Typography
The 40-Character Column: Reading Gravity and the Economics of Attention
In 1963, Miles Tinker published the results of more than three decades of legibility research at the University of Minnesota. Among his most robust findings: optimal line length for 10-point type fell between 3 and 3.5 inches, yielding approximately 55 to 70 characters per line. Lines shorter than 40 characters fragmented reading into a staccato rhythm. Lines longer than 75 characters caused return-sweep errors — the reader's eye landing on the wrong line after traveling from the end of one line back to the beginning of the next. This single variable, line length, governs more of the reading experience than any other typographic parameter.
The Return Sweep Problem
The return sweep is the saccadic eye movement from the end of a line to the beginning of the next. It is the most error-prone moment in the reading process. When lines are too long, the return sweep covers too great a horizontal distance, and the probability of the eye landing on an adjacent line — either the one just read or the one two lines below — increases sharply. The reader must then locate the correct line, a process that interrupts comprehension and consumes cognitive resources.
Tinker's data showed that return-sweep errors increased measurably when line lengths exceeded 75 characters. At 90 characters per line, the error rate was high enough to produce a statistically significant decrease in reading speed. This is why newspaper columns are narrow: not to save space, but to ensure that the return sweep is short enough to be completed accurately at scanning speed.
The Short-Line Penalty
If long lines cause return-sweep errors, one might expect that shorter lines are always better. They are not. Lines shorter than 40 characters force too many return sweeps per unit of text. Each return sweep introduces a micro-pause — a moment of reorientation that interrupts the reading rhythm. With very short lines (25 to 30 characters, common in mobile layouts and narrow newspaper columns), the reader spends a disproportionate amount of time executing return sweeps rather than reading.
Bringhurst places the ideal range at 45 to 75 characters for a single-column layout, with a "comfortable" target of 66 characters — a number that derives from traditional typographic practice as well as legibility research. For two-column layouts, he recommends a minimum of 40 characters per column, acknowledging that the shorter measure will reduce reading speed slightly but will be offset by improved return-sweep accuracy.
Character Count as Grid Constraint
In a modular grid system, column width is not set in arbitrary units — it is derived from the intersection of the typeface's character width, the desired characters-per-line count, and the page dimensions. For 11-point Source Serif Pro (a face with a moderate x-height and character width), a line of 66 characters requires approximately 26 picas of column width. On a 6×9-inch page (36 picas wide) with 4.5-pica margins on each side, the resulting text block of 27 picas accommodates 68 characters — near the ideal.
This relationship between character count, type size, and column width is deterministic. Given any two of the three variables, the third is fixed. The grid system formalizes this relationship by making column width a function of the type specification rather than an independent design decision. This is what separates systematic typography from layout-by-eye: the column exists to serve the measure, not the other way around.
Attention Economics: Longer Lines, Faster Abandonment
Nielsen Norman Group research on web reading behavior has found that users read only about 20% to 28% of the text on a typical page. This percentage increases when text is well-formatted and decreases sharply when lines are long and paragraphs are dense. While web reading differs from book reading, the underlying principle is the same: the reader is constantly making cost-benefit calculations about whether to continue, and every moment of reading difficulty tips the calculation toward abandonment.
The 40-to-75-character measure is, in economic terms, a way to minimize the per-line cost of reading. Each line is short enough to be captured in one or two fixations, long enough to convey a meaningful phrase, and connected to the next line by a return sweep short enough to execute without error. The reader who encounters this measure does not notice it — and that invisibility is the highest compliment a typographic decision can receive.
The Actionable Rule
Calculate your column width from your typeface and size, targeting 45 to 75 characters per line, with 66 as the ideal. For a standard book page (6×9 inches) with 11-point serif type, this typically yields a single-column text block of 25 to 28 picas. For two-column layouts, accept a minimum of 40 characters per column. Derive the grid from this measure, not the reverse.
Line length is not a detail. It is the fundamental parameter that determines whether your reader's eyes move through the text efficiently or struggle against the geometry of the page. Get the measure right, and every other typographic decision becomes easier. Get it wrong, and no amount of typographic refinement can compensate.
Put this into practice
Every principle above is built into PagePerfect.
Baseline grids, proportional type scales, and 15 professionally engineered templates. Preview for free, export KDP-ready PDFs from $19.99.