Any critical reader of Karl Popper's The Logic of Scientific Di.scove<y (1934) must be struck, once Section 19 is reached, by the parallel between the description of conventionalism and Popper's own methods. If conventionalism consists in procedures taken to save a theory, then the 1972 edition of this work appears as an elaborate saving operation [ 11. It is full of stars: new footnotes, new appendices, as well as the most recent addenda. It is remarkable that this operation began even before the publication of the first edition in 1934, especially for the weakest part of Popper's theory, and that which posed the greatest problems for him, the application to probabilistic scientific theories: " For although probability statements play such a vitally important role in empirical science, they turn out to be in principle impervious to strict falsification " (p. 146).
Popper would perhaps object that his work is philosophical and not scientific, and hence that his own criticisms are not applicable.
So be it. However, two essential questions stand out: Why is such an elaborate saving operation necessary? And is it successful?
In this paper, a demonstration that Popper's theory of empirical falsifiability is not logically consistent is attempted. This demonstration begins with that self-admitted weakest link, i.e., probability theories, and then proceeds to show that, on Popper's own terms, his theory does not hold even for deterministic scientific theories. On the other hand, Popper's critique of induction, especially in its logical positivist form, is accepted as valid; it is not affected by this demonstration.
However, Popper equates such an inductive approach by verification, with all positivism. More common definitions of positivism, for example, the four points listed by Kolakowski (1958) most definitely include Popper's theory of empirical falsifiability as a member.
In order to place this critique in historical context, Popper's work is compared to that of another "philosopher of science". An elaborated theory of falsifiability of scientific theories was proposed at least a dozen years before Popper by Fisher (1922) for that specific case which gave so much trouble to Popper, probabilistic theories. To my knowledge, this predecessor