Why couldn’t accounts in natural science simply change ad infinitum? Perhaps the very nature of natural science is that it approaches fixed truth though the attainment of it involves contradiction. The notion of truth as unchangeable would be borrowed from some higher science and inappropriate to natural science. There are, in fact, several very good arguments that this is in fact the case:
1.) From the definition of nature. Nature is a ratio of the divine art, so that things might move and achieve their own ends. It is thus a way of taking part in the divine intellect, and to understand it requires an intellect that is proportioned to understanding such a thing. But the human mind is not such, at least not in its present sense. The same argument can be made by understanding nature as creation.
Note that the argument applies only to the attaining of the nature ultimately, with an exhaustive knowledge of what is innermost in it, where “innermost” is understood as the unifying source of all observable phenomena. Such a term is a point of contact with the divine action. But a failure to attain to this term (or to attain to it with human knowledge as human) does not eliminate the possibility of the mind attaining some final truths and knowing them as such. But it does require that the advance in the natural sciences will more and more alienate it from any sort of final knowledge. Paradoxically, though in a familiar enough way, the reduction of more and more phenomena to a single theory will involve more and more pluralism in the details of the theory; and the approach to the truth by negation of false theories will involve a greater plurality of explanatory theories.
2.) From way we know the proper object of natural science. The history of natural science is of a gradual separation from the sensible: Aristotle based his whole theory on the properly sensible; Newton and Galileo shifted science to the imaginable magnitudes of Euclidean space; and the last great revolution showed the inadequacy of quantity as imagined. The connection to the sensible has become more and more remote from the empirically verifiable, even while natural science can never wholly separate itself from empirical verification. Based on the progression of the sciences so far, a final or perfect science would require transcending the very principle that constitutes the science itself – a pretty straightforward contradiction.
Notice that the difficulty is not with the sensible as such, but that we know the sensible by sensation and yet desire to give an intellectual account of it.
3.) From the imperfect intelligibility of the object itself. The objects of natural science all have an intrinsic principle of unintelligibility. All the complaints against Aristotle’s account of form in the Metaphysics – Geach somewhere says that Aristotle rarely says one thing about it that he doesn’t deny somewhere else – arise from matter being a real principle lacking intelligibility in itself. The dual role of matter, namely that it enters into the essence of a thing (and so must be in the definition) even while it is unintelligible and in this sense cannot enter into the definition, requires a double approach to the principles of natural being. The clarity of categories will diminish as we advance in knowledge of the sensible.
With human beings, the unintelligibility of matter also arises from another principle – that we know by abstraction. Our knowledge consists in being acted upon, and only forms can do such. Scotus, however, goes too far in saying that this is all there is to the unintelligibility of matter.