> It added that three trains that had gone over the tracks at 17:21 on Sunday, 19:01 and then 19:09 had similar notches "with a compatible geometric pattern".

Then the crashed train passed at 19:45.

I don't see why an overnight inspection must have caught this, it could have happened just before the 17:21 train, or even have been caused by it.

We will need to wait for the investigation to continue, and I hope Japan's rail people will not be so arrogant as to assume they can't learn something from it.