The statistical hypothesis about probability asserts so little that it cannot be falsified. One could, however, falsify the assertion that the probability of ‘tails’ is one by just tossing one ‘head’; but not if a lower probability is asserted. One could construct mathematical models for all strange sequences, e.g. a regular series 100, 100, 100, 1..... for millions of times, and then suddenly it goes malignant and alters type. Could have a mathematical law covering this behaviour.
This illustrates that a merely statistical hypothesis can never be falsified. It is not scientific, but is metaphysical. It becomes scientific only when we adopt an attitude towards it in order to falsify it. One must use (as in physics) a statistical hypothesis in order to deduce physical effects from it, which can in their turn be tested. Let us take the statistical theory of light. There is a statistical bombardment of photons. Luminosity is just a measurement of the probability of the hits of photons on that point, hence one can deduce effects of relations of distance from source to brightness, or angular relationship, etc. Since these light hits are irregular there is always the probability that the photons will miss the area for a time, etc. The rule in the conversion of a statistical law into a physical law is to convert it into the production of mass effects, and test these effects without the possibility of retreating back into statistical law. If you don’t do this, you can explain anything and therefore nothing.