From Casual to Textile to Marker: the story of a font

IN SHORT

In 1993, Bigelow & Holmes designed a font named Lucida* Casual Italic, based on informal handwriting. In 1998, B&H designed for Apple a font named Textile* which looked like a sumo wrestler version of Lucida Casual Italic - bigger, bolder, brawnier. In 2014, Bigelow & Holmes released a font named Lucida Marker, which is a dead ringer for Textile. Here’s the story.

DESIGN PROCESS

In 1990-91, Bigelow & Holmes created Lucida Handwriting, an informal joining script. The font became widely used after its release by Microsoft in 1992. Some two years later, we designed two informal, non-joining fonts that looked like they’d been written rapidly with a partly worn-out fiber-tipped marker. These relatives of Lucida Handwriting are Lucida Casual roman and italic. They evoke a distant echo of Renaissance humanist handwriting but were made in the modern age with modern tools. Lucida Casual fonts were released on Hewlett Packard ink-jet printers in 1993 and were later licensed for use on many kinds of laser and inkjet printers. 

  Ill1-calligCasItalCompare-01
A few years after the launch of Lucida Casual, B&H were asked by Apple to design a fun-loving, playful font to contrast with the retro-futuristic look of Chicago*, which nevertheless had a “fun” look for many users, perhaps because of its geometric-engineered insouciance. We were familiar with every feature of Chicago because we had designed for Apple System 7 the TrueType outline version of Chicago, based on Susan Kare’s original bitmap design. (1) 

We began work on the fun-loving new font when Apple’s “Copland” operating system was being developed to include an “Appearance Manager” that let users switch between different graphic themes in the user interface, changing system fonts, graphics, and effects. The new font was to be part of one of the themes.

We felt certain that a casual italic could have the desired playful look, but a problem was that the font had to be “cabined, cribbed, confined,” as Shakespeare’s Macbeth complained about his plight. It needed to look as big as Chicago and had to conform closely to the average width of Chicago, so that substituting the new font in the system would not cause line length overflows in menus, messages and texts. We were puzzled about how to fit a loose, playful design into a procrustean metric bed, but then we thought of Mark Twain’s remark that, “A round man cannot be expected to fit in a square hole right away. He must have time to change his shape.”

On the low resolution computer monitors of the 1990s, a font needed a big x-height to be easily readable at screen reading distances that were 25% to 50% greater than for reading on paper. (2)

We began by experimenting with Lucida Casual Italic, which had the free spirit we wanted and a big x-height (same as Lucida Grande), but it did not have the bigger x-height, greater boldness, and narrower width needed for a system font in the Chicago class. 

KHDossier_Typeface_20120117pm.03-01

So, we transformed Lucida Casual Italic from a light, sprightly alphabet into a heavyweight contender we called “Lucida Big Casual”. The transformation was not accomplished all at once, but by a process of progressive modification and refinement over time (following Twain’s dictum about giving a round man time to change his shape). 

Our first task was to increase the x-height.

LucidaBigCasual-640
Many readers feel that italic faces are less legible than roman, and some legibility studies in the 20th century supported that feeling. In some studies, at small type sizes, reading italics was significantly slower than reading roman. (3) 

We thought, however, that slower reading of italic may not have been a fault of italic per se. The vast majority of modern readers, including college student subjects in reading studies, are more familiar with roman. At times in the past, however, people read long texts in italic. In the 15th century, Italian Humanists invented, read, wrote, and made fashionable the cursive handwriting that eventually became our so-called “italic” type. (4)

Vespucci_1460-5

Handwriting of Giorgio Antonio Vespucci, circa 1460-5 [4].

In the first quarter of the 16th century, a refined descendant of Humanist cursive, the elegant “cancelleresca corsiva” was written by Italian chancery scribes in Rome, Florence and other Italian cities and was adopted in Humanist circles in other countries. Queen Elizabeth I wrote a good chancery hand. 

We designed a modern chancery cursive called Lucida Calligraphy for the Lucida family, and designed another interpretation of Italian chancery cursive for Apple, released as Apple Chancery®. It was not designed as a system font, but as a way to demonstrate the power of Apple’s advanced typography. Compared to Lucida Calligraphy, Apple Chancery has a smaller x-height and a lesser degree of slant, more in keeping with classical chancery cursives. 

BigNCalligAndChanc-01
Blue 'n' = Lucida Calligraphy. Black 'n' = Apple Chancery.
 
Ill3-calligApChanceryCompare-01
 

We can’t go back to the 15th century to test reading of italic (though fascinating it would be), but in favor of italic is that in 1501, Venetian printer Aldus Manutius printed the first book in italic type, cut by Francisco Griffo. Aldus’ italic type size was roughly 11 point, whereas the roman types Aldus had used until then were roughly 16 point. Aldus’ pocket books were composed entirely in his small italic, and were were so popular that they were soon widely imitated and the type style plagiarized by rival printers in other countries. Hence, the relatively small size and italic style of the Aldus-Griffo font seem not to have been impediments to reading or book buying.

A 1998 study of reading speed and comprehension of roman versus italic (actually, slanted roman) on computer screen, found no significant difference between reading speed and comprehension, although users preferred the roman version. (5)

Despite the attractions of 15th and 16th century Humanist writing, we were designing a typeface for millions of computer users at the end of the 20th century, so we based the proportions of our design on how modern readers read on screens (circa 1995). 

Desktop computer screens were usually viewed at a greater distance than text on paper (we estimated 25% to 50% greater, based on ergonomic guidelines). Greater reading distances cause type on screens to appear smaller than type on paper, because greater reading distance reduces the size of text on the retina of the eye. This can be compensated by simply making text bigger on screen, but in a user interface, screen real estate is at a premium, so an alternate approach is to increase the x-height of a typeface without increasing its body size. 

In practice and in reading studies, bigger x-height has been comparable to bigger body size (6).

Figure 1a_Legge-Bigelow_640whbg

Accordingly, we made the x-height of Big Casual a whopping two-thirds of the body. The result was that Big Casual looked even bigger than Chicago, and substantially bigger than most common sans-serif designs. We estimated that Big Casual would be set at around 14 point on screen, but that its big x-height would make its size seem more like 18 point. 

LucBigCasualCompareAppleChicago-01
Big x-height is not a legibility solution in itself. It is part of an ensemble of design features. In a face with a big x-height, the ascenders and descenders are necessarily reduced, but because they are important distinguishing markers, if they are made so short that a reader cannot distinguish ‘h’ from ’n’, ‘b’ from ‘p’, ‘d’ from ‘q’ and so on, legibility is impaired. Minimum ascender/descender lengths vary depending on other features of a design, and on the size at which it is read. When we raised the x-height of Lucida Casual Italic, the skimpy ascenders and descenders flirted with being too short for the x-height of the letters. But as we condensed the design, the ascenders and descended seemed to become more evident. 

Some typefaces, especially sans-serifs, tend to assimilate letter shapes, but instead we strove to differentiate them. The reading eye tends to fixate near the x-line, so distinguishing features have accumulated near the x-line in the evolution of typefaces. Although there is some assimilation of lower-case 'a' and 'g' with the 'bdpq' set, italic forms can also help differentiation. In letters like ’n’ with a hairline and an arch joining a left stem to a right, the join often falls lower in italic than roman. Accordingly, the Big Casual ’n’ has a deeply cut join on the left and a peaked arch on the right; strongly differentiating ’n’ from ‘o’. 

After we increased the x-height, we emboldened the font by increasing the stem thickness until it was 25% of the x-height. In many text faces, normal stem thicknesses range between 16% to 19% of x-height, and demibold weights are often around 150% of normal, so Big Casual Bold was equivalent to a demibold weight. Of course it then needed a new work name, “Big Casual Bold”.

To keep widths from expanding too much after raising the x-height and emboldening the weight, we reshaped, narrowed, and refitted all the letters. In narrowing the face, we tried not to eliminate hand-written effects, like the slight serif-like traces on the entrance and exit terminals. Nor did we abandon the essentially cursive structure of the design. We tried to keep the counters (the internal forms) clear and well differentiated even as we narrowed them. The internal counter of the ‘b’ is different than that of the ‘d’; the ‘a’ different than the ’n’, ’n’ different than ‘o’, and so on. Of course, we called the next stage of the modified font, “Big Casual Bold Narrow”.

KHDossier_Typeface_20120117pm.03-01
The original Lucida Casual had a slightly free approach to metrics. Vertical metrics were not rigidly locked to certain values, so that slight variations conveyed some of the freedom of handwriting. For Big Casual, however, we believed that metric regularity was important for easier hinting and predictable rasterization, so as we adjusted x-height proportion, weight, and width, we also regularized the exact measures of x-heights, capital heights, ascender heights and descender depths to specific values, although we kept some variations in round x-height letters. 

All these modifications were made by hand - there was little software then, and still not much today, that could make all these modifications automatically. Ikarus software, which we used for the development, had capabilities for regularizing stems and vertical metrics, but its regularizing modules worked better with sans-serifs or regular seriffed faces. Big Casual was an italic with barely any straight lines in its contours, and all its strokes swelled near their terminals, so we found it easier to make adjustments by hand.

TrueType’s resolution of 2,048 units per em is more than 100 times greater than the original Macintosh font bitmap screen resolution of 72 pixels per inch or 14 pixels for 14 point type, so we could tighten spacing between letters, compared to Chicago, which has loose letter spacing based on the original Mac screens. When finished, Big Casual Bold Narrow turned out to be slightly narrower overall than Chicago.

Ill4-ChicagoTextileCompare-01

EXPRESSIVENESS

It is hardly surprising that many type designers and nearly all type marketers claim their types are designed for legibility. Exceptions include curiosities like the psychedelic lettering of 1960s poster artists, grunge types of the 1990s, and the distortions called “captchas” intended to help sort out humans from net bots. 

Typefaces are more than legible or illegible; they are, importantly, expressive. Despite claims that certain faces are “neutral”, we believe there is no such thing as an inherently neutral type design. Neutrality is not a graphical feature but a relationship between form and context. A given typeface may be perceived by readers as neutral in some contexts but non-neutral in others.

Typefaces can allude to an era, evoke an emotion, suggest a theme, promote an ideology, frame an anthropomorphic analogy, clue a reader that the designer is a hipster (or an old fogey), and express whatever else a designer has crafted into a font when it appears in the context where a typographer puts it.

Take novels, for example. In English language publishing, nearly all novels are composed in seriffed types, and roughly half of those are “old style” (typefaces designed before the early 18th century, or revivals or re-interpretations of them). No matter whether a novel is classical literature, mystery, romance, thriller, avant-garde, science-fiction, or fantasy, set in the present, past, or future, in a magical realm, a galactic empire, utopia or dystopia, in a world of ordinary men and women, peasants or lords, vampires, trolls, aliens, robots, cops or robbers, whoever - its story is told with serifs. In the context of novels, seriffed fonts are a “neutral” style. (But if you know of a novel composed in sans-serif - please send us a photo!) 

Yet, in other kinds of publications, such as photography books, art books, museum catalogs, sans-serif types can become a neutral style, while old-style seriffed types may stand out as anachronistic. 

By designing a screen font in a cursive style, distantly derived from Renaissance handwriting and more directly based on modern marker handwriting, adjusted to reading requirements on computer screens, we tried to make a bold statement in favor of font expressiveness. 

The increase in weight combined with the narrowed letter widths gave Big Casual Bold Narrow a more emphatic boldness than either modification alone. The huge x-height made the face seem 16% bigger than Chicago and 25% bigger than Lucida Grande (which later became a long-lived, successful system font on its own). The cursive, bounding arches made the Big Casual look active. In the well-known pangram (a sentence containing all the letters of the alphabet), “A quick brown fox jumps over the lazy dog,” Big Casual really looked like it was jumping. 

Ill5-quickbrownfox-01

FONTOLOGY

Our discussion until now has been about design: graphical solutions to the twin problems of readability and expressiveness in typefaces. But, design is only a part (albeit the most important part) of the making of a font. There is also technology: bits, bytes, outlines, splines, hints, character codes, and so on. Those we lump together under the rubric of “fontology”, the technical lore of fonts. 

In the 1990s, computer engineers wanted fonts to be small in memory but fast in rasterization. Early in the decade, software was distributed on floppy disks with limited storage capacity. Later in the decade, CD-ROM media had much greater capacity, but processors were slow and RAM expensive, compared to today’s hardware. Hence, fonts and rasterization placed a non-trivial burden on processors, RAM, and storage media. 

For most of the design and development process of Big Casual, we used the Ikarus system of Peter Karow as our basic digitization tool. We had begun using Ikarus ten years earlier and felt comfortable with it, a tool we knew well. It offered very fine resolution, some 15,000 units per em square, seven times the 2,048 units of TrueType and fifteen times the 1,000 units of PostScript Type 1. Output fonts don’t need the very high resolution of Ikarus, but Karow developed Ikarus so it could produce an archival format with resolution higher than any practical output device would ever need. So far, he’s been right. 

After we had digitized the outlines the way we wanted them in Ikarus, we converted them, still using Ikarus, to Apple’s TrueType B-spline curve format, adjusting tables and other data with software called "Ingredients" by Henry Pinkham and Ian Morrison. 

TrueType fonts usually rasterized faster than Adobe’s PostScript Type 1 font technology, which Apple had licensed before developing TrueType, but TrueType, which uses quadratic B-splines instead of Adobe’s cubic Bézier curves, typically needed more outline points (spline knots and control points) to describe curved contours faithfully. 

At printer resolutions below 600 dpi, and at most screen resolutions up to the past few years, both Type 1 and TrueType had to be “hinted” with program code that adjusted character outlines to conform to output digital raster grids, so as to make pleasing bitmap patterns. The hinting instructions for TrueType were more verbose than those in Type 1 — explicit instructions instead of the parsimonious “hints” of PostScript Type 1 format. In TrueType, carefully written hinting code could double the file size of a font. For italics, the hint burden could be even greater, although we had some doubt about how useful it would be for a font like Big Casual. 

There were engineering pressures to reduce font data and file size, but there were opposing pressures to add more characters, which increases font and hint data. To enable Big Casual to function as a system font, we added more symbols, arrows, and signs, all in the cursive style of the typeface. These additional characters increased the data size of the font, but we wanted the look of the font to be consistent throughout all characters, so we didn’t use data-economical geometric shapes for signs and symbols. 

Big Casual outlines were almost entirely composed of curves, responsible for its relaxed, supple, fun look but also adding a lot of data. We had made TrueType Chicago economical in outline data by building it mostly of straight lines and circular arcs at corners; it needed few hints to rasterize like the original bitmap Chicago at the standard Macintosh user interface menu size. Compressing the data of Big Casual Narrow required an entirely different approach.

To compress outline data, we considered building characters from primitive parts, an ingenious concept that computer scientist Philippe Coueignoux had developed in his MIT thesis in 1975. This could have resulted in extreme compression, but would have been a major challenge. We had previously worked out a decomposition structure for an early design of Lucida, but it was never used in practice. We soon decided that the Big Casual development schedule wouldn’t be able to sustain a delay if a decomposition experiment failed, and we felt that TrueType’s implementation of character composition probably wasn’t up to the task. Early in the development of TrueType, when it was still called “Royal”, we hand-hinted a Lucida Sans demo font for Apple, perhaps the first full font hinted in TrueType. Based on our early experience writing hinting code by hand, we believed that structural decomposition into strokes and pieces would be too difficult with TrueType, at least as it was then implemented.

However, standard TrueType composite functionality was an easy way to reduce font data by building accented characters from separate letter and accent glyphs. The composited characters are “virtual” and thus don’t add much data beyond for a small amount of compositing and positioning information. In a font of 400 Latin alphabet characters, perhaps as many as half may be composites. Hence, compositing the Big Casual font saved a fair amount of data, with only a slight performance cost for the compositing operations. Yet, we wanted more compression than that. 

Another way to compress font outline data is to simplify glyph outlines by using fewer on-curve contour points (also called spline “knots”) and off-curve control points. TrueType uses quadratic B-splines, essentially arcs of parabolae, to represent curves, and usually a closed curve like an ellipse needs at least four on-curve points and eight off-curve control points. Hand-drawn curves like those in Lucida Casual usually needed more points. When the number of points on an elliptical contour is reduced to only four on-curve points and four off-curve controls, the “corners” of the ellipse suddenly bulge out into a semi-squarish shape. We called this the “superellipse catastrophe” because the resulting shape resembled a superellipse (a term coined by Piet Hein, which he said was a compromise between circle and square). Superelliptical shapes have sometimes been used in type designs, notably by Hermann Zapf, who based several of his typefaces, including Melior in 1952 and Edison in 1978, on superellipse-like forms.

Edison640

Conversion from Ikarus to TrueType involves a tolerance setting for curve fitting. Higher tolerance results in fewer points but greater divergence from the curves digitized from drawings. At highest tolerance - fewest points, most divergence - we weren’t happy with the results, so we tested other software tools for curve simplification and data compression but didn’t find one that rendered the feeling we wanted in the Big Casual forms. However, noticing that each tool gave different results, we tried pipelines of different permutations of curve simplification tools: A, followed by B, followed by C; or C followed by B followed by A; or B, A, C, and so on. After we had reduced many of the contours to their most economical representations in TrueType B-splines, the font looked more relaxed and uninhibited than when we forced it to conform to our hand-drawn “design” sensibility. 

A technical pay-off was that the resulting TrueType font was substantially smaller in file size than when produced in PostScript Type 1 format (both formats unhinted). Although a special case, this showed that TrueType outlines did not necessarily require more data than PostScript outlines, and could actually be more economical, which pleased the TrueType cheerleaders.

Tools influence type design, but by how much has been argued for decades at least, and in some ways for centuries. Typefaces diverged from handwriting and from stone carving in the 1400s, but some scholars have argued that the processes of punch-cutting and casting influenced the shapes of type forms. For this font, we let digital tools, in particular a certain class of digital curves, take over as the dictators of shape. It evoked a disquieting feeling, like in Jack Williamson’s dystopian novel, “The Humanoids,” in which allegedly benevolent robots take over humanity, ostensibly for its own good, and the hero and heroine ultimately learn to live happily, more or less, with that outcome. 

Today, after two more decades of computer progress following Moore’s Law, the font file size limitations and processing bottlenecks of the mid-1990s are negligible for Latin-based fonts, so most of our efforts at data compression are no longer needed. Nevertheless, the look of the font, having emerged from a particular set of circumstances and pressures, keeps its panache and swagger.

On the screen side, there has also been hardware progress, although the magnitude is much less than for processors and memory. Display resolutions have increased by factors of four to six over the past 20 years, from 72 pixels per inch to more than 400 ppi on some hi-res mobile phone screens. Moreover, hinting is no longer needed at higher screen resolutions. Anti-aliased gray-scaling and sub-pixel rasterizing smooth the edges of letters, eliminating the annoying “jaggies” of yesteryear and rendering delicate details better. At resolutions above 300 pixels per inch, fine stroke patterns and features can be rendered near the acuity limits of human vision. (There are arguments about exactly what screen resolutions are beyond human vision, but 300 pixels per inch seems to be a fairly good estimate.) 

Our final working title, “Lucida Big Casual Bold Narrow”, was way too long and cumbersome a name to suit Apple, which renamed the font “Textile”®.* When Apple’s Copland system was discontinued, it seemed like the newly made and named Textile might never appear on screens, but some Copland features were included in System 8. An Apple support document for System 8.5 described the Appearance Control Panel and the set of “Large System Fonts” available, including “Textile”. (7) Textile continued to be a part of Systems 8 and 9 and then “Classic”, bundled with OS X until the release of OS X 10.5 “Leopard” in 2007. 

After that, Textile made only sporadic OS appearances bundled with a few applications. Even so, the font has remained popular, but instead of a work-horse system font, it caught the eyes of packaging designers and its fun personality placed it not on screens but in supermarket aisles, where it enlivened wrappers of tasty snack foods, happy home cleaning products, bouncy comfort accessories, sporty outdoor outfits, and other brightly marketed products. Even now, when we wander down the aisles of a supermarket, we see its face winking out at us.

For those who have missed the font in Apple systems, we have retrieved the digital outlines from our archive, and revived and upgraded the font, expanding its international character set and modifying a few details. To revive our original conception of it, we have renamed it “Lucida Marker”.

 

* Textile®, Chicago®, Geneva®, Monaco®, and New York® are registered trademarks and Apple Chancery is a trademark of Apple Computer, Inc. Lucida® is a registered trademark of Bigelow & Holmes Inc.

(1) “Notes on Apple 4 Fonts” by Charles Bigelow and Kris Holmes, in Electronic Publishing, vol. 4(4) 171-181 (September 1991)

http://cajun.cs.nott.ac.uk/compsci/epo/papers/volume4/issue3/ep050cb.pdf

(2) Ergonomic guidelines for viewing monitor screens varied, but most specified distanced greater than for reading on paper. Some recommended 18 to 24 inches, others, 20 to 28 inches. A common distance to paper in reading studies is 16 inches (40 centimeters), so reading distances for screens were roughly 12% to 75% greater than for paper. We assumed the middle screen distance range from 20 to 24 inches as typical, which meant that a font on screen needed to be roughly 25% to 50% bigger than the font on paper.

Big x-heights were also important at lower screen resolutions because the bigger the x-height, the better the rendering of shapes and details. We designed TrueType versions of Apple’s Geneva*, Monaco*, and New York types*, based on Susan Kare’s bitmaps, and our TrueType versions all had big x-heights. Our Lucida Grande, the OS X system font for 14 years, also has a big x-height, but not as big as Chicago’s or Textile’s. 

(3) See: Legibility of Print by Miles A. Tinker, Iowa State University Press, 1963; pp. 54-56. Tinker found that italic faces slow reading speeds by small but significant amounts, roughly 4% to 6%, and by greater amounts in suboptimal conditions, such as small sizes. 

(4) See: The Handwriting of Italian Humanists, by Albinia Catherine de la Mare. Printed at the Oxford University Press for the Association Internationale de Bibliophilie, Oxford, 1973.

For Apple, we created Apple Chancery, a formal cursive italic based on Ludovico degli Arrighi’s Renaissance italic handwriting, as taught by our teachers at Reed College, Lloyd Reynolds and Robert Palladino, the latter also Steve Jobs’ calligraphy teacher. Apple Chancery reproduces the look of hand-written formal calligraphy with more than a thousand characters, ligatures, swashes, and other variants. It was designed to show off Apple’s GX glyph-substitution typography (later called AAT), but was not intended to be a system font. 

ReynoldsPage

Comparison of page from Lloyd Reynolds' instruction manual to Apple Chancery.

(5) See: “A study of fonts designed for screen display,” by Dan Boyarski, Christine Neuwirth, Jodi Forlizzi, Susan Harkness Regli, in CHI '98 Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, ACM Press/Addison-Wesley Publishing Co. New York, 1998, pp. 87-94 

(6) See: “Does print size matter for reading? A review of findings from vision science and typography”, by Gordon E. Legge and Charles A. Bigelow

Journal of Vision, 2011;11 8 http://www.journalofvision.org/cgi/content/abstract/11/5/8

See also: How and Why We Designed Lucida

And: "Font Design for Personal Workstations" by Charles Bigelow. Byte magazine, Vol. 10, No. 1, January, 1985. https://archive.org/stream/BYTE_Vol_10-01_1985-01_Through_The_Hourglass#page/n255/mode/2up

And: "Principles of Type Design for the Personal Workstation," by Charles Bigelow, in Gutenberg-Jahrbuch 1986, Gutenberg Gesselschaft, Mainz, 1987.

(7) See: “Mac OS 8.5: Appearance Control Panel Settings”

http://support.apple.com/kb/TA25751?viewlocale=en_US&locale=en_US

 
 © Bigelow & Holmes Inc. 2014

 


Older Post Newer Post


Leave a comment

Please note, comments must be approved before they are published

Sale

Unavailable

Sold Out