Michael Huemer has a post from a few weeks back, Theory of the Four Elements Finally Refuted!; it's partly tongue-in-cheek, and I take it that the real point of the post is at the tend, to raise the question of what in our view of the physical world might eventually go the way of the theory of four elements. But it does raise some questions -- and I think some common errors -- with regard to how the theory of the four elements worked, and I'd like to look briefly at them.
Huemer raises a few questions, but before getting to them, there are a few general principles that need to be considered.
A. The elements are not explanatory in and of themselves; they are the possible combinations of active and passive dispositions (historically known as wet, dry, hot, and cold) that are (very roughly) kinds of attraction and repulsion. For the most part, it's these active and passive dispositions that actually do the explanatory work when we are talking about changes. The primary exception is when we are dealing with the contrast between light and heavy, which is a phenomenon that in the theory of the four elements is primarily treated at the level of elements, being a very complicated matter of how a body is composed, how its surrounding bodies are composed, and where the bodies are in the universe.
B. Contrary to what is usually assumed, and is certainly assumed by Huemer, elements are not immutables. This is seen in two ways. First, you can make one element act somewhat like another element by introducing the right kind of active and passive dispositions extrinsically. For instance, you can make water act more like air if you introduce a lot of 'dry' action; you can make it act more like earth if you introduce a lot of 'cold' and 'dry' action. It still is water with its natural tendency, but you've modified its behavior. Second, if you do this enough, you actually turn one element into another. In Aristotle's account of the elements, in fact, elements turning into each other is a fairly common occurrence. Any element can turn into any element, but some alterations are easier than others -- e.g., it's easier to get air to change into fire than to get it to change into earth; it's easier to get earth to change into water than into fire.
C. While there is a link to our experience, the pure elements themselves are never actually experienced, but only inferred based on the possible combinations of dispositions. Everything we actually experience has a mixture of elements; wind is mostly air, but it will also have a lot of earth, water, and fire in it, while the water you drink is mostly water, but it will also have a lot of air, fire, and earth in it.
D. How elemental motion works differs a lot depending on which version of the theory of elements you are considering. Any element can be moved in any direction by an external cause. In some later accounts, every element has an inherent actual motion to its natural place in the universe. In Aristotle's own account, on the other hand, it seems clear that elements only move when something moves them -- earth has a natural tendency to be at the center, so it (arguably) has a preferentiality for centerward movement and (certainly) will only naturally come to a rest at the center, but it doesn't have a natural tendency to move to the center, and will only do so if something actually moves it in that direction, and if something stops it, it will not head toward the center unless something moves it again. Other theorists, again, have somewhat different accounts; it's one of the more disputed areas of the theory.
Given these, let's look at Huemer's questions, and how a generic theorist of the four elements might approach them.
Where Do Trees Come From?
The rigidity of trees obviously comes from the element of earth; but Huemer errs in assuming that the only source of earth is the soil. In fact, by (C), air and rain are also sources of earth. Most wood floats on water; it is quite clear that this is because wood has a lot of air in it. (In the case of woods that don't float on water -- ebony is the usual example given in the historical texts -- the air is said to have evaporated off, leaving a greater proportion of earth than in most woods.) Fire would also explain it, and there is certainly some fire in the wood, but as one increases the amount of fire in something, the less stable it becomes, so this suggests that we should give a preference to air over fire in our account of the composition of wood. The theory of the four elements is not a rigorously predictive theory, but in many ways a classificatory one; that is to say, it doesn't tell us what will happen, per se, but on its assumptions, it can tell us what the possibilities are. So it would not be quite right to say that the theory would predict that the tree gets its earth-element from the ambient air, but this is a genuine possibility, and we have reason to think that the tree is in fact getting a lot of its composition from the ambient air, because it has not just a lot of earth but a lot of air as well. The elemental theorist would also recognize rain as a possible source -- but it is usually sporadic, and so could not explain most of the composition -- and the soil as a perhaps more likely possible source; nothing in the theory would exclude either of these explanations, because again, the theory just gives you the possibilities, not the actual course of events. But given further facts about the composition of the wood, it is easy to argue that a considerable portion of its composition comes from the ambient air.
Why Is There Fire Down Here?
This is an easy question to answer: because there is fire everywhere in the universe. By (B), throughout the universe other elements are turning into fire due to various causal actions. Further, by (D), fire is shunted and pushed in all sorts of different directions, although fire's particular active and passive dispositions mean that it's harder to make deviate than other elements, and therefore this mechanism is perhaps less often an explanation for fire than for the other elements. Nonetheless, even in the highest regions of fire, fire can crash into fire and shove it downward quite forcefully; for instance, the sun by the force of its influence is a candidate for constantly pushing fire down onto the earth (although for a number of reasons, elemental theorists would usually have preferred an explanation of the first type, in which the sun is always through its influence turning everything in our world a little bit fiery).
What Are Ice and Steam?
Also relatively easy to answer. We are never dealing with pure water-element, so 'water' as we normally think of it is a very water-rich mixture of all four elements. Ice is when this composite acts as if it were more earth-rich and steam is when this composite acts as if it were more air-rich; by (A) and (B) this happens when external action on the composite temporarily changes the active and passive dispositions of the whole. The water-element itself will stay the same, and the water-composite will mostly stay water (although its proporitions may shift somewhat), but the new active and passive qualities it receives from outside itself will make the water-composite act either more earth-like or more air-like than the water itself will be able to explain. Remove this external action, and slowly the water-composite will converge back to a state of action and passion that is primarily explained by its constituent water.
Where's the Evidence?
This is a much trickier question. The evidence for the four elements is simply that these are the possible combinations of active and passive dispositions:
Hot Dry (fire)
Hot Wet (air)
Cold Wet (water)
Cold Dry (earth)
Thus the primary evidence for the theory is not concerned with the elements themselves, but with the active and passive dispositions. As noted in (A), the active and passive dispositions do the major explanatory work. Despite the names, hot, cold, dry, and wet are not here understood to be sensible qualities but (very roughly) kinds of attraction and repulsion. I say 'very roughly' because they actually aren't pure attraction and repulsion, but have something to do whether the attraction/repulsion is for the similar or the different, and the precise formulation of how this works is a difficult issue. But the point is that when we are talking about evidence for the active and passive dispositions, we can interpret this as talking about evidence for the physical changes we find the world being explained by attractions and repulsions, cohesions and disruptions. And of course that evidence is quite good, and we still do explain physical changes this way, at least to a great degree, although the way we organize our understanding of these attractions and repulsions, cohesions and disruptions, has massively changed. You can pick up a rock and recognize that some of its properties in changes are due to the fact that it holds together fairly well, and that others are due to the fact that it is not very active in affecting its environment; you can study flames and see that it is in some ways the opposite, being very active in disrupting its environment and not having the same kind of coherence that rock does.
It's perhaps best to think of it less as a matter of evidence than as a matter of explanatory need, although there is some connection between the two. There are physical changes in the world; these are clearly not at random, so there has to be something that explains them; physical changes are clearly not unlimited, so the explanation has to work in some way combinatorially; physical changes clearly affect how things hold together or break apart, and this is what one would expect given combinations of a limited number of explanatory factors, so the explanatory factors have to have something to do with attraction, cohesion, repulsion, disruption, and similar things. All of this up to this point clearly applies just as much to the Standard Model of modern physics as it does to the theory of four elements. It is meeting these criteria that make it possible for both the theory of four elements and the Standard Model to be explanatory of physical changes. The evidence for both the theory of four elements and the Standard Model will largely, then, overlap -- it will consist of ordinary physical changes, and the fact that the theory has some account of them. There are very precise experiments that you can do to argue that the Standard Model is better than the elemental theory, but for the most part the evidence for both is that they can give you an account of ordinary things like 'this rock heats up and eventually melts when in flames' and 'hitting this metal with a hammer makes it bend rather than splatter'. It's not glamorous or startling evidence, but it's a lot of evidence, and if the Standard Model regularly failed to give us accounts of these ordinary things of experience, we would not consider it a good theory, because it wouldn't be. The Standard Model is a better theory not because it does something entirely different but because, among the things it does are all the elemental theory does, but reconceived so as to be more useful and cover more phenomena.
But the other fact involved comes down to practicality, and the most practical aspect of explanation is that you can't explain anything with a theory you don't have. What other non-elemental theories were actually on the table for most of its run? None, really. People recognized that there were a lot of unanswered questions about the theory of the four elements -- you have only to look at the history of people talking about it to see that they had a lot of puzzles. But to get an alternative you need something that meets the desiderata but understands the combinations of attractions, etc., in a completely different way. Through the ancient and medieval periods, people proposed lots and lots of modifications of the theory of elements to try to deal with the various puzzles that kept coming up, but they couldn't find anything that was fundamentally different. In the early modern period, there was some attempt to try to find ways of doing this entirely with locomotion; all those attempts failed. Getting the theory we have now required working out theories of forces, of chemical composition, of energy, of electromagnetism, and of fields over a period of several centuries. And if you went to physicists today and said, "OK, suppose you are as wrong about the Standard Model as the ancients and medievals were about the four elements; what then?", physicists would know that an alternative would still have to do all the general kinds of things that the Standard Model does, but in a completely different way, and would be somewhat at a loss about where to go from there. You're not going to be able just to pull it out of a few considerations off the top of your head.
So was there evidence for the theory of four elements? Yes, lots of it. Was it of a kind that could distinguish between the theory of four elements and some alternative? Mostly no, but there wasn't any known alternative until very late. Was it such as to remove everything puzzling about the theory? Not at all, but that's not surprising.
When we look at how the theory of four elements worked, the shape of the theory is the right general shape. The primary limitation was in how the attractions, etc., were understood, which allowed for only limited precision, being kind of fiddly and difficult to measure, and for a flexibility that, while considerable along some lines, was very restricted along others. The secondary limitation is that physical changes were, considered in themselves, treated as entirely distinct from locomotive ones; our current theory relates the two very differently. A result of overcoming these limitations is that we get a theory that both does better at classifying the possibilities for change (which the elemental theory did fairly well for a much more limited range of phenomena) and at predicting how changes would likely go (which the elemental theory did not really do except in a very generic way). Obviously these are improvements. But even allowing for its limitations in precision and flexibility, the theory of the four elements had the right general kinds of features to explain the phenomena it was put forward to explain.