Serif or sans serif?
« previous post | next post »
Most people care about their typefaces
Appearances matter, especially whether fonts have serifs or not.
"Font Wars Spread After State Department Replaces Times New Roman with Calibri
"'I'm banging my head against the wall;' camps divided in fallout from government efforts to make documents easier to read"
By Katie Deighton, WSJ (3/14/23)
One wonders whether it is a matter of functionality and efficiency or esthetics and taste. Whatever motivates the confrontation, one thing is evident, and that is that people have deeply held opinions in favor of / against one side or the other.
What sounds like a typeface tempest-in-a-teapot has boiled over in the U.S. and U.K., where changes in document requirements have set off a war of words among cantankerous font factions.
The State Department announced in January that Calibri would replace Times New Roman on official documents to make them easier to read. U.K.’s Home Office, for similar reasons, x-ed out the 83-year-old Times New Roman, which has the wings and feet on letters known as serif style.
…
“For every study out there that says that sans serifs are more legible than serifs, you’re going to find an opposite study,” said Maria Lindenfeldar, the creative director at Princeton University Press in New Jersey.
A 2017 study published in the scientific journal Annals of Dyslexia found that text in Dyslexie, a typeface designed to make reading easier for people with dyslexia, didn’t test any better—whether measured by speed or accuracy—than words in Times New Roman.
Some typeface executives say that, in fact, the serif flourish makes letters easier to distinguish. They cite the identical appearance in some Calibri fonts of the lowercase “l,” as in look, and uppercase “I,” as in India.
The serif versus sans-serif debate extends to nonverbal communication.
Typographer Sarah Hyndman, author of the book “Why Fonts Matter,” found that people saw serif typefaces such as Times New Roman embodying “traditional,” “conventional” and “trustworthy” values, she said; Calibri and other sans-serif typefaces were seen as “confident,” “friendly,” and “honest.”
…
Rebecca Creed, a Florida-based appellate attorney, had in the past used Times New Roman or Courier New for court briefs and other legal documents. In 2021, the Florida Supreme Court adopted a rule requiring Arial or Bookman Old Style, chosen for their readability on screens, for computer-generated documents.
At Ms. Creed’s law firm, Bookman Old Style won out.
“We just liked the way it looked,” she said. “That sounds dumb, but it’s really just what it came down to.”
Chacun à son goût.
Selected readings
- "Choose your font carefully" (8/22/22)
- "Decolonizing Chinese fonts by probing the past" (9/11/21)
- "Character building is costly and time consuming" (12/22/15)
- "Font adjustment: Times Beef Noodle" (11/14/18)
- "Ye Olde English katakana" (9/11/14)
- "Phonetikana)" (10/22/15)
- "Washirweng" (3/26/14)
- "An inconclusive psycholinguistic take on post-period spacing" (5/1/18)
- "Font rage" (7/5/07)
- "The interpretation of stones and typefaces" (7/11/04
- "Do what to the switch?" (5/4/16)
- "Idea: A new typography term" (Ironic Sans, 2/19/08) — about bad kerning, aka "keming" (David Friedman's coinage), with plenty of references to "flick" in the comments
- "FLICK BAD KERNING" (Notes from a Linguistic Mystic, 11/1/12)
- "Fuck Yeah Keming" (Flick Yeah Kerning!) — many hilarious examples
- "Why kerning matters" (Wug Life, 12/25/13)
- "Another chapter in the history of the Chinese typewriter" (8/14/21)
- "Agbègbè ìpàkíyèsí" (12/15/08)
- "Charaelerislie" (5/13/21)
- "How Michael spent his summer vacation" (9/14/08)
- "@Everybody" (7/26/20) — in the comments
- "I kept doingn it wrongn" (5/26/10)
Dick Margulis said,
March 19, 2023 @ 5:56 am
I have a proposition: Lawyers should be allowed to make decisions about enforcing typographic standards exactly to the extent that typographers should be able to make decisions about enforcing legal standards.
Maria Lindenfeldar at Princeton is correct. I suspect she went on to say this: the vast majority of so-called studies of legibility published in the last forty years or so have been conducted by undergraduate psych majors for class credit and have been so poorly designed as to be laughable. In virtually no cases did anyone actually speak with a professional typographer to understand even a few of the dozens of confounding factors, thus rendering their data meaningless. For lawyers and bureaucrats to rely on meaningless conclusions based on terrible data to make decisions way outside their lane is just ridiculous.
Thomas Lee Hutcheson said,
March 19, 2023 @ 6:04 am
FWIW, I switched to Ariel for personal use which I find similar to Calibri, but with slightly thicker (to me more readable) lines.
Thomas Lee Hutcheson said,
March 19, 2023 @ 6:07 am
Chat GPT has a very measured take:
Both Calibri and Times Roman are considered readable fonts, but they have different characteristics that can affect their readability in different situations.
Calibri is a sans-serif font that was designed specifically for digital use. It has a modern and clean look, with a rounded shape and even spacing between letters. This can make it easier to read on screens and in small sizes, as it is less likely to become blurred or pixelated. Calibri is also a popular choice for presentations and documents because of its contemporary appearance.
Times Roman, on the other hand, is a serif font that has been used for print media for centuries. It has a more traditional and formal appearance, with serifs that help guide the eye along the lines of text. Times Roman is generally considered more legible in printed materials, particularly for longer passages of text. It is often used in academic and professional writing, as well as in books and newspapers.
Ultimately, the readability of a font depends on a variety of factors, including the size and spacing of the text, the medium in which it is presented, and the preferences of the reader. It may be helpful to experiment with different fonts and settings to find the best option for your specific needs.
Dick Margulis said,
March 19, 2023 @ 7:23 am
@Thomas: As one would expect, ChatGPT is not a reliable source on the topic, given the amount of myth and misinformation in the corpus it trained on.
Robert Coren said,
March 19, 2023 @ 9:59 am
I note that the typeface I see on Language Log, whatever it is called, lacks serifs except for "1" and "I", which makes it acceptable even for a Times Roman fan such as myself.
Jerry Packard said,
March 19, 2023 @ 11:30 am
I think that, generally speaking, the simpler the font (fewer embellishments) the more quickly it will be processed. Now, that varies as a function of font size, so anything over 8-point will not yield a significant difference. My basis for saying this is a study on the relationship between reading speed and number of strokes in Chinese characters, which found strong effects with very small font sizes (simpler characters read more quickly) — an effect that disappeared as the font size approached 8-pt or greater. (Huang, K.-C., and Hsu, C.-F. 2005. “Effects of Numbers of Strokes on Chinese Character Recognition during a Normal Reading Condition.” Perceptual and Motor Skills 101: 845–852.)
Kimball Kramer said,
March 19, 2023 @ 12:34 pm
Robert Coren and Jerry Packard are on the right track. Devise a font that uses serifs only on lower-case "L" and upper case "i"—I've used the opposite to be clear—and increases serifs as necessary as the point size is reduced below 8 (?) points. A complex combination? Yes, but computers can handle it VERY easily. A bastardization? Yes, but very many fonts use letters (say, "o") that are identical to the same letter in another font. What matters is to be a clear and easy to read as possible; not to adhere to some made-up or evolved standard that has no importance compared to clarity and reduced probability of
understanding.
James Wimberley said,
March 19, 2023 @ 1:13 pm
The Merovingian Frankish kings are usually thought of as murderous, illiterate and ineffectual barbarians, but the abbeys they patronised did succeed in spreading the great typographical invention of ascenders and descenders (b/p, d/q, f/j (?), g, h, t, y). These gave us a great gain in readability from the earlier uncial script, beautiful but hard going. At least four slots corresponding to the inverses are still vacant, and available to the designers of Klingon and Elvish typefaces. Cf. the Tengwar inscription on Tolkien's one Ring ( https://tolkiengateway.net/wiki/Ring_Verse ).
Chas Belov said,
March 19, 2023 @ 2:33 pm
The font specification for this blog is "Verdana, Arial, Helvetica, sans-serif", and Verdana indeed does have a serif only on the capital I. I actually find Calibri quite difficult to read, and override with Verdana whenever I have the means to do so. On my work website, we use Graphein Pro, which I find even easier to read than Verdana.
Not a naive speaker said,
March 19, 2023 @ 3:51 pm
My preferred font for readability and distinctiveness of almost similar looking characters as in "1IiLl0Oo" is Source Code Pro
Robert Bloomfield said,
March 19, 2023 @ 5:20 pm
I've been using Atkinson Hyperlegible font, which is reviewed here . I had assumed that there was actually some evidence behind the claims that this is more accessible for people with vision impairments, but maybe not!
David L said,
March 19, 2023 @ 6:43 pm
I believe (but am too lazy to check) that part of the rationale for using Times in newspapers is that it combines good readability with density — i.e. you get a lot of words per column inch.
Like the Florida Supreme Court, I admire Bookman Old Style. But it takes up a lot of space. If you change something on your screen from Times to Bookman at the same font size, it will expand markedly. Not a good thing for newspaper and magazines that are worried about the cost of paper.
Timothy Rowe said,
March 19, 2023 @ 6:49 pm
I am greatly disappointed that my browser didn't display this discussion in Comic Sans. Or at least Papyrus.
Dick Margulis said,
March 19, 2023 @ 8:33 pm
@David L:
You're headed in the right direction with your comment. The fundamental variables for readability, in this order of importance are
1. Line length in average characters per line*
2. Leading (pronounced ledding), that is, line spacing
3. Ascender height†
4. Evenness of color‡
Once you solve for those, you can choose any reasonably well designed text face and be assured of acceptable readability for fluent readers. Nonfluent readers require a different set of considerations. This used to be a significant problem when all reading was on paper. But nowadays, swapping in a CSS file that accommodates a given reader's needs will generally do the trick.
* Times New Roman was designed specifically for the Times (London), with its narrow columns. Few other newspapers adopted it, and the Times no longer uses it. But yes, you're correct. The font is semi-condensed, maximizing the number of characters per line in a narrow newspaper column. Details of the font were designed to accommodate the spread of ink on low-quality paper in a high-speed rotary letterpress. It has since found wide use (thanks largely to Microsoft), but it's a terrible face for a six-and-a-half-inch line length on a sheet of US Letter paper. What we're aiming for is 55 to 65 characters per line in pretty much all situations where we expect sustained reading.
† Other than young learners, we do not read letter by letter. We read word by word. We accomplish this largely by scanning word shapes (in conjunction with our statistical expectation of the next word, much the way predictive text works on your phone). Typefaces that feature exaggerated x-height have short ascenders, and this flattens out word shapes, which slows down reading. This is why all caps is a bad idea for running text. The reason I don't mention descender length is that descender length had to be standardized across all fonts during the age of mechanized typesetting, and we're still stuck with that legacy for standard text faces.
‡ Here's where serifs can come into play. If you take a word like illumination and start squeezing the letters together by tightening the kerning, you start to get denser and looser collections of vertical stems. The ill get all clumped together and min is thick in the middle and open toward the ends, etc. With metal type, this was not an issue, but with digital type, that kind of abuse is possible with a sans serif face. Serifs literally keep the stems separated so you don't see that clumping. Bottom line, though, is that a competent compositor can give you as good readability with a sans serif face as with a serif face.
Erik F said,
March 19, 2023 @ 8:37 pm
For some discussions on the interwebs, I would suggest "Adobe Blank" (https://github.com/adobe-fonts/adobe-blank), which conveniently uses very little space and treats all text exactly the same. :-)
Brett Altschul said,
March 19, 2023 @ 9:11 pm
@Kimball Kramer: Multiple master fonts—in which the specific glyph shapes can depend on the size and style—are a tremendously useful tool, especially in physically printed material. I think it's unfortunate that they have not become much more widely used.
Chester Draws said,
March 19, 2023 @ 11:38 pm
Context is everything.
I prefer Calibri for small font sizes and spreadsheets.
Times or similar for larger bodies of print, unless I have a lot of footnotes.
But I use Verdana a lot, because I am a Maths teacher, and it has the clearest numbers. (A lot of fonts have ridiculously small and hard to read indices.)
We have a lot of very good fonts available now, so why stick to one?
I confess that Arial is a very clear font, but it is so ugly that it one of the few I can't use.
KeithB said,
March 20, 2023 @ 8:11 am
I have no idea if this is true or not:
https://notalwaysright.com/she-finally-has-the-words/284627/
but if there are fonts that are easier for dyslexics to read, maybe that should be a consideration, too.
Dick Margulis said,
March 20, 2023 @ 4:09 pm
@KeithB:
There are many fonts that help some people with dyslexia to read. Dyslexia is not a single disorder and there is no font that helps all people with dyslexia.
At the same time, most fonts designed to address some form of dyslexia—and indeed most of the strategies we use to accommodate nonfluent readers in general (such as setting text ragged, avoiding end-of-line hyphenation, using simpler letterforms) work against readability for fluent readers, meaning the vast majority of people.
The beauty of modern technology is that we no longer have to print a single copy of a textbook in a special font for Janelle and another single copy in a different special font for William. We can simply make the text available to them both in an html page, and they can each apply their own css to it, and Bob's your uncle. Meanwhile, the rest of us can read the print book that's designed for readability.
Jenny Chu said,
March 21, 2023 @ 8:25 am
Having worked for many years in big corporations, I can state confidently the decision of the DoD has nothing whatsoever to do with readability, and everything to do with ease of technical administration.
Calibri comes as the default font with Microsoft Office. It is troublesome to get everyone to change.
Bloix said,
March 21, 2023 @ 5:58 pm
Thomas Lee Hutcheson-
Chat GPT's take may be measured, but it's wrong. There is no such font as "Times Roman." Times New Roman has not been around "for centuries." It was designed in the 1930s for the London Times, and was given that name because, first, it's a "roman" (small r) font, meaning it's part of the large family of traditional serif fonts (the other major font families are blackletter and italic), and second, it's "new" because it was intended to have a bolder, more modern look compared to the prior fonts used by the Times (these were rather meager and spidery in order to fit more words on the very short lines the Times used when it would run seven columns on the front page).
Kent McKeever said,
March 22, 2023 @ 10:18 pm
As I realized my eyesight was worsening, I consciously looked for the the font that was most legible to me within the MS Word offerings. I settled on Rockwell. For texts that include numbers, I use Consolas, because its representation of zero has a slash through it, so I don't confuse it with the letter "O.".
Chas Belov said,
March 23, 2023 @ 9:18 pm
@Dick Margulis re "setting text ragged, avoiding end-of-line hyphenation…work against readability": I like these. How do they work against readability? In particular, I find full justification hellishly hard to read; online I will copy it to another app where I can read it ragged.
Mike Anderson said,
March 29, 2023 @ 9:21 am
Too bad no one at State checked with the Braille Institute to find their FREE Atkinson Hyperlegible (sans serif) font. You can get it at https://brailleinstitute.org/freefont
Dick Margulis said,
April 1, 2023 @ 7:12 pm
@Chas Belov: I think there's a huge confusion in this discussion between best practices for print and best practices for reading on a screen. Screen technology is changing so rapidly that even well-designed studies of only a few years ago are already obsolete. So I have nothing to say about reading on screens. You may be entirely right about readability in that environment, especially as few web designers take advantage of modern hyphenation capabilities, believing—falsely, I think—that hyphenation is a problem for most readers. It's a problem for some readers, to be sure. You may be one of those. Or you may be visiting terribly designed web pages (from a typographic point of view, most web pages are terribly designed).
Chas Belov said,
April 3, 2023 @ 10:47 pm
You have a good point the print and on screen are different. I do prefer serif in print and have a strong preference for sans-serif on screen. I am seriously thrown off by full justification in either mode. And that said, I'm probably OK with hyphenation in either one, unless there's an error or they break the word differently than I do, in which case it would throw me off for several minutes well I obsess about it.