The second book of Aristotle's Physics lays out his four causes. The most important of these to the natural scientist, he says, is the final cause, which is the purpose or end that causes a change. The eighth chapter deals with a question especially pertinent to the modern perspective: what is to say that change has a purpose at all? That is, it is obvious that it does not rain so that crops will grow, it just rains anyway (due to water vapour condensing in the atmosphere) and coincidentally, crops use this rain to grow; it would rain anyway. Or, if the tide comes in and gets your feet wet, it is not said that the tide came in in order to wet your feet. That your feet are now damp is a coincidental result of the tides (which are caused by lunar gravity) and more directly caused by your foolish insistence on standing at the beach for hours on end.
(Going by Aristotle's theory of four causes, that is only the efficient cause; the material, formal and final causes are separate. The material cause, for example, is the water from which the sea is made.)
Aristotle claims this is not the case for natural things, which are pretty much living things. The previous question might be phrased, "Why are front teeth sharp, as if for biting and tearing, and back teeth flat, as if for crushing?". Might this not be also a coincidence, as he points out:
Why should there be any purpose behind this? Why should it not just be an accident? And the same question could be asked of any other part of the body that appears to have a purpose.
Another example would be: must your heart be the way it is for the purpose of pumping blood, or is it just that way anyway, and by coincidence ends up powering your circulation? He continues with a striking remark:
So where every part turned out just as it would have been if it had had some purpose, the creatures survived because, spontaneously [a], they happened to be put together in a useful way. But everything else has been destroyed and continues to be destroyed, as Empedocles [b] says of his "cow like creatures with the heads of men".
On the face of it, Aristotle here appears pretty much to lay out the Darwinian theory of natural selection. He uses a pretty absurd image to make the point, some sort of cow-griffin hybrid, but it is fundamentally this: why can we not suppose that random chance does in fact play a part in generating the form of living things, and that the unfit products of this process aren't found because they're all dead and gone?
At its core, this is the Darwinian hypothesis that random changes in genes create more or less fit variations on a species, and the more fit variations outcompete and outbreed the less fit, but stated backwards and in more crude terms of destruction and generation.
Infuriatingly, Aristotle goes on to reject this theory on the grounds that, since teeth "always or usually" grow in the same way, and the heart always or usually grows to be able to circulate blood, and so on for all other natural things; their forms therefore cannot be spontaneous (i.e. random). From an earlier definition of chance and spontaneity in the same book of the Physics he points out something can either have a purpose or be spontaneous, and since we've decided these "always or usually" events are not random, they must have a purpose.
What I think is missing here is that the hypothetical random event does not have to happen on the same "stage", so to speak, as the growth of the organism. I feel like I'm cheating a bit here, with two thousand or so years of intervening science on Aristotle feeling like a bit of an unfair advantage, so without having to come up with it myself I can comfortably propose that the safe passage of genes from parents to offspring sits in the "usually" part of "always or usually" (he actually raises this later, see next quotation), and this aberration on the original formal cause of the parent creates a new formal cause the child now grows into.
The problem with that explanation is that it verges on contradicting with the use of ascribing a formal cause to the animal in the first place. The formal cause really should be like the species or genus of animal we want to talk about, e.g. so that we can always talk about "dogs" as one thing, which is rendered very unhelpful if every animal with a slight mutation must sit in its own species.
In principle that problem can be dealt with by considering the nature of the heirarchy of forms at play, however, so I would not consider it insurmountable.
Aristotle also notes out how for living things the formal, efficient and final causes all share the same form (a man grows as he does because he has the form of a man, and this form is what causes him to grows as a man, perhaps think of DNA here, and the end of his growth will produce a man) and in this case the change in formal cause indicates a change in final, but not efficient cause. This is because the efficient cause isn't retroactively changed by aberration, although you could consider it added to or expanded since we'd want to include whatever caused the change in form, which would presumably be some genetic mechanism.
The upshot being that the final cause, the purpose, of the natural thing in question — of the part that is for the purpose of crushing food or whole that (let's be uncontroversial) is towards the end of reproduction — doesn't have to remain the same across every generation, that's the point here.
Now, if it is possible for there to be products of human skill which correctly serve some purpose, and mistakes in this province constitute failed attempts at some purpose, then the same should go for natural things too, and monstrosities would constitute failures to achieve that natural purpose. In the beginning, then, any combinations like those "cow-like creatures", which were incapable of achieving some definite end, must have arisen because of some defect in their source, just as defective seed is responsible for the births of such creatures nowadays.
[...]
Moreover, chance would also have to have been a factor in their seeds [sources]. But this idea totally subverts any notion of nature and natural things. The point is that those things are natural which undergo continuous change, starting from an intrinsic source of change and concluding at a particular end. Starting from a given source of change does not result in the same end in every case, but it is not just any chance end either; there is in fact always a tendency towards the same end, unless something intervenes.
He's right here, of course, and later compares a thing's nature or purpose to a doctor healing himself — it's what keeps that thing on the path set for it. However, we can see (again with a lot of work done for us in the intervening centuries) that genetic mechanisms can in fact intervene in the manner he describes, and constitute the causes of the "monstrosities", which I think we can fairly identify with genetic abnormalities like growing an extra finger, or being born with only one lung — but the monstrosities can randomly (that is, by chance) end up being better at reproduction, or fitter, than the originally more natural organism of which they would have otherwise emulated the form!
Ultimately, all Aristotle was missing was a less rigid concept of "destruction", in the form of competition, and a theory of genetics. Altogether rather small conceptual leaps given the overall scope of his work.
Addenda
Aristotelian Chance
It's important to define exactly what Aristotle means when he talks of "chance" and "spontaneity" (let's call the category containing both of them "randomness"), because he does give them specific meanings that may differ from what you or I instinctively think of when we see them.
Randomness according to Aristotle is a coincidental cause of something that does have a purpose. What this means is that, when something happens to some end, for some purpose, we might ordinarily expect the cause of it to have also had the purpose in mind. So, someone who builds a house sets out with the goal of building a house, it doesn't just fall into place that way when he chucks down some wood and bricks.
However, there are many times where something happens that does in fact end up fulfilling some purpose but this is totally coincidental as a cause. So maybe our foreman builds his house, and the presence of the house blocks your view of whatever's behind it. Unless he's particularly mean, the builder probably did not intend this as a consequence of the house, but, nevertheless, events have transpired such that it fulfils that purpose.
This is chance, according to Aristotle, and spontaneity is the same sort of argument but applied to domains where speaking of the thought or intent behind something is inappropriate i.e. nature; maybe the wind blows a tree down, which falls on and destroys this house, but the wind did not blow for the purpose of restoring your view, but had this effect coincidentally.
So above, when I said that the monstrosities can end up fitter than the natural thing by chance, hopefully it's clear now I meant they end up that way (fitter), but by coincidence, since that was not the purpose of their being monstrosities, but rather a coincidental result of it.
Nature, 1908
When I was about finished with this page I thought I had better make sure I was writing something at least halfway original, and searched the topic online. I didn't carry out an exhaustive search or anything, but I didn't find much and it convinced me this isn't a widely talked-about topic. There is a short article from 1908 that appears in the journal Nature, by one F. A. Dixey, which appears to be more concerned with setting the record straight against some people who I assume were advocating the view that Aristotle in fact did propose Darwin's theory rather than reject it.
Notes
a:
The word that is translated as "spontaneously" seems to be very
close in meaning to our "randomly", or "arbitrarily", if it were
not for that arbitrariness requires arbitration of some kind that
natural processes cannot make. See the first addendum for more.
b:
Presocratic philosopher/mystic who lived about a century before
Aristotle.