Drawing Columba
A text typeface for three writing systems with italics, and weight and optical axes.
The typeface Columba began as my final project on the MATD 18/19 typeface design course in Reading. The font went on to win the Granshan Grand Prize for Non-Latin Scripts in 2019 and in 2020 was published by Colophon Foundry.
1. The plan
At the outset of the project, I set myself the challenge of designing a variable multi-script type family for use in complicated tight-set printing environments like newspapers, dictionaries and bibles. These kind of documents use footnotes, referencing and captions alongside standard sized reading text. The font would also need a micro size (now called ‘ruby’ — more on that later) as well as a text size. The ruby size would require not only column-line efficiency but also legibility at tiny sizes. So, the font family was devised along two axes — weight and optical size (italic and roman masters were kept separate) — and would cover three scripts — Latin, Greek and Cyrillic (Hebrew was originally in the family but didn’t make it to the finished files). The project was driven by the challenge of making three design spaces — optical size, transition of weight and different script styles — all work together on a printed page.
2. Inspiration
A feature of newspapers, dictionaries and bibles is the need for line-efficient columned text. For some historical inspiration I looked at the narrow letterforms of 16th-17th type-cutters like Hendrick van den Keere and Miklós Kis. Van den Keere’s 2-Line Double Pica was originally cut slightly narrow in order that it would be line-efficient and still readable to choral singers in poorly-lit churches. Stories like this were a big inspiration for Columba. I also looked at contemporary typefaces like Fred Smeijers Quadraat, Robert Slimbach’s Arno and Joshua Darden’s Freight as references for proportional widths of characters across multiple scripts and how to employ features like ink traps effectively. I also read and re-read Tim Ahrens thesis on optical sizing multiple times. The ruby sizes of Columba (and the inspiration for my subsequent MATD dissertation on designing bolds) would not have been possible without Ahrens brilliant research.
As I was drawing, I realised Columba was more of an old-style typeface than a transitional one — this wasn’t intentional but not unexpected given I had set out to make a printed text face, and many are widely used text types (Garamond for example) are old-style faces. I had unconsciously started to draw a style of font that I was seeing and reading everyday. Sometimes this kind of accidental inspiration is welcome. Once I realised, I was happy that Columba would be an old-style typeface.
Columba was designed for running printed text at a range of sizes and for all sorts of type hierarchies in books, magazines, newspapers and everything in between. I did also draw a display optical size, but due to time constraints the display hasn’t been released with the first version of Columba. It does however exist in part and may be released one day. What follows are some details about the design process and design decisions that went into some aspects of the font (but not all — I’m wouldn’t want to bore you to death).
3. Ruby (née micro)
Ruby was drawn alongside the text size so as these two optical sizes could work together. However, ruby follows a set of different metric rules and systems from the text size. Ruby was designed to work at 5–6pt, run in short paragraphs and/or list reference information — for miscellaneous tables, stock and shares reports, racing and sports results, etc. When set at the same size, ruby regular and text regular look quite different. But as the ruby shrinks, so does amount of visual information available to a reader. As the letters become smaller they become less elegant and more robust.
One of the last decisions I made was to drop the name ‘micro’ as the descriptor for the smallest optical size and use ruby instead. Ruby is a historical term referring to the smallest size of text used in newspapers (about 5.5pt, or 13 lines of text to the inch). As a term, micro is perfectly effective and easily understandable. However, I was aware that microprinting and micro-text has its own history on banknotes and printed cheques. What I had drawn for Columba was not this and I was keen to avoid overlapping terms. So I did some research and found ruby.
From the beginning, it was obvious that ink traps would be important in the ruby size. Across the Latin, Greek, Cyrillic (essentially) straight strokes intersect. Ink traps in this case are made with corner components composed of four points which expand in the ruby creating a square notch. Using a virtual master within the font file, ink traps do not ‘kick-in’ until the optical size is below 7 (on a scale of 5–10 indicative of the point size suitable for the optical size.
Ruby size characters with counters needed to be specifically enlarged. Following Tim Ahrens suggestions on optical sizing, space within the characters was divided more equally creating more vertical balance in the letterform. Where possible stems, strokes and apertures were squared up in order that printer rasterization did not over-ink angled strokes. Characters were also made wider and along with squarer counters more light was let into the letter form.
The ruby size heavy weights posed specific challenges. During print testing it became noticeable that the ruby bold weight would be too light when printed at 5pt if the same stem width as text bold were used. So I increased the stem width of the ruby bold then also applied heavier more block-like serifs. This distributed the weight more evenly around the ruby letter, reducing contrast while making the letter shapes more defined at 5pt. Then to counterbalance the added weight, more spacing around each ruby bold letter both helped with legibility and evened out the colour of the paragraph relative to the text size bold.
So, when text bold and ruby bold are compared at the same size, the ruby bold looks slightly darker but more spaced out than its optically larger counterpart. However, when compared at their intended-use sizes (10pt and 5pt approx. respectively) they look similar enough to be related.
It was also necessary to rotate the angle of stress in the Greek across both the optical and weight axis in order that weight be effectively added to the letter. Compared with Latin and Cyrillic, this seemed like a strange decision, but after talking with Gerry Leonidas I was assured that changing the stress like this would not disturb a Greek reader.
4. Relative proportions and line spacing
I wanted to find space-saving solutions for lines of text. I made modest changes to characters and their design, from first to final draft, with the specific aim of saving a few percent of line space across all three scripts — Latin, Greek and Cyrillic. The relative proportional width of characters became an important aspect of Columba, as I tried to figure out what was both space-saving but looked natural.
One trick was rotating the stress of curves and lightening-up counter shapes. The angle of stress on the e was initially too upright. Throughout the revision process I rotated the stress from about 10º to 30º, giving much more weight to the downward curving stroke. This was inspired by the e in Times New Roman as I realised that this angle of stress meant I could make the letter more compact while still opening up the eye of the e. The angle of stress is different to the c (20º) and very much different from the o (13º), but this discrepancy only seemed to add to the natural texture of the paragraph.
Another trick for saving line space was modifying serifs on angled strokes. The outer serifs on angled letters like v w y etc are shorter than the inside serifs. This aids spacing and kerning, and stop the letters appearing too characterful and thorny. As weight increases in serifed letters, the serifs become proportionally smaller. This is to stop them becoming abnormally large, allow weight to expand inside the letter and help with letter spacing. Lots of text typefaces do this, but I’d not noticed these things until I started looking for them.
I also redesigned the curves and their relationship to straight stems a few times. The breakthrough was a revision made to the G. The first drawing of the letter had a long thin stroke in the bottom right curve, as if to mirror form of the C. The revision was to remove this expanding stroke by shifting the angle of the letter (as with the lowercase e) so as to make the curve stroke in one motion. This also then mirrored the returning strokes in the b, d, p and q. Reducing the number of strokes in a letter not only tidied up the design but also meant the letterforms could be easily squeezed to be slightly more condensed and save line space. This change was useful for letters which traditionally allow angled axes of stress — predominant in Latin and Latin derived characters.
In order to aid spacing and rhythm in running text, lowercase and small cap letters with direct relation to characterful uppercase letters were made more upright and less quirky. The outer stems of the small caps m and n were made more upright while the Cyrillic м was also drawn perpendicular to the baseline. As these letters would most often be used within lines and paragraphs of text it seemed sensible to draw their stems in a similar pattern.
5. Italic
I began drawing the italic with two goals — firstly, I thought a more upright italic would be best suited to saving line space. Secondly, that the Latin and Cyrillic italic would bear some relation to a true-italic or more calligraphic form common in the Greek. For example, my first draft of an n reflected a true italic showing a continuity of stroke through a bulge on the baseline of the lower left stem. With some revision and work the first of these goals succeeded. The second did not for several reasons.
The bulging forms of the exiting thin strokes interrupted the rhythmic flow the letters, making a line of text seem unsettled and as if it were jumping around on the page. This was particularly noticeable in Latin and Cyrillic, but not so much in Greek where the rhythm of a line of text relied more on rounded closed-counter shapes. All three scripts however seemed to still work together. So, in Latin and Cyrillic letters like n, r, н, ю etc the bulges along the baseline (which pushed out to the left of the stems) were removed. However I kept the bulges along the x-height as they had less of an effect on the rhythm of reading, for example in the а, ц, ч, я etc.
Without the bulge on the baseline indicating a true italic form, the thin to thick connecting strokes were now free to be moved around. To add a bit more softness to the italic and connect it visually with the upright, bending the expanding strokes was key to make the italic feel at home with the upright. Every adjustment of the thin to thick adjoining strokes meant small adjustments to the width of the character in order that the rhythm of the line of text was maintained.
The Greek kept its bulges and its closer connection to a true italic written form. This kind of flowing form was also complimented by a set of Greek only ligatures which mimicked patterns of handwriting. To draw these, open and closing strokes had to feel naturally possible and not overly contrived.
7. Naming
Coming up with a name isn’t easy. A lot of the first ideas for names were too abstract, un-spellable, lacked fullness in their sound or didn’t have the punch to be memorable — Eliade, Deist, Liitus and others. But then I read a book called the Secret Pigeon Service by Gordon Corera. A history of Operation Columba and the use of pigeons by the British in WWII to ferry messages back and forth into occupied Europe. I’ve always liked pigeons and birds in general (if you are in Reading look up and you’ll see some beautiful big Red Kites gliding around the sky). After some further reading I also found out about St Columba and given my original brief had involved text setting in Bibles the Christian reference was more than appropriate. So, I settled on Columba as a name.
8. Bold
Unlike the regular and italic I considered the bold as less of style of script in-and-of-itself and more as a necessary aberration. It doesn’t do anything different or new, but exists solely in relation to the lighter, more commonly used and readable weights. It was fun to draw despite my initial sketches being too dark and/or condensed. After plenty of print testing I found the right weight. But in the process of comparing other typefaces regular and bold weights, I struck upon an idea which evolved into my MATD dissertation — Why and how to draw a bold.
In that paper, I wrote about the context and usage of bold as a secondary style for emphasis and distinction. Then devised a formal analysis of regular in relation to bold, what those design differences are (measured by rates of change — known as differentiation), then concluded with a write up of my findings and a set of guidelines aimed at other type designers looking to make a standard text bold. Instead of banging on about that here, for anyone interested in drawing bold, the dissertation can be found here.
8. Acknowledgements
Thanks to Gerry Leonidas, Victor Gaultney, Fred Smiejers, Ricardo Olocco, Michele Patanè, Frank Griesshammer, Toshi Omagari, Sahar Afshar & José Miguel Solé Bruning for their help. Not to mention all the other great type designers who stopped by the Reading type design department in 2019 who took the time to give feedback. Plus everyone on the course that year for all the beers, the chats and good suggestions. Also, thanks to Granshan for the Grand Prize for Non-Latin Scripts 2019 award and their recognition. And a big thanks to Colophon for taking a risk and publishing Columba.
