Sufficient contrast

Claim: text has a contrast ratio of at least 4.5:1 against its background, dark text on a light ground by default, with a dark theme offered and never forced.

• Grade: measured for dark text on a light ground; the 4.5:1 floor is a convention (a standard), and so is an offered dark theme.
• Evidence: typography and layout (strand 4), baseline, terms.
• Check: sufficient-contrast, on the site's stylesheet with its variables resolved in every theme. It is exact for a text colour declared with its background, and warns for one that inherits it, measured on the page and on the surfaces a repository names. The lines of a drawn figure reach 3:1 against the page and their fill (figure-contrast, exact; WCAG 1.4.11).
• Reader styles: none override it.

What it asks

• 4.5:1 for body text, 3:1 for large text and for the borders of controls (WCAG 2.2, 1.4.3 and 1.4.11).
• Colour never the only cue (WCAG 2.2, 1.4.1).
• Light background by default; a switch for readers who want dark.

Why

Light-on-dark text cost acuity and proofreading accuracy at every age (Piepenbrock 2013), and more as type shrank (Piepenbrock 2014).

Limits

By day the polarities did not differ. In a dark room (about 0 lux), white text on black was less legible (Dobres 2017, measured: 34 adults, single Italian words read at a glance). The claim that dark mode improves reading is not supported.

A dark theme is still offered, as a reader's choice and not for legibility. That is a convention, which no study here tests.

The 4.5:1 floor is a convention under revision. APCA makes the threshold depend on the size of the text, and on its weight in principle (APCA 2026); WCAG 3 has not chosen a method (W3C 2026). It is also widely broken: on 83.9% of the million home pages WebAIM scanned, an automated AA check found low-contrast text (WebAIM 2026, measured, descriptive). APCA is its author's proposal; its page cites no independent test.