Despite what your dialogs and panels may say about the size of your type, some faces simply look smaller or larger than others. Some of this is an optical illusion, but often they actually are smaller or larger than “normal,” obliging you to make unexpected layout and typographic adjustments.
Some faces, for example, are simply small in the body, as the typographic expression goes. In the era of metal type, this referred to how large the character images of a given font were in relation to the metal block—the body—they were cast on. Those blocks are gone, but digital characters are still drawn in an equivalent standard space based on the em square. This em square is essentially a canvas on which the type designer works, and these days its width and height are divided into increments of 1/1000. While the width of that canvas may vary according to the width of individual characters (the Q is 823 units wide), its height always remains the same, a full 1,000 units. Encoded within each digital font is the relative location of the baseline, so programs can assure that all the type on a line base-aligns even though the font or point size may change along the way.
In a font-editing program, this Q sits in a bounding box 1,000 units tall, like all its fellow characters. At 832 units wide, the Q’s width is unique.
Although the canvas for drawing characters may be standard, how much space those characters take up is not. In this example, all the samples are set at the same point size. The most petite of the faces shown is Perpetua, which by any standard is small in the body. By comparison, Trump Mediæval ranges to the other end of the spectrum.
All of these alphabets are set in the same size. Trump sets almost 30% larger than Perpetua; to make Perpetua appear the same size as 18-point Trump, it has to be set at more than 23 points.
Apart from perhaps having to be set in larger sizes to be comfortably readable—especially over wide measures—faces that are small in the body may also need relatively less leading than larger faces.
The top sample here, set in Centaur, seems generously leaded, but it’s set solid: 11 on 11. To get a similar spacing feel, the Stempel Garamond sample below it gets another point-and-a-half of lead: 11 on 12.5.
One very practical impact of all of this is that you can’t define a page grid until you’ve chosen a typeface. A baseline grid that will work well for Centaur will not suffice for Palatino.
The x factor
Complicating your choice of leading value is the x-height of the face, which is a measure of the height of its nonascending lowercase letters. Faces with small x-heights tend to give the appearance of a lot of white space between lines of text. Conversely, faces with tall x-heights tend to make conventional leading values look insufficient because they cheat into the normally empty space between lines.
A lesser cause of the apparent size of type is the height of its tall characters: its cap height and ascender height—the heights of characters such as “l” and “f.” This is simply because there aren’t as many ascending characters in text as there are x-high characters. Cap heights—the heights to which capital letters range—tend to be proportional to ascender height, although in some faces these two are quite different. Ultimately, you can only lead type as tightly as cap and ascender height will allow before the tall characters on one line collide with descending characters (“g,” “j,” etc.) on the preceding line.
Although Goudy Old Style (left) and Times New Roman (right) have virtually identical cap and ascender heights, Times looks bigger because of its larger x-height.
The roles typefaces were designed to play also help determine their apparent size. Times New Roman, for example, appears slightly bulky for its size because it was designed for newspaper use. It has a relatively tall x-height for its somewhat crimped width because it was designed to save space and compose well across narrow margins. But these characteristics make it a dismal choice for book work. Popular book faces, by contrast, are typically smaller in the body but with more generous character widths that are well adapted to wide book columns. In the narrow columns of magazines and newspapers, they fare badly.
Similarly, the proportions of characters—and their resulting apparent size variations—can also come from efforts to design faces for use in specific size ranges. In metal types, it was common for each point size to have a slightly different cut, with the smallest sizes being broader, bolder, and with taller x-heights than those used for body text. Text faces, in turn, were sturdier and more generously spaced than those used for display purposes.
This practice is just taking hold in digital fonts. For example, certain fonts from Adobe now include so-called “opticals,” which are typeface variations designed for different point-size ranges. These variations tend to be far less dramatic than the ones used in metal type of yore. This example shows the difference, using classified ads in so-called agate type: 5.5-point type, set 14 lines to the inch.
Two very different approaches to reproportioning type for use in small point sizes: On top, Times New Roman is almost illegible at classified ad size, but its Small Text complement is radically wider with much a taller x-heights. The difference between Arno Pro Regular and Arno Pro Caption (below) is far more subtle.
Contemporary styles in type tend toward larger x-heights than in the past. Although arguments are made in favor of this on readability grounds, it’s more a matter of taste. One place where larger x-heights do matter is on webpages and text made to be read onscreen. In these cases, smaller x-heights may mean that text-size type is rounded down an extra pixel or two when rendered for the screen, making for pinched characters and harder reading in general.
For type destined for print, though, readability hinges on your ability to take the face of your choice and set it well. In this, there’s no one size that fits all.