According to epistemic requirement of scientific realism the truths – or near truths – about objects posited by scientific theories should be knowable. However, the optimistic view that a scientific theoretical system can be shown to be true is blocked by the familiar Hume’s arguments against induction. The paper shall not deal with them primarily but only marginally in order to compare Hume’s conclusions with the Duhem-Quine thesis. For the common reading of this thesis is that it shows the powerlessness of negative instances to disprove scientific theories, just as Hume’s critical arguments against induction have shown the powerlessness of positive instances to prove scientific theories. The paper aims to expose erroneous aspects of the analogy and to explain what the errors imply for the epistemic requirement of scientific realism, even if it is weakened from knowability of truths to knowability of near truths.