Without any measure of reproducibility it's a study that's really only good for fun.
Best to not-so-good differed by about 20%. The top ratchets were so close it's hard to imagine any difference.
But without testing the same model say, three, five, or even ten times, one has no idea of the measurement's precision.
I'd have thought an assessment of which ratchets were *rebuildable* after desctruction might have been helpful.
In my world, the long term life when used repeated at torque levels much, much lower than ultimate strength is more useful.