For an English version of this article – click here
“How do you design a typeface?” – it’s a question that occupies me quite a bit, and one I am asked fairly often. Unfortunately, or fortunately, I don’t have a simple answer, and looking back at letters I have designed, there is no single formula either. Instead, I suggest taking a close look at a type design project I did between November 2015 and March 2016: Assistant – a Hebrew version of Source Sans Pro for the Google Fonts library.
The brief was simple: design a Hebrew typeface that matches the existing Latin. The aim was to understand the character of the typeface, not to copy and paste the Latin shapes and build Hebrew ones out of them. A bit like translating a text – you need to understand the meaning of the sentence, not the individual words.
The Latin version, designed by Paul Hunt at Adobe, is a modern sans serif meant as a text face for screens. The family has six weights, from thin to black. It is a typeface that aims to be transparent and serve as a default font. It is carefully designed, and its forms are built from disciplined, geometric calligraphic strokes with a hint of movement.
Hebrew
In general, the model that influenced me as a type designer and helped shape how I see the Hebrew letter is the Aleppo Codex. The structure of the letters, and the rhythm of the script, hold triangles in the inner spaces and in the roofs (as shown in the picture). Although I did not base Assistant directly on the Aleppo Codex, I felt I was designing in its spirit.
Under the hood
I design typefaces in an app called Glyphs. The design process is divided into masters, each master representing an extreme of the typeface: the thinnest or the boldest, the narrowest or the widest. Working with masters makes it possible to interpolate between two of them – that is, to generate intermediate weights.
I draw the thin weight and the bold weight, and ask the software to generate however many weights I need between them. For this to work properly, the letter must be built correctly in both masters – the points and curves have to match at the extremes of the typeface (more on interpolation in the post on variable fonts).
For Source Sans I drew two masters: one for the lightest weight and one for the boldest, and then added four more intermediate weights by interpolation. I usually prefer to start with the light weight, where it is easier to check the basic shape of the typeface and its color in a block of text.
Getting started – a basic Hebrew letter
I started with a simple letter, ד, built from a straight meeting of stem and roof. First I decided on the proportions between the two – in Hebrew sans serifs the stems are usually designed thicker than the roofs – and so it is in Assistant.
The proportions of the Latin Source Sans are not classical: the letter is relatively tall and narrow, and so the Hebrew too, usually a square letter, got the proportions of a narrow upright rectangle. For the two scripts to live in harmony, the height of the Hebrew letter (the height of ם) is slightly above the Latin x-height, so that the roof of the Hebrew letter rises a little above the top of the x.
When I design a type family I think of letters according to the kind and direction of the movement that creates them, so that the alphabet in effect rearranges itself: אצץ דז גנ יון רהחתץף בכפמם טסשעלק. This is a good way to examine a typeface, but it is not a rule one must stick to: in Assistant, for example, the pair “לק” was split – the ל joined the “אצ” group and the ק joined the “טסעש” group.
I knew I wanted to bring triangular shapes into the typeface. At first I chose to “neutralize” the most fun letter to design – the א – and drew it and the צ as strokes of straight lines that create sharp triangles in the letters’ inner spaces, inspired by the Aleppo Codex of course.
The Latin is rounder than the Hebrew, and Source Sans has a “double-story” g – built from two circles, the upper a perfect round and the lower an ellipse. In Hebrew sans serif we don’t have that richness of form, and because I wanted to match the Latin I had to bring round shapes into the Hebrew, which is square by nature.
This led me to the “טסעשק” group, where I decided to draw an arc that joins a geometric stroke at an angle (to make a triangle). This is where the solution for ט comes from: I didn’t want to break the arc flat and give the ט a straight roof, so I continued the stroke from the base of the letter upwards and then cut it at a calligraphic angle. I repeated the same movement in ס, ש, ע, ק and ג. This brought a lot of round inner white space and consistent movements into the typeface, which help create an even rhythm.
I added a rise in the roofs of the “גנ” group: a calligraphic movement that climbs above the straight roofs and echoes the rising movement in ט.
The מ, on the other hand, got as round a shape as possible: an open letter that brings in a lot of roundness, especially next to the sharp “יוון” group.
Working on details
When I thought about the design of the typeface I imagined the tool that drew it: the stroke of a calligraphic nib cut by straight geometry. I looked for a way to embed the feeling of a calligraphic stroke in a simple geometric letter like ו. I added a stroke at the head of the letter, and now I could write “יוון”:
From macro to micro
After drawing all the letters as a quick sketch, I set some text to see how the typeface works at different sizes, how the optics behave, and what feeling the letter conveys.
A type family needs to be harmonious within itself – if one letter is too narrow or too wide it will jump out of the block of text and disturb reading. Much of the work is spacing each letter in the most correct way, so that the right white spaces form with the letter before and the one after, and in the end the text has an even rhythm and a pleasant texture.
Some letters call for kerning (pair spacing) – adjusting the space where two letters meet. For example, the meeting of ד and ב needs negative kerning so that the space between them is not too large. The blue area represents the negative spacing between the letters:
Even in the first sketch you can see Assistant’s character. The design of the letter is raw and not all the forms are right, but the statement of intent is clear.
Drawing weights
Having improved the first master, the thin weight, I move on to draw the second master, the bold weight. What would happen if the typeface ate a lot and put on weight?
To match the bold weight of the Hebrew to the bold weight of the Latin, I had to give the Hebrew the same text color and reach an optically correct result – one that looks right to the eye even though it is not necessarily right geometrically. I increased the height of the letter by a few percent relative to the Latin and used blacker proportions of stems to roofs, something I did not do in the light weight. Broadly, these are the two main parameters you can change to reach a text color that feels matched to the Latin, and you have to play with these proportions until they are precise.
I duplicated the thin master and drew the bold on top of it in wider, blacker proportions. Moving to a bold letterform, the curved meeting of stem and roof changes: in the thin weight the arc is wide inside and out, and in the bold it is short on the inside and wide on the outside, which helps bring white space into the letter and define its shape.
In drawing the bold weight I kept all the features of the thin weight – I cut the letters at the same angle (the leg of the ג, the trunk of the פ). In ש, to keep the narrow proportion of the typeface, I slimmed the middle stem. I let more white space into the meetings of the strokes: where the leg of the ג meets the stem, for example, or where the arm of the מ meets the arc.
All of these are done for optical reasons and have to do with how we see, and how details read when they are enlarged or reduced. Without optical corrections, letters can look clogged or too black at small sizes.
Legibility tests
After finishing the design of the typeface in all its weights I put it through a whole range of legibility tests. Assistant is meant, as mentioned, to be a text face for screens, so the tests were done accordingly. I checked how the typeface behaves at different sizes and formats, refined it and prepared it for use: I spaced the regular weight, for example, to suit its most common use, which is reading running text on the web.
Each weight in the typeface gets different spacing: the light weights, used for running text at small sizes, are spaced openly. The bold weights are spaced more tightly – they are meant for headlines, where there is less need for space between the letters.
Summary
Designing a typeface takes time, like a puzzle with a great many pieces. It helps me to think of the design process as sculpture – you have to sculpt the letter in your head and “carve it in stone” until it appears. Carving is an action that takes time, and there are no shortcuts.
The design process took about four months. It was not daily work, and part of the process was a group conversation held in the Google Fonts type design group. I was also helped by designer friends for critique (and this is the place to thank Meir Sadan, Yotam Hadar, Amir Avraham, Oded Ezer, Omer Ziv and Dave Crossland).
Assistant can be downloaded from Google Fonts for free.
Assistant in use
