On of my favorite works by C. S. Lewis is his last published academic book, The Discarded Image: An introduction to Medieval and Renaissance Literature. The book was published in 1964, the year following his death on November 1963.

Part of the book features some interesting thoughts on Science:

The business of the natural philosopher is to construct theories which will ‘save appearances’…. A scientific theory must ‘save’ or ‘preserve’ the appearances, the phenomena, it deals with, in the sense of getting them all in, doing justice to them. Thus, for example, your phenomena are luminous points in the night sky which exhibit such and such movements in relation to one another and in relation to an observer at a particular point, or various chosen points, on the surface of the Earth. Your astronomical theory will be a supposal such that, if it were true, the apparent motions from the point or points of observation would be those you have actually observed. The theory will then have ‘got in’ or ‘saved’ the appearances.

But, if we demanded no more than that from a theory, science would be impossible, for a lively inventive faculty could devise a good many different supposals which would equally save the phenomena. We have therefore had to supplement the canon of saving the phenomena by another canon– first, perhaps, formulated with full clarity by Occam. According to this second canon we must accept (provisionally) not any theory which saves the phenomena but that theory which does so with the fewest possible assumptions. Thus the two theories (a) that the bad bits in Shakespeare were all put there by adapters, and (b) that Shakespeare wrote them when he was not at his best, will equally ‘save’ the appearances. But we already know that there was such a person as Shakespeare and that writers are not always at their best. If scholarship hopes ever to achieve the steady progress of the sciences, we must therefore (provisionally) accept the second theory. If we can explain the bad bits without the assumption of an adapter, we must.

In every age it will be apparent to accurate thinkers that scientific theories, being arrived at in the same way I have described, are never statements of fact. That the stars appear to move in such and such ways, or that substances behaved thus and thus in the laboratory– these are the statements of fact. The astronomical or chemical theory can never be more than provisional. It will have to be abandoned if a more ingenious person thinks of a supposal which would ‘save’ the observed phenomena with still fewer assumptions, or if we discover new phenomena which it cannot save at all.

This would, I believe, be recognized by all thoughtful scientists today. It was recognized by Newton if, as I am told, he wrote not ‘the attraction varies inversely as the square of the distance’, but ‘all happens as if’ it is so varied. It was certainly recognized in the Middle Ages. ‘In astronomy’, says Aquinas, ‘an account is given of eccentrics and epicycles on the ground that if their assumption is made (hac positione facta) the sensible appearances as regards to celestial motions can be saved. But this is not a strict scientific proof (sufficienter probans) since for all we know (forte) they could also be saved by some different assumption.’ The real reason why Copernicus raised no ripple and Galileo raised a storm, may well be that whereas the one offered a new supposal about celestial motions, the other insisted on treating this supposal as fact. If so, the real revolution consisted not in a new theory of the heavens but in ‘a new theory of the nature of theory’.

On the highest level, then, the Model was recognized as provisional. What we should like to know is how far down the intellectual scale this cautious view extended. In our age I think it would be fair to say that the ease with which a scientific theory assumes the dignity and rigidity of fact varies inversely with the individual’s scientific education. In discussion with wholly uneducated audiences I have sometimes found matter which real scientists would regard as highly speculative more firmly believed than many things within our real knowledge. The _imago_ of the Cave Man ranked as hard fact, and the life of Caesar or Napoleon as doubtful rumour. We must not, however, hastily assume that the situation was quite the same in the Middle Ages. The mass media which have in our time created a popular scientism, a caricature of the true sciences, did not then exist. The ignorant were more aware of their ignorance then than now.

p. 14 – 17