Accessibility
What is guaranteed, what has been tested, and what has not. The third list matters as much as the first.
- One main heading per page, with the headings beneath it descending in order rather than skipping levels.
- A skip link as the first focusable element on every page, visible when focused.
- Every expandable section uses the browser’s own built-in control, so keyboard operation, focus order and how it is announced to a screen reader all behave the way your browser behaves everywhere else rather than the way we invented. The evidence layers still open and close if scripting is switched off.
- The focus outline is never removed. Wherever you are on the page when navigating by keyboard, there is a visible ring around it.
- Colour is never the only signal. Every evidence badge, verdict and peer-review label pairs its colour with a word.
- Search results and copy confirmations announce through polite live regions.
- Wide tables scroll inside their own container; the page body never scrolls sideways.
- If you have asked your device to reduce motion, smooth scrolling is switched off and transitions collapse.
- Dark mode follows your device setting, and its colours are checked separately.
Every page on this site is one of a small number of designs, and 63 of them are checked automatically against WCAG 2.0 and 2.1 at levels A and AA, plus a set of best-practice rules, before anything is published. They currently report no failures.
That check is not decoration. The first time it ran it found seven real faults, including headings that jumped a level and two navigation areas a screen reader would have read out under the same name — all fixed.
These are genuinely open, and are not claimed:
- Colour contrast is calculated, not observed. Every foreground colour on this site is checked arithmetically against every background it is used on, in both the light and the dark palette, and a pair below the WCAG AA ratio fails the build. What that cannot see is a colour landing on a background nobody wrote down — the automatic page check cannot work out final on-screen colours without actually drawing the page, so it reports that test as switched off rather than quietly counting it as a pass.
- Whether tap targets are large enough, and whether every scrollable area can be reached by keyboard — unverified for the same reason.
- No screen-reader testing has been performed. Not with VoiceOver, not with NVDA, not with anything.
- No keyboard-only walkthrough has been recorded by a person.
- No testing on real phones or tablets. The layout has been checked at a range of screen widths, which is not the same thing as holding the device.
Accessibility problems are treated as defects, not suggestions — and NatureHQ currently has no channel through which you can report one. That is stated plainly on the corrections page rather than dressed up as a process. When a channel exists, the most useful report will describe what you were trying to do and what assistive technology you were using.