A Typography Sampler

This piece is a little unusual for me. Most essays here pick one idea and follow it in a straight line. This one is closer to a cabinet of curiosities — a working demonstration of everything a paragraph of prose can be asked to carry, strung together loosely around a single claim: that the shape of a text on the page is not decoration layered on top of meaning, it is part of the meaning¹. A block quotation reads as testimony. A footnote reads as a hedge, or a confession, or a joke the main text is too polite to make outright. A table reads as a claim to objectivity, whether or not the numbers in it deserve one². None of this is neutral, and a writer who ignores it is making choices by accident instead of on purpose.

Headings are argument, not decoration

A heading is not a label stuck on top of a section after the fact — or it shouldn’t be. It is a compressed claim about what the paragraphs beneath it are for. When a heading says “Objections” rather than “More Thoughts,” it has already told the reader how to read what follows: skeptically, as something to be weighed rather than absorbed.

A subsection is a smaller promise

Nest a heading one level down and you are promising something narrower: not a new topic, but a facet of the one just introduced. Readers trust this nesting instinctively, which is exactly why breaking it — putting an unrelated tangent under a heading that promised a facet — is one of the more disorienting things a writer can do to a reader who is skimming in good faith.

And a fourth level is a private aside

By the time you are four levels deep, you are no longer structuring an argument so much as annotating one. This level is for asides, for the “one more thing” that does not deserve its own section but also does not belong buried in a paragraph. Use it sparingly. A document that is all subheadings has stopped being prose and started being an outline that never got written up.

What a blockquote claims to be

Setting a passage off as a block quotation is a small act of deference: it tells the reader, this is not my sentence, and I want you to notice that before you read it. Longer quotations, especially ones that run several paragraphs, ask for something closer to patience — the reader is meant to sit with someone else’s argument for a while before the essay’s own voice returns.

The eye, in reading a well-set page, does something closer to gliding than to fixing. It does not linger on individual letters; it hops from word-shape to word-shape, filling in gaps from memory and context faster than it could ever read letter by letter. This is why bad kerning is so exhausting in a way readers often cannot name — the eye’s predictions keep failing, quietly, hundreds of times a page.

This is also why a typeface can be technically legible and still tiring to read at length. Legibility is about whether a letter can be told apart from its neighbors in isolation. Readability is about whether a whole paragraph of those letters lets the eye’s guesses succeed often enough to feel effortless. The two are related but not the same, and a great deal of bad type design comes from optimizing the first while ignoring the second entirely.

— Attributed, in the absence of better evidence, to a printer’s apprentice’s notebook, circa 1690s

Lists, in both directions

Some sequences have an order that matters. Others are just a pile of related things, and pretending they have an order would be a small lie. Both belong in an essayist’s toolkit, and mixing them up is a common tell of writing that has not thought carefully about its own claims.

An ordered list, for a sequence where the steps genuinely depend on one another:

1. Choose a typeface whose x-height and default line spacing you actually trust at body size.
2. Set a measure — the width of a line of text — somewhere around 60 to 75 characters.
3. Pick a line height that gives each line room to breathe without floating loose from the one below it.
4. Only then start worrying about color, weight, and all the things a beginner worries about first.

An unordered list, nested, for a set of related but non-sequential concerns — the kind of thing a type designer keeps in mind all at once rather than in sequence:

• Legibility at the sizes the typeface will actually be used at
  • Small caps and old-style figures, if the face has them
  • Behavior at the extremes: a very bold weight, a very light one
• Compatibility with the languages the text will actually appear in
  • Diacritics that do not look bolted on as an afterthought
  • Punctuation that matches the era and register of the prose
• How the face behaves badly, because every face has a failure mode
  • Some blur at small sizes on certain screens
  • Some feel cold in long-form settings no matter how well they test in a specimen

A table, because some comparisons want rows and columns

Prose can describe a comparison, but a table lets a reader compare two things without holding either one fully in memory. Here is a rough, opinionated map of a few classical serif faces and the register each one tends to suit:

Typeface	Era	Typical use	Personality
Garamond	1540s	Long-form book text	Warm, quiet, humane
Caslon	1720s	General printing, English	Sturdy, unshowy
Bodoni	1780s	Title pages, fashion print	High-contrast, formal
Georgia	1996	Screens at small sizes	Robust, plain-spoken

Code, when the subject is technical

Not every essay needs a code block, but this one is partly about how text is set, and some of that is genuinely a matter of markup. A footnote reference and a definition, in the flavor of Markdown this theme expects, look like this in source:

Here is a claim that needs support.[^1]

[^1]: Here is the support, tucked out of the main flow of argument.

And an inline mention of a specific value — say, the font-variant-numeric: oldstyle-nums; declaration that swaps in old-style figures for a serif face — reads differently set as code than it would as plain prose. The monospace treatment signals this is a literal token, not a word choice, which is exactly the distinction a reader needs when a sentence starts mixing prose with syntax.

A small checklist, and a small glossary

Some sequences are neither strictly ordered nor simply related — they are a set of conditions to be satisfied, which is its own kind of list:

• Pick a typeface with real italics, not a synthetically slanted roman
• Set a measure the eye can travel without losing its place
• Resist the urge to justify text without also enabling hyphenation
• Stop tweaking kerning pairs nobody will ever consciously notice

And a few terms worth pinning down plainly, in the manner of a glossary entry rather than a sentence:

Kerning — the adjustment of space between two specific letter pairs, done by eye, to correct the mechanical spacing a font provides by default.

Tracking — a uniform adjustment to the spacing across a whole run of text, used sparingly, and almost never on body copy.

Leading — the vertical space between baselines of text, named for the strips of lead type compositors once inserted between lines of metal type.

---

None of these devices — the headings, the quotes, the footnotes, the lists, the table, the code, the checklist — are here because an essay needs to prove it can use them. They are here because each one makes a small, distinct promise to the reader about how to hold what follows: as sequence, as comparison, as testimony, as aside. Learn what each shape promises, and you can say things in the gaps between sentences that the sentences alone never could. For further reading on the mechanics behind all of this, the Elements of Typographic Style remains the best single guide, decades after its first printing.

Footnotes

1. This is, admittedly, an easy claim to make and a hard one to prove; the best evidence for it is usually negative — you notice the shape of a text only once someone has gotten it wrong. ↩

2. A table with three rows of made-up numbers still looks more authoritative than a paragraph making the same argument in words. Use that power carefully. ↩