We can, in our thoughts, separate Laws of Nature from the Facts which conform to them. When we do this, the Law is represented by the Ideas and Conceptions which it involves. Thus the Law of a Planet's motion round the Sun, as to space, is represented by the conception of an Ellipse, the Sun being its Focus. Laws so abstracted from Facts are Theories.
(William Whewell, Elements of Morality Including Polity, section 9.)
In other words, to translate roughly out of Whewell's own phil. sci. jargon into something nearer that used today, a theory is an abstract modeling of a set of facts. 'Theory' in this sense is, according to Whewell, closely related to what we always do in understanding anything, so it bears no hints or suggestions of tentativeness or weakness. (The only place any sort of tentativeness would enter in is if you had good reason for thinking the motion might not properly be modeled by an ellipse at all; even if it were an imperfect ellipse, correcting the model to accommodate these perfections more perfectly wouldn't suggest any tentativeness about the corrected theory, just a degree of idealization and approximation, which is different altogether.) And, indeed, it probably would not have at the time at all; most of Whewell's readers would have had at least some Greek and Latin under their belt, so would probably have recognized the Greek root of the word 'theory' and associated with higher-level contemplation or thought. A lot of Whewell's contributions to and clarifications of scientific terminology were excellent in this way; when Faraday wrote to him asking him for better names for eastodes and westodes, Whewell suggested 'anodes' and 'cathodes' as less confusing (again, this would have made sense for most educated people who, unlike the largely self-taught Faraday, would usually have had some Greek). One of Whewell's philosophical concerns was putting scientific terminology in order so that it not only optimized usefulness for scientists themselves, but so that it also optimized clarity (and minimized confusion) for the interested public. The service he performed in this way is immense, although largely unacknowledged. It would be nice if there were someone capable of doing this today; but alas, we philosophers only rarely produce the likes of Whewell, and if we did today, I'm not sure scientists would be very pleased to have him scolding them about the words they used. He certainly wouldn't have the respect Whewell had with people like Faraday and Maxwell. And he would have a harder task, anyway, since there is less of a useful common linguistic ground like Greek and Latin for building such terminology. A troubling issue, that; but it doesn't seem we can do much about it.