The paper aims to show that turning to the original works of William Whewell, the author of the hypothetico-deductive model of substantiation of scientific knowledge, can force us to return once again to the conversation about the content of the inference leading to knowledge. Positivists have consolidated what Larry Laudan calls «consequentialism» – the idea that the only significant form of evidential support for a theory is the empirical confirmation of its consequences. And this generally corresponds to the declared logic of the project – to focus on the logical aspects of the analysis of knowledge and to shift the focus from explaining phenomena to confirming hypotheses. In this sense, turning to the history of science, the analysis of W. Whewell’s ontology of knowledge, can be considered as a rejection of the absoluteness of the positivist’s vision and the basis for restoring the discussion of the «metaphysical» premises of the inference, but, obviously, taking into account the results obtained by the philosophy of science over the past one and a half century. The main idea is not only to focus attention on (a) which elements of Whewell’s scheme of hypothesis confirmation were rejected by positivists in the course of forming the now canonical interpretation of the hypothetico-deductive model, but also to emphasize (b) the independence of W. Whewell’s original concept of constructing scientific theories, part of which is literally the «meta-physical» interpretation of inference to explanation. W. Whewell’s theory of knowledge combines three types of inferences – induction, deduction and abduction – and as an example of a project for describing and theoretically generalizing the idea of a scientific theory that corresponds to the great era of scientific discoveries of the late 19th century, sets the task of answering questions of how a phenomenon occurs and why it occurs. In particular, abduction in W. Whewell’s theory did not imply the choice between the best hypothesis, – his criteria for the truth of the explanatory hypothesis are aimed at additional verification and correction of the hypothesis already accepted at the first, hypothetico-deductive, stage. This reflects the idea of the dynamics of the scientific process – it provides the opportunity not only to test the generalizations that have been made, but also to make adjustments to the theory based on new evidence.