Will Hill

This paper was given at the Edward Johnston Seminar at Ditchling, Sussex in 2016

To many designers the ligature is an attractive typographic curio, or a minor refinement in the character set: a form which resolves an aesthetic problem with some elegance, and may be interesting in itself.
Traditionally sited at the extremities of the job case or the compositor’s keyboard, these are marginal forms, whose value has been contested by the rationalising influence of several phases of type production technology.
Their development, decline and resurgence, form one of the ‘underground rivers’ of type history, and provide us with some insights into where type has been and where it might be going.
The concept of the ligature highlights some key questions around type’s relationship to writing. A response to the limitations of movable type, it marks some of the boundaries of typographic thinking, and thus helps illuminate our understanding of type itself. Though engaging and often beautiful, these forms reflect the unfulfilled promise that the printed page makes to the written word, and highlight some important contradictions between the continuum of handwriting and the unitisation of the cast letter
Some typographic forms originated as ligatures to then become distinct glyphs in their own right. Some familiar examples would be the ampersand and the w, but we should also include not only the German ‘sharp -s’ or eszett, and the contractions that give us the exclamation mark and question mark. These fall outside the scope of this study, as they have come to denote distinct sounds and meanings in their own right. My concern in this talk is with those ligatures that occur for aesthetic reasons; as a means to resolving a stylistic problem that occurs in translating aspects of the written letter to the constraints of cast type.
Strictly speaking the ‘ligature’ is the linking stroke that connects two letter shapes, but the usage has developed over time, to denote a ‘ligatured combination’. I’m going to narrow that definition down in one respect and open it out in another. On the one hand, I’m not including those forms sometimes described as ‘lexical’, ligatures, which exist for specific linguistic purpose and denote a specific sound-value. These would include the Latin ae, the french oe, and language-specific digtaphs including the dutch ij and spanish ll. Though they have been subject to some of the same technological constraints, theirs is perhaps a history for linguists and lexicographers than typographers.
This definition therefore extends beyond those letters that are actually connected forms, to include letter pairs that although not actually linked, would have had to be cast on a single body.
The digital version of Klingspor Shrift from the digital foundry Alter Littera in Madrid, reinstates original ligatures from Rudolf Koch’s 1925 design. As a transcription of an eclectic historical design within a specific tradition, it prefigures several different aspects of my talk.
Although Klingspor Schrift is itself a synthesis of several stylistic tendencies, as a typeface based in the textura idiom it serves to identify some characteristic ligature issues from the very origins of western movable type.
Analysis of Gutenberg’s bible pages suggests that very large numbers of sorts were manufactured in order to replicate those variations of form prompted by different letter combinations, and to aid justification. These are clearly defined in the letters of Peter Schoeffer for his 1462 bible.
The more closely a typeface aims to mimic the properties of handwriting, the more problematic letter pairs will occur, necessitating larger numbers of ligatures. This was particularly characteristic of the early italics; Harry Carter notes that the original Aldine italics cut by Fancesco Griffo at the turn of the 15th century included at least 65 ligatures, many of which ‘did little to improve the flow of the text’.
These included double c, double t, ct and st ,while c, t, f and long s were linked to succeeding vowels and other letters. We can see this as the experimental period of italic typefounding, in which common standards of practice were yet to be established and a consensus had not yet determined which forms were practical and strictly necessary.
These ligatures might be divided into further categories: those whose purpose was an entirely stylistic one reflecting the scribal approaches to particular letter-pairs, and those in which the form of the letter created a mechanical problem, requiring the cast letter to overhang the body with a fragile kern.
Most notable among these problem forms is the lower case f and – for the time it continued in use – the long s, which shared a common overhanging loop. Some letters required a back-facing ligature or kern: a consideration for the italic f and long-s
During the first 400 years of type production, while punches were cut by hand, different ligatures would be produced to meet the needs of different languages. In his account of the Fell types, Stanley Morison quotes a correspondence between Fell and Thomas Marshall from 1672 about ligatures in Dutch black letter types, in which Marshall says: “Van Dyck was willing to make f and long s”
and later : “neither have the Dutch any fh (long sh), or some double letters, which in english we cannot well misse”
The Fell Pica black letter (Fell English type 119) includes a ligatured ee,and ll ,while the Fell pica italic no.26 has has ll, as, is and us. The Small pica ital no 17 includes is ,us, ij ,ll . Morison notes the sh ligature as a later ‘intruder’ while as, is, us, ll seem to be accepted as standard for italics in this period; a legacy from Griffo
Over time, as professional consensus and user expectation become more consistent, we see the emergence of the ligatures fi fl ff ffi ffl as a ‘standard ’ provision. Before it fell into disuse use similar ligatures were necessitated by the long s which presents the same technical kerning problem. But why fi, fl, ff, ffi, ffl and not fb, ffb, fh, fj or fk? Surfboards may not have been widely known in the eighteenth century, but snuffboxes were.
Bringhurst says:
“Most of the early European fonts were designed primarily for setting latin, in which the sequences fb, fh fj and fk do not occur but the sequences ff,fi, fl, ffi and ffl are frequent. The same set of ligatures was once sufficient for English and these five ligatures are standard in traditional roman and italic fonts”
This seems a very broad generalisation for a period in which vernacular printing fast overtook Latin without any corresponding changes in ligature provision.
More problematic still are the linked ct and st now commonly known as ‘archaic ligatures’. Unlike the f ligatures they do not address a technical problem but refer more directly to conventions of written form. I’ll assume the problem is this: the handwritten letters s and c do not provide a basis for a linking outstroke without reversing the stroke to leave the overhang intact. Their survival is an anomaly that begs a further question: why only those letter pairs, when c and s occur in conjunction with so many other ascender letters?
The incidence of these letter pairs either in Latin or across European languages does not seem to explain their continued use up to the eighteenth century, and I’m forced to accept the rather unsatisfactory explanation of stasis and conformity: that the continued use of these sorts served to establish user expectation. The survival of archaic ligatures ct and st, into the 20th century along with the continued casting of the long s after it had ceased use, can be liked to the Caslon revival.
The advent of mechanical composition served to reinforce the concept of ‘standard’ ligatures, consolidating the status of the five f-ligatures while marginalising all others.
The Lanston Monotype keyboard shows the ligatures at the left edge, (a position that corresponds to their place in the California job-case for hand-set metal). The configuration of the Linotype keyboard had some variations over time and territory, and some keyboards offered only three or four of the five ‘standard’ ligatures.
These practical limitations in the Monotype and Linotype systems were in turn to inform the design of most new types. (The exceptions being those types designed exclusively for hand setting) The stylistic features that had necessitated ligatures, were moderated by the requirements of industrial typefounding. While the overhanging f of traditional old face types could be accommodated by kerning or ligatures, the matrix for machine composition encouraged a more radical approach. Line-casting systems such as Linotype and Intertype did not allow for the casting of kerned letters (f, j etc), and so letters had their extremities truncated to fit the procrustean bed of the matrix.
At this point therefore, a stylistic shift in letter design takes place in response to the conditions of new technology. Line-casting systems such as Linotype required a non-kerning f. Times New Roman, as designed for Monotype setting, has an f loop that necessitates only and fi and an fl ligature; it is restrained enough that the ff pair does not cause problems. Adapted for Linotype setting as Times, however, the free-standing f is abbreviated very severely so as not to require kerning, yet it was furnished with a full set of five standard ligatures, in which the f-loop is necessarily much more fully extended.
The Ludlow system did not use a keyboard but relied upon manual composition of matrices prior to the casting of a slug of type, As a consequence, supplementary ligatures could still be as easily accessed as ‘standard’ ones, but as a line-casting system it did not allow for kerning. This explains the character set and design of Ernst Detterer and Hunter Middleton’s Jenson revival Eusebius, in which we see a relatively generous provision of ligatures but a severely restrained overhang to the regular f.
In viewing a period in which both approaches co-exist, we can identify two contrasting philosophies of problem-solving. I’d suggest the idea of a ligature reflects a renaissance methodology rather than a modernist one; it is an additive, inventive solution. Recognising the integrity of the problem form, such as the overhanging loop of the f, it creates a further sort to substitute and address the problem.
The industrial approach is quite different: addressing the problem at its source by shortening the stroke, and where ligatures are unavoidable, reducing them to a standardised minimum. The introduction of non-kerning f, seen in the Ludlow Eusebius and the Linotype Times, is a characteristic example.
While some useful ligatures were lost during this period, the idea of an established ‘standard’ also prompted the provision of ligatures that seem unnecessary. In some explicitly twentieth-century faces, most notably sans serifs, ligatures can be redundant, illogical and intrusive.
As the non-kerning f gained currency as a requirement of mechanical casting, we see a peculiar divergence in new faces to which a short f is perfectly integral but which nevertheless retain a ligatured form that is increasingly divergent in style and structure. These are an odd anomaly; like the Cheshire cat’s smile, they remain as a surviving ‘solution’ to a problem that has already been removed.
Bringhurst describes them as ‘important’ to Gill Sans. I can’t agree. The face was originally released with a full ligature set; some re-cuts have only two (fi and fl), and they actually seem more intrusive than useful, an awkward conformity to an expectation that had already become superfluous. Other releases offered instead the option of kerning and non-kerning fs.
The original release of Futura adds ft, and also digraph ligatures ch and ck, to the ‘standard’ set. This seems to have been a widespread user expectation in Germany. Also, rather surprisingly for such a forward-looking design, it includes a long s, and the basic ligatures this required. This makes more sense when considered in relation to the esszett as a compound form derived from long and short s.
Jan van Krimpen’s unpublished Romulus Sans, though abandoned at an experimental stage, was designed with a full set including fb and fk. These seem very fully integrated, logical and coherent.
The technology of photosetting allowed for the close nesting of letter pairs into alignments that would have otherwise required special matrices. This might have been expected to prompt an extension of the ligature set, but the reductive influence of metal composition had already established norms and limits which were transferred into the new medium. Types adapted from machine-composed metal type had already suffered the standardisation of ligature set.
In this phase of technology, type design was ‘device-dependent’ and the limits of the character set were determined by the machinery used. If your keyboard doesn’t allow you to set an fb ligature, there’s not much incentive to design one.
The original capital ligatures of Herb Lubalin’s Avant Garde are a defining feature of a typeface designed for close headline setting. They were dropped from the digital form of the face, which then saw very widespread use in the early phases of desktop technology. This innovation, constrained by limits of computer memory, tended to reinforce the rationalising tendency we’ve already seen in photosetting; ligatures were limited, and those reductions of character set that had occurred in the transition from metal were retained or indeed taken further. Anyone still using digital fonts from before Open Type will find that original ligatures are missing or have been consigned to a separate font as an ‘expert’ feature. Jonathan Hoefler’s Requiem from the early 1990s offered over eighty italic ligatures but the constraints of font storage necessitated packaging these as a separate font and they could only be introduced manually.
As digital technology advanced and memory capacity increased, more sophisticated revivals and re-cuts prompted first the reinstating of traditional ligatures, then a variety of approaches to type design that developed the ligature into new directions
It was this point of historical confluence that prompted Hermann Zapf to revisit a long-abandoned project that became Zapfino. This calligraphic typeface makes ingenious use of ligatures. It also introduces a large number of variant forms for many letters and letter combinations, and anticipates the automated substitutions made possible by contextual alternates.
I’ve avoided discussion of calligraphic script typefaces, as an area where the concept of the ligature tends to blur or collapse completely, but Zapfino signalled possibilities that were to influence and inspire the use of extended ligature sets in more typographically-based designs.
Matthew Carter’s Sophia proposed a different strategy, using something we might call a ‘modular ligature’ system. It provides for a variety of ligatured letter pairs through the use of joining letters. In this we can see ligature design moving on from reference to a material ‘body ’ to prompt a set of digital variant forms. This idea is played out further in his Walker, where both serif and link stucture form part of a system of option variants, and Mantinia, which offered a range of ingenious capital ligatures and nested letter pairs. From this point the distinction between ‘standard ’ ligatures and others is no longer absolute; digital type design is ‘device independent’ and no longer constrained by the mechanical factors of a type production system.
The cap ligatures in Zuzana Licko’s Mrs Eaves propose a re-reading of history. Though this is ostensibly a Baskerville revival, it is most notable for its anomalies and inventions; fanciful features that have no historical precedent . Less biography than imaginative fiction, this is post-modern type design in magic-realist mode.
The increased flexibility of digital type opened up a range of reappraisals of the ligature, and of letter linkage in general. In Jeremy Tankard’s monoline semi-script face Aspect, the ligature set is so extended as to suggest an alternate alphabet in its own right.
Further, the 1990s saw the emergence of a ligature that is without significant precedent in the mainstream of Roman types, the ‘Th’ favoured by Robert Slimbach in the Adobe typefaces. This divides opinion, particularly when it occurs as standard in digital revivals of historic faces, as in Adobe Garamond, reflecting different typographers’ personal concerns for the opposing arguments of optical pragmatism or truth to history.
A retro-fit form like the Th can be seen as speculative design, proposing a fictitionalised reading of the past. If Caslon had access to today’s technology – if he had been designing for Adobe – might he have opted to make one?
Unlike many ligature ideas, it does not open up questions about further implementation; we’re short of words begining Tl, Tb, Tk.
This also brings into view two opposing considerations or philosophies: a calligrapher’s aesthetic, versus the type user’s expectation…
Through detailed revisiting of past practice and the experiment, designers have redefined basic expectations and set new standards., both in enhanced revivals and new designs.
John Hudson’s Constantia, designed for the Microsoft Cleartype project, expands the ligature set from 5 to 13, adding b, j, h and k. Jos Buivenga’s Calunna provides a nice example of contemporary approaches within a functional text face. Once-marginalised ligatures are reinstated, and in addition extended a-historical ligatures are introduced to remedy the inconsistencies of tradition in wholly new pairings. Kafka can take his surfboard to the fjord without any awkward collisions…
Our story so far traces a classic narrative arc:
Letter meets letter (in a fond embrace that lasts 400 year or so)
Letter loses letter (in the days of machine composition, photosetting and early digital).
Letter gets letter back…
so: can we expect them to live happily ever after?
The advances of Open Type technology might therefore be seen as the saviour of the ligature, but they also prefigure its eventual redundancy.
On the one hand, this technology has enabled the reintroduction of ligature forms that had passed out of use in the previous century or earlier, alongside the almost limitless scope to incorporate or invent more new ligatures, to extend the ligature set in line with contemporary usage.
But on the other hand, the scope and flexibility of this technology question the premise upon which the concept of the ligature depends. The problems out of which the concept arose, no longer have any material substance. New media have given us unparalleled solutions, but they are solutions to problems that no longer exist.
At the outset I suggested a definition of the ligature designed to encompass compound letter groups cast in metal upon the same sort. The closest equivalent we have is the glyph box, which has no material properties. The pairing of overhanging and touching letters no longer requires a single specific design to be cast. It may, as with Ed Interlock or Zapfino, be addressed in several different ways. The typeface designer is now working with full knowledge of the capabilities of contextual alternates, the near boundless range of ligatured forms and the dissolution of ‘definitive’ letters into a variety of variants.
A future based around responsive ‘intelligent’ type design is one in which we can expect the concept of the ligature to become superfluous.
In exploring the use of multiple ligatures, it’s clear that we have moved on from the ligature as a solution to a finite mechanical problem, to instead propose different ways of looking at letter groups and their relationships.
Faced with an infinite variation of alternate glyphs, the increasing mutability of previously ‘definitive’ forms, and the capacity to pair and combine multiple permutations and allographs, we can see the scope to repair the 500-year divorce between writing and type, assimilating the historic problems of the ligature into the fluid typography of the future.

©Will Hill 2016