How many peer-reviewed studies would it actually take to falsify any given law of physics? Say for instance that you did an experiment where you dropped a carrot to the ground repeatedly. But once every ten trillion tries or so, it remained in the air, seemingly unaffected by gravity.

Let us say that some scientists observed this, and published their data in a peer-reviewed journal. Would that be sufficient evidence for the claim that carrots can float in the air unaffected by gravity sometimes, in some circumstance, for some reason, at some frequency?

If not, what about if there were ten such studies? If not then, what about if it were a hundred such studies? If not then, what about if it were a thousand such studies? Etc. At what point would we justified, based on the data, in claiming to know that this can happen?

Of course, claiming that something can or has happened is very different from saying that something is the general rule. No one would say that it is the general rule that carrots can float in the air when released because of such data. For such a claim, we would indeed need to balance the evidence against its exact antithesis, that is, all the evidence that carrots generally fall to the ground when they are being dropped. So there is huge difference between a claim that something is the rule, or a law of nature, and saying that something can or has happened, or that it is an exception to a general rule. This question only concerns the latter.

I do not have the answer. But at the same time, it seems absurd to say that we would never be justified in knowing that it can sometimes happen, no matter the data, just because it generally does not happen.