This was text originally presented by Dr Fiona Ross as a talk at the First annual Friends of St Bride conference, 24 & 25 September 2002. It had since been hosted on the website of the St Bride Library London, yet recent changes in its structure, management, and website made this important resource vanish. I am honoured and pleased that Fiona agreed to have it made available again here. For this purpose, Fiona has asked me to add: “Since the writing and presentation of this paper I am delighted to note that new facts in this field have emerged through more recent rigorous research – such as those presented in Titus Nemeth’s commendable PhD thesis.”
References in blue refer to slides accompanying the paper and can be downloaded as a PDF here.
By Fiona Ross
The Department of Typographic Development at Linotype-Paul Limited was established by Walter Tracy in the early 1970s at Kingsbury in London. Already at its inception, the typographic department focused on the design and development of non-Latin fonts, alongside work on developing such typefaces as Tracy’s renowned Times Europa. 2
Linotype has a long history of involvement with Arabic typesetting. In 1911, Linotype shipped the first 15 of its machines to Cairo with matrices for Arabic fonts. Keyboards, character sets, as well as the design of non-Latin typefaces, have been of concern and of commercial interest to the Linotype group of companies for almost a century. 3
During the 1920s and 1930s the company also shipped typesetters to India with fonts for setting Indian vernacular scripts. A variety of writing systems was translated into type by Linotype, but it is the adaptation of the Arabic and Indian scripts to the new typesetting technologies of the latter part of twentieth century that characterises the work of the typographic department and exemplifies its changing nature.
In the 1950s, the Arabic typeface design Yakout was developed. It was produced in 1956 by Linotype & Machinery for hot-metal typesetting, being specifically intended to function as a newspaper text face (dispensing with diacriticals and ligatures). With the dual intention of fitting the Arabic script onto a Linotype linecasting machine for setting type for rotary printing, and of maximizing keying speeds in creating copy for daily newspapers, much effort was concentrated on reducing the normal Arabic character set of over 100 characters. Yakout was designed in a similar manner to Arabic typewriter fonts created during this period: used a limited range of letterforms to represent the full Arabic character set. The resultant style of type design became known as ‘Simplified Arabic’. 4 The number of characters was reduced to 56, which enabled the typeface to fit into one 90-channel magazine. A brochure at the time claimed that ‘the output of work may be increased by as much as 30 per cent’. Yakout was manufactured in six different point sizes and became, indeed remains, one of the most popular Arabic typefaces. 5
When I joined Linotype in 1978 as research assistant, the typographic department, under the management of Tony Bisley, was converting existing type designs such as Yakout, or implementing new designs like Badr, for film composition. New Arabic in-house designs, such as Lotus, were being developed under Walter Tracy’s consultancy.
According to Timothy Holloway, a type designer who worked in the typographic department for about 18 months, the non-Latin fonts adapted for filmsetting were never wholesale conversions of existing metal designs. The policy of reshaping type designs to make best use of new technology continued throughout the department’s life. In other words, improvements in the quality of non-Latin typography overrode concerns for maintaining compatibility with previous font formats.
Embracing technical developments in typesetting meant a high degree of experimentation by the department. My first task was to work on a scheme named High Speed Arabic for the V-I-P photo-typesetter. Traditional Arabic fonts needed four film strips to generate a full character repertoire. Setting Arabic was considerably slower than setting copy in English. The brief for High Speed Arabic was to rearrange the layouts of the V-I-P film fonts so that the most frequent characters occurred in the central positions of the first fonts. Faster setting speeds could be achieved due to fewer lens movements and less frequent font changes.
Specific software designed for setting Arabic was first developed by Linotype for the Linotron 505 typesetter. Innovative programming enabled the selection of contextual forms for Simplified or Traditional Arabic, the kerning of some characters (by setting them in the reverse direction) and the placement of diacritical marks above and below letterforms. It was used for the typeface Osman (subsequently renamed Badr) which was originally drawn and of which negatives (friskets) were made at Linotype & Machinery in Manchester.
The 505 Arabic software was enhanced for composing Arabic on the V-I-P. It allowed the setting of additional ligatures; bi-directional composition no longer presented a problem; improvements to kerning were introduced; and above all, additional Arabic script languages could be composed, Farsi, Jawi, Pashto, and Urdu (Naskh). A collection of Arabic script typefaces were designed for implementation on the V-I-P.
Bisley’s management of the typographic department was brief, and in 1979 only two people remained. Although Walter Tracy had retired, he continued his weekly visits – even when the department relocated to Chelham House in Cheltenham, which became the head office of Linotype. After the relocation, the structure of the department changed. Graduates in fine arts, rather than apprentice letter-drawers, were employed in the design studio and other graduates were employed for research and development work. The staff grew to ten, most of whom remained for over 12 years. 6
This continuity of staff enabled the undertaking of large projects, and provided the opportunity to increase the overall responsibilities of the department. By 1980, the brief for the design studio, headed by Georgina Surman, was to produce finished artwork from in-house designs, or from commissioned designs, or for conversions to new formats. The studio was also to produce negatives (usually hand-cut friskets) and to proof the final fonts. The fonts were manufactured either in Germany by D Stempel (later absorbed into Linotype-Hell), or in America by Mergenthaler Linotype.
Meanwhile the responsibilities for the R & D section (staffed by three people including myself) had increased considerably. We did linguistic research and feasibility studies for potential projects. The artwork analysis and supervision became my responsibility; but the tasks of font layouts, keyboard layouts, font encodings, software specifications, screen fonts, font specification tables (for accent placements, character selection, etc.) were shared amongst us. This section, whose staff included Ros Coates and Gillian Barrett (and also for several years Sarah Morley) was also responsible for quality control and font testing, sample setting, and release documentation.
Assistance came in different forms. In the case of Arabic, the linguistic input and customer feedback came from Linotype’s Middle East Liaison Office (MELO, located first in Beirut and then in Cyprus). This was staffed by 13 native Arabic speakers from different parts of the Middle East and Africa, who negotiated with clients and also conducted applications testing. Technical support was also available from the font manufacturers Stempel and Mergenthaler Linotype. In addition to the services of such consultants as Walter Tracy and Tim Holloway, both designers of Arabic type faces, good relationships were created with Arabic calligraphers and designers whose type designs were adapted and unitised (re-fitted) to work within the Linotype type design guidelines.
Vital assistance was provided by a team of programmers situated at Linotype in Cheltenham who were dedicated to non-Latin support work. Film composition was giving way to digital photocomposition and Arabic and South Asian software requirements were energetically discussed and incorporated into the new non-Latin programs for the next generation of typesetters.
Yakout was one of the first Arabic typefaces to be digitised. The design was revisited by the typographic department, and additional forms were introduced since the Light and Bold fonts no longer needed to be ‘simplified’ for the Linotron 606 machine. Perhaps this was the only design which was treated fairly conservatively in its initial adaptation to digital technology: the typeface was in daily use by major newspapers which did not want a significantly different appearance or word count to affect their columns.
However, the Linotron 202 typesetter led to a radical reassessment of all the Linotype non-Latin fonts and the adoption of a new approach to non-Latin type design that was particularly useful for Indian scripts. The popularity of Linotype’s 202 machine in the world of Latin typesetting encouraged demand for its use in the Middle East and India for non-Latin composition. With hindsight it could be said that the Linotron 202 was essential to the company’s success in the Middle East and also in India.
The history of Linotype’s involvement in Indian writing systems does not date as far back as that in Arabic. But its implementation of Indian scripts on the Linotype linecaster was significant, not least by enfranchising part of the population (by enabling the printing of electoral registers and voting slips) and by making newspapers more accessible. However, the constraints of the Linotype hot-metal machines seriously compromised Indian vernacular typography. 7 The severely restricted character sets, the lack of kerning, and the inability to position the subscribed or superscribed vowel signs when combined with poor quality Indian newsprint tended to produce barely readable text. The first step in redressing the effects of hot-metal composition was the design of a new Devanagari typeface for filmsetting by Matthew Carter. 8 A well-designed text font was essential to regional newspaper setting: Devanagari is the script used for writing Hindi, Marathi, Rajasthani, Sanskrit, and other Indian languages.
The resulting V-I-P Devanagari fonts were able to kern characters and ‘float’ vowel signs fairly accurately. The character set, though larger than its hot-metal counterpart, was still limited and needed characters to be created from two or more components. However, the overlapping headline of this joining script and the separate character parts fused well. Above all, the design based on a Nirnaya Sagar Foundry font was far superior to its predecessors. 9
At this stage the Indian scripts had not benefited from the Arabic system of selecting contextual forms. The keyboard was large, and the compositor needed to combine different elements to form all the required characters. This also incurred a great deal of font changing, which rendered Devanagari slow to key and slow to process.
When I arrived in the department, fresh out of university with a postgraduate degree in Sanskrit, I was not slow to criticise the cumbersome V-I-P Devanagari keyboard layout, and I was soon given the task of adapting this font to the even more limited Linoterm machine. This salutary experience led to the search for alternative solutions to keyboarding Indian scripts for photocomposition.
The dissatisfaction expressed by clients with current keyboarding practices in India, coupled with a strong desire by major Indian newspapers to use 202 machines for newspaper production, prompted us to do a new feasibility study for implementing Indian scripts on the 202 to improve both the quality of the typography and methods of composition. Experiences with Arabic typesetting were encouraging.10
The initial project for the typographic development of an Indian script for the 202 was by the newspaper group and book publisher Ananda Bazar Patrika. Its first requirement was for a Bengali text face in two weights for its daily newspaper; subsequently Devangari fonts would be required. Problems encountered and the skills learnt during projects for the V-I-P informed this project; and one of the first items that came under discussion was the keyboard layout.
We soon found that the concept of using software to select contextual forms could be used to an even greater extent for Indian scripts than for Arabic. This concept, and the notion of writing software to merge fonts, led to the invention of Linotype’s phonetic keyboard for Indian scripts in 1978, which effectively revolutionised keyboarding practices for Indian scripts. Based on the Indian phonological system, the phonetic keyboard simplified the keying of texts whilst optimising keying speeds. It also enabled all Indian scripts to use standard keyboards. In other words, no special hardware was required. 11
The pioneering software for Indian scripts, devised by Mike Fellows, allowed the invention of the phonetic keyboard and also transformed the way in which Indian typefaces could be designed. Once the keyboard layout had been established, the artwork could begin. Knowing that the font could contain over 256 characters, and that character components (half-forms) were no longer a necessity, designs closer to Indian orthography could be drawn. The ability to kern characters to the left and to the right, and the remarkably accurate placement system by the use of X and Y co-ordinates, provided the design studio with the opportunity to work with an unprecedented degree of freedom. The anachronistic hot-metal Bengali designs were discarded. In the case of Devanagari, the typographic department worked with Matthew Carter to revise some of the letterforms and to extend the character set to around 300 characters per weight for the first digital Devanagari fonts. 12
As well as the continued and crucial support of the non-Latin software group at Linotype, the department had additional advice from the subsidiary company Linotype India Associates, and from the India Department, notably the late Dr Tarapada Mukherjee, of the School of Oriental and African Studies at London University.
The favourable reception of the new Linotype Bengali and Devanagari 202 fonts in 1982 led to a succession of other projects for Indian newspapers. It saw over the next four years the in-house design of fonts for a further ten Indian scripts alongside the development of Arabic typefaces. 13 The design and typographic development of each writing system comprising two typestyles (usually a light and a bold weight) took around nine months. The success of the 202 machine stimulated the demand for extra typefaces in other areas: Sinhala, Thai, and Amharic fonts were also developed along with corresponding keyboard layouts, screen fonts, and appropriate software. It also laid the foundations for the development of a novel and successful system for composing Nasta’liq (a highly calligraphic Arabic script for writing Urdu, which possesses a multi-level baseline), for which Linotype obtained a patent (no GB2208556B).
According to a former marketing director of Linotype, the company’s market share in the Middle East amounted to 95 per cent of Arabic script newspapers and an 80 per cent share of the Arabic commercial press. In India, Linotype had 90 per cent of the vernacular press market. The status of type design was raised to the extent that the fonts were no longer regarded as peripheral to machine sales; so the company invested in type development. The department grew and software support increased, and by the late 1980s it was ready to take on PostScript font development, and font production in Cheltenham.
Linotype had an early entry into PostScript (Adobe’s page description language) for Latin and non-Latin scripts. By 1988 the production of Arabic and Indian PostScript fonts had begun. The phonetic keyboard enabled non-Latin scripts to participate in the DeskTop revolution, and WYSIWYG was a tremendous boon to the editing of scripts that employed contextual forms and a large array of ligatures. Not only could an Arabic PostScript system set graphics and bi-directional text on the same page concurrently, but in the case of Indian scripts using the South Asian software it was now possible to mix several writing systems at the same time.
From 1988 onwards the Linotype non-Latin library, aside from Armenian, was converted to PostScript format. Again, the typographic department took advantage of the new font production techniques that provided increased flexibility in character design. Yakout and Linotype Devanagari were amongst the typefaces that were revised to exploit the new technology by the inclusion of additional characters, the improvement of spacing, character joins, and the like. The typographic department continued to add new designs and scripts to its non-Latin Library. In-house font production facilitated and accelerated typeface development times. 14
Much as the advent of PostScript and the consequent DeskTop revolution was welcomed, it precipitated the demise of typographic departments as that of Linotype’s. In 1997 the non-Latin master fonts, drawing collection, extensive reference library, and archives were despatched to Linotype Library in Germany.
What of the type designs? During my time at Linotype, where since 1983 I was responsible for the design and development of the non-Latin digital fonts and typesetting schemes, the department produced a digital collection of 31 Arabic and 26 Indian typestyles in a variety of formats and languages. 15 Additionally, Thai, Hebrew, Armenian, and Amharic digital fonts were developed. In order to create high-quality non-Latin type designs, the department worked as a team in conjunction with outside support. For the successful typographic rendition of all these scripts, the Linotype fonts depended on proprietary software designed in-house, software that is no longer supported. But as the quality of the Linotype designs is still appreciated, the fonts are enjoying another lease of life through OpenType technology. 16
Owing to the multi-disciplinary nature of non-Latin type design and development, a team approach to non-Latin typography, as initiated by Walter Tracy in 1974, still ensures the best result. However, in the 21st century the members of the team may be situated at very different parts of the globe, and communication is effected through the Internet rather than by walking into Linotype’s design studio at Cheltenham, or by sending a telex to India or a fax to the Lebanon.
Fiona Ross works as a consultant, author and visiting lecturer, specialising in non-Latin type design and typography