Accessibility

Read it the way that suits you

Display in the header changes the theme, the text size, the line spacing, and the typeface. The choices are saved on your own device, need no account, and are there on the sign-in page as well as inside — you should not have to get past a form you cannot read in order to fix the way it reads.

• Theme. Light, dark, or follow your system. Following the system is the default, because you have already made that choice once and we should not ask again.
• Text size. Four steps. Everything scales together, including the spacing around it, so larger text does not collide with itself.
• Line spacing. Tight, standard or roomy.
• Legible typeface. Atkinson Hyperlegible, designed by the Braille Institute to pull apart the letters that ordinarily collide. Italics become bold while it is on, because a slanted face undoes what the upright one is for.

What is true on every page

• Keyboard throughout. Every menu, form and control is reachable and operable without a pointer. Menus close on Escape and hand focus back to where you were. Focus is always visible.
• Skip to content. The first thing Tab reaches on any page, so the header is one keystroke to pass rather than a dozen.
• Motion follows your system. If you have asked your device to reduce motion, animation and transitions stop. There is no autoplay and nothing moves on its own.
• Contrast is measured, not eyeballed. The brand teal was darkened twice because the original failed against white. Colour never carries meaning by itself — a stream is always named beside its colour, and a status always says what it is.
• Forms work without JavaScript and say what went wrong next to the field that went wrong, not in a summary at the top.

How it is checked

Every change runs an automated scan of every page — signed out, signed in as staff, and signed in as an applicant — against WCAG 2.2 AA, using two independent engines. A page that fails does not merge. That catches a great deal and it does not catch everything: automated testing finds roughly a third of real barriers, and no scan can tell whether a sentence makes sense.

Where it falls short

• It has not been tested with real screen reader users. Automated checks and keyboard testing are not the same thing as someone who uses JAWS or NVDA every day trying to get a job done.
• Some tables scroll sideways on a narrow screen. They are readable, but a phone is not the easy way to use the run of show.
• Reading preferences are per device. Signing in on a different machine means setting them again.