Accessibility

Scope

This statement covers this website and the eleven pages it holds.

Conformance

No level of conformance is claimed. The site has been assessed against WCAG 2.2 AA by its author and has never been audited by anyone else, and claiming a level nobody independent has tested would be worth nothing. What follows is what was measured and what was not.

Measures

Every one of the eleven pages carries a single main heading, a site header, a navigation, a main landmark and a footer.

Every page opens with a link that skips straight to the content, and focus stays visible on each element that can take it.

All thirty-seven images on the site carry a text alternative. Four of them are decorative and say so with an empty one, which is how you tell a screen reader to skip a picture that only repeats the words next to it.

Movement runs only where the browser reports no preference for reduced motion, and the resting state of every page is visible and still.

Body text is Carbon on Parchment at 15.7 to 1. The quieter grey measures 6.6 to 1 and the accent 5.5 to 1. All three pass the AA threshold of 4.5 to 1, and the body text alone passes AAA at 7 to 1. The pale fields behind the contents list and the figures of each case study raise those numbers rather than lower them: text on them measures between 14.5 and 16.4 to 1.

The small drawn signs measure 3.3 to 1 against the paper, above the 3 to 1 the standard asks of a graphic that carries meaning.

Tested by hand

Three checks that used to sit under known limits have now been run, in August 2026, with the site served over local HTTP. I ran them myself, on my own machine, and nobody independent has repeated them.

A screen reader. Reading from the top with VoiceOver on Safari, the skip link, the menu and the page title are announced in that order. The three case cards can be told apart in the links rotor, which was the doubt worth testing, given that the whole card is a single link. The headings rotor gives the chapters in order and without repeats, and every figure is announced with a description that says something.

Text at two hundred per cent. Zoomed in a real browser, the cards stay square on all eight components that draw one, no piece of the small drawn signs leaves its frame, and the text and the layout hold on the home page, on the first case study and on Contact. The reasoning behind that was already on record: no type size on the site grows by more than two and a half times between the narrowest window and the widest, which is the condition that keeps the zoom safe. The browser confirmed it.

More than one browser engine. Everything above was run twice, in Safari 26.5.2 and in Firefox 153 on macOS. The work that came before had been looked at in Chromium alone.

Known limits

Two things have not been checked, and until they are I will not say they work. The site has never been opened on a real handset, and it has never been served from an address, so how the type renders on a phone is not known. And no laboratory scores have been measured: the weights on record were taken from the files themselves, and nobody has run the site over a network to see how it behaves there.

Feedback

If something here gets in your way, write to federico.rita@tuta.com. I read everything that arrives there, and I usually answer within a few days.

Method

Self-assessment by the author, not an independent audit. The checks were run with scripts written for this site: every image tested for a text alternative, every heading level walked for gaps, every page loaded at eight widths from 320 to 1920 pixels and searched for anything crossing the edge of the window, and every pair of colours computed rather than judged by eye. What is listed under tested by hand was run in real browsers instead of by script. An automated engine, axe-core 4.10.2, was run over the eleven pages and reported nothing. Tools of that kind catch between a quarter and two fifths of what is actually wrong, so a clean result sets a floor and leaves the rest to the checks above. Reviewed in August 2026.