30th ANNIVERSARY OF LUCIDA TYPES
2014 is the 30th anniversary of the first showing of Lucida, the first family of original, digital typefaces for laser printing and screen displays, so we think it is time to write more about our approach to the design.. We first wrote about “The Design of Lucida” in 1986, and on “The Design of a Unicode Font” (Lucida Sans Unicode) in 1993. To observe the 30th anniversary of the family, we offer these notes on salient features of the original Lucida fonts, and what we thought when we designed them in the early 1980s.
Over the past 30 years, some of our reasons have been proven well founded, while others have been irrelevant to trends in typography, and some remain inconclusive. After these initial paragraphs, we offer further notes including pros and cons, alternative views, personal recollections of our teachers, and a bibliography.
1. BIG X-HEIGHT
The x-height of Lucida Grande, Lucida Sans, and all other Lucida fonts is large, approximately 53% of the body size. Lucida’s large x-height has two functions that help it adapt to reading on screens and printing on modest resolution devices. First, a big x-height makes the typeface appear perceptually bigger, aiding legibility when text is viewed at greater than average reading distances or at small sizes, or both. Text on monitors was read at distances 50% greater than on paper, according to ergonomic recommendations of the 1980s. Second, the big x-height provides more pixels for better definition of features in the x-height region, which typically carries more information than ascenders and descenders, thus helping distinguish letter shapes for better recognition.
Before designing the Lucida outline characters, we made experiments with hand-edited bitmap renderings of letters at various resolutions to study how high resolution forms devolve into minimalist pixelations at low resolutions. The x-height always seemed most important for letter recognition. We produced a series of bitmap fonts, which we called “Pellucida” for screen displays, including on a Smalltalk workstation and on the operating system Plan 9 from Bell Labs.
2. GENEROUS SPACING
Lucida Grande, Lucida Sans, and the original Lucida seriffed faces have more space between letters than most modern types. The generous spacing improves legibility at text sizes (8 point to 14 point) at low resolutions on screens and printers.
Lucida spacing was influenced by the generous spacing of early roman typefaces, like Jenson’s roman of 1470, which remained legible despite the “noisy” environment of rough paper, easily worn types, and uneven pressures of early printing technology. Also, Lucida spacing is partly based on adjustments for three visual phenomena: “optical scale”, a traditional craft adjustment of letter shapes for different sizes, which included wider spacing for small sizes; tuning the base spatial frequency, the alternation of black strokes and white spaces, to the peak sensitivity of the human visual system; and compensation for “crowding”, a problem recognizing letters set close together. Beyond these global adjustments, generous spacing also prevented some local problems when errors in rasterization and fitting make adjacent letters accidentally merge, as often happened with a popular grotesque sans-serif in early laser printers, when ‘r’ touched a following ’n’ and made a spurious ‘m’, turning words like “fern” into “fem”.
Counter-forms are the spaces inside letters; some are totally enclosed as in ‘b’, others non-enclosed as in ‘c’. A few letters have both enclosed and non-enclosed counters, as in roman ‘a’, ‘e’, and ‘g’. Enclosed counters can clog-up in printing, so we tried to make enclosed areas big. The nearly enclosed counter-forms of ‘c’ and ‘e’ in “grotesque” style faces, while stylish at big sizes, appear to close up if the gap (aka channel or aperture) separating the two terminals of ‘c’ gets clogged or blurred, making ‘c’ confusable with ‘o’.
4. DISTILLED LETTER FORMS
Lucida Grande, Lucida Sans, and original Lucida seriffed have forms and thick-thin proportions derived from pen-written letter shapes written and read in the 15th century by Italian Humanists, whose handwriting was the model for the first roman typefaces. We had studied Humanist handwriting as students of Lloyd Reynolds and others. The Humanists based their writing on what they thought was the most legible ancient handwriting, written by scribes in the court of Charlemagne 600 years earlier. Early Humanist letter forms were simple and unadorned, crafted to be easy to write and easy to read, even by older scholars with declining vision, in an era when eyeglasses were rare. The Humanist style was therefore extensively “user tested” in two different historical eras. Of course, nearly all roman types descend from one or another era in the long evolution of type forms that began with Humanist bookhands, but Swiss designer, Hans Ed. Meier, showed us that a modern sans-serif type could be created by going all the way back to Humanist forms.
5. DIFFERENTIATED DETAILS
A problem at low resolutions is that letters start to look alike because there often isn’t enough information to distinguish shapes easily. Type styles that assimilate forms, like geometric and grotesque sans-serifs, are particularly prone to this problem, especially along the upper region around the x-height, where traditional typefaces rely on details of shaping to differentiate letters. Even in an ostensibly simple sans-serif ’n’, there is a white cut or crotch where the arch joins the left stem. This cut, along with the square corner of the left stem, helps keep ’n’ from being confused with ‘o’. At low resolutions, these differentiating details can be obscured, so, we lowered the arch join, cutting more deeply into the shape than in normal grotesque sans-serifs. This also tended to increase the thickness of the arch, further distinguishing ’n’ from ‘o’. We cut off terminals of curved strokes and diagonals vertically, to align with the vertical axes of digital rasters. However, we kept the serif-like terminal on ‘a’, to differentiate it from other letters. We tried to use the elegant Humanist ‘g’ with closed lower loop, but our bitmap tests showed that the letter shape did not survive at low resolutions and small sizes, so we settled on the “grotesque” style ‘g’ in Lucida Sans and Lucida Grande. As in Aldine humanist typefaces, we drew ascenders taller than capitals, to distinguish lower-case ‘l’ from capital ‘I’, and also to de-emphasize capitals slightly so that all-capital composition like acronyms, common in high-tech prose, and texts with frequent capitals, as in German orthography, did not unduly interrupt the pattern of text.
6. ADJUSTED CONTRAST OF THICK/THIN STROKES
In the original Lucida seriffed faces, the thickness ratio of main stems to hairlines is 2 to 1, thicker than in a face like Times Roman. We observed that in early laser printing and screen displays, thin hairlines were often “broken” by white gaps because of errors in rasterization. Such breaks made letters difficult to recognize and text annoying to read, so we thickened hairlines and serifs to avoid breaking or loss. Yet, for Lucida Sans, and Its twin sibling, Lucida Grande, we maintained some thick-thin contrast, roughly 4 to 3, following the ductus of pen-written roman and italic. We felt that near-monoline sans-serifs like some “grotesque” styles tended to look a bit stolid and dull. We wanted to give our sans-serif slightly more graphical modulation in its lines, and the slightly thinner lines also helped keep the text from darkening too much on write-black laser printers.
We drew Lucida by hand but digitized it with the Ikarus software system developed by Peter Karow. We edited the digital outlines to achieve precise regularity of base-line, x-line, capital line and other alignments, and to ensure that repeatable letter elements like stems, bowls, and serifs were digitally identical. This made it easier for software to recognize and adjust outlines, as was first done in Peter Karow’s Ikarus programs, and later in the “hints” of PostScript Type 1 and “instructions” of TrueType font rendering. Following research by Philippe Coueignoux, we experimentally decomposed Lucida letter shapes to a small set of repeatable component parts from which all the letters could be assembled, in case extreme data compression was needed for some implementation. This extreme data reduction was never needed in commercial font formats for Latin fonts, so we didn’t continue it.
The ratio of vertical stem thickness to x-height in Lucida normal weight fonts is 1 to 5.5. This is slightly heavier than most seriffed text faces. Although Times Roman has about the same stem to x-height ratio, it has thin hairlines and serifs that lighten the overall tone. The normal or regular weights of some popular grotesque sans-serifs are slightly lighter than Lucida normal or regular weights. When we designed the first Lucida fonts, we chose a slightly dark weight to compensate for erosion around the edges of black letters on white background screens and on write-white laser printers, which visually reduce weight, making text look weak in small sizes. The slightly dark weight made Lucida well adapted to most screen displays for almost 30 years, but printing on 300 dot-per-inch write-black laser printers had a slightly darker tone than we desired. When printer resolutions increased to 600 dpi, this darkness was mostly corrected because the percentage of weight added by write-black laser technology was reduced at the higher resolution. As screen and printer resolutions increased we wanted to offer a broader gamut of tones for the medium weight range, to adapt text to different conditions, colors, and backgrounds. We finally did so in 2014, producing Lite (1:7.3), Book (1:6.3), Text (1:5.9), Normal (1:5.5), Thick (1:5.2), and Extra Thick (1:4.9) weights for Lucida Sans, Lucida Casual, Lucida Calligraphy, and Lucida Handwriting. To complement the extended gamut of text weights, we also produced a broader range of bold weights. For example, the “Retina” weight of Lucida Grande Bold is 4% bolder than the original Lucida Grande Bold, to give the bold weight more visual emphasis on high resolution LCD screens. [see Lucida Basic Font Weights and Visual and Semantic Functions of Typeface Weight]
A full showing of Lucida Sans weights is in the Lucida Fonts store.
Lucida was first shown at a meeting of the Association Typographique Internationale (ATypI) in London, September 1984, in the form of a type specimen booklet from the Imagen Corporation, a Silicon Valley laser printer manufacturer. The Lucida booklet was designed by Michael Sheridan, Imagen’s type director whose appreciation of fine typography stemmed from his prior work experience at Grant Dahlstrom’s Castle Press, a Pasadena, California printing firm renowned for fine typography and printing.
Today, new, original typefaces are released every day, so it may be hard to believe that three decades ago, there were nearly none. As typography shifted from analog to digital technology in the 1970s and 1980s, typefaces for digital typesetters and printers were, with very few exceptions, digitizations of existing typeface from the previous eras of metal or photo-typography. (Among the few instances of original designs for high-resolution digital typesetters were the Marconi (1976) and Edison (1978) news face type families designed by Hermann Zapf for the Hell-Digiset firm, which had invented and demonstrated the first digital typesetter.) In the article “Digital Typography” by Charles Bigelow and Donald Day in Scientific American, August 1983, we wrote that the initial, imitative phase of digital typography would eventually be followed by a creative phase of original design, but that had not happened in laser printing and screen displays by 1984, so one reason we developed Lucida was to show that original digital designs could be effective and successful.
The x-height of Latin typefaces has been known to be important for legibility of text and economy of paper since the mid-1500s, when printers began to use typefaces cut in the style of Garamond but with larger x-heights, by Pierre Haultin, Robert Granjon, Hendrik van den Keere, thus beginning a gradual but sometimes intermittent trend toward larger x-heights over ensuing centuries. In 18th century France, Pierre-Simon Fournier cut faces of different x-heights on the same body sizes, demonstrating how x-height influenced the perceived size of type, and the Caslon foundry in England began to offer selected fonts with larger x-heights. In the 20th century, newspaper economy led to typefaces of successively larger x-heights. Stanley Morison, co-designer of Times New Roman for The Times newspaper, proposed in a 1963 essay “On the Classification of Typographical Variations” that the history of type design involved the reduction of the real size of letters while maintaining their apparent size. Harry Carter, in an influential on the “Optical Scale in Type founding” first published in 1937, observed that x-height, as a proportion of body size, was greater for small sizes of type, less for larger. Decades after the design of Lucida, Gordon Legge, a reading psychophysicist, and Charles Bigelow summarized several arguments in favor of x-height as the main indicator of perceived type size. In a paper in the Journal of Vision: “Does print size matter for reading? A review of findings from vision science and typography” (J Vis 2011;11 8.), Legge and Bigelow cite a preponderance of large x-heights in newspaper and computer screen typography, environments where visual “real estate” is at a premium, for the former, because of the cost of newsprint, and for the latter, because of limitations on screen sizes and resolutions. Other scientists, however, (cf. Aries Arditi) have argued that capitals are more legible than lower-case. Independently, typographic measures in in Germany often assume capital height as a standard, presumably because of the frequency of initial capitals on nouns in modern German orthography.
Early roman types of the Renaissance, in particular those of Nicolas Jenson in Venice in the 1470s, are generously spaced. This equalization of spaces between letters to counter-spaces inside letters makes a regular and enjoyable rhythm of alternating strokes and voids in the pattern of text. A technical bonus was that generous letter spacing maintained distinctiveness of letters when ink squash, rough paper, type wear, and uneven printing pressures tended to degrade the text image. We thought that generous spacing would likewise help in the early days of digital type, when coarse resolutions and laser printing tended to degrade the look of well known printing types. As type technology improved in the Renaissance, letter spacing was slightly reduced in printing by Aldus Manutius with the types of Francesco Griffo at the end of the 15th century. Spacing was slightly reduced again in the second half of the 16th century and in the 17th century, as paper costs became an important factor in the printing business, and when publishers and readers began to favor economy over luxury. In the second half of the 20th century, tight letter spacing became common in advertising typography, not for economy but for fashion. Although tight spacing of grotesque style typefaces had occasionally been done by hand-trimming of sidebearings, photographic typesetting technology enabled much easier tightening letter spacing. At display sizes, tight letter spacing attracted attention, and this trend eventually affected text typography. A common rationale for tight, “sexy spacing” was that when letters were crammed together, they made distinctive word images that were easier to read, based on an hypothesis that reading is done word-by-word. Numerous reading studies have since shown, however, that words are read by recognizing letters, not as unitary chunks or gestalts, so the up-tight rationale is demonstrably false. The actual situation is more complex and still not fully understood.
We used generous spacing for Lucida fonts not only because of humanist and typographic antecedents, but also because of some findings in vision science.
The first interesting, to us, finding was the contrast sensitivity function (CSF) explained by Campbell and Robson in “Application of Fourier analysis to the visibility of gratings,” published in 1968 in the Journal of Physiology. Determinations of the CSF by Campbell and Robson, and other researchers, indicated that the human visual system is most sensitive to spatial frequencies in the range of three to six cycles per degree of visual angle. The contrast sensitivity function also indicated that the highest perceivable spatial frequencies are around 50 to 60 cycles per degree of visual angle. This upper threshold comes up in discussions about whether the 300+ pixels per inch resolutions of recent smart phone screens are above the resolving limits of the human eye. (Vision scientists often use visual angle as a measure of type size instead of physical size, e.g. typographic points, because visual size at the retina varies according to the eye’s distance from the text, whereas degree of visual angle is constant. Text in 12 point type read at 16 inches has the same visual angle as text in 18 point read at 24 inches.)
Today, it is reasonable to ask if the more generous spacing of Lucida really makes it more legible. Anecdotally, yes. Lucida Grande functioned well as system screen fonts on Macintosh OS X for 14 years, at sizes ranging from 10 to 14 point. Also, a variant, Lucida Console, has been used a terminal font in Windows for 20 years. But, the jury is still out on whether the hypothetical advantage of generous spacing actually improves reading speed and comprehension. It is also reasonable to point out that the contrast sensitivity function is not the same as the modulation transfer function for human vision, so using the CSF may not be the optimal way to determine the ideal stroke frequencies for a font.
Another finding from vision science was termed a “visual crowding effect” by Herman Bouma In a paper entitled “Interaction Effects in Parafoveal Letter Recognition”, published in 1970. Bouma investigated the crowding effect further in another paper, “Visual interference in the parafoveal recognition of initial and final letters of words”, in 1973, Bouma found that the ability to perceive fine details is impaired when contours are close to the details to be recognized. In particular, recognition of letters is impaired when flanking letters are close by, and impairment worsens the farther the letters to be recognized are from the fixation point of central vision. Bouma’s observation that perception of details was impaired by close contours caused us to think that generous spacing could ameliorate some problems in recognizing type on screens. We already thought that the tight letter spacing of popular grotesque faces was a hindrance to reading at text sizes, and Bouma’s papers reinforced our visual impressions. In the 30 years since we designed Lucida, there has been a great deal more research on crowding, including by Denis Pelli and other vision researchers. It now appears that loose letter spacing is not a cure-all for crowding; the more space is added between letters, the farther the outlying letters are from central fixation, and thus the worse the crowding effect, which is proportional to distance from central fixation. Although there have been some studies suggesting that wide letter spacing is helpful for dyslexic readers, other studies have not found this benefit. Still, the slightly more generous spacing of Lucida has seemed to be helpful on screens.
Today, it should be possible to to improve readability by adjusting letter spacing automatically, depending on parameters of type size, design, and reading distance. But, in 1984, it was hard enough to make digital type at all, frankly, so we spaced Lucida to work reasonably well within the probable text size range. It was intended to be a text face, and that’s how we spaced it. It is possible to create optical size masters to tune a design to different size ranges, as has been effectively done for a few families of digital fonts. We experimented with optical masters in the 1990s, but instead of releasing them, we created more scriptal variations for Lucida, including calligraphic, casual, and handwriting styles.
Wherever possible, we opened up counters, as in ’a’, ‘c’, ‘e’, ‘g’ in the roman styles. In the italics, we adopted a chancery cursive style with a characteristic counter-form for ‘a’, ‘’d’, ‘g’, ‘q’ in one orientation, and a rotationally contrasting counter in ‘b’ and ‘p’, to help distinguish letters that can be confused. These can be seen in italic text samples here.
Because we were designing Lucida for text sizes and, often, coarse resolutions, we distilled the letter shapes to minimalist forms that we felt would be recognizable under most imaging conditions. We wanted Lucida typefaces to be without distracting details, essentially transparent as conveyors of information. As some fellow typographers commented, Lucida is a “workhorse” design. We took that as a compliment. The true italics are somewhat showier, exhibiting traces of the fast Humanist handwriting styles still called “cursive” or ‘running’. Maybe not Lipizzaner stallion show horse style, but obviously more sprightly than the romans. For simplified italics, we assumed that most operating systems and font rendering engines could automatically turn the romans into “obliques”, which are minimally italic, differing from roman only in slant. This was not always the case, however, so in 2014, we produced “Oblique” versions for most of the Lucida families. For example, the difference between true italics and obliques for Lucida Grande can be seen here.
BEFORE LUCIDA - INFLUENTIAL TEACHERS & MENTORS
We learned the art and craft of letter design from several teachers and mentors, whose influence we wish to acknowledge.
Charles Bigelow and Kris Holmes studied calligraphy with Lloyd Reynolds at Reed College in Portland, Oregon. Reynolds, Calligrapher Laureate of Oregon, taught calligraphy not as a decorative art but as the foundation of civilization. He taught italic handwriting in the chancery style of Ludovico degli Arrighi, as exemplified in Arrighi’s little manual, La Operina, of 1522, and in 20th century revivals of chancery cursive. He taught pen-written roman capitals in Italian humanist style, emphasizing that the classical capitals had several width groups, instead of the modern tendency to make most capitals of similar widths. and a series of historical alphabets - rustics, uncials, half-uncials, gothics, and more. For Reynolds, the graphical shapes of letters and the means of making them were not merely artistic forms or kinaesthetic exercises, nor simply functional marks, but keys to unlocking the wisdom of the ages. Kris Holmes studied brush-written roman capitals with Robert Palladino, Reynold’s successor at Reed, and with Palladino’s teacher, Edward Catich, the master of roman capitals, when he was invited to teach a workshop at Reed.
After college, Charles Bigelow studied typography as a student and teaching assistant for Jack Stauffacher at the San Francisco Art Institute. Stauffacher was a well-known book designer and much of his typographic passion was focused on legibility and clarity of text. His most famous and controversial work was a hand printed, limited edition of Plato’s dialogue, Phaedrus, separately described in his “A Search for the Typographic Form of Plato’s Phaedrus”. Stauffacher also excelled in large, abstract type compositions, arranged spontaneously and hand printed on his proof press. He had worked with Hermann Zapf a decade earlier to design a “private press” face, Hunt Roman, for the Hunt Botanical Library at Carnegie Mellon University, where Stauffacher was then professor of typography and director of the New Laboratory Press. (Unconnected with typography, Stauffacher was an outstanding bicyclist and bicycle polo player.)
Bigelow studied visual perception and color vision with psychophysicist Gerald Murch at Portland State University in the mid-1970s. Studying an intriguing visual phenomenon, the McCollough Effect, with Murch led Bigelow to related literature in edge detection and spatial frequency perception, particularly a series of enlightening papers by Fergus Campbell, John Robson, and their associates and students, about spatial frequency and contrast sensitivity in the human visual system. Bigelow believed these explained some aspects of the perception of typographic forms, which are rife with edges and spatial frequencies. Thus began a continuing interest in the relevance of scientific reading studies and vision theories to type design. Long after the design of Lucida, Bigelow had occasions to write papers with Gordon Legge and Denis Pelli, who had been a post-doctoral associate and student, respectively, of Campbell and Robson. (Unconnected with the McCollough effect Jerry Murch raced cars.)
HANS ED. MEIER
For his private press in the 1970s, Bigelow purchased foundry type of Hans Ed. Meier’s Syntax-Antiqua, a modern sans-serif based on Renaissance handwriting and early typefaces, as part of a study on how to present texts in Native American literature. Meier was a long-time teacher of lettering at the Zurich School of Arts and Crafts, and widely respected for his fine book on the history and development of writing, in which he recreated and hand-wrote all historical examples himself. Bigelow & Holmes enlisted Meier’s help in designing special phonetic characters for Syntax, to represent sounds in Native American languages. Working with Meier was like a graduate education in type design, an introduction to the search for pure forms, distilled from the vast range of historical styles. (Unconnected with lettering art, Meier in his youth flew gliders off Swiss mountains. We don’t know why some of our influential teachers and mentors had a need for speed, but Chuck as a teenager had a similar propensity, as the traffic court of his home town often recorded.)
Kris Holmes and Charles Bigelow studied calligraphy and type design with Hermann Zapf at the Rochester Institute of Technology in the summer of 1979. Zapf told his type classes about his work in digital type, such as Marconi and Edison, and emphasized his conviction that type artists should create new designs for new technologies, and avoid making, as he put it, “warmed over” versions of older designs. In his calligraphy classes, Zapf emphasized the importance of the pen in shaping letters, in particular the legible, handwritten humanist writing of the Renaissance. (Often in his class demonstrations, Zapf swiftly wrote graceful script capitals on a blackboard, but so far as we knew, he didn’t play bicycle polo, race cars, or fly gliders. Speed and agility were in his hands.)
“Does “Print Size Matter for Reading? A review of Findings from Vision Science and Typography,” by Gordon E Legge and Charles Bigelow. Journal of Vision, (2011).
“Crowding and eccentricity determine reading rate,” by Denis G. Pelli, Katharine A. Tillman, Jeremy Freeman, Michael Su, Tracey D. Berger, and Najib J. Majaj, in Journal of Vision (2007) 7(2):20, 1–36.
“The Science of Word Recognition or how I learned to stop worrying and love the bouma,” by Kevin Larson. 2004. http://www.microsoft.com/typography/ctfonts/WordRecognition.aspx
“The remarkable inefficiency of word recognition,” by Denis G. Pelli, Bart Farell & Deborah C. Moore, Nature, Vol. 423, 752-756, June 12, 2003.
"The Design of a Unicode Font,” by Charles Bigelow and Kris Holmes, in RIDT'94, special issue of Electronic Publishing, Vol. 6, No. 3
"The Design of Lucida: an Integrated Family of Types for Electronic Literacy," by Charles Bigelow and Kris Holmes in Text Processing and Document Manipulation, J.C. van Vliet, ed., Cambridge University Press, 1986
"Principles of Type Design for the Personal Workstation," by Charles Bigelow, in Gutenberg-Jahrbuch 1986, Hans-Joachim Koppitz, ed., Gutenberg-Gesellschaft, Mainz, 1986
"Digital Typography and the Human Interface,” by Charles Bigelow, (Proceedings of the Typography Interest Group at CHI '85)," in SIGCHI Bulletin Vol 17, No. 1, July 1985.
"Introduction" to "The Computer and the Hand in Type Design," Special Issue of Visible Language Vol. XIX, No. 1, Charles Bigelow & Lynn Ruggles, editors, Winter 1985
"Digital Typography," by Charles Bigelow & Donald Day, Scientific American, Vol. 249, No. 2, August 1983.
“Visual interference in the parafoveal recognition of initial and final letters of words,” by Herman Bouma, Vision Research Volume 13, Issue 4, April 1973, Pages 767–782.
“Interaction Effects in Parafoveal Letter Recognition,” by Herman Bouma, Nature 1970 Vol: 226 (5241):177-178.
“Application of Fourier analysis to the visibility of gratings,” by F. W. Campbell and J. G. Robson, Journal of Physiology (London) 197: 551-566, 1968.
CSF Chart: http://ohzawa-lab.bpe.es.osaka-u.ac.jp/ohzawa-lab/izumi/CSF/A_JG_RobsonCSFchart.html
“On the Classification of Typographical Variations,” by Stanley Morison, in Type Specimen Facsimiles 1-15, John Dreyfus, ed. Bowes & Bowes and Putnam, 1963.
“Optical scale in type founding,” by Harry Carter. Printing Historical Society Bulletin, 13, 144–148, 1984. (Original work published 1937, Typography 4, 2–8)