The NY Times Magazine Relaunch
The New York Times magazine launched a re-design on February 22, 2015. Editor Jake Silverstein discussed the new design in the NY Times on February 18, “Behind the Relaunch of The New York Times Magazine.”
The article mentions new fonts and shows four of them in 2-letter (‘Gg’) samples, asserting, “Not a single letter in this relaunch issue has ever seen the light of day.” Three of the fonts illustrated are primarily display faces: they are used in headlines, subheads, pull-quotes, call-outs, or other short pieces of text, and are set at distinctively bigger sizes than the body text. For example, the new, condensed, bold, slab-serif display face is set at approximately 13 or 14 point, with its x-height around half the body size, at 6.75 points. A new, geometric sans-serif (like great-grand-child of Futura by way of Avant Garde Gothic) is used at text sizes, as small as 6.5 to 7 point with x-height of 3.46 points.
(Throughout this essay, the emphasis is on x-height as the indicator of visual size, rather than body size or font size. The reasons are explained in a paper, "Does print size matter for reading?" by Gordon Legge & Charles Bigelow (see reference below). Measuring physical x-height size on paper is easier than measuring body size, which is a little more elusive, comprising ascender + descender length + a “gap” of white space, which may be confusable with inter-line spacing or "leading". In what follows, I try to avoid confusion by distinguishing between physical x-height measured on actual type in print, versus x-height ratio or percentage of body size, determined for a given font by dividing its x-height by its body size, using the EM units of the font. For example, if a given font, say Times Roman, has an EM of 1,000 units for its body, and an x-height of 450 units, then its x-height ratio is 0.45 or 45%. To figure out what size that will be on paper, multiply the x-height ratio by the point size of the type. For our Times Roman example, when set at 10 point, it will have an x-height of 4.5 points (10 x .45 = 4.5).)
The “relaunch” article does not mention or show a sample of the new text font, but as body text, it is by far the dominant font in the magazine. It is what anyone who actually reads the magazine for the articles reads most. I estimate that of approximately 35,000 words in the magazine, at least 95% of them are composed in the text font. In print, the new text font has an x-height of 4.12 points, and body size around 9.5 points. In x-height, that is approximately what Times Roman would look like set at 9.25 to 9.5 point, depending on the version of Times.
Short reason why the relaunch is not more legible than before.
If you don’t have time to read all this long article, here’s the short answer.
The text type size in the relaunch is nearly at the “critical print size” threshold where text become harder to read. Studies of the relationship between type size and reading speed show that reading speed slows down substantially when the x-height of the text font falls below the equivalent of 4 points, read at an average distance of 16 inches. For some readers, and under some reading conditions, critical size may be bigger, like 4.5 or 5 points, so text type needs to be correspondingly bigger to maintain legibility. If critical print size is equivalent to an x-height of 4 point, then the size of the new text typeface in the NYT Magazine does not actually fall in the sub-optimal range, but it is close enough to flirt with it. Critical print size may vary among different readers, and if it is bigger for some readers, say 4.5 to 5 point, then a print size of 4 point will fall in the suboptimal range for those readers. The previous text face in the NY Times magazine was slightly bigger in x-height than the new one, so the new one is probably not more legible than the old one.
The Long Story
If you have time to read the long story, here is the reasonings, history, science, measurements, and calculations. (Disclosure: In my 30+ years of teaching typography, the most common student comments were that my digressions, side-tracks, and tangents were more interesting than formal lectures, so I’m including plenty of side-courses.)
How much do we read?
A lot. Denis Pelli, who researches reading and neuroscience at New York University, has observed that someone who has reached the age of 50 has read one billion letters (roughly 200 million words). I estimate that more than 90% of those letters were composed in small sizes in text fonts. The bigger sizes in display fonts are just thin icing on the big cake of a lifetime of reading small text type.
Analog versus digital
Reading, especially among younger readers, has been migrating to electronic devices for the past two decades. An increasing percentage of younger readers are reading news, magazine articles, blogs, tweets, text messages, and books on computers, tablets, and even phones. Older readers, however, formed their reading habits back in the old days when print was king, and are likely to be a large cohort of the loyal subscribers to the New York Times Sunday edition and magazine.
Older readers and aging eyes.
The age of readers should be considered in the design of print publications, because readers older than 40 to 50 are afflicted by vision deficits, the most common of which is called “presbyopia” or “far-sightedness”. Presbyopia is a loss of flexibility of the lens of the eye, with consequent difficulty focusing on near objects. It is usually noticed first in reading but also affects other activities like sewing and handwriting that involve seeing small objects up close. Readers can compensate by holding pages farther away, but presbyopia progresses as we age, so eventually, our arms aren’t long enough. Moreover, the farther away a page is held, the smaller the type is optically, until it is too small to read. A common correction for presbyopia is eyeglasses or magnifiers that enlarge the type image and bring it into focus. Other visual impairments and eye diseases like cataracts and macular degeneration also occur more often in the eyes of older readers.
Renaissance printers, readers, and type sizes
Except for its name (literally, “old man vision”, although women may be affected more than men and at earlier ages), presbyopia is hardly a new discovery. Italian Renaissance printers took it into account in the late 1400s. They used roman type sizes that were roughly 15 to 17 point in today’s measure. Book buyers of that era included humanists, lawyers, bureaucrats, clerics, aristocrats and others who could afford to buy books, often because they had higher incomes from being further along in their careers. Hence, many of them were presumably presbyopic. Accordingly, printers used fonts big enough that their customers could read the text without eyeglasses, which were rare at the time.
It could be argued that the technology of Renaissance printing wasn’t good enough to cut, cast, and print small sizes of type. While that is plausible, the Venetian printer, Aldus Manutius, provided a counter-example. In 1501, Aldus produced the first portable or pocket books, which by virtue of being smaller were less expensive (paper being a major expense to printers then as now). Smaller size made the books more affordable to a greater number of readers, broadening the market.
For his small books, Aldus used a new typeface, the first italic ever, cut by Francesco Griffo, a Bolognese punch-cutter who worked for Aldus in Venice. Griffo’s new italic type was cast at roughly 11.3 point, slightly less than 70% of the size of Aldus’ famous roman type, also cut by Griffo, which was around 16.5 point. The Griffo-Aldus italic resembled the fashionably prestigious humanist handwriting of the era, and this association appears to have helped promote sales, so much so that Aldus’ book format and his italic type style were soon plagiarized by other printers, who produced fake Aldine editions.
Humanist hipsterism to Italic handwriting & type
If wealthier, older readers preferred bigger types in the Italian Renaissance, why did Aldus’ pocket books become so popular? Lower cost is one reason. Cheaper books appealed to younger people, who would probably have had better vision and wouldn’t have been bothered by smaller type size. Another reason is that Aldine portables were the height of humanist style. In 15th century Italy, humanism was as fashionable as modernism was in 20th century Europe and America. To those who may say that modern modernism is, well, more “modern” than humanism, we must point out that in their time, the humanists were modern, and besides, they knew a lot more Latin than modernists today.
Another factor that may have enabled reading of smaller type is the use of eyeglasses, which were becoming more affordable in the 1500s. With glasses, small type would have been readable even for older readers.
What about the common belief that italic type is harder to read than roman? It may be true for today’s readers, but it doesn't seem to have been true in 1501 and for a few decades afterwards. Aldine portable books were composed entirely in italic type, but still they were very popular. The notion that italic is less legible than roman may be due more to cultural preference and familiarity than to basic factors in the visual system. The hip handwriting of the humanists was italic: if you couldn’t write italic, you weren’t cool; probably couldn’t even get a date. Beginning some twenty years after Griffo’s italic for Aldus, popular manuals on how to write italic were published by writing master like Ludovico degli Arrighi and Giovanantonio Tagliente, so everyone who read and practiced could become cool, at least in handwriting. Even Queen Elizabeth I of England wrote a fine italic hand as a young woman. As queen, one of the coolest ever, she had men do that work for her.
Surviving copies of Aldine portables are often elegantly bound and decorated with hand painted ornamentation and rubrication, so despite lower-cost, book buyers obviously valued the little books very highly indeed. A good thing, too, because they are like readable little jewels, pleasures to hold and behold, a tribute to a humanist printer-publisher-promoter and his brilliant type cutter. Of course, to actually read them, there’s that Latin issue again.
Important numbers, for later use
I will note now, for future reference, that the x-height of the Aldine roman was around 6.85 points, equivalent 0.343 degrees of visual angle. The x-height of the Aldine italic was around 4.72 points, equivalent to 0.236 degrees of visual angle. That 0.236 degrees will be important later, as I will explain.
Returning to the present era, display types, like the ones used in the redesign of the NY Times magazine, are usually set at sizes big enough for readers of any age. Some display types look weird (and at least one of the new fonts doesn’t disappoint in that respect), but perceived weirdness in display typefaces is usually an aesthetic and semiological impression based on cultural expectations, not on basic legibility. Display types are designed to attract, arrest or assault readers’ visual sensibilities, to get people to look at the text. If an odd-looking display type seems hard to read, well, apart from its shock value and “look at me” attitude, it is more for ornamentation than easy reading, and its size is usually big enough to overcome most legibility limitations. Moreover, titles and headlines are normally only a few words or short sentence, so display types evade the problem of uncongeniality in extended reading.
Despite the beliefs of many readers and a surprising number of editors and graphic designers, real weirdness in type lies not in display types at large sizes, where we see pictorial or iconic weirdness: big hulking slabs of font concrete, scribbling sarabands of font ribbons, fuzzy font teddybears, creeping font caterpillars bristling with spines, or sans-serifs which would be dull except for being eaten away by acid like the walking dead of fontology. The sort of stuff young designers love because such monstrosities repulse staid adults. Ha!
Small text types
No, true font weirdness is in text types at small sizes. That is where things get non-linear and unpredictable. If permitted to compare the humble world of typography to the lofty realm of physics, small type is to big type as quantum physics is to Newtonian physics. In Newton’s normal sized world, things are continuous and reasonable in our normal experience of sensible things, but when physics delves into the quantum world of the extremely small, things get discontinuous, weird, and irrational to our common sense. A remark attributed to Richard Feynman is, “If you think you understand quantum mechanics, you don’t understand quantum mechanics.” A similar remark by Niels Bohr is, “Anyone who is not shocked by quantum theory has not understood it.”
So it is, albeit to a vastly lesser degree, with small text type. If you look at a typeface at a large size and think you know how it will look small, you are likely mistaken. The visual appearance of type at small sizes has little to do with the fair curves, crisp edges and clever details visible at large sizes. The reasons are both subtle and complicated, and I think neither vision scientists nor type designers understand them fully, although type designers have been dealing with the problem a lot longer, at least since Griffo in 1501, and before him, scribes were writing even smaller letters with crow quill pens. Marginalia in early books are often smaller and more crisply crafted than the types. Throughout the history of typography, only a handful of designers knew how to cut type that worked well at small sizes. Of course, there were only a handful of designers in most eras. Today we are blessed with many more type designers, thanks to digital technology and more education in the art of type design, but it seems that relatively few modern designers know how to make good tiny type, or good type tiny.
An advantage for type designers is that they can use 20/20 hindsight to study previous designs and assess their own trials and errors to see what works and what doesn’t, and thus develop better approaches to designs for small sizes. The disadvantage is that designers often lack scientific methods to measure and evaluate their success or failure objectively, so they rely more on received craft lore, which is sometimes very good but sometimes not so hot.
An advantage for scientists is that they use controlled experiments, measurements, and calculations to establish testable values for what is going on between the graphical objects, eye, and brain at small sizes. A disadvantage is that scientific findings alone don’t provide enough guidance for designing practical types that are acceptable to the majority of readers, in part because cultural history and aesthetic standards are sometimes arbitrary. But, science can guide designers, and designers can create forms for scientists to study, so there is reciprocity.
Critical Print Size
One of the most important scientific findings about legibility has come from studies by Gordon Legge, a psychophysicist and reading researcher at U. of Minnesota. He calls this finding the “critical print size”; it is the threshold of fluent reading. Above the critical print size, people read fluently at their top or easy reading speed. Surprisingly, increases of type size much above critical print size barely increases reading speed. Speed remains nearly constant, along a long plateau even as type size increases, until, at a very large size, roughly ten times the critical print size, reading speed eventually starts to decrease.
Below critical print size, however, reading speed falls sharply as size decreases. A graph plotting reading speed against type size looks like a cliff - a long, horizontal, nearly flat plateau with a steep, nearly vertical drop-off at one end. Below the critical print size, a reader’s vision falls off the cliff and plunges toward illegibility, as seen in the drop at the left of this graph.
Reading speed (words per minute) vs. print size (x-height in degrees). Data are replotted from four experiments. Legge et al. (2007) used the RSVP method (open circles). The other three studies used scrolling text—filled circles: Legge et al. (1985), matrix sampling; filled squares: Legge et al. (1985), blur; and triangles: Akutsu et al. (1991). The data points are means across subjects. RSVP is known to yield much higher reading speeds than scrolling text, accounting for the vertical shift in the curve for Legge et al. (2007). From Legge & Bigelow, see References below.
Professor Legge has been publishing papers on this phenomenon since the mid-1980s, but he points out that it was recognized by the psychologist Edmund Burke Huey in 1908. Huey in turn credited some earlier studies for similar findings. For fluent reading, Huey recommended that the minimum size of printed type x-height should not be below 1.5 millimeters. That number turns out to be important to the present study, and I will come back to it later.
The reasons why reading speed decreases precipitously when size falls below the critical print size are not fully understood, but presumably are a combination of factors and limitations in the human visual system. One limitation is acuity - how well we can see very small objects. As objects, like letters, get smaller, we have more trouble seeing them. The familiar Snellen eye chart with its slab-serif capital letters at different sizes is a test of visual acuity. Another limitation on reading is a phenomenon called “crowding”- how well we can recognize separate symbols when they are crowded together. Crowding is rarely tested by doctors or ophthalmologists, but it turns out to be an important factor that limits legibility at small type sizes. Another limitation is contrast between letters and background. Most type is nominally black on white, but often, the black isn’t really black, nor the white really white. Contrast usually doesn’t slow reading speed much until it is quite low.
Another possible factor in the NY Times magazine is the contrast sensitivity of the human visual system in relation to the rotogravure printing process of the magazine. Rotogravure renders images with a grid of tiny cells, the same in two-dimensional cross-section like a pixel pattern, but of varying depths. The cells get filled with liquid ink when the gravure cylinder is charged, and depending on how much ink fills a cell, that little part of the image or text is a lighter or darker tone of gray on the page.
For type, this also means that the letter image is dithered around the edges by gray pixels where letter features do not align exactly with the gravure cell boundaries. This is a little like anti-alias gray-scaling of type on screens. The gray-dithered pixels occur even on straight line edges, as well as on curves and diagonals, giving them a dithered, feathered, or chewed appearance when viewed in magnification. When type edges are dithered and grayed, some of the finer details of the letterforms are lost while some noise is added, but it is unclear (at least to me) how much this affects legibility. Reading can be remarkably resistant to blur and noise.
Other factors affecting reading speed are discussed in Professor Legge’s book, Psychophysics of Reading in Normal and Low Vision (Lawrence Erlbaum Associates, Mahwah, NJ, 2007).
At very small type sizes, letters are crowded together, our visual system gives up trying to recognize the letters, and text stops making sense, becoming “texture”, as Sarah Rosen and Denis Pelli have written (see References below). As a typographer, I use the word “texture” somewhat differently. I would say that body type in readable text sizes has a recognizable texture. Is “texture” necessarily inscrutable? This is a matter of definition, not of phenomenon. As a typographer, I think of texture as an emergent quality that can be used to recognize different typefaces, qualities, and aesthetics in text that is still legible. Years ago, I was lucky enough to be a student of Hermann Zapf, and later to work for a time with Zapf and Max Caflisch, a great Swiss typographer who had been an assistant to Jan Tschichold. These guys were in their 60s then, and they could look at a page of text and recognize not only which type it was, say Garamond, but which version of Garamond from which foundry or composing machine. They could tell by the texture.
[For more about texture in typography, see my essay “Form, Pattern & Texture in the Typographic Image” after this article in the blog: http://bigelowandholmes.typepad.com/bigelow-holmes/2015/03/form-pattern-texture-in-the-typographic-image.html]
NY Times Magazine text.
Now again we return to the legibility question of the relaunched NY Times Magazine.
The text type design is not specifically shown or mentioned in Mr. Silverstein’s article. The text typeface design appears to be a “text” variant of one of the new display typefaces, but because it is set small, it is not easy to distinguish all its features.
The text type features
It may well be, as Mr. Silverstein writes, that this face has never before seen the light of day, but its parts certainly have. It is a hybrid of historical precedents that are separated by hundreds of years and imitate features from several different designs cut by different historical designers. One batch of features is derived from types cut by Johann Michael Fleischmann, an 18th century type punch-cutter who was born and trained in Germany but spent his adult career working in the Netherlands. Another bunch of features comes from types cut by Robert Granjon, a 16th century French punch-cutter, who flourished two centuries earlier than Fleischmann. Some of the Granjon-esque features are not exactly imitative of Granjon’s own cuttings, but appear to derive from a 1913 reworking of Granjon’s types by the English Monotype corporation under the direction of its works manager, F. H. Pierpont.
The Monotype-reworked Granjon of 1913 is called Plantin, after Christophe Plantin, a renowned printer of the 16th century, who used types cut by Granjon. French by origin, Plantin spent most of his illustrious career in Antwerp, where a museum of printing is now housed in his former establishment.
Another set of borrowed features in the new type is its near-triangular serif structure, typical of “Latine” style French types of the 19th century. A good modern revival of Latine is Adrian Frutiger’s Méridien of 1957, a revised version of which is called Frutiger Serif.
When I look at a page of the relaunched NY Times magazine, I cannot, unlike my tutors Zapf and Caflisch for types of their era, recognize the face by its texture. Instead, I feel a queasy sense of presque vu - as if I’ve seen it somewhere before but can’t quite remember exactly what, where, or when. I feel like the hapless time traveler jerked around in Alain Resnais’ disorienting classic movie “Je t’aime, je t’aime”, in a past out of control.
Nevertheless, despite its time-traveling melange of borrowed features, the new text typeface is crisply drawn and well fitted. Its legibility problem is not its design per se, but its size.
Measuring type size by visual angle
To realize this, we need to return to the “critical print size” studied by Gordon Legge and his colleagues. Critical print size is measured as a visual angle subtended at the retina of the eye. Although a very useful measure, it is unfamiliar to most typographers, designers, and editors, which is may be why the new text in the NYT magazine is no more legible than the old. Actually, the new text may be slightly less legible, despite the effort of creating a new font.
Point size of type is a numerical measure that developed over the past 300 years. First attempts to determine numerical sizes of type began around 1694, long predating the metric system, and the most recent integration of point sizes with the English inch system occurred in 1984, with the PostScript measure of exactly 72 points per inch. Point sizes are good for general use, but not really accurate for measuring the perceived size of type. The problem with a single physical measure of type size, like nominal point size, is that visual size changes with viewing distance. If you hold a page half as far away as normal, the type will look twice as big, If you hold the page twice as far away as normal, the type will look half as big. To the eye, closer means bigger, farther means smaller. With text type, the reader’s eye is, or should be, be the primary thing we care about. This is something that art directors and graphic designers must to consider - they need to see things through their readers’ eyes.
Up close, type subtends a bigger angle at the retina; farther away, a smaller angle; this is why vision scientists use visual angle — it is a measure that takes both physical/graphical size and viewing distance into account. Recall that when you take an eye test at a doctor’s office or department of motor vehicles, you have to stand or sit at a prescribed distance for which the test is calibrated. The test measures visual acuity for a certain viewing distance, and certain letter sizes.
Visual size, as measured by visual angle, helps us understand the relationship between type size and reading distance. To get an idea of visual angle in relation to type size and viewing distance, it is helpful to recall some high-school geometry. (Wait, don’t stop reading yet; you aren’t really going back to high-school; I promise things will clear up again after this rough patch.)
Imagine a thin right triangle with its sharpest angle at the eye, and the triangle side opposite that angle as the x-height of a font on a page perpendicular to the line of sight to the eye. OK? Try sketching it. Got it?
So now we have a long straight line from the eye to the baseline of some type. That long line is one leg of a right triangle, and the x-height perpendicular to it is the other leg of the right triangle. So now we have two sides forming a right angle.
Now imagine the third side, called the hypotenuse in geometry (don’t worry, there is no test on the material), running from the top of the x-height to the eye, where it joins the other long leg to make that sharp angle of the skinny right triangle. So now we have a triangle. What we want to do is measure that sharp angle in degrees. That is the visual angle that we use to estimate legibility.
To repeat (after that harrowing brush with geometry), visual angle uses two perpendicular measures: the height of the type and the distance to the eye. That enables us to compare visual sizes independent of distance. If we halve the viewing distance but keep the type size, it will look twice as big, but if we halve the viewing distance and also halve the type size, it will appear the same size to the eye, at it did at normal distance and size. And, if we double the viewing distance and also double the type size, it will appear the same size as normal distance and size.
This can be grasped visually by printing out text at different type sizes and viewing it at different distances. Compare, for example, a typeface printed at 12 point and viewed at 8 inches to the same face printed at 24 point and viewed at 16 inches, and one printed at 48 point and viewed at 32 inches.
If you know the size of the font and the distance from the page to the eye, you can calculate the angle at the eye, using trigonometry. What is best to measure? In measuring the size of a font in relation to reading, for the Latin alphabet, the x-height is more useful than body size or capital height. Gordon Legge and I offered arguments in support of using x-height in our paper, “Does print size matter for reading? A review of findings from vision science and typography” in the Journal of Vision, 2011. (See references below.)
In arguing in favor of x-height measurement, we don’t claim that a big x-height is always better for type, but rather, that x-height is usually the best indicator of perceived size. If someone prefers to measure capital height, for example, which may be more useful in German-literate countries where all noun begin with capitals, the visual angle for capitals can be found the same way as for x-height.
Of course, it is important when comparing measurements of visual angle, to be sure that all items in the comparison are measured the same way, all x-height, or all capital height, or all body size.
The critical print size found by Legge is approximately 0.2 degrees of visual angle for the x-height of the type ion print. In his book, Psychophysics of Reading in Normal and Low Vision (Lawrence Erlbaum Associates, Mahwah, NJ, 2007), he states that in some studies, the critical size is as large as 0.3 degrees, while in other studies, as small as 0.17 degrees, and that a visual angle of 0.2 degrees is a consensus size for normally sighted readers across several studies. Different findings from different laboratory studies suggest that critical print sizes may vary for different people in different situations. In the everyday world of practical typography, critical print size can be expected to vary at least as much as in the laboratory. Some readers may have keen eyesight, others may have uncorrected vision deficits; some lighting conditions may be bright, some dim; some readers will be younger readers, some older. A 1991 study by Akutsu, Legge, Ross, and Schuebel (see critical print size graph above) showed that not only does reading speed slow down for text below 0.20 degrees of visual angle, but that it slows more among older readers than younger. At angular sizes between 0.3 degrees and 1.0 degrees (approximate x-heights of 6 point to 20 point), older readers without eye disease read as fast as younger readers, but at very small or very large sizes, the older readers read only 70% as fast as young readers. Because many young people have moved their reading to electronic media, the remaining loyal readers of the print edition of the NY Times Magazine are likely to be older and to need a larger print size for maximum reading speed.
Therefore, to support a wide readership, type sizes should not be aimed at the youngest readers with the best vision under optimal conditions, but to a wider range of average readers under various conditions. Remember that the successful Italian Renaissance printer, Aldus Manutius, used roman types with x-height around 6.85 points or 0.343 degrees, well in the range for older readers, and italic type with x-height around 4.72 point or 0.236 degrees of visual angle, somewhat smaller than the range for older readers but still well above the 4 point and 0.2 degrees of the critical print size. The type designs that Aldus used have been deservedly praised, but the sizes at which he used them were also well-adjusted to his readership.
We have learned a lot about typography and vision over the past 100 years, but Edmund Burke Huey’s 1908 recommendation of minimum x-height of 1.5 millimeters is equivalent to 0.21 degrees of visual angle and very close to the 0.20 degrees given by Legge in 2007. It appears that our vision and reading have not changed much over the past century, nor much since the era of Aldus, 500 years ago. (Aldus died in 1515.)
To visually satisfy a large readership like the 2.5 million people who subscribe to or purchase the Sunday New York Times and its magazine, the text type size should be bigger than the critical size of 0.2 degrees or 4.0 points, to make reading easier for older readers who remain loyal to print but benefit from slightly bigger text. As for younger readers with good vision, they may have already embraced reading at small sizes on their iPhones without difficulty and won’t be eager to return to cumbersome print.
Over the past ten years, several paperback publishers have increased type sizes to appeal to older baby-boomers with aging eyes. This trend was reported in the NY Times in 2005:
Among popular print publications, the greatest pressure to reduce type size is in newspapers, where the cost of newsprint dwarfs other production costs. Newspapers have been reducing page sizes to save money, but publishers need space for advertising, designers need space for photos and graphics, and editors need space for words. When everyone need space, but page sizes are reduced, what has to give way? Type size. But not as much as one might expect, because newspapers cannot shrink type down to minuscule sizes and still keep readership.
The critical print size resists type shrinkage. I doubt that newspaper production managers read the psychophysical literature and calculate font size accordingly, but the collective, practical experience of many newspapers, production managers, and designers shows clear trends.
For the Legge & Bigelow paper on print size, we did an “ecological” study of type sizes in the real world of commerce instead of the laboratory, by measuring body text x-heights in hundreds of newspapers. We found that almost half the papers used type with an x-height of 0.23 degrees, roughly another quarter used x-heights of 0.24 degrees, and the remainder were mostly either 0.22 degrees or 0.25 degrees. Hence, the critical print size of 0.20 appears to be a lower bound to which newspapers add a safety margin of 10% to 15%. to cover a wider range of readers, not just those with good vision. Sizes greater than 0.25 were rare, apparently unnecessary as well as uneconomical. In the narrow span of sizes, the small end is limited by vision, and the large end by economics.
Print Size, Visual Angle, and Arithmetic
Those measures are in visual angle, which is useful to scientists, but how can editors, publishers, or designers take advantage of relevant reading and legibility studies without being scientists in laboratories?
Turns out, it’s not hard to figure out visual angle. Legge and I pointed out a simple rule-of-thumb for a standard reading distance of 16 inches (equivalent to 40 centimeters): to calculate degree of visual angle, just divide the printed size of an x-height, measured in points, by 20.
For example, assume that for some font, you measure a physical x-height of 4 points on paper. Divide that by 20, and the result is 0.20, the famous critical print size of 0.2 degrees of visual angle, assuming the text is read at 16 inches.
An x-height of 5 points results in a visual angle of 0.25 degrees, and an x-height of 6 points results in a visual angle of 0.3 degrees. As an important reminder, those measures are x-height, not body size, and the reading distance is assumed to be 16 inches. For closer distances, the visual angle will be greater, and for farther distances, the angle will be less.
Tips with Tools
For anyone who often uses or inspects digital type, it is not difficult to calculate degree of visual angle without printing the type. It can be calculated by using the digital metrics of the type. Using our familiar Times Roman example:
1) Crack open a digital font with an editing tool (some vendors offer free, limited functionality or limited time versions) or by looking at “Adobe font metrics”, which exist for some common typefaces.
2) Find the height of the lower-case ‘x’. In Times New Roman in PostScript or OpenType CFF format, it is usually 450 units. Also find the Em (body size) of the font. For PostsScript or OpenType CFF, the body size is usually 1,000 units. So, the x-height is 45% of the body size. For TrueType versions of Times New Roman, the x-height is usually around 916 to 920 units and Em size is 2,048 units, which rounds off to an x-height 45% of the body.
3) Determine the body point size (the font size you would select in a page layout applications) at which the font will be composed. Let’s say 10 point for Times Roman. Multiply body point size times x-height percentage (45% or 0.45 in decimal). Thus, the x-height of Times Roman at 10 points is 4.5 points.
4) Divide the x-height in points by 20. Our Times Roman x-height at 10 point body size is 4.5 points; divided by 20 that equals a visual angle of 0.225. Easy, OK?
I should point out that digital calculation usually gives a minimum size, because there is none of the ink spread, blur, or gravure feathering that tends to increase actual printed size. Also, measuring the x-height of the 'x' itself gives the minimum value of the x-height. The height of a lower-case ‘o’ will be somewhat larger, because it under-hangs the base-line and overhangs the x-line slightly.
Without access to a digital font, it is necessary to measure actual size in print. To determine the visual angle of the new text type of the NY Times Magazine, I measured its x-height in print. For rotogravured type, measurement of the exact dimensions on paper is subject to some ambiguity, depending on how much of the gray dithering is included.
I made several measurements of the new NYT Magazine text using a digital measuring microscope, and arrived at a rounded average x-height of 1.45 millimeters. That’s for the letter ‘x’ including a touch of gray dither at top and bottom. Round letters like ‘o’ and ‘e’ with slight overhangs of the base-line and x-line are slightly larger. Depending on readers’ contrast sensitivity and lighting conditions, the gray dither may add more or less to the perceived size of the letter.
My digital microscope measures millimeters, which I converted to printers’ points, and points to degree of visual angle. Over several measurements, I determined the average print x-height of the new NY Times magazine text type to be approximately 0.206 degrees of visual angle. (The third decimal place is probably not significant but I leave it in.) At 0.20 or 0.21, the x-height is close to the edge of the critical print size cliff, but not falling off it.
New text type: NYT Magazine, March 1, 2015, text type "boxes": x-height = 1.445 millimeters = 4.096 points = 0.205 degrees visual angle
For comparison, I measured x-heights in the issue from a week earlier, February 15, composed in the previous text font. That issue had a long cover story about a Stanford undergraduate student and her relationship with an off-campus Silicon Valley mentor that resulted in an allegation and lawsuit. The frequency of words like “sex” and “sexual” in the article made it easy to find ‘x’s to measure - almost twice the frequency as in average text. I don’t believe I’ve ever seen so many x’s in an article in a family newspaper. As a former Stanford professor, I was shocked, shocked.
Older text type: NYT Magazine, February 15, 2015, text type "sexua": x-height = 1.524 millimeters (average of 2 measurements) = 4.319 points = 0.216 degrees visual angle .
Comparison of visual size of former and current text types
Surprisingly, the average degree of visual angle for the x-height of the previous text font in the NY Times Magazine in print turned out to be approximately 0.215 degrees of visual angle, around 4% bigger than the visual angle of its replacement in the relaunch.
So, the x-height of the new font in print is apparently a little bit smaller than the x-height of its predecessor. In legibility, the size of the new font is closer to the edge of the cliff. It is still above the drop-off into sub-optimal legibility, but close enough to look down and wave to it.
I should point out that because of slight variations in the accuracy, magnification, calibration, and handheld manipulation of my little digital microscope and measuring instrument, the third decimal place in the measurements is probably not significant. If my measures are rounded to two decimal places, they differ by only one-hundredth of a degree of visual angle, probably not enough to matter. Someone else performing the same measurements with different tools might come up with slightly different measures. An actual digital copy of the font, plus the nominal text body size at which it is composed, would provide an exact measure, although not the actual size of the image in print.
In terms of x-height and critical print size, then, the old and new text faces for the NYT magazine may be equivalently legible. The new face is probably not more legible in terms of print size, and may be even slightly less.
The proximity of both text typefaces to the critical print size suggests that some readers, perhaps many readers, whose vision or reading circumstances are suboptimal, will find the text a little too small.
New sans-serif font at smaller size.
The new geometric sans-serif font is used in small sizes in recipes in the "Eat" section and in the "Puzzles" section. I measured the x-height in those sections and found that the x-height in the paragraphs of the "Puzzles" section is 1.22 millimeters, equivalent to 0.173 degrees, which is below the consensus critical print size. Some readers will not find that small a size problematic, but others will. If you have trouble reading those sections, you are likely not alone.
Eye strain, visual fatigue, and readership.
What do readers feel when the text is a little bit too small? Legge in his 2007 book cites studies on “visual comfort” in reading. My reading of studies of visual comfort found that readers may express dissatisfaction with legibility in somatic terms. Various studies cite complaints of “eye strain”, “tired eyes”, “visual fatigue”, “blurred vision”, “neck aches”, “headaches”, and related physical complaints. Some of these studies were of reading on CRT and computer displays, where vision complaints are sometimes termed “computer vision syndrome”.
A 1983 study by Gerald Murch found measurable but temporary loss of acuity from reading on raster display terminals compared to paper. (See References.) His study didn’t include LCD displays, so it isn’t clear how it would relate to today’s devices and screens.
We need more studies of legibility and reading on smart phone and tablet screens, which are much more easily adjustable in reading distance than the old, heavy computer CRT monitors. Also, on some smart phones, tablets, and e-readers, type size can be adjusted.
If the readers of a print magazine complain about visual discomfort, eye strain, visual fatigue and other problems, it seems likely that some of them will become less committed to reading the magazine. After all, popular magazines are read for pleasure and entertainment.
News typefaces as a special genre.
Interestingly, both of the NY Times Magazine text faces in print appear to be smaller in x-heights than the text in many common newspapers.
One could argue that the design of a font influences legibility and makes a significant difference in reading speed. I love that argument with my heart, but alas, my head doesn’t agree. In-well controlled experiments, it is difficult to demonstrate that type design makes a significant difference in reading speed, except for x-height, and perhaps letter widths. When two typefaces are similar in many design features, as are the current and previous NYT Magazine text faces, both partially based on Granjon’s old-style types, it may be even harder to find statistically significant differences in reading speed.
Over the past 100 years, designers of news faces gradually increased x-heights, as a way of maintaining, or sometimes increasing the visual print size of text, while reducing its body size. This trend may have reached its peak with faces like Hermann Zapf’s Edison, a news face with an x-height 52.7% of the body, and Robin Nicholas’ Nimrod, with an x-height 53.9% of the body. Those faces do look very big indeed, especially compared to Times roman, with its 45% x-height, which, although designed as a news face, and bigger than, say, traditional Garamond, now looks small and delicate compared to the sumo class news faces.
Some yet newer face designs have backed down from the extremes of Edison and Nimrod, but most still have x-heights at least 47% of the body, and some larger than that.
Ways to increase legibility at small type sizes
The designers and editors today are fortunate to have wide range of alternatives for improving the legibility of print.
1) An obvious alternative is to increase point size slightly. For a typeface like Times Roman with an x-height ratio (x-height ÷ body size) that is 0.45 or 45% of the body size, a composition size of 9 point will yield an x-height of 4.05 points (0.45 x 9 = 4.05), with a visual angle of 0.2025 degrees (4.05 ÷ 20 = 0.2025), which rounds to 0.2, right on the cusp of the critical print size. A one point increase in body size to 10 point would boost the x-height to 4.5 points (0.45 x 10 = 4.5) and a visual angle of 0.225 degrees (4.5 ÷ 20 = 0.225), nearly 0.23, like the most common news text sizes. Although size is easy to change, it has consequences on the rest of the page layout. If size scaling is linear - the same in both horizontal and vertical dimensions - the increase would also widen the text proportionally, so more design and layout adjustments would be needed, such as less blank space around pull quotes, or less space around titles, just as examples.
2. Modify an existing design to increase its x-height around 10%. If the x-height ratio is 45% of the body, raising it to 50% would yield a print x-height of 4.5 points when the type is set at 9 points (0.5 x 9 = 4.5), which would be equivalent to using 10 point type with an x-height ratio of 45%, as in the previous example. Because the text typeface of the NY Times Magazine is a custom design for the NY Times, a little re-tailoring and re-fitting for a re-relaunch would likely not be too difficult, albeit more work than just increasing the body type point size. Raising the x-height without widening the letters, however, would make the type look visually look narrower, something that not all readers are fond of, thought it would not actually be narrower than before the refitting. Leaving the width metrics unchanged would avoid problems of refitting columns and other aspects of the layout, because both vertical and horizontal measures of the type would be the same. Only the internal x-height would be bigger.
3. Find a different but visually appropriate existing newspaper font that would fit in with the rest of the design. This would probably be less satisfying than having a coordinated set of custom designs. It is, however, a decision made by many newspapers.
At last we have reached the end of the long story. I hope there were enough side-tracks, tangents, digressions and distractions to entertain you. Congratulations for going the distance.
The graph of critical print sizes and the graph of newspapers and book x-heights are from: Legge, G., Bigelow C. (2011). Does print size matter for reading? A review of findings from vision science and typography. Journal of Vision (2011) 11(5):8, 1–22
A discussion of readable texture is: Charles Bigelow, “Form, Pattern & Texture in the Typographic Image” first appeared in Fine Print: A Review for the Arts of the Book, in 1989. The text of that article follows this one on the B&H blog.
Gordon E. Legge. Psychophysics of Reading in Normal and Low Vision. Lawrence Erlbaum Associates, Mahwah, NJ, 2007.
Gerald Murch. "Visual fatigue and operator performance with DVST and raster displays". Proceedings of the society for information display, 1983. (Sorry, I don't have a full citation or url, and Jerry was my teacher of visual perception, too, how embarrassing!)
Rosen S, Pelli D, 2011, "When text forms a texture, grouping induces crowding and slows reading" Perception 40 ECVP Abstract Supplement, page 35.
[Note: A handy and fun visual angle calculator by Nick Sherman and Chris Lewis can be found at: http://sizecalc.com
For example, you can find out that a 12" traffic light lens viewed at 288 feet is the same size as an x-height of 4 points viewed at 16 inches, that is: 0.198943 degrees of visual angle, just under the critical print size, and just a hair smaller than the x-height of the text type in the NY Times magazine. Of course, we don't read traffic lights. As the Reverend Gary Davis used to sing, all we have to do is: "Stop on the red, and go on the green, don't mess around with Mister in-between."]