The one point of philosophical interest about this doctrine appears alike in the scheme of the "Categories" in the presence of a category of "substance," and in the list of "Predicaments" in the sharp distinction drawn between "definition" and "proprium." From a logical point of view it does not appear why any proprium, any character belonging to all the members of a class and to them alone, should not be taken as defining the class. Why should it be assumed that there is only one predicate, viz. man, which precisely answers the question, "What is Socrates?" Why should it not be equally correct to answer, "a Greek," or "a philosopher"? The explanation is that Aristotle takes it for granted that not all the distinctions we can make between "kinds" of things are arbitrary and subjective. Nature herself has made certain hard and fast divisions between kinds which it is the business of our thought to recognise and follow. Thus according to Aristotle there is a real gulf, a genuine difference in kind, between the horse and the ass, and this is illustrated by the fact that the mule, the offspring of a horse and an ass, is not capable of reproduction. It is thus a sort of imperfect being, a kind of "monster" existing contra naturam. Such differences as we find when we compare e.g. Egyptians with Greeks do not amount to a difference in "kind." To say that Socrates is a man tells me what Socrates is, because the statement places Socrates in the real kind to which he actually belongs; to say that he is wise, or old, or a philosopher merely tells me some of his attributes. It follows from this belief in "real" or "natural" kinds that the problem of definition acquires an enormous importance for science. We, who are accustomed to regard the whole business of classification as a matter of making a grouping of our materials such as is most pertinent to the special question we have in hand, tend to look upon any predicate which belongs universally and exclusively to the members of a group, as a sufficient basis for a possible definition of the group. Hence we are prone to take the "nominalist" view of definition, i.e. to look upon a definition as no more than a declaration of the sense which we intend henceforward to put on a word or other symbol. And consequently we readily admit that there may be as many definitions of a class as it has different propria. But in a philosophy like that of Aristotle, in which it is held that a true classification must not only be formally satisfactory, but must also conform to the actual lines of cleavage which Nature has established between kind and kind, the task of classificatory science becomes much more difficult. Science is called on to supply not merely a definition but the definition of the classes it considers, the definition which faithfully reflects the "lines of cleavage" in Nature. This is why the Aristotelian view is that a true definition should always be per genus et differentias. It should "place" a given class by mentioning the wider class next above it in the objective hierarchy, and then enumerating the most deep-seated distinctions by which Nature herself marks off this class from others belonging to the same wider class. Modern evolutionary thought may possibly bring us back to this Aristotelian standpoint. Modern evolutionary science differs from Aristotelianism on one point of the first importance. It regards the difference between kinds, not as a primary fact of Nature, but as produced by a long process of accumulation of slight differences. But a world in which the process has progressed far enough will exhibit much the same character as the Nature of Aristotle. As the intermediate links between "species" drop out because they are less thoroughly adapted to maintain themselves than the extremes between which they form links, the world produced approximates more and more to a system of species between which there are unbridgeable chasms; evolution tends more and more to the final establishment of "real kinds," marked by the fact that there is no permanent possibility of cross-breeding between them. This makes it once more possible to distinguish between a "nominal" definition and a "real" definition. From an evolutionary point of view, a "real" definition would be one which specifies not merely enough characters to mark off the group defined from others, but selects also for the purpose those characters which indicate the line of historical development by which the group has successively separated itself from other groups descended from the same ancestors. We shall learn yet more of the significance of this conception of a "real kind" as we go on to make acquaintance with the outlines of First Philosophy. Over the rest of the formal logic of Aristotle we must be content to pass more rapidly. In connection with the doctrine of Propositions, Aristotle lays down the familiar distinction between the four types of proposition according to their quantity (as universal or particular) and quality (as affirmative or negative), and treats of their contrary and contradictory opposition in a way which still forms the basis of the handling of the subject in elementary works on formal logic. He also considers at great length a subject nowadays commonly excluded from the elementary books, the modal distinction between the Problematic proposition (x may be y), the Assertory (x is y), and the Necessary (x must be y), and the way in which all these forms may be contradicted. For him, modality is a formal distinction like quantity or quality, because he believes that contingency and necessity are not merely relative to the state of our knowledge, but represent real and objective features of the order of Nature.
In connection with the doctrine of Inference, it is worth while to give his definition of Syllogism or Inference (literally "computation") in his own words. "Syllogism is a discourse wherein certain things (viz. the premisses) being admitted, something else, different from what has been admitted, follows of necessity because the admissions are what they are." The last clause shows that Aristotle is aware that the all-important thing in an inference is not that the conclusion should be novel but that it should be proved. We may have known the conclusion as a fact before; what the inference does for us is to connect it with the rest of our knowledge, and thus to show why it is true. He also formulates the axiom upon which syllogistic inference rests, that "if A is predicated universally of B and B of C, A is necessarily predicated universally of C." Stated in the language of class-inclusion, and adapted to include the case where B is denied of C this becomes the formula, "whatever is asserted universally, whether positively or negatively, of a class B is asserted in like manner of any class C which is wholly contained in B," the axiom de omni et nullo of mediæval logic. The syllogism of the "first figure," to which this principle immediately applies, is accordingly regarded by Aristotle as the natural and perfect form of inference. Syllogisms of the second and third figures can only be shown to fall under the dictum by a process of "reduction" or transformation into corresponding arguments in the first "figure," and are therefore called "imperfect" or "incomplete," because they do not exhibit the conclusive force of the reasoning with equal clearness, and also because no universal affirmative conclusion can be proved in them, and the aim of science is always to establish such affirmatives. The list of "moods" of the three figures, and the doctrine of the methods by which each mood of the imperfect figures can be replaced by an equivalent mood of the first is worked out substantially as in our current text-books. The so-called "fourth" figure is not recognised, its moods being regarded merely as unnatural and distorted statements of those of the first figure.
Induction.--Of the use of "induction" in Aristotle's philosophy we shall speak under the head of "Theory of Knowledge." Formally it is called "the way of proceeding from particular facts to universals," and Aristotle insists that the conclusion is only proved if all the particulars have been examined. Thus he gives as an example the following argument, "x, y, z are long-lived species of animals; x, y, z are the only species which have no gall; ergo all animals which have no gall are long-lived." This is the "induction by simple enumeration" denounced by Francis Bacon on the ground that it may always be discredited by the production of a single "contrary instance," e.g. a single instance of an animal which has no gall and yet is not long-lived. Aristotle is quite aware that his "induction" does not establish its conclusion unless all the cases have been included in the examination. In fact, as his own example shows, an induction which gives certainty does not start with "particular facts" at all. It is a method of arguing that what has been proved true of each sub-class of a wider class will be true of the wider class as a whole. The premisses are strictly universal throughout. In general, Aristotle does not regard "induction" as proof at all. Historically "induction" is held by Aristotle to have been first made prominent in philosophy by Socrates, who constantly employed the method in his attempts to establish universal results in moral science. Thus he gives, as a characteristic argument for the famous Socratic doctrine that knowledge is the one thing needful, the "induction," "he who understands the theory of navigation is the best navigator, he who understands the theory of chariot-driving the best driver; from these examples we see that universally he who understands the theory of a thing is the best practitioner," where it is evident that all the relevant cases have not been examined, and consequently that the reasoning does not amount to proof. Mill's so-called reasoning from particulars to particulars finds a place in Aristotle's theory under the name of "arguing from an example." He gives as an illustration, "A war between Athens and Thebes will be a bad thing, for we see that the war between Thebes and Phocis was so." He is careful to point out that the whole force of the argument depends on the implied assumption of a universal proposition which covers both cases, such as "wars between neighbours are bad things." Hence he calls such appeals to example "rhetorical" reasoning, because the politician is accustomed to leave his hearers to supply the relevant universal consideration for themselves.
Theory of Knowledge.--Here, as everywhere in Aristotle's philosophy, we are confronted by an initial and insuperable difficulty. Aristotle is always anxious to insist on the difference between his own doctrines and those of Plato, and his bias in this direction regularly leads him to speak as though he held a thorough-going naturalistic and empirical theory with no "transcendental moonshine" about it. Yet his final conclusions on all points of importance are hardly distinguishable from those of Plato except by the fact that, as they are so much at variance with the naturalistic side of his philosophy, they have the appearance of being sudden lapses into an alogical mysticism. We shall find the presence of this "fault" more pronouncedly in his metaphysics, psychology, and ethics than in his theory of knowledge, but it is not absent from any part of his philosophy. He is everywhere a Platonist malgré lui, and it is just the Platonic element in his thought to which it owes its hold over men's minds.
Plato's doctrine on the subject may be stated with enough accuracy for our purpose as follows. There is a radical distinction between sense-perception and scientific knowledge. A scientific truth is exact and definite, it is also true once and for all, and never becomes truer or falser with the lapse of time. This is the character of the propositions of the science which Plato regarded as the type of what true science ought to be, pure mathematics. It is very different with the judgments which we try to base on our sense-perceptions of the visible and tangible world. The colours, tastes, shapes of sensible things seem different to different percipients, and moreover they are constantly changing in incalculable ways. We can never be certain that two lines which seem to our senses to be equal are really so; it may be that the inequality is merely too slight to be perceptible to our senses. No figure which we can draw and see actually has the exact properties ascribed by the mathematician to a circle or a square. Hence Plato concludes that if the word science be taken in its fullest sense, there can be no science about the world which our senses reveal. We can have only an approximate knowledge, a knowledge which is after all, at best, probable opinion. The objects of which the mathematician has certain, exact, and final knowledge cannot be anything which the senses reveal. They are objects of thought, and the function of visible models and diagrams in mathematics is not to present examples of them to us, but only to show us imperfect approximations to them and so to "remind" the soul of objects and relations between them which she has never cognised with the bodily senses. Thus mathematical straightness is never actually beheld, but when we see lines of less and more approximate straightness we are "put in mind" of that absolute straightness to which sense-perception only approximates. So in the moral sciences, the various "virtues" are not presented in their perfection by the course of daily life. We do not meet with men who are perfectly brave or just, but the experience that one man is braver or juster than another "calls into our mind" the thought of the absolute standard of courage or justice implied in the conviction that one man comes nearer to it than another, and it is these absolute standards which are the real objects of our attention when we try to define the terms by which we describe the moral life. This is the "epistemological" side of the famous doctrine of the "Ideas." The main points are two, (1) that strict science deals throughout with objects and relations between objects which are of a purely intellectual or conceptual order, no sense-data entering into their constitution; (2) since the objects of science are of this character, it follows that the "Idea" or "concept" or "universal" is not arrived at by any process of "abstracting" from our experience of sensible things the features common to them all. As the particular fact never actually exhibits the "universal" except approximately, the "universal" cannot be simply disentangled from particulars by abstraction. As Plato puts it, it is "apart from" particulars, or, as we might reword his thought, the pure concepts of science represent "upper limits" to which the comparative series which we can form out of sensible data continually approximate but do not reach them.
In his theory of knowledge Aristotle begins by brushing aside the Platonic view. Science requires no such "Ideas," transcending sense-experience, as Plato had spoken of; they are, in fact, no more than "poetic metaphors." What is required for science is not that there should be a "one over and above the many" (that is, such pure concepts, unrealised in the world of actual perception, as Plato had spoken of), but only that it should be possible to predicate one term universally of many others. This, by itself, means that the "universal" is looked on as a mere residue of the characteristics found in each member of a group, got by abstraction, i.e. by leaving out of view the characteristics which are peculiar to some of the group and retaining only those which are common to all. If Aristotle had held consistently to this point of view, his theory of knowledge would have been a purely empirical one. He would have had to say that, since all the objects of knowledge are particular facts given in sense-perception, the universal laws of science are a mere convenient way of describing the observed uniformities in the behaviour of sensible things. But, since it is obvious that in pure mathematics we are not concerned with the actual relations between sensible data or the actual ways in which they behave, but with so-called "pure cases" or ideals to which the perceived world only approximately conforms, he would also have had to say that the propositions of mathematics are not strictly true. In modern times consistent empiricists have said this, but it is not a position possible to one who had passed twenty years in association with the mathematicians of the Academy, and Aristotle's theory only begins in naturalism to end in Platonism. We may condense its most striking positions into the following statement. By science we mean proved knowledge. And proved knowledge is always "mediated"; it is the knowledge of conclusions from premisses. A truth that is scientifically known does not stand alone. The "proof" is simply the pointing out of the connection between the truth we call the conclusion, and other truths which we call the premisses of our demonstration. Science points out the reason why of things, and this is what is meant by the Aristotelian principle that to have science is to know things through their causes or reasons why. In an ordered digest of scientific truths, the proper arrangement is to begin with the simplest and most widely extended principles and to reason down, through successive inferences, to the most complex propositions, the reason why of which can only be exhibited by long chains of deductions. This is the order of logical dependence, and is described by Aristotle as reasoning from what is "more knowable in its own nature,"[#] the simple, to what is usually "more familiar to us," because less removed from the infinite wealth of sense-perception, the complex. In discovery we have usually to reverse the process and argue from "the familiar to us," highly complex facts, to "the more knowable in its own nature," the simpler principles implied in the facts.
[#] This simple expression acquires a mysterious appearance in mediæval philosophy from the standing mistranslation notiora naturæ, "better known to nature."
It follows that Aristotle, after all, admits the disparateness of sense-perception and scientific knowledge. Sense-perception of itself never gives us scientific truth, because it can only assure us that a fact is so; it cannot explain the fact by showing its connection with the rest of the system of facts, "it does not give the reason for the fact." Knowledge of perception is always "immediate," and for that very reason is never scientific. If we stood on the moon and saw the earth, interposing between us and the sun, we should still not have scientific knowledge about the eclipse, because "we should still have to ask for the reason why." (In fact, we should not know the reason why without a theory of light including the proposition that light-waves are propagated in straight lines and several others.) Similarly Aristotle insists that Induction does not yield scientific truth. "He who makes an induction points out something, but does not demonstrate anything."
For instance, if we know that each species of animal which is without a gall is long-lived, we may make the induction that all animals without a gall are long-lived, but in doing so we have got no nearer to seeing why or how the absence of a gall makes for longevity. The question which we may raise in science may all be reduced to four heads, (1) Does this thing exist? (2) Does this event occur? (3) If the thing exists, precisely what is it? and (4) If the event occurs, why does it occur? and science has not completed its task unless it can advance from the solution of the first two questions to that of the latter two. Science is no mere catalogue of things and events, it consists of inquiries into the "real essences" and characteristics of things and the laws of connection between events.
Looking at scientific reasoning, then, from the point of view of its formal character, we may say that all science consists in the search for "middle terms" of syllogisms, by which to connect the truth which appears as a conclusion with the less complex truths which appear as the premisses from which it is drawn. When we ask, "does such a thing exist?" or "does such an event happen?" we are asking, "is there a middle term which can connect the thing or event in question with the rest of known reality?" Since it is a rule of the syllogism that the middle term must be taken universally, at least once in the premisses, the search for middle terms may also be described as the search for universals, and we may speak of science as knowledge of the universal interconnections between facts and events.