> Hadal's WCAG 2.2 AA commitment, the one accessibility figure actually measured, the measures built into the site, the open gaps, and how to report a problem. - Canonical: https://hadalinstruments.com/accessibility/ --- Standards # Accessibility statement Hadal builds measurement instruments, so a page about accessibility that contained no measurements would be the wrong document from the first line. What follows is kept in three separate parts, and they are separate on purpose: what we are committed to, what has actually been measured, and what has *not* been done. The third part is the one worth reading. ## What we are committed to We are working to meet **WCAG 2.2 Level AA** across every page on hadalinstruments.com — the marketing surface, the instrument pages, the glossary, the research articles, the docs and the methodology receipts alike. **This is a commitment, not a conformance claim.** Under WCAG, claiming a conformance level means a full evaluation of complete processes has been carried out and can be produced. No such evaluation has been run on this site. So this page states a target and a work list, and it will keep doing that until an evaluation exists to point at. ## What has actually been measured One figure. Not a programme of testing, not a site-wide result — one figure, from one automated tool, against one page. Lighthouse — accessibility category 100 / 100 Scope: the homepage (/) only, under mobile emulation, using Lighthouse's default accessibility audit set. One page, one run profile. SHA-256 NOT YET PUBLISHED The digest above reads **NOT YET PUBLISHED** because it is not published: the Lighthouse report artifact behind that 100 has not been posted, so you cannot currently re-derive it. By this site's own standard that makes it a weak number, and it is printed here with its weakness attached rather than polished. **What that figure does not establish** - **It does not describe the rest of the site.** The homepage was scored. No other template — instrument page, glossary term, research article, docs page, receipt page — has a published accessibility score. - **It does not describe a real device.** Mobile emulation is a desktop browser pretending, in a lab. It is not a phone, and it is not a phone being used by a person. - **It is not an audit.** Lighthouse checks a fixed list of machine-detectable conditions. Whether a link makes sense out of context, whether a reading order is logical, whether an error message actually helps the person who hit it — no automated tool decides any of those, and they are where real accessibility failures live. - **A perfect automated score and an unusable page are compatible.** We would rather say that ourselves than have it discovered. ## What the site is built to do Everything in this section is structural, present on every page, and checkable by anyone with a keyboard and a browser. - **Skip link.** The first focusable element on every page jumps straight to the main content, and the main region takes focus when it does — with its focus ring kept and turned inward, because a ring drawn outside a full-width region is drawn off-screen, which is the classic silent skip-link failure. - **Landmarks.** Every page ships a banner, a named navigation, a main region and a contentinfo footer. Breadcrumb navigation is named separately so it is not confused with the primary menu. - **Visible keyboard focus.** One focus rule is applied through the universal selector, so every interactive element — including anything added later — inherits a visible ring by default and can only lose it by actively opting out. Under Windows High Contrast the ring is re-expressed in system colours, so it survives the operating system discarding our palette. - **Reduced motion is honoured throughout.** With `prefers-reduced-motion: reduce` set, animations and transitions collapse site-wide, and the scripted effects — the heading reveal, the particle field, the tick stream — are never instantiated at all rather than running invisibly and still costing you a busy main thread. - **Touch targets of at least 44 × 44 CSS pixels** on interactive controls. Small and icon-only controls recover the area with an invisible inset region instead of being inflated visually. - **Tabular figures on every number.** Numbers render in monospace with tabular lining figures, so a column aligns without a table and a value that changes does not shove its neighbours sideways. Layout that holds still is an accessibility property, not only a performance one. - **Contrast computed, not eyeballed.** Every text and accent step was computed (sRGB relative luminance) against every surface it can land on, and the resulting table is kept in the stylesheet next to the tokens themselves — including the two entries that fail, which are listed in the next section rather than quietly left out. - **Forms that do not destroy their own disclosures.** On the waitlist form the data-handling sentence is static copy and is what the field points at for its description; progress goes to a polite live region and failures to an assertive one, as three separate elements. An earlier build reused one element for all three, so the first submit erased the privacy notice. The email field is also set at 16 px so iOS Safari does not force a zoom the visitor then has to undo by hand. - **Charts owe a data table.** The figure component this site publishes charts through requires a data-table alternative alongside the graphic, and treats the table as the authoritative version. See the next section for the honest status of that. - **Language is declared** on every page, so a screen reader picks the right pronunciation rules rather than guessing. ## What has not been done Stated plainly, because a statement that lists only what went well is an advertisement wearing a standards badge. - **No full manual audit has been performed.** Nobody has walked every WCAG 2.2 AA success criterion against every template on this site. Until that happens, no conformance level is claimed here. - **No assistive-technology user testing has been done.** No screen reader user, no voice-control user, no switch user and no magnification user has tested this site. Automated checks and a developer's own keyboard pass are not a substitute for that, and we are not going to imply they are. - **No independent accessibility audit has been commissioned**, and no third party has reviewed any of the above. - **Two known contrast shortfalls, with their measured values.** The tertiary text tone clears AA ( 4.5:1 ) on the darker surfaces — between 5.4:1 and 4.8:1 — but measures 4.4:1 and 3.9:1 on the two lightest ones, which are hover and raised states. That means the shortfall appears *on interaction*: a caption legible at rest can drop below AA once the pointer is on its card. Separately, the deepest step of the accent ramp measures 4.2:1 on the lightest surface and clears AA everywhere else. Neither is fixed by changing the token — the only value that would clear AA collapses two text levels into one — so the repair is a call-site change, and it is in progress rather than complete. - **The chart data-table rule has never been exercised.** The measurements published so far ship as downloadable artifacts rather than charts, so no chart on this site yet carries measured data and no data-table alternative has had to exist. The rule binds the first chart that does. Until then, treat it as untested. ## Reporting an accessibility problem Accessibility contact [accessibility@hadalinstruments.com](mailto:accessibility@hadalinstruments.com) This address is monitored, and it is the channel this statement names — a statement nobody can reply to is a monologue. It sat as a stated absence until the route existed, because printing an inbox nobody reads is worse than admitting there is none. **A useful report includes** - the address of the page, and what you were trying to do on it; - your browser and operating system, and the assistive technology and version if you use one; - what happened instead — including "nothing happened", which is a real answer and often the most useful one. A response-time commitment will publish here once it is a figure we can actually keep rather than a comfortable one. ## How this statement is kept Accessibility defects enter this project's standing defect register like any other defect, and are worked in the same queue rather than a separate one that is politely never reached. This page changes when that register changes — when a gap above closes, or a new one is found — and not on a publishing schedule. If a claim on this page ever stops being true, the claim comes off the page. It does not get softened. - **Statement drafted** — 2026-08-01 - **Operator review** — NOT YET PUBLISHED - **Next scheduled review** — NOT YET PUBLISHED - **Standard targeted** — WCAG 2.2 Level AA [About Hadal](https://hadalinstruments.com/about/)[Methodology](https://hadalinstruments.com/methodology/)[Privacy](https://hadalinstruments.com/privacy/)[Terms](https://hadalinstruments.com/terms/)