Overshoot
The small amount a round or pointed letter is drawn past the line it sits on, so that it looks the same height as a flat one.
Near it
Baseline
Cap height
x-height
Optical correction
Optical alignmentWhat it is made of
| measured on this shelf | the CAPITAL O's top above the H's, over the cap height, across 23 families, from Drew Unicase at -0.2314 to Drew Poster at 0.0886, with eight families at 0.0186. THE LETTER IS NAMED BECAUSE THE ANSWER DEPENDS ON IT: the lowercase o against the x is a different axis, and on this shelf it is measuring the x more than the o (the last row) | measured over all 23 house families on the fixed shelf, 2026-09-24, and confirmed in rendered ink: Drew Grotesk's O stands 13px above its H at a cap of 700px, 0.0186, the outline's figure exactly. The quiet and loud groups this record carried until then (Grotesk at 0.1000 among them) were measured before the baseline fix, when the H's own foot and top were where the ink happened to land |
| the drawn correction, sixteen families | Drew Engrave 0.0014, Terrace 0.0086; Condensed, Extended, Grotesk, Mono, Monoline, Outline, Slab and Square 0.0186; Press 0.0371, Gazette 0.0414, Compact 0.0429, Micro 0.0457, Round 0.0586, Poster 0.0886 | one to nine per cent of the cap: the round letter drawn past the flat one so that it does not look short, and at the top of the range, Poster and Round, a heavy round letter needing more of it |
| six that read short, and why | Drew Perforate -0.0591, Stencil -0.0541, Extrude -0.0486, Hairline -0.0143, Pixel -0.0125, Display -0.0071 | not O's drawn short. Perforate and Stencil cut their strokes into pieces with round ends, so the H's top runs half a stroke past the cap line; Extrude's H has its depth; Pixel's grid rounds the cap up to 722. The O is measured against an H that is itself taller than the line |
| and one that goes the other way by design | Drew Unicase at -0.2314 | Unicase's O is a unicase form drawn to the lowercase, so the negative number is the face working as designed rather than a fault |
| the lowercase axis measures the x | the o against the x runs from Drew Terrace at -0.0981 to 0.0248 in eight families, and eleven families read BELOW zero | which says the x's top stands above its own line in those eleven, not that their o is short: the flat cut on a shallow diagonal is taken only at the baseline, so the x rises past the line its u sits on. Found 2026-09-24 re-measuring this record and owed to the foundry; until it is fixed this row cannot report an o overshoot for those faces |
Where it came from
An optical correction, not an error. A circle touching a line at a single point reads smaller than a square resting on it along a full edge, so the o is drawn taller than the x at both ends.