Typography.

Typography is the discipline that turns written language into a read object. A sentence carries its meaning in its words; it carries its speed, its rank, and its degree of authority in its setting. Before a reader has parsed a clause, the typeface, the size, the weight, and the measure have established how quickly the clause is to be taken and where it stands in the order of the page. Typography is not a surface applied to text once the writing is done. It is the design decision that assigns each passage a compositional position.

A typographic vocabulary is a set of those decisions held still long enough to be relied upon. When a heading is set in one serif at one step of the scale, and held there across a body of work, the reader recognises a heading as a heading before reading it; the form has become a position the page is navigated by. This is what the paper means by compositional accountability in the typographic case. A vocabulary that can be relied upon can also be audited: each setting has a defined role, and a setting used outside its role is legible as an error. A page with no fixed roles cannot be held to anything, because nothing on it has a role to depart from.

The discipline is old, and its mature position is restraint. Bringhurst opens his account of typographic style with the principle that typography exists to honour and elucidate the content it serves (Bringhurst, 1992). Warde, writing earlier, held that the most accomplished typography is the setting the reader does not notice; she compared it to a crystal goblet, a vessel chosen so that the wine it carries is seen and not the glass (Warde, 1955). Neither describes typography as ornament. Both describe it as a service performed well enough to disappear.

The corollary is that a typographic vocabulary, held over time, becomes the institution's hand. Handwriting is recognised before a signature is read, and a typographic system the institution keeps to is recognised on the same principle. A reader who has read the institution once and returns later meets the same setting and knows the source before any mark confirms it. This is the recognition the Abstract names: an institution becomes legible without a logo when its typography has been held long enough to be known. The vocabulary is therefore not a matter of preference. It is the institution's continuity, made legible one page at a time.

A small vocabulary is harder to hold than a large one. The available typefaces number in the thousands, and each new occasion presents a reason to introduce one more: a face for an announcement, a face for a heading the present face does not flatter. Adding is the path of least resistance. The discipline argued here is the opposite path. The institution holds two typefaces, a serif and a sans-serif; one typographic scale; one measure; two systems of number; and a small set of conventions. Everything it publishes is composed from them.

Two typefaces are enough because the work of a page divides cleanly in two. There is text to be read continuously, and there is everything that orients the reader to that text: headings, labels, figures, the apparatus of a page. A serif drawn for the considered statement and a sans-serif drawn for continuous text and working surfaces cover both. A third typeface does not extend the vocabulary; it blurs the division the first two were chosen to make. The reader can no longer tell, from the typeface alone, which kind of passage is in front of them, and the system has stopped doing its first job.

A small typographic vocabulary is also auditable, in the sense the Materials chapter gives the word. Two typefaces, six elements, a fixed scale: the whole system can be held in view at once, and a departure from it is visible. A system of twenty-five settings cannot be audited, because there is no held standard for a departure to be measured against. Restraint is not the absence of typographic range. It is the condition under which range becomes legible, because the reader has a constant to read the variation against.

Industrial design reached the same position. Rams held that good design is as little design as possible, and that a reduced vocabulary, each element better understood, outlasts a large one (Rams, 1995). The typographic case is stricter, because type is read and not only seen: an unfamiliar typeface costs the reader a measure of attention before the first word is taken. The institution spends that attention once, on two faces, and then spends less of it with every page that follows.

The institution's display typeface is a serif, Newsreader. It is set where the institution makes a considered statement: the headline of a page, the title of a chapter, the standing claim, the line a section is signed with. Its forms carry the modulation of the drawn and the cut letter, the transition from thick stroke to thin that the eye reads as deliberate. At display size that modulation gives a line presence; the headline holds the top of the page because the letterforms themselves have the weight to hold it.

The institution's headlines are declarative and end in a period. A headline left without terminal punctuation is open, and an open headline reads as a banner, a phrase the reader is invited to complete. The period closes the statement. It marks the line as a sentence the institution has finished, a claim rather than an invitation. The mark is the smallest in the typographic vocabulary, and it carries the chapter's argument exactly: a typographic decision is the institution committing to a position a reader can hold it to. The headline that ends in a period has committed.

The serif appears in one register that is not display. The line a section closes on is set in the serif at the size of body text. This is a deliberate departure from the division the vocabulary otherwise holds, where the serif carries display and the sans-serif carries text. The departure is the signal: a serif at text size, set alone beneath a passage, is read as a different kind of sentence, the institution stepping out of the working voice to make a measured close. A typographic system earns the ability to mean something by a departure only because it holds the rule everywhere else.

The institution's text typeface is a sans-serif, Manrope. Its strokes are close to uniform, without the modulation of the serif, and that evenness is what suits it to the body of a page. A paragraph is read in saccades, the eye landing several times a line; an even texture offers the eye no incident to catch on, and the reading runs without friction. The same evenness suits the sans-serif to the working apparatus of a page: labels, the fields of an intake surface, the small functional text a reader uses rather than reads.

The division can be tested by violation. Set a paragraph in the serif, and the modulation that gave a headline its presence becomes, across many lines at text size, a coarsened texture; the transition from thick to thin repeats hundreds of times, and the eye negotiates it at every word. Set a heading in the sans-serif, and the line goes flat; the evenness that carried a paragraph gives a heading nothing to hold the top of the page with. Figure 2 sets one passage each way. Neither setting is wrong as a letterform; each is wrong as a position. The pairing is the resolution, each typeface placed where its drawing already wanted to be.

The sans-serif also carries the institution's working surfaces, and there it meets the materials of Chapter I. The same body text is set on the photographic ground, on the document register, and within a specimen. The typeface does not change; its weight and its tone are tuned to the material beneath it, because text on a luminous photographic field and text on a dark working surface do not need the same contrast to hold. Typography is set on a material, and the material is the first thing a typographic decision answers to.

The typographic scale is the fixed set of sizes the institution sets type at. Every passage on every page is set at one step of that scale. The scale is small, and its steps are spaced widely enough that two adjacent steps are told apart at a glance. The reader reads rank directly from size: a passage one step above another is, reliably, one degree higher in the order of the page.

The scale is a discipline because it removes a lever. Any passage can be made to feel important by being made large; size is the easiest claim a designer can make. A fixed scale forecloses that claim. A passage cannot be enlarged into importance; it can only be placed at the step its role earns. Hierarchy then has to be carried by position, by weight, by the space around a passage, and by the typeface it is set in, which is to say it has to be composed rather than asserted. The size a passage is set at is a consequence of its rank, never the cause of it.

A scale held this way reads, to a designer accustomed to a free range of sizes, as too quiet. The institution accepts the quiet. A display line one step smaller than expected asks the reader to attend rather than be addressed, and attention given is steadier than attention seized. The live pages are set at restrained steps throughout, and the restraint is not an absence of confidence. It is the form confidence takes when a system trusts its reader to read.

The measure is the length of a line of continuous text. It is held, not improvised, and it is set close to the length at which a line returns to the reader without strain. Bringhurst gives the readable range as roughly forty-five to seventy-five characters a line, and names the value near sixty-six as the setting a single-column page is most comfortably built on (Bringhurst, 1992). A line much longer tires the eye on its return, and the reader loses the start of the next line. A line much shorter breaks the rhythm of reading, the eye snapping back too often for the prose to gather momentum.

The measure is held even when the surface it sits on is wide. A broad page does not widen the text to fill it; the line stays at its measure, and the surface carries the additional width as margin and space. This is a decision a reader rarely notices and always feels: the column of text on a wide page is the institution declining to make the reader's eye travel further than reading wants. Figure 4 sets one passage to three measures, the held measure between a line too long and a line too short, so the held value can be read against its failures.

Where the measure ends, Composition begins. The length of the line is a typographic decision; the arrangement of lines into columns, the placement of the column on the page, and the rhythm of text against space belong to Chapter IV. The measure is named here because it is set in type and answers to reading. The page it is then composed into is the next chapter's subject.

Number is set in two systems, and the institution holds them apart. Roman numerals classify: which edition, which chapter, which section. Arabic numerals quantify: a sum, a count, a rate, a figure of record. To classify is to state where a thing stands in an order; to quantify is to state how much of a thing there is. The two are different acts, and a typographic system that set them in one form would let the reader confuse a position with an amount.

A quantity is set in one of three modes, according to the work it does. In running prose a figure is written toward words, softened into the sentence it serves. As the anchor of a section a single figure is set large, held up as the subject rather than read in a line. Within a specimen, where figures are to be audited, they are set in tabular numerals: every digit on a fixed width, so that figures stack in true columns and the decimal points align down the page. A reader can sum a tabular column by eye, and cannot sum a column whose digits do not align. Tufte makes the case for the table on the same ground: the arrangement earns its place because it lets the eye perform the comparison the figures were assembled for (Tufte, 1983).

This is what the specimen of Chapter I rests on. A specimen is a reconciled unit of financial logic offered to the reader to verify, and the alignment of its figures is the first audit it offers, available in the setting itself before the verify affordance is opened. Number is the point at which typography stops orienting the reader and begins to furnish evidence; the discipline that aligns a column is continuous with the discipline that reconciles it.

The eyebrow is the tracked uppercase label set above a heading to name the section beneath it. It is the most reduced setting in the vocabulary: small, letterspaced, set in the sans-serif, held in a recessive tone. It does not address the reader and it does not make a claim. It names. The reader uses it to know where they stand, the way a reader uses a running head, and then reads past it to the heading.

Uppercase text is letterspaced because uppercase forms set tight are read slowly; the even rhythm of capitals needs added space between the letters to stay legible at a small size. The tracking is a correction, not a decoration, and the amount of it is a detail the vocabulary fixes once. Hochuli locates the quality of typography in exactly these decisions, the spacing of letters and words and the treatment of the small (Hochuli, 2008). The eyebrow is the chapter's clearest case of the detail carrying the discipline: a setting almost beneath notice, fixed with the same care as the headline above it.

Held across every page, the eyebrow becomes one of the institution's most recognisable marks, precisely because it is so reduced. There is little to it: a size, a tracking, a tone, a typeface. A reader could not describe it and would know it anywhere. It is the small end of the argument the chapter opened with. A vocabulary held long enough is recognised, and it is recognised first in its quietest elements, because those are the ones held most exactly.

The vocabulary is defined as much by what it refuses as by what it holds. Three refusals carry the most weight.

The institution refuses typography that performs. A line set large for impact, a heading given a treatment that draws attention to the setting rather than the statement, a typeface chosen because it is novel: each spends the reader's attention on the type itself. Warde's argument stands behind the refusal. The setting noticed as a setting has failed; it has put the goblet between the reader and the wine (Warde, 1955). Typography in this vocabulary does its work and returns the reader to the content.

The institution refuses size as the instrument of importance. A passage is not enlarged to be made to matter; it is placed at the step its role earns, and its importance is composed from position, weight, and space. The fixed scale is the standing form of this refusal. It is harder than the alternative, because it forecloses the easiest move a page offers, and the difficulty is the point.

The institution refuses the third typeface. The vocabulary divides the work of a page in two, and a third face blurs the division rather than extending it. A new occasion is met by composing the two faces and the held scale differently, not by admitting a new element. This is the typographic form of the discipline of few, and it is the refusal most often under pressure, because a new face is always the easiest answer to a page that is not yet working. The discipline is to find the answer in the vocabulary already held.

Typography does not stand on its own. It is set on the materials of Chapter I, and the material beneath a passage is the first thing the setting answers to; the same text is weighted differently on a photographic ground than on a dark working surface. It is arranged by the Composition of Chapter IV, which takes the line at its measure and builds it into the column and the page. It carries the Voice of Chapter V; the register a sentence is written in and the typeface it is set in are two halves of one act, and the reader receives them together. Its refusals are part of the larger discipline of Chapter VI.

This chapter is set in the vocabulary it describes. Its headings are in the serif and end in a period; its text is in the sans-serif at the held measure; its sections are classified in Roman numerals and its figures quantified in Arabic; its eyebrow names it. A reader testing the chapter's claims has the evidence in hand, which is the condition the paper set itself in the Introduction. A chapter on typography that was not itself well set would have refuted its own argument before making it.