I think the implication is not that both X and ~X provide evidence for the hypothesis, but rather something like, "yes, there are a few exceptions to the rule, but if you look at what the exceptions are they're so unusual that they just underline the fact that the rule is generally (though not universally) applicable."