medieval speculation. It need not concern us here.

Taking the form of the species for granted, common-sense biology proceeds to ask how it comes to be in a given instance, say

in the dog Toby. Before this dog was born or thought of, his form or species was displayed in each of his parents. And now it looks as though the form of dog had detached itself from them through the generative act, and set up anew on its own account. How does it do that? By getting hold of some materials in which to express itself. At first it takes them from the body of the mother, afterwards it collects them from a wider environment, and what the dog eats becomes the dog.

What, then, is the relation of the assimilated materials to the dog-form which assimilates them? Before assimilation, they have their own form. Before the dog eats the leg of mutton, it has the form given to it by its place in the body of a sheep. What happens to the mutton? Is it without remainder transubstantiated from sheep into dog? It loses all its distinctively sheep-like characteristicsm but there may be some more basically material characteristics which it preserves. They underlay the structure of the mutton, and they continue to underlie the structure of the dog's flesh which supplants it. Whatever these characteristics may be, let us call them common material characteristics, and let us say that they belong to or compose a common material nature.

The common material nature has its own way of existing, and perhaps its own principles of physical action. We may suppose that we know much or that we know little about it. This one thing at least we know, that it is capable of becoming alternatively either mutton or dog's flesh. It is not essential to it to be mutton, or mutton it would always be; nor dog's flesh, or it would always be dog's flesh. It is capable of becoming either, according as it is captured by one or other system of formal organization. So the voters who are to go to the polls are, by their common nature, Englishmen; they are essentially neither Socialist curs nor Conservative sheep, but intrinsically capable of becoming either, if they become captured by either system of party organization.

According to this way of thinking, there is a certain looseness about the relation of the common material nature to the higher forms of organization capable of capturing it. Considered in itself alone, it is perhaps to be seen as governed by absolutely determined laws of its own. It is heavy, then it will fall unless obstructed; it is solid, then it will resist intrusions. But considered as material for organization by higher forms, it is indeterminate. It acts in one sort of way under the persuasion of the sheep-form, and

in another sort of way under the persuasion of the dog-form, and we cannot tell how it will act until we know which form is going to capture it. No amount of study bestowed on the common material nature will enable us to judge how it will behave under the persuasion of the higher organizing form. The only way to discover that is to examine the higher form itself.

Every form, then, will really be the object of a distinct science. The form of the sheep and the form of the dog have much in common, but that merely happens to be so; we cannot depend upon it, or risk inferences from sheep to dog: we must examine each in itself; we shall really need a science of probatology about sheep, and cynology about dogs. Again, the common material nature has its own principles of being and action, so it will need a science of itself, which we may call hylology. Each of these sciences is mistress in her own province; but how many there are, and how puzzlingly they overlap! So long as we remain within the province of a single science, we may be able to think rigorously, everything will be 'tight'. But as soon as we consider border-issues between one province and another, farewell to exactitude: everything will be 'loose'. We can think out hylology till we are blue in the face, but we shall never discover anything about the entry of material elements into higher organizations, or how they behave when they get there. We may form perfect definitions and descriptions of the form of the dog as such, and still derive no rules for telling what elements of matter will enter into the body of a given dog or how they will be placed when they do. All we can be sure of is, that the dog-form will keep itself going in, and by means of, the material it embodies—unless the dog dies. But what happens to the matter in the body of the dog is 'accidental' to the nature of the matter; and the use of this matter, rather than of some other equally suitable, is accidental to the nature of the dog.

No account of material events can dispense with accidental relations altogether. We must at least recognize that there are accidental relations between particular things. Accident in the sense of brute fact had to be acknowledged even by the tidiest and most dogmatic atomism of the last century. That atomism must allow it to be accidental, in this sense, that the space surrounding any given atom was occupied by other atoms in a given manner. It belonged neither to the nature of space to be occupied by just those atoms in just those places, nor to the nature of the atoms to

be distributed just like that over space; and so in a certain sense the environment of any atom was an accidental environment. That is, the particular arrangement of the environment was accidental. The nature of the environment was not accidental at all. It was proper to the nature of the atom to be in interaction with other atoms over a spatial field, and it never encountered in the fellow-denizens of space any other nature but its own. It was not subject to the accident of meeting strange natures, nor of becoming suddenly subject to strange or unequal laws of interaction. All interactions, being with its own kind, were reciprocal and obedient to a single set of calculable laws.

But the medieval philosophy had asserted accidental relations between distinct sorts of natures, the form of living dog and the form of dead matter, for example. No one could know a priori what effect an accidental relation would produce, and all accidental relations between different pairs of natures were different: at the most there was analogy between them. Every different nature had to be separately observed, and when you had observed them all, you could still simply write an inventory of them, you could not hope to rationalize your body of knowledge. Let us narrow the field and consider what this doctrine allows us to know about the wood of a certain kind of tree. We shall begin by observing the impressions it makes on our several senses, and we shall attribute to it a substantial form such as naturally to give rise to these impressions, without, perhaps, being so rash as to claim a knowledge of what this substantial form is. Still we do not know what its capacities of physical action and passion may be. We shall find them out by observing it in relation to different 'natures'. It turns out to be combustible by fire, resistant to water, tractable to the carpenter's tools, intractable to his digestive organs, harmless to ostriches, nourishing to wood-beetles. Each of these capacities of the wood is distinct; we cannot relate them intelligibly to one another, nor deduce them from the assumed fundamental 'woodiness'.

We can now see why 'substantial forms' were the bêtes noires of the seventeenth-century philosophers. It was because they turned nature into an unmanageable jungle, in which trees, bushes, and parasites of a thousand kinds wildly interlaced. There was nothing for it, if science was to proceed, but to clear the ground and replant with spruce in rows: to postulate a single uniform nature, of which there should be a single science. Now neither probatology