In 2005, the Universal Web Project set out to answer a practical question: how should text be set on screen so that people with low vision — and, by extension, everyone — can actually read it? The original guidelines drew on twentieth-century graphic design principles and the display technology of the day: coarse CRT and early LCD panels, a handful of screen-optimized typefaces, and browsers that treated type size as a suggestion at best.
Two decades later, nearly every assumption underlying that research has shifted. This second edition revisits every finding, keeps what held up, and revises what didn't.
The motivation has not changed. Low vision remains an uncorrectable sight deficiency, most often the result of age-related conditions such as glaucoma and macular degeneration. As populations age, the number of readers who depend on well-set screen text grows rather than shrinks. What has changed is that screen reading is no longer a degraded substitute for paper. For many people it is the only reading environment that matters — and, configured well, it can now be the more accessible one.
What Changed Since 2005
The original guidelines were written for screens that rendered type at roughly 96 pixels per inch, where a letterform had perhaps a dozen pixels to work with. Under those conditions, the research favored sturdy sans serifs, generous sizes, and defensive simplicity. Four developments have forced a rethink.
First, display density. High-DPI panels — commonly two to three device pixels per CSS pixel — render letterforms with detail approaching print. Fine serifs, subtle stroke modulation, and true italics survive on screen in ways they simply could not in 2005. Several of the original recommendations were compensating for rendering limitations that no longer exist for most users, though not for all.
Second, the type itself. Variable fonts put weight, width, optical size, and sometimes grade on continuous axes controllable from CSS. Decisions that were once binary — regular or bold, one family or another — are now tunable, and can respond to context: dark mode, small sizes, user preference.
Third, the standards landscape. The Web Content Accessibility Guidelines matured through 2.0, 2.1, and 2.2, giving us testable success criteria for text resizing (1.4.4), reflow (1.4.10), text spacing (1.4.12), and contrast (1.4.3 and 1.4.11). Meanwhile the WCAG 3.0 draft and the APCA contrast research behind it are rethinking how contrast, size, and weight interact — a question the 2005 guidelines could only gesture at.
Fourth, CSS grew up. Rem-based sizing that respects user defaults, fluid type with clamp(), unitless line-height, ch-based measure, font-size-adjust, text-wrap, and media queries like prefers-color-scheme and prefers-reduced-motion give designers direct, standards-based tools for the accommodations the original project could only recommend in principle.
What Held Up
Not everything needed revision. The core thesis — that legibility is built at the level of the letter and the word, and that choices invisible to most readers are decisive for readers with low vision — has only strengthened. Large x-heights, open apertures, distinct letterforms, adequate size, moderate weight, and generous spacing remain the load-bearing recommendations. Research on reading and visual acuity continues to support them.
What changed is the reasoning. In 2005 we recommended sans serifs largely because serifs rendered badly; today the serif/sans question matters far less than x-height and letter distinction. We recommended 14 points because smaller text fell apart on coarse screens; today the argument for a 16px floor rests on user defaults, zoom behavior, and WCAG conformance rather than pixel grids. Where a recommendation survives, this edition tries to give it its modern justification — because a guideline whose rationale has expired tends to be applied badly.
How This Edition Is Organized
The structure follows the original's progression from smallest unit to largest, with one addition.
Part 1, The Letter, covers the legibility of individual letterforms: style, size, weight, slant, width, and spacing, updated for high-density rendering and variable font technology.
Part 2, The Word, moves to the recognition of words in continuous text: case, spacing, line length, and the mechanics of reading, where eye-movement research has refined much of what we assumed.
Part 3, The Page, is new in this edition. In 2005, page-level control was too crude to standardize. Today, responsive layout, fluid type scales, user preference media queries, dark mode, and reflow requirements make the page a first-class accessibility surface, and it deserves its own treatment.
In CSS
/* The reading defaults every page can start from */
html { font-size: 100%; } /* honor the reader's browser setting */
body {
font-size: 1rem; /* never below the 16px default */
line-height: 1.5;
}
article { max-width: 65ch; }Recommendations
- Treat the 2005 guidelines as historically grounded but rendering-specific; verify each against current displays before applying it.
- Build to WCAG 2.2 success criteria as the testable floor, and watch the WCAG 3.0 draft and APCA research for where contrast guidance is heading.
- Use the platform's accessibility machinery — rem units, user preference media queries, reflow-friendly layout — rather than reinventing accommodations in script.
- Remember the constituency: what is now optional polish for most readers remains essential infrastructure for readers with low vision.