To coincide with our newsletter featuring the work of Akira Kobayashi, we publish his essay on the making of FF Clifford™. It is a wonderfully personal story about the struggles and discoveries of a five-year journey in type design. This text and its illustrations are published in the book "Made with FontFont", which is filled with other histories and visual examples of the typefaces in FontShop's house foundry. — SC

Designing a Latin text typeface was quite a task for me, having been born and raised in Japan where people use daily a couple thousand ideographs plus two kinds of phonetic scripts and write and read both horizontaly and vertically.

FF Clifford is the first roman text type I designed for use in Western countries. (I had designed more than a dozen latin alphabets for two Japanese type foundries from 1990 to 1994. The alphabets were made to match the Japanese text typefaces.) The Clifford family has 25 styles and took me five years to complete. I sometimes had to look into historical resources, which helped me greatly. There is an old Chinese saying: looking into the old, learning something new. That was how I designed the type.

My first impression of digital type was not good. I remember very clearly when I first used a digital version of Bembo typeface. I had never touched a computer until I worked for a type design studio in Tokyo in 1991, right after I came back from London where I learned calligraphy, read books on typography, and studied the English language. The studio bought type design software from a European company. The software company offered ten digital fonts from their library and I was responsible for choosing the fonts. The fonts I chose included Bembo, Bembo Italic and Bembo Bold. The fonts arrived and I immediately installed them in the computer on my desk. Bembo Roman was the first font I opened with the type design software. The font layout appeared, and I double-clicked the lowercase 'g' just to check if the software worked all right. What I saw was a letter 'g' about ten-inches high, and I thought,

"Oh, I opened the wrong font, silly me!" because it did not look like a Bembo 'g'. Then I re-selected Bembo from the pull-down menu, and opened the 'g' but the result was the same. After seeing the 'a', and another letter, I got into a slight panic. None of the letters looked like Bembo! For a moment I froze in front of the computer, thinking about writing a letter of complaint to the company for sending us the wrong font.

After a while I checked the Bembo Italic and I slowly began to realise that the fonts were Bembo. I calmed down enough to recall that the typeface was originally designed for metal type, and most of the specimens and texts I saw were set in metal type in text size. That was why the images of the characters did not overlap. I knew that a metal typeface was cut or designed separately for each size, but a film composition or digital face is a kind of compromise having proportions designed for reduction and enlargement. I was overwhelmed to see the huge gap. Then I looked into the types used in Western offset-litho prints to see the digital Bembo types in use (this type was and still is unpopular in Japan). The types that were originally designed for hot-metal often looked too light and feeble. I even purchased a digital rendering of a famous foundry type that cost ¥30,000, but it did not quite satisfy me. I began thinking about the need for a new digital text typeface suitable for offset-litho printing. The idea was still vague but it had to be conventional or classical in design and robust, such as Monotype Plantin.

My idea was further developed when Barry McKay, a friend of mine who I first met at the ATypI conference in 1990, gave me a book. It was a 16mo book printed by the Foulis brothers of Glasgow in 1751. The type used in that book was Wilson's Long Primer roman type. The book was in Latin, which I could not read. But at first sight, I could tell it was extremely legible. The type was so beautiful that I liked it immediately. The colour of the text was even, the letterspaces were consistent.

Inspired by Wilson's type, I started a few drawings of my new typeface. I had basic ideas to start with. The type was to be an oldstyle face with some robustness. The words set in the type were to be comfortable to read, each character would be beautiful. All the subtle curves and nuances would not be ironed out. However, I was not interested in rendering a faithful image of the printed letters, with all the bumps and warts. No photographic enlargement was made. What I wanted to do was to distill the letterform from the prints by hand. My goal was designing a new text type of even letterspace and colour, not an imitation or a reproduction of the metal type.

I drew carefully several lowercase half-inch-high characters while thinking of a way in which they might be used. The drawings were scanned and traced by hand. The rest of the characters were drawn directly onscreen. Adjusting the width and letterspacing followed. It was a long and slow process; I assume I spent more time working on letterfitting and kerning than drawing the characters.

When the design process of the roman type on the Mac was nearly finished, I began drawing the italic type. There was no italic type shown in the Foulis book. The Joseph Fry & Sons' Pica Italic No.3, shown in their specimen of 1785, was chosen for the model. Although the model was not made by the same hand nor in the same peroid of time, I believed it would match the roman. Again there was no photographic enlargement. I retained all the vigourous swashes on the letter 'Q', 'T', and 'Y' that were typical of the Anglo-Dutch oldstyle types.

When a set of roman and italic was completed, I set various texts in different sizes. The type was not bad, but it obviously had limitations. In larger than 18 pt, it already started to look too bold and clumsy, and too dark and compressed when used in 6 or 7 pt. I remembered being disappointed the very first time I bought a digital type. As long as the type has one master design, the limitation is inevitable. Usually a serifed typeface that looks all right in 10 point cannot look refined in 72 point.

The ultimate solution was "optical scaling", or to design different masters for different ranges of size. The idea was not innovative at all. In fact it had been the ordinary method of making a series of typefaces for five hundred years. Because the punches, the original letters prepared for mass-production, had to be cut one by one and size by size. Most filmsetting and digital types currently available have only one letterform and the types are enlarged or reduced mechanically. The modern method was efficient from an economical point of view but rather a disadvantage for the readers, as many type critics had pointed out. Also, there was an excellent forerunner in the digital type industry who had employed optical scaling. ITC Bodoni™ was designed in three variations: Seventy-two, Twelve, and Six. That was appropriate for the "modern serif" typeface. Otherwise, the delicate hairline serifs would fade away in small size and look blunt in displays and headlines.

The greatest disadvantage of optical scaling is that the type designer has to spend more time completing the family. But I had been designing the type for fun so I did not have to rush. The next drawback is the number of font families. The more variation the type has, the more it takes up computer memory. But in today's environment of type and graphic design, the amount of memory that a font family requires is very small when compared to that of a full-colour image. So I decided to make the Clifford type a size-sensitive text type family. First I had to seek after the letterform that was appropriate to the function of the size of footnotes, but there was no written criteria of designing variations for smaller size. I looked into metal type specimens to see how the designers of the metal type era overcame this task. Particularly, I referred to the way Caslon Old Style exhibits certain irregularities depending on its size as a part of its distinctive character, and the Monotype Bembo series, the 6 point of which is extremely well designed. What I found in the specimens during the research excited me. In smaller size, the contrast of thick-and-thin was reduced, x-height was set a little higher and the extenders were designed shorter, the fit of the letters was looser. The overall shapes of the individual characters were sometimes altered as well. It was most descernible in the Caslon Old Style type. The curlicues in some swash characters, which often appear too complicated, were made simple in small size, and gorgeously flourished in larger sizes. The modifications were made so cleverly that I could not take my eyes off the old type specimens!

All the modifications were made for the small size variation, and the result satisfied me. However, there was one question remaining. Should I add a bolder version to the already dark text type? I tried to design a bold variant but it looked awkward so I abandoned it. Instead, I started designing the larger size which was meant to be used for chapter headings, small headlines or large texts. At this point the type had three optical size variations: type for text, another for footnotes, and another for headings.

In 1995 I had a chance to write the renowned calligrapher and type designer Hermann Zapf about the Clifford typeface in progress. He wrote me a favourable reply about the design of the type, but he also warned me that it was not the right period of time to submit a traditional text face to font distributing companies. I also showed specimens of the type to Juzo Takaoka who was running a letterpress printing workshop in central Tokyo. He also liked my typeface, but he said that I should choose my time well and meanwhile, try to design something different. So I decided to freeze temporary the Clifford and wait for the appropriate time. I started to design a sanserif typeface family based on my handwriting in 1996 and I submitted the type to the International Typeface Corporation® (ITC) in New York. It was named ITC Woodland™ after my family name Kobayashi (which means small woods).

In autumn 1997, I saw an advertisement for a typeface competition organised by U&lc, the graphic magazine published by the ITC. The competition had three categories: Display, Text, and Picture fonts. It was a good oppportunity for me to have my text type design assessed. All of the six variations were reviewed before the submission. I reviewed and revised almost every letter. It took longer than I had first imagined. Eventually the Clifford type received the Best of Category (text) and the Best of Show awards at the same time. In summer 1999, the type was finally published by the FontShop International under the name of FF Clifford. The total number of the family is 25, including fonts with lining figures, expert sets, and borders. In 2000, Clifford received the TDC2 Certificate of Excellence in Type Design from the Type Directors Club in New York.

The Clifford type is a bridge between the past and the present. I learned from old fellows through their type design, and implemented their ideas to my digital type when necessary. Electronic enlarging and reducing of a single letterform seems to be the mainstream, but it may look monotone. A printed page set in variations of slightly irregular quality will be lively and more pleasant to read. A size-sensitive family of 25 fonts may sound like too many, but thanks to today's technology, it is easier to handle than metal type, since the cases of which require storing.

Another typeface inspired from the past is FF Acanthus™, published in 1998. It started when Clifford was almost finished. FF Acanthus is more faithful to its source of inspiration: Henri Didot's (1765-1852) type used in De Imitatione Christi, Paris 1788. It may be classified in the "modern" category but there is little straight lines and razor-sharp corners. Again I did not intend to imitate the original printed surface, but I thought the moderns should look that way. The serifs of Acanthus are extremely thin, but they are very carefully emphasized at the ends. Acanthus was designed for display work or large text. To me Henri Didot's italic did not seem to match Acanthus roman. It was somewhat too stiff and mechanical. I had to draw a new italic in the spirit of the period. Recently, I added the "text" variations for use in around 10 point.

I could not stop designing. In 1999 I came up with another typeface, this time based on the fifteenth-century type by Sweynheym and Pannartz: the two German printers active in Rome at that time. They produced a unique, slightly-out-of-balance-but-attractive type. This time the original print acted merely as a reference. The distinctive lowercase 'a' and some other letters were inspired by Sweynheym and Pannartz's second roman type, but I revived the type in a more informal way. Here I used the historical type as a springboard. The resulting type looks different, taking on a rather contemporary and lively look. It received a prize in the text type category in the third Linotype International Type design Contest in 1999 and has been released by Linotype®. I assume that Linotype Conrad™ is the first revival of the Sweynheym and Pannartz type, though it does not closely resemble the original.

Designing new types is fun. It is also stimulating to look into the old type specimens and rediscover how clever the old fellows' designs were. Reviving old types in the digital era does not necessarily mean designing cheap imitations. Sometimes they help me create traditional-looking types, and sometimes the types that are fresh and lively. I am seeking another good type from the past waiting to be rediscovered.

Read more first-hand accounts from type designers and see more typographic eye candy in "Made with FontFont", a new book in celebration of FontFont's 15th anniversary.

» Download a PDF sample of FF Clifford and print on a laser to see it at its best.