The second step
You come out of a meeting room in a building you have never been in before and take the stairs down. You look at the first step, and maybe the second. After that you are looking at your phone, or at the person you're talking to, or at nothing, and twelve more steps go by under you without being checked. You have never seen this staircase before and you put your whole weight on it, blind, fourteen times.
This atlas has stood near here already. In the light switch you can find in the dark the body knew the count of your own stairs at home, and the phantom stair at the top of them got filed under the complaints department. That was memory: a flight you have climbed ten thousand times. This one you have climbed once, and you have only half climbed it. What the foot knows here it learned in two steps. The first gives the height, the second confirms it, and the rest of the flight is assumed to agree.
Something has to make that assumption safe, and in most countries it is written down. The American residential code sets a maximum height for a step, but the line in it that matters more is a different one. Within a single flight, the tallest riser may not exceed the shortest by more than three-eighths of an inch, which is about nine and a half millimetres. The English regulations ask that every step in a flight have the same rise and the same going. Neither rule is mainly about the size of a step. The law allows a fairly wide range of heights. What it hardly tolerates at all is a step that differs from its neighbours. In effect it is regulating what the second step tells you about the tenth.
Where the regulation doesn't reach, a warning does the same work. In an old pub, the single odd step down to the toilets is painted with a yellow stripe and labelled MIND THE STEP. It is a confession that this step breaks the promise the others keep. The stripe is there because the foot will not see the problem by itself. It stopped looking a long time ago.
The number deserves some suspicion, because of where it came from. Tolerances like that one were set by people studying falls: accident statistics, slowed-down film of people losing their footing, expert reports for court cases. The architect John Templer spent years cataloguing how people come off stairs. All of that evidence is evidence of failure. It can tell you how much variation is enough to bring someone down. It cannot tell you how little the foot actually notices. A flight that is two millimetres out and has never tripped anyone leaves nothing to study. So the nine and a half millimetres is a line drawn at the edge of the accidents, not a measurement of the foot. What the foot really assumes, below that line, has never been recorded, because getting it right leaves no trace.
And so the trust runs in a chain. You trust the builder, someone you will never meet, to have made the tenth step agree with the second. The builder trusts an inspector, and the inspector trusts a number derived from people who fell. None of them is on the stairs with you. What is on the stairs is the second step, which you looked at once and believed.
Every staircase you have ever gone down without looking was built to keep a promise made by its second step.