvetica Medium.
On top is a normal setting in Helvetica Narrow. Below it is a setting in Helvetica Medium, condensed electronically by 19%; it’s identical to Helvetica Narrow. Helvetica Condensed, at the bottom, shows what a real — that is, designed — reduced-width version of Helvetica looks like.

A close look at Helvetica narrow shows why there was so much eye-rolling in the typographic community at the time of its release. Suffering a 19% compression, its vertical stems are noticeably narrower than the crossbars of characters such as the H, whose widths they should match. The tops and bottoms of rounded characters are bulgy in comparison to the vertically oriented parts of the same characters. When compared with Helvetica Condensed (the true reduced-width member of the Helvetica family), you can see that these features are backward, and that in a properly designed version of the face, the vertically oriented parts of rounded characters are thicker than the rest. What slight stress Helvetica Condensed displays is vertical; in Helvetica Narrow, it’s horizontal. In Helvetica Condensed, the vertical and horizontal strokes of rectilinear characters such as H are equal in weight.
Altering character widths for design reasons, then, is not a good idea, unless distortion is the message you’re after, and that may well be the case. If you want to catch the eye by unabashed weirdness, this trick fits the bill.
That said, though, there is one situation when I allow myself to visibly distort the widths of characters: to create fractions when using fonts that contain no numerator or denominator characters. Figures needed for fractions are small, only about 60% as tall as the integer numerals they appear next to. But when reduced to this size, full-size numerals appear too thin and too narrow. Using a semibold face can often compensate for their wan appearance, but they’re still too narrow for type at that size to be comfortably legible. Increasing their width by 15% or so helps a lot. (It’s hard to see this in natural size online, but try it with a high-resolution printout, and the difference is obvious.)
The fraction on the left has been set using reduced-size numerals. The one in the middle has been set using specific numerator and denominator characters, which are bolder and somewhat wider, for better legibility. On the right, the characters in the original reduced-size fraction have had their widths increased by 20%, matching those of the real numerator and denominator. They’re still paler than the designed fraction numerals, but they’re much more legible than the originals.

Because small caps are also about 60% of their full-size cousins, they too benefit from widening in cases where they need to be faked. And given the relatively small number of fonts out there with fraction characters and small caps, that’s pretty often.