Nathon John Hay
February 19th, 1999

Reading and Typography

Reading is unavoidable. Students read textbooks; fathers read newspapers; engineers read manuals; technicians read webpages; politicians read bills; Christians read the Bible, and the list goes on. Everyone reads something. Seeing, perceiving, and recognizing lines and dots as a form of language is a process that is extremely complicated yet necessary. Scientists have researched many aspects of the visual reading process, and one of the most immediately applicable areas of concern is in the field of typography. Researchers are attempting to answer two questions posed by publics such as graphic artists, magazine editors, résumé writers, and even standardized test publishers: What typestyle is best for what situations?, and How do different characteristics of a font affect different audiences?

The term font is a generic word used to express the general computer category of typewritten characters. Similarly, a type or typeset refers to a complete family of sets of characters having a certain fundamental design or structure. For example, the Courier type may include the variations Courier New and Courier Bold. Other typesets are Caslon, Quill, and Old English. Typestyle is used to categorize types by attributive similarities. Two of the most recognizable, and most researched, typestyles are distinguished by the presence or absence of serifs and by fixed width (FW) and variable or proportional width (PW) pitch. Types which display the serif feature add short, decorative lines to the tips of the characters; this line of print (12pt PW) is in Garamond and has serifs. Types such as Arial, as in this line (12pt PW), do not have the serif addition and are thus called sans serifs. A fixed width font may be like this Andale Mono (12pt FW), and a proportional width font may look like this Times New Roman (12pt PW).

Other attributes for fonts include size, letter width, stroke width, and leading. All of these qualities combine to produce a great number of options for users. Likewise, there are as many types and variations of users as there are fonts, and consequently these audiences have unique needs for reading and for font usage. The elderly prefer a different font than younger people (Keller 1997). Normal vision readers have a greater advantage in seeing than low vision persons, so special considerations must be made for those less fortunate (Mansfield, Legge, & Bane 1996), and persons reading from print are affected differently than those viewing a computer monitor. Researchers have considered these differences and have put fonts and their traits to the test.

Thomas Keller of Nonprofit World Magazine noted a survey conducted by the Disabled American Veterans (DAV). The DAV sent out mailings with a new type, a serif type, and reportedly claim that their bottom-line profits increased by $500,000 due solely to the change in typestyle. Keller also attests that the DAV's findings support the "conventional wisdom" that serif types are more legible and effective than sans serif faces. Unfortunately, his "conventional wisdom" contradicts that of graphics professionals, and his conclusions from the mailing survey are improperly made, as scientific research indicates that other equally subtle factors than the face are involved in readability (Arditi, Knoblauch, & Grunwald 1990). Nevertheless, articles like Keller's printed in widely read magazines could have far-reaching, negative implications. To claim from one survey that sans serif fonts are, on the whole, useless is both arrogant and unfounded, and, also, the "conventional wisdom" he promotes is too limited in scope to be wisdom. Gratefully, a fine point was made when he emphasized how important it is to know when and when not to use certain typestyles.

Normal vision readers, for instance, have little trouble reading any font, with or without serifs. The human eye and its cortical organization allows us to analyze and comprehend a near infinite diversity of variations in type. Because we can read so many texts with so many alterations, it is not surprising that most reading material is in a serif typestyle. Truly, serifs are more aesthetically pleasing, are the most common and therefore most quickly readable or recognizable, and it has been suggested by Rolfe F. Rehe as quoted in Keller's article that the serifs create a more distinctive shape for letters causing serif faces to be more easily perceptible than sans serif faces. Not so truly; Yager, Aquilante, and Plass (1998) show that at normal luminance at regular size, reading speed for serif fonts is equal to that of sans serif fonts thus challenging Keller's, and perhaps catholic America's, "conventional wisdom" that serifs are far superior.

Not so amazingly then, most other research presents serifs in a bad light. Yager et al. (1998) also show for low luminance levels the sans serif was read approximately 11.5% faster than the equally sized serif sentences. They also indicated that at a smaller size, in normal lighting, both normal and low vision readers read the sans serif faster than the serif. Leat, Li, and Epp's (1999) experimental pretest indicated that, in eccentric (non-foveal) vision, sans serif letter Cs allowed for a smaller acuity level (the smallest size that can just be read) than serif Cs, providing a slight explanation for Yager et al.'s discoveries about the greater versatility of the sans serif typestyle.

Too, varying stroke thickness and serif size may account for serif's low applicability at dim light levels and small sizes. The additions of lines and changes in thickness in the visual reading field may clutter, rather than help, our processing of data (referred to as crowding), especially with large amounts such as paragraphs in a novel or with small print as in a dictionary. Regan and Hong (1998), while studying recognition and detection of texture-defined letters, discovered that as more "noise dots" were added to a letter cell, letter recognition by participants was less accurate for serif Cs than for sans serif Cs. In other words, serifs may make letters harder to distinguish, and it may be extended that other slight influences on letters can change their identifiableness as well.

Another overt variable in the function of legibility is kerning or pitch (spacing between letters), and reading research has focused much on this area. Overall, results indicate that one pitch style is not better than the other; rather, they have complementary functions serving in one situation where the other lacks. Arditi et al. measured subjects' reading rates for sentences of differing pitches and sizes. For medium to large sized fonts (about 18pts), the proportional width (PW) type was read faster. Mansfield et al. replicated the results showing maximum reading speed at normal size of PW fonts over fixed width (FW) fonts was 5%. For font sizes nearing the reading acuity limit (about 9pts), Arditi et al. found the FW type to have been read faster. Mansfield and colleagues reproduced similar results as well. Arditi et al. explains these paradoxical outcomes stating that crowding of the PW characters at small sizes makes them illegible. Interestingly, Yager et al. reported FW serifs as having been read faster than PW serifs (remember Regan & Hong's suggestion that serifs may cause crowding). Also discovered by Mansfield et al. was a size differential. For PW typestyles to be as equally readable as FW styles, the PW font had to be 15% larger.

In summary, the following data were discussed (better refers to an increase in reading speed or legibility):

• for normal luminance and normal size, serif is better than sans serif
• for normal luminance and smaller size, sans serif is better than serif
• for low luminance, sans serif is better than serif.
• serifs may cause a crowding effect
• for smaller size, PW may cause a crowding effect
• FW serif is better than PW serif.
• for medium to large size, PW is better than FW
• for smaller size, FW is better than PW.
• sans serif has a lower acuity level than serif
• PW must be 15% larger than FW to share equal acuity levels

As such, the following conclusions can be presented. The best general type for all users may be a serif font of a large size with a fixed width. This combination allows for all audiences to receive the best of these style options.

More specifically, normal vision readers have few major problems reading any type, but may prefer a sans serif, fixed width font when in dim lighting or with small sizes, and a serif, proportional width font at other times. Low vision persons, on the other hand, are at a distinct disadvantage to normal vision readers and require special consideration. These individuals may prefer or need a large sized sans serif fixed width font because of the ease of readability and reduction of the crowding effect. They may also prefer a sans serif over a serif for any pitch, size, or lighting because of the sans serif's increased acuity.

Other users require unique typestyles, too. For example, very little formal research has been done on the typographical needs of the large population of computer users. One researcher (Geske 1996) examined the legibility of a sans serif font on computer screens and showed that bolding significantly increased legibility. Knowing this, though, is just skimming the surface. Readers viewing text on monitors are receiving direct light, as opposed to print which reflects light, and accordingly have heightened eye strain. Also, computer screens do not have the resolution capabilities that printers do, and as a result acuity is reduced. There is much left for others like Geske to investigate in order to expand our knowledge of the reading needs of computer users.

More fundamentally, though, any amount of research in the general area of typography would be welcomed. Many others like Keller make inaccurate assumptions without consulting proper data. Vrest Orton (1993) introduced a controversy about illiteracy among children by blaming US book and magazine publishers for creating "unnatural and monotonous" texts by using sans serif type. Behavior like this can only be counteracted by reversing the effects of misinformation. Until more robust, empirical studies are conducted analyzing all aspects of types and their characteristics and their affects on various classes of people, persons will continue to base their typographical decisions more on aesthetic pleasantry and convenience rather than on informed, accurate depictions of what works best for whom in which environments.

References

Printed in 12 pt Book Antiqua.
Some specific types may not appear correctly due to individual differences among computers.

Return to frontpage for SL 1999.