A large amount of data would allow conclusion whose accuracy exceeds the precision of the TIMER, provided the TIMER is accurate to the given precision (which I believe is the case so several significant figures). This could be tested; would be interesting to devise and experiment to do so.