ABSTRACT
The logic of analogy, and of the formation of hypotheses, is not tight; and no definite rules can be laid down for it. But this does not mean that there is no logic at all; and the paradigm accepted among scientists can guide towards helpful analogies. The success of hypotheses which unified masses of facts into a coherent structure leads to them being called theories; and has meant that in the twentieth century we have become much more tolerant of hypotheses than Newton was. But what we do demand of them is that they are testable; hypotheses which are not can be described as ‘ad hoc\ invented to save an awkward situation, or as metaphysical, and scientists may use such things but would hate to admit it.
