Sheet 02No. 0777 entries

Type classification, from Venetian to Grotesque

Classification is not trivia. It is the shortest way to say what a face will do in a paragraph before it is ever set in one, because the four features that name a family (stroke contrast, serif form, the bracket where serif meets stem, and the axis of the round letters) are the same four features that decide how it holds up at nine point and how it behaves next to another face.

Six blank filing cards laid in a grid, each with a paper index tab, one tab violet.
No. 077 Six families, one card each. The tabs are the only thing that tells them apart, which is the whole problem.

Entries in this section

Register
  • No. 214 Best fonts for graphic designers 18 Aug 2026 Open Best fonts for graphic designers Forty-one faces grouped by the job they do, each with its class, its designer and the licence you actually get.
  • No. 408 Anatomy of a typeface 20 Aug 2026 Open Anatomy of a typeface Every part named, and the five lines measured out of the font file this page actually serves.
  • No. 519 Font pairing 14 Aug 2026 Open Font pairing The four working rules, and a bench that sets any pair at the size, leading and tracking it will really be used at.
  • No. 132 Font licensing 12 Aug 2026 Open Font licensing Open, free-for-personal-use, and the separate desktop, web and application rights, and which one governs.
  • No. 265 Free display typefaces 24 Aug 2026 Open Free display typefaces The twelve free display faces one designer released, what the catalogues still list, and what the terms permit.
  • No. 190 Type in practice 16 Aug 2026 Open Type in practice Body size, ratio, leading and measure: how the decisions on paper become a scale a real page can hold.
  • No. 331 Typography glossary 19 Aug 2026 Open Typography glossary Around seventy working terms, grouped by what they describe and defined by what each one changes.

Why the families are named after printers and not after shapes

Every classification in common use descends from the same accident of history: type was cut by hand, one punch at a time, and the tools available in Venice in 1470 were not the tools available in Paris in 1784. The families are therefore periods first and shapes second, which is why the shape rules have exceptions and the period boundaries do not. A face is Venetian because it inherits what Nicolas Jenson could cut, not because every Venetian face shares a measurement.

That bears saying outright, because the useful part of classification survives the exceptions. The four features below separate the families in almost every case, and where they fail to, the failure is itself informative: a face that will not sit in a box is usually a face doing something deliberate.

A row of blank paper index tabs fanned in a shallow arc, graduated from bone white to warm grey.
Fig. 02 Classification is a gradient before it is a set of boxes. The names mark places on it.

The four features that do the work

Stroke contrast

The ratio between the thinnest and the thickest part of a letter. A pen held at a constant angle produces contrast automatically; a broad-nibbed pen held flat produces a great deal of it. Venetian faces sit near one to one and a half. Didones reach one to six and beyond, which is why they shatter at small sizes and glitter at large ones. The six skeletons below are drawn to the ratio printed beneath each, so the drawing answers to its own caption.

Serif form

A serif can be a wedge, a slab, a hairline, or absent. The wedge belongs to the oldest faces and still carries the mark of the pen that drew it. The hairline belongs to the Didones and is the visible result of engraving rather than writing. The slab arrives with the nineteenth century and with advertising, where a letter had to survive being printed badly on cheap paper at speed.

The bracket

The curve where the serif meets the stem. A bracketed serif flows out of the stem; an unbracketed one is stuck on at ninety degrees. This is the single quickest test in the whole system, and it is visible at a glance once it has been pointed out. Garaldes bracket. Didones do not.

The axis

Draw a line through the thinnest points of a lowercase o. If it leans left, the face descends from a pen held at an angle. If it stands upright, the face descends from an engraver working with a burin, or from a designer thinking in geometry. The axis moves from roughly fifteen degrees off vertical in the Venetians to dead vertical in the Didones, and the whole history of the roman letter is contained in that rotation.

The six families in order

Venetian

Low contrast, heavily bracketed wedge serifs, a strongly angled axis, and, the giveaway, a sloped bar on the lowercase e. Warm, slightly irregular, and easy to read in long settings. Centaur and Jenson revivals are the reference points.

Garalde

The refinement of the Venetian. Contrast rises, the axis is still angled but less so, the e bar levels out, and the proportions tighten. Garamond, Bembo and Caslon live here. If a text face feels like a book, it is probably a Garalde or borrowing from one.

Transitional

The eighteenth century, and the moment the letter stops remembering the pen. Contrast rises again, the axis moves close to vertical, serifs sharpen. Baskerville is the name in every textbook because Baskerville is the point where the change is unmistakable.

Didone

The end of that road. Extreme contrast, a perfectly vertical axis, and flat unbracketed hairline serifs. Bodoni and Didot. Magnificent above about thirty point and structurally unsound below about ten, which is not a criticism of the faces but a description of what they were cut to do.

Slab

Contrast collapses back to nearly nothing and the serifs become rectangles the same weight as the stems. Built for posters, timetables and anything that had to be legible while being printed badly. Clarendon adds a bracket back in and is the reason the family is not as uniform as its name suggests.

Grotesque

The serif is removed entirely. Early grotesques keep a trace of contrast and a slightly awkward, gathered feeling: the name was originally an insult. Later neo-grotesques regularise everything, and geometric sans faces abandon the written skeleton for a compass and a ruler.

Where classification stops being useful

Three cases break the system, and all three are worth knowing because they are common.

The first is the humanist sans: a face with no serifs at all whose skeleton is unmistakably Garalde, with an angled axis and calligraphic terminals. It classifies as a grotesque by the serif rule and as an old-style by every other rule, and the second answer is the one that predicts how it behaves in a paragraph.

The second is the display face, of which the free-font catalogues hold many thousands. A display face is designed for one size and one job, and its features are chosen for effect rather than inherited from a period. Classification tells you almost nothing about it. The anatomy of the letterform tells you a great deal.

The third is the variable font, where contrast, weight and width are axes a reader's browser can move at runtime. A single file can cross two families between its extremes, and the classification applies to an instance rather than to the font.

None of that makes the six families obsolete. It makes them what they always were: a fast, approximate vocabulary for saying what a letter is going to do, to be replaced by measurement the moment the answer matters.

Questions this raises

Queries

Is there an official type classification system?

There is a standard (Vox-ATypI, adopted in 1962) but it was formally disavowed by its own association in 2021 for being built around Latin type and unable to describe most of the world's writing systems. What survives in practice is the six-family core described here, used as shorthand rather than as taxonomy.

What is the quickest way to tell a Garalde from a Didone?

Look at where the serif meets the stem. If there is a curve between them, it is bracketed and the face is a Garalde or a Transitional. If the serif is a flat line stuck on at a right angle with no curve at all, it is a Didone. The test takes about a second and needs no reference.

Does classification matter when choosing a font for a website?

Indirectly. It predicts behaviour: a Didone will lose its hairlines on a low-density screen at body size, a slab will hold its weight almost anywhere, a humanist sans will read longer without fatigue than a geometric one. Those are the decisions that actually get made, and the family name is a fast way to reach them.

Where do display and script faces sit in the six families?

Outside them. Both are usually treated as separate categories because their features are chosen for a single use rather than inherited from a period. A display face is judged on how it performs at the one size it was drawn for, which is a measurement question rather than a classification one.