Comment by JoeAltmaier

Go further: reproducible experiment can tease out details to many decimal places of likelihood. Something can be known so thoroughly that we can make statements like "this is certain to behave as predicted millions or billions of times more often that not", which is pretty close to certain knowledge.