Typeface Mechanics: 001

DEN HÄR ARTIKELN KOMMER FRÅN Frere-Jones »Square and round shapes: mathematically equal, or optically equal

Our conscious minds want to draw one shape, but our eyes need to see another. Part of typeface design is managing this eternal friction between logic and optics. It’s always there, no matter the style.

This new series of posts will explore what I call “typeface mechanics”, the behind-the-scenes work that makes typefaces visually functional. It is what placates the stubborn oddities of human perception, helps or hinders the user, and informs long-standing conventions of design.

The typeface design process has many counterintuitive moments. One of the earliest pertains to vertical position and size, which we expect to be consistent among letters. We could simply pick a measure and apply it everywhere. But this straightforward and logical plan would fail, thanks to our eyes and brains.

Square shapes like H have a simple and stable relationship to the baseline and cap height. Their upper and lower edges coincide with these boundaries and stay put. But only a narrow sliver of an O is the full height, and the rest of the shape falls away. The parts that are too short greatly outnumber the parts that are big enough, so we conclude — wrongly, but very reliably — that the round shape is too small.


If the “correct” height appears inadequate, “too much” will look right. So the O is made taller and deeper than the H, even if the most stringent mathematical reasoning would declare it incorrect. But we read with our eyes, not with rulers, so the eye should win every time. Typefaces from any period will demonstrate this compensation, often called “overshoot”.

“Romaines Droites” by Fonderie Turlot, 1880 specimen
“Overshoot”, extending round shapes beyond flat shapes

If curves need overshoot because they don’t behave like squares, pointed shapes are even less like squares, and accordingly get more overshoot. But with enough care, readers won’t notice the multiple sizes and positions. All of it will feel equal.

Futura Medium by Paul Renner, Bauer 1928
Progressive overshoots for round and pointed shapes

But like many aspects of type mechanics, overshoot comes in degrees. Slow curves behave more like a flat shape, lingering near the baseline longer than a rounder, sharper curve would. Below are two typefaces from the same designer and foundry, at the same point size and roughly the same proportion. The difference in overshoot is driven by the difference in the curvature: one is fully round (trending towards a diamond, even) while the other is emphatically square.

Recta Nera Stretta (above) and Metropol Nera Compatta (below). Both by Aldo Novarese, Nebiolo ca 1960 and 1967
Diminishing overshoot

Even within a single family, variations of weight and width can alter a shape’s exterior, and require a recalibration of alignment. For instance, a V might present a sharp exterior point in lighter weights. But accommodating the extra mass of heavier weights could force a blunter apex and change its apparent depth, and therefore require a new alignment.

Overshoot decreasing as bluntness increases: FB Nobel Light, Book and Regular by S H de Roos, Amsterdam 1928. Revival for Font Bureau 1992

Lowercase alignments are often more difficult to negotiate, because many letters have flat shapes immediately adjacent to round shapes, and both need to be optically correct. A mixed alignment should look “natural”, and show no sign of the struggle involved.

Helvetica Medium by Eduard Hoffmann, Stempel 1956
Mixed alignment in lowercase

Sloping serifs can make matters harder again. This feature, a hallmark of an oldstyle lowercase, leaves only a handful of letters with a flat shape on top. In this genre, the literal height of the lowercase appears only once in a while, and the majority of letters — including many of the most common ones — are a mix of fickle curves and angles.

Caslon No. 540, American Type Founders 1906

The lowercase x is rare case, in that its exterior doesn’t change much between one style and another. With thick serifs, thin serifs or no serifs, high weight contrast or low, vertical or titled axis, the x will very likely have a totally flat top and bottom. With such a reliable benchmark, we use “x-height” to refer to the central zone of the lowercase (W. A. Dwiggins sometimes used the term “z-height”, but I haven’t found evidence of anyone else using that convention.)

Orpheo (unfinished)

Serifs mark the top and bottom boundaries, but not always precisely. Designs with a calligraphic derivation or some other kind of organic quality may have concave serifs, so these numerous “flat” shapes are not actually flat. The conflict is unusually localized, with the center and ends of a serif suggesting different alignments. To broker a visual peace, these cupped serifs often straddle the baseline, rising and falling equally.

Establishing an optically consistent size is just one puzzle among very many. The entanglements of weight, width, and spacing happen at the same time, as do the predicaments inside each letter.

But those come next.