Signifier design information
Signifier is a Brutalist response to 17th century typefaces.
Signifier’s digital immateriality draws on a deeply material past. Acknowledging the processes and tools of digital form-making, I worked consciously with the computer to recast the lead, antimony, and tin of the 17th century Fell Types into ones and zeros. Signifier emerged from this alchemy with Bézier curves and sharp vectors determined by machine logic and a Brutalist ethos.¹
The true source
English Roman from the Fell Types. Stanley Morison, “The Roman Italic & Black Letter Bequeathed to The University of Oxford by Dr. John Fell”, Typophile Chap Book Nº 25, (1951): 10-11.
Signifier began as a revival of English Roman from the Fell Types. The Fell Types are a collection of metal fonts that currently reside at the Oxford University Press. Their storied history is fascinating. In 1668 John Fell installed printing presses at Oxford University. He wanted good fonts for the press, but there wasn’t anything suitable or available in England.² In the 1670s he sent Thomas Marshall to Europe to obtain type. Marshall returned with several different fonts, subsequently identified as cut by many greats of typography: Voskens, van Dijck, Garamont and Granjon. These fonts partially comprise what is now known as the Fell Types. For a long time the fonts remained “untouched and unnoticed”, until 1863 when they were re-cast because ‘Old style’ fonts were back in fashion. Seeing the actual printed forms is magnificent. One of these fonts caught my attention: English Roman.
“A Star Chamber decree of 1637, and the subsequent Printing Act of 1662, had limited the practice of letter-founding to four printers and the University presses of Oxford and Cambridge”. Ould & Thomas, “The Fell Revival”, (2000): 2. Caslon was the first competent cutter and caster of typefaces in England. His success virtually stopped the importation of European, particularly Dutch, type from about 1720.⤴
Signifier, version 1.
English Roman was cut some time in the 17th century. Stanley Morison identifies it as Dutch, but the punchcutter is unknown.³ The font is about 13.5pt by today’s measure. Its atmosphere and aesthetic are dark and assured on the page. Naturally I started to digitize it, hoping to capture these qualities. The first version of Signifier was a close copy, I sympathetically digitized the best version of each letterform. Because of the analogue origin I drew lots of curves, and only straight lines where I imagined the original punchcutter would want them.
A “punchcutter” is analogous to a “typeface designer” today. Punchcutters engraved a letter in steel. Each of their steel punches was then struck into a metal bar to create a matrix. The finished matrix was placed in a mould, which molten lead was poured into. The result was a piece of type. This was the “type founding” process. This is an excellent punchcutting video.⤴
55 English Roman lowercase a from a single page.
Once again, I came up against the age-old revival problem of “which is the true source?”⁴ One of the decisions revivalists face is which font to base their design on. Before digital fonts each point size was cut specifically to create necessary optical adjustments in letter shapes, spacing and proportions. A typeface might be made up of several sizes of regular, for example. Even if a letterform is exactly the same shape, printing ink on paper creates variation.
The numerous variations provide options, but only within the parameters of the original metal shape. Often the original letterforms in English Roman weren’t quite right, so I borrowed from other Fell Types like Great Primer Roman and Double Pica Roman. I realised I was starting to Frankenstein.⁵
Great Primer Roman.
Great Primer Roman.
Great Primer Roman.
Double Pica Roman.
Double Pica Roman.
Double Pica Roman.
Many modern type designers frame a revival by imagining what the original punchcutter would have done with digital tools. It’s a workable proposition if you’re interested in performing an act of contemporary design fiction. As soon as I noticed what I was doing, I fell into a philosophical rabbit-hole trying to understand the meaning of revivals. What is authentic? What is the source? What is the intent? My previous typefaces have clear remits and intent. Feijoa replicates letterpress by using curves where possible; Pitch captures the aesthetic of typewritten text; and Söhne the analogue materiality of Akzidenz-Grotesk.
I didn’t want Signifier to be a revival of English Roman.⁶ Instead of emulating the stylistic quirks of the original, I interrogated the essential nature of digital fonts. What makes a font truly digital? What is an honest approach to working with Bézier curves? What is the essence of English Roman’s letterforms, and how can I translate them to the 21st century?
Thinking about the materiality of digital fonts lead to Brutalism. According to architectural historian and critic Michael Abrahamson, “the word ‘Brutalism’ has lost its meaning. At present, it equates to: large buildings, sometimes of concrete, constructed sometime between World War II and the end of the 1970s.”⁷ Abrahamson clarifies and re-orientates the meaning, quoting Peter Smithson:
“Revive” is a curious word for typeface design. It’s always sounded hubristic to me, like the designer is somehow channeling divinity, resurrecting the dead. It’s natural to anthropomorphise the process but typefaces don’t actually die, so it’s impossible to bring it back to life. I suppose “reviving” sounds better than “re-drawing someone else’s work”.⤴
Michael Abrahamson, “Brutalism: The Word Itself and What We Mean When We Say It”, Critic Under the Influence, (November 2011). Retrieved 20 February 2020.⤴
Brutalism is not concerned with the material as such but rather the quality of the material, that is with the question: what can it do? And by analogy: there is a way of handling gold in Brutalist manner and it does not mean rough and cheap, it means: what is its raw quality?
Signifier version 1.
Signifier final version.
My question now became, “what is the raw quality of digital fonts?” Digital fonts are strings of co-ordinates rasterized as images on screens and surfaces. Bézier curves are visualised equations. When we manipulate Bézier handles, we change the parameters of the equation. I stopped working against what I was seeing and started to work with the tools and screen. I deleted details, replaced curves with lines and snapped to the grid. Consciously making things consistent, I practiced a sort of typographic quantization.⁸
Quantization is the rounding of analogue signals to digital values, like turning soundwaves to data or images to pixels. It’s a good description for the transition of fonts from analogue to digital media. What once was infinitely variable is now infinitely repeatable. Each serif, letterform, letterspace can be simplified and regularised. It feels like the logical and natural thing to do — thousands of digital typefaces display these qualities. Why would you draw each serif slightly differently when they could (should?) all be the same? Why would you try to emulate the irregularities of past analogue methods unless you’re trying to be deliberately anachronistic?
After re-working my original drawing I compared the two versions. At large sizes the differences in detail were obvious. But at small sizes, on both screen and laser print, their gross characteristics and feel were almost indistinguishable. Signifier’s true nature finally snapped into focus: I had successfully translated ancient analogue printed forms to the high-resolution pixel grid. It felt contemporary.
Interfaces and pixel grids
Computer interfaces were once constrained by the underlying physical pixel grid of the screen. Windows, buttons and icons adhered to the coarse raster. As screen technology matured the underlying grid became finer and the colour gamut expanded. Many interface styles arrived, like Aero for Windows and Aqua for OSX. These eventually fell out of favour, now considered kitsch and ornamental.⁹ ‘Flat’ interfaces are currently de rigueur, codified in aspirational guides like Google’s Material Design and Apple’s Human Interface Guidlines.¹⁰
The development of digital fonts took a similar path. Once bound by the coarse raster, letterforms are now freed by a finer grain. They will never have the materiality of inked metal pressed into paper, but digital’s flexibility makes up for this perceived deficit. Two early digtal fonts influenced Signifier: Zuzana Licko’s Matrix and Matthew Carter’s Charter. They both dealt with the screen and memory constraints of the nascent digital medium.
Matrix flourished from the pixel grid. Released in 1986, it was Emigre’s first PostScript typeface. Based upon Licko’s earlier bitmap font, Emigre 14, its outlines were simplified to reduce the amount of point data within the font files, thus saving precious memory. In the Matrix II specimen Emigre writes, “as well as respecting the nuances of our traditional typefaces and their evolution in conjunction with reading habits, the forms of computer fonts must result as an integral part of the digital process, not in spite of it.”
Matrix’s distinctive triangular serifs use one point less than a square serif, and the 45° angle was the smoothest diagonal early digital printers could render. For the first time serifs are — mathematically — infinitely sharp. One could zoom in forever and they’d never blunten. In general this is something that is not done with modern digital fonts. Sharp details usually have small flat sections to take the ‘digital’ edge off, to make it feel more ‘analogue’. This was a piece of unchallenged ‘received wisdom’ that influenced how I drew fonts for a long time.
“Fig. 8. Comparison of the contour points of r from Times, Helvetica and Charter. The required storage for Times is approximately twice as large than Helvetica and Charter.” Typographische Monatsblätter 5, (1989).
Charter’s distinctive terminals. Typographische Monatsblätter 5, (1989).
Released in 1987, Charter attempted similar strategies of outline data optimisation for memory efficiency. Ironically, Carter tells a story about engineers solving the memory problem before the font was finished, making his design solution ‘redundant’.¹¹ Charter’s main legacy is a ‘chopped’ ball terminal. This terminal has become a kind of de-facto digital signature, emulated across many digital typefaces since.
Memory efficiency is no longer an issue for fonts, but Matrix and Charter both feel fresh today. Licko and Carter are masters of their craft. Skilfully working within the contraints of the medium, they distilled traditional forms into modern fonts. Their pioneering digital efforts continue to provide raw material and philosophy for contemporary designers.
Domaine Display Regular.
Heldane Display Regular.
Tiempos Fine Light.
Matrix allowed the infinite sharpness I’d previously avoided.¹² Making sharp points in a font seems insignificant, but it was ingrained, I’d never before questioned it. Now this golden rule was unnecessary, even redundant. Signifier’s sharpness is almost imaginary at small sizes. Scaling and rasterisation slightly softens the points, even with offset printing — the gold standard for font reproduction.
Domaine Display (above) & Text (below).
Epicene Display (above) & Text (below).
Tiempos Fine (above) & Text (below).
Signifier’s sharpness has the unexpected bonus of behaving like a set of optical masters. Many digital serif fonts have distinct “text” and “display” versions. The details in text versions are robust to handle small sizes, display versions have fine detail appropriate for large sizes. Signifier’s details work at all sizes, it doesn’t need text and display variants. The qualities of both are built into the same letterforms.
1200 dpi laser print. Uncoated stock, 10pt.
Screenshot, MacOS 10.15.6, 50px.
Offset print. Uncoated stock, 10pt.
Offset print. Coated stock, 10pt.
The four example n above demonstrate that all font reproduction methods rasterize. The sharpness in Signifier (and all vector-based fonts) is theoretical. Only math is infinite, the physical world has limits. When a letterform’s detail is magnified, it can be imagined at small sizes even though, perhaps, it cannot be actually perceived.
From Charter I took the iconic chopped terminals and applied them where appropriate throughout Signifier. Terminals on letters like c f j r y are very hard to get right. Garamond-era fonts don’t offer a lot of guidance, their terminals are organic and blobby. Small sizes and rough printing further obfuscate their detail. Translating their smooth continous curves into digital is hard. I always feel like I’m fighting Bézier curves trying to shape these terminals. That’s why Carter’s solution is genius — the combination of curve + cut is elegant and easy to draw digitally. It feels like working with the grain.
A typeface is a collection of fonts, fonts are made of letters, letters are built from Bézier curves.¹³ Every detail in a single letterform should bind harmoniously to the concept and aesthetic of the entire typeface. Signifier’s language of details like rectangular baseline serifs, triangular head serifs and sharp terminals are bound by the connective tissue of taut curves. They work with each other to create a vivid, sharp impression in the romans. The next challenge was translating this aesthetic language to a related italic.
Translating to italic
Pica Italic from the Fell Types. Stanley Morison, “The Roman Italic & Black Letter Bequeathed to The University of Oxford by Dr. John Fell”, Typophile Chap Book Nº 25, (1951).
Signifier’s italics return to the Fell Types. Granjon’s Pica Italic is the raw material, processed with the same digital rationalism of the romans. Before the 1800s italics tended to be narrow and lively. They differentiated themselves from romans by slope and texture. Derived largely from hand-written forms, italics from the Fell Types era are dynamic and full of detailed flourishes. The near-constant slope binds them together in harmony, but sometimes they can feel too fussy.
As letters get narrower, curves tighten and are harder to draw continuously. The solution for Signifier italics contrasts smooth outer curves with sharp internal corners. This alleviates pressure from tight internal curves, sharp corners keep the forms narrow and the impression crisp. In more extreme cases, straight lines replaces curves where possible. The interesting result is the stems still feel curved at small sizes, like £.
The sharp internal corner comes from one of the fundamental letters: n. The lowercase n is very important for me as it establishes the aesthetic, proportion, details and curve quality for the majority of the character set. Because a typeface is a system of interrelated details and parts, a small change to a serif, for example, has wide-ranging implications across the entire family.
Signifier Italic, v.1
Signifier Italic, v.2
Signifier Italic, v.3
Signifier Italic, v.4
Signifier Italic, v.5
Signifier Italic, v.6
This exploration of Signifier’s Regular Italic n shows my progression of thinking, weighing up decisions and details. If the stem is round (v.1) what other glyphs will need to be rounded? Is a smooth (v.2) or abrupt (v.5) stem-to-curve transition more “digital”? If the stems are flat but angled (v.3) will the baseline become jarring? Does a triangular head serif (v.4) diminish the italic-ness? Will a flat exit stroke (v.6) align nicely with the entry stroke or look dull?
All of these instances of n could make a serviceable italic. My personal preference, in isolation, is v.1. I turned it into an entire lowercase, but abandonded it in deference to v.5 — a stronger concept with better systematic application. Some ideas survived the process, like the flat exit stroke. As the weight interpolates from Thin to Black, it moves from 38° to 0°. This makes the Black Italic exit stroke shorter, allowing better spacing between characters.
Traces of the tools
Interpolation has been a vital process of digital typeface design since Ikarus.¹⁴ It’s essentially blending one shape with another. With typeface design for example, interpolation facilitates blending from a light to a bold. The typeface designer manually draws the light and bold fonts, and interpolates intermediate font weights like regular, medium and semibold. Intepolation is integral to modern typeface production: nobody draws every single weight manually.
Interpolation is powerful, but it’s not perfect. It gives the best results in between masters (M). These are called instances (I), and usually end up as the named weights in a typeface family. When it’s pushed outside of the masters, it’s called extrapolation (E). Interpolation works between known quantities, the masters. Extrapolation takes the trajectory of the masters into the unknown and outlines become distorted and broken. Despite this, extrapolation is very useful during the design process because it’s easy to experiment with.
Fig 30, 31. Visible Language Vol. 19.
Fig 33. Visible Language Vol. 19.
Matthew Carter’s Galliard was “the first original type design to make substantial use of a computer in its creation”. In 1977 he experimented with extrapolation during Galliard’s design. He writes, “the italics were all drawn by hand, although we did do trials on the Ikarus — pushed to deliberately absurd extremes in the case of Figure 33.”¹⁵ The outlines are wonky, but give a tantalising glimpse of a possible thin Galliard weight.
Signifier Thin, raw extrapolation.
Signifier Thin, reworked and finalised.
Signifier Thin Italic, raw extrapolation.
Signifier Thin Italic, reworked and finalised.
Signifier’s thin weights are derived from extrapolation. Bethany Heck was an early Signifier tester and suggested a weight lighter than Regular.¹⁶ I was reluctant because I’m not a fan of thin serifs. But she’s one of the best typographers working today, so I gave it a stab. My first move was a quick and dirty extrapolation in Superpolator — and the results were a revelation.¹⁷ Extrapolation’s caricature and distortion is either cleaned up or abandoned with digital work, but Signifier embraced it. The effects of the extrapolation algorithm are exposed and synthesised into workable letterforms.
Using the production process to define the aesthetic — to leave a trace of the tool — is a perfect extension of the digital concept. I have previously argued that a typeface is not a tool because, “a tool is part of the process of making; it’s not visible or part of the end product. You can see vestigial traces of its use: the sliced vegetable, the smooth wood, the hemmed skirt.” Signifier is not a tool, but Thin, Extralight and Light show traces of the digital tools used during production.
The end product of digital font design holds no secrets. Crack one open in a font editor or ‘convert to outlines’ and you’ll see the vectors. This is a great leveller and teaching aid. But it’s subtly elusive — letterforms rarely reveal why they were made. And they never reveal the time, practice and craft for the placement of nodes and points necessary to create beautiful, harmonious letterforms.
Brutalism wasn’t a specific material or style, it was an attempt to be true to the raw qualities of materials. It was an ethic.¹⁸ Signifier adheres to this ethic, Brutalism’s core concepts framed my working process and thinking rather than pre-determining the outcome. There’s a sense of the vector, the grid, the underlying digital nature. For instance, you can’t see that Adobe Garamond’s a is digital, but you can see it in Signifier’s a.
“The difference is not merely one of form of words: ‘Neo-Brutalist’ is a stylistic label, like Neo-Classic or Neo-Gothic, whereas ‘The New Brutalism’ is, in the Brutalist phrase “an ethic, not an aesthetic. It describes a programme or an attitude to architecture.” Reyner Banham, “The New Brutalism”, (1966): 10.⤴
Around the time I started designing Signifier, mum was diagnosed with cancer. In mid 2019 she accepted palliative care. During her final weeks my family were privileged to sit with her as she died. I finished Signifier next to her death bed. A few days later, on the morning of August 31, mum died. That evening, at her earlier insistence and with the family’s blessing, I flew to Auckland to accept a 2019 Arts Laureate from the New Zealand Arts Foundation.
In the following weeks and months I read books to help cope with my grief.¹⁹ The best ones were by Buddhist practitioners, gently advocating acceptance of impermanence. I finally realised that searching for the essential materiality of digital fonts was misguided. Bézier curves aren’t anything physical, their “raw quality” isn’t material at all. Their essential nature, like all things digital, is immaterial.
Fonts once existed as physical things used to make more physical things. Even if they were melted down to make bullets or forgotten like the Fell Types, they still existed materially. Digital fonts only exist fleetingly. They are experienced, mediated by a screen. Once the power is switched off they cease to exist. A craft history with five centuries of physical output replaced by virtual output takes some reconciling. We have retained our sense of line, spacing and form. We have lost the physical, material touch, as Ruskin once railed against.²⁰ What we have gained is speed, flexibility and reach. Raymond Gid elegantly expresses the transition when he wrote, “Letter, you are to be married to Light. That is the wedding that we shall celebrate to-morrow”.²¹
What is the nature of craft when the thing being crafted does not physically exist? Richard Sennett offers an answer:²²
Rather than get lost in this philosophical forest, it might be better to focus on what makes an object interesting. This is the craftsman’s proper concious domain; all his or her efforts to do good quality work depend on curiosity about the material at hand.
We don’t have material anymore, simply forms mediated by screens. Our input is mediated, our output is mediated. We shape letters from numbers and draw curves with equations. Letters are no longer things, but pictures of things transmitted by light.
Form is the void and void is the form.