Even, mathematically precise spacing can still look wrong to the eye, and understanding why reveals one of typography's most useful lessons.
Designers who lean on evenly measured spacing often end up with type that looks clumsy, even though every value checks out on the grid. The reason comes down to a simple mismatch: the eye judges spacing by perceived space between letters, not by ruler distance. Understanding that difference changes how you think about type from the ground up.
Two different definitions of "even spacing"
Metric spacing sets identical or formula-based distances between character bounding boxes. It's easy to automate, but it treats every letter shape the same. Optical spacing, on the other hand, aims for equal perceived white space between letterforms, which means the actual distance has to change depending on each glyph's shape. A round "O" next to a straight "H" needs a smaller physical gap than an "H" next to another "H" to look equally spaced. The eye estimates the visual space around a letter, not the measured gap, and that's where mathematically even spacing starts to fall apart.
Why straight, round, and diagonal letterforms break uniform grids
Straight-sided letters like H, I, L, E, and F create a hard vertical edge, so an identical numeric gap next to them reads as tighter than the same gap next to a curved letter. Round letters like O, C, G, and Q curve away from the body of the letter, leaving extra visual space even at the same sidebearing value, so they usually need tighter fitting to look balanced. Diagonal letters such as A, V, W, Y, and X create gaps that shift depending on their pairing, which is exactly why kerning tables exist. Spacing handles the general rhythm; kerning cleans up the specific pairs that still look off. This straight, round, and diagonal framework shows up consistently in type design references, including Karsten Luecke's writing on spacing and kerning.
Spacing vs. kerning: they solve different problems
Spacing, meaning the sidebearings set for every glyph, establishes a typeface's overall rhythm and should make most combinations look acceptable on its own. Kerning is a secondary layer, meant to fix specific pairs that still look wrong after spacing is set, like "AV," "To," "LT," or "Yo." Type design documentation from tools like Glyphs and FontLab describes a clear workflow: space the full character set first for optical evenness, then kern only the exceptions. Leaning too hard on kerning to patch bad spacing is a known trap, because it multiplies the number of exceptions instead of fixing the root problem.
Why grid-perfect layout software can still produce bad type
Design tools can apply uniform tracking across an entire string, but that's a blunt adjustment that ignores the shapes of adjacent letters. Auto-kerning features estimate optical spacing algorithmically, but foundry-set kerning, built directly into the font by its designer, tends to be more reliable because it reflects deliberate visual judgment about specific pairs. That's why the same tracking value can look fine in one word and awkward in another. The letterforms change, but the numeric adjustment doesn't.
How type designers actually judge optical evenness
The real test isn't measurement, it's the eye. Type designers use representative word and letter strings, often nonsense combinations mixing straight, round, and diagonal letters, viewed at actual size. Writers like Frank Grießhammer describe this as an iterative process: squinting, stepping back, testing in context. Sidebearing values are typically set by eye first, then checked numerically afterward, which reverses the instinct most designers start with.
The takeaway for designers
Good type isn't built from a formula, it's built from perception. Next time a headline or logotype looks slightly off despite "correct" spacing, trust your eye over the grid. That instinct, more than any measurement, is what separates readable type from type that simply measures well.
