ABSTRACT
At first sight, this claim seems to be justified as a matter of simple logic. Let ‘T’ stand for a scientific theory, and ‘O’ for a statement of the observable results of some test-procedure. If we can deduce O from T, and show that O is false, it follows logically that T is also false. If, however, O is shown to be true, it does not follow that T is true: thus the impossibility of conclusive verification. But several problems arise when this simplified logical model of falsification is applied to the actual process of testing a scientific theory. The degree of certainty with which we can assert the falsity of T depends upon the corresponding degree of certainty of the falsity of O. Logically, all that can be shown is that if O is false, then so is T: any uncertainty about the former is necessarily reflected in the latter. Further, what is in fact deducible from a theory is not a statement describing the observable results of a test-procedure, but a hypothetical, or conditional statement, asserting that if the relevant procedure is carried out, such results will occur. Failure to get the predicted results does not, therefore, directly falsify the theory, since it is always possible that the test-procedures have not been carried out satisfactorily.
