Before a new airliner carries anybody, one of them is built to be destroyed.
The aircraft is fixed to the floor of a hall and cables are attached along both wings. Then the wings are pulled upward, slowly, past the load of a normal flight, past the load of the worst turbulence anyone expects, and on to about one and a half times the greatest load the wing should ever meet in service. The wingtips rise several metres. Eventually something gives, and the engineers go to look at exactly where it gave.
Nobody learns anything from a wing under a normal load. It holds, which was already known. A design only reveals its limit at the extreme, and the extreme is the one place a weakness cannot hide.
The habit is worth borrowing. A rule that has only been tried on easy, ordinary cases has not really been tried. Push it to the edges — the empty case, the case of exactly one, the negative case, the case that is slightly less than one — and it will either hold or show you precisely where it does not. Either answer is useful, and only one of them can be found in the comfortable middle.