That ad hoc hypotheses are both repugnant and useful is a known fact.
The joke of the biologist about teleology, that like a mistress, one wants to have it but not be seen with it, is more characteristic of ad hoc hypotheses. And, indeed, for mechanists, every teleological hypothesis is terribly ad hoc; Spinoza called teleology the shelter of ignorance because it was ad hoc.
When Newton said his gravity was not occult he argued from the fact that it was not ad hoc but a powerful explanation. Copernicus was indignant about Ptolemy’s epicycles but had some himself, of course. The reduction of purpose to cause is ad hoc all too often. And Newton’s optics was ad hoc, as William Whewell argued at great length. Now Copernicus’ complaint is not in itself unreasonable: Ptolemy’s epicycles were old hands and gained the legitimacy of regular customers, whereas his own were stop-gaps. This idea can be generalized.It is clear that one epicycle in Copernicus’ system had a special privileged position, namely that of the moon - as Sir Francis Bacon observed, not knowing whether this was quite kosher. Now, clearly, whether it is so, and more specifically, if yes why, is easier to feel than to express.
Let us assume that what is rational is what has a general rationale rather than has no better cause than a whim. Now the ad hoc is neither due to a whim nor due to a general rationale. And so it can be either irrational, an excuse to stick to a rationale which turns out to be less general than originally assumed, or non-rational, either in being an exception or in being a sui generis or anything in between, e.g. a case of an unstudied territory.
The general rationale is at times a scientific theory, at times (and more often) a metaphysics. Let us take them separately beginning with scientific theory.
I.