These, then, are the reasons for my conviction that universals must be immanent in their particulars, if the case for universals and particulars is to depend upon the argument from similarity. I see no reason from the argument to suppose that there might not be universals that are never particularized, nor, for that matter, particulars that are never characterized. But the argument certainly does carry with it the implication that if a universal is particularized, that universal is immanent and manifested fully in the particular.

I should like now to add a few further consequences, which may account in part for the rather surprising absence of a flat-footed realistic theory of the immanence of the universal and the bareness of the particular. This realistic combination proves to be untenable, and any philosopher who carried its consequences through would simply abandon the theory. My thesis then is that this is the only realistic theory countenanced by the argument from similarity, and that this theory is untenable.

What are the undesirable consequences? The first is that only entities capable of functioning in perception as the grounds for similarity immanent in a number of particulars are genuine subsistent entities, or universals. Redness might be a genuine universal but not color nor the character of being a scarlet tanager. Color is not a character that can be immanent in any particular. It is the name of a limited disjunction of characters. Color is redness or orangeness, or yellowness or greenness or, etc. Two things said to be similar in that both are colored are, consequently, not necessarily similar at all. One may be yellow and the other blue. There is no paradox of similarity in one particular being yellow and the other blue, and accordingly no applicability of the argument from similarity, and no proof that color is a universal. Color is adequately accounted for as a name for a certain disjunction of characters.

The character of being a scarlet tanager is not a genuine universal for another reason. Different particular scarlet tanagers are genuinely similar, to be sure, in contrast to different colors which are not. But the similarity of scarlet tanagers is not based on the discrimination of a single character of scarlet-tanagerness but of a set of characters -red and a certain shape and a certain sound and, etc. Scarlettanagerness is a name for a certain conjunction of characters. Manifestly, there would be no way of proving by the argument from similarity that a conjunction of discriminated universals was itself a genuine universal. But more than that, by the argument such a conjunction of universals could not be a genuine universal. For if

a genuine compound universal could be perceived in two particulars as an immanent simple common character, which would be the only proof of its genuineness by the argument from similarity, and if it should subsequently be discovered that the character was not simple but compound-not just red but perhaps red-and-hard-what guarantee would there be that in the earlier perception the two particulars supposed similar were not really different in character, the one red and the other hard?

Hence no compound entities can be genuine universals in a tenable theory based on the argument from similarity. This consequence considerably depopulates the realm of subsistence, but is not in itself very alarming. It prepares the ground, however, for another consequence that is devastating.

The second consequence is that relations cannot relate anything by this theory. They cannot relate genuine universals in subsistence and so develop compound universals which could be particularized as a whole in existence, because, as we have just seen, genuine compound universals are not compatible with the argument from similarity. Neither can any means be found of making relations relate particulars. Clearly, by our theory, they could only relate particulars by the process of particulars participating in them. Suppose we say that S is a relation in which the particulars a and b participate, as a result of which a and b become related through S. But this desired result will not eventuate. This S, which we hoped would be a relation, would turn out to be indistinguishable from any simple quality. For the multiple participation of two or more particulars in one universal is our hypothesis of similarity; so that the participation of a and b in S does not relate them, but only makes them similar. All we have succeeded in showing is that a is S and that b is S, not that a Sb. S, then, would necessarily be a quality and not a relation.

How could we definitely distinguish the particularization of a relation from that of a quality? There seems to be only one means open to us. We might say that in the transaction between particulars and a relation, two or more particulars participate in one subsistent entity in one participation, whereas in the transaction between particulars and qualities only one particular can participate in one subsistent entity in one participation. The distinction between qualities and relations is thus tracked down to a distinction between two sorts of participations. Straight participations indicate that the subsistent entities are qualities, forked participations indicate that

they are relations. The trouble with this distinction is that there is no provision for different sorts of participations in the argument from similarity. Moreover, if per impossibile in the immanent view of the transaction demanded by the argument a difference of character could be discriminated between two sorts of participations, then the characters discriminated would automatically become subsistent entities and the participations themselves particulars, and the structural rôle of these participations in the system would be dissipated. And the new participations which permitted the old participations to be qualitied would be homogeneous. Or if these also were said to be qualitied, these by the same argument would be dissolved into particulars and characters, demanding for them some genuine participations and so to infinity with no hope of ever attaining a set of heterogeneous participations. Of course, the point is that the argument from similarity provides for and permits of homogeneous participations only.

If now some obstinate realist insists that the very phrase 'homogeneous participations' implies that participations have similar characters and if similar why not just as easily dissimilar, my reply is, 'Very well! splendid! but if you insist that any participation is to be analytically broken apart, like a perception and in accordance with the argument from similarity, into a particular and a universal, then your theory is wrecked at the start, for you can never bring particular and universal together again. I am inclined to agree with you, but I was going to grant you this point for good measure. I did not want to sink your ship at the launching. I wanted to let her float for you, so you could see for yourself she was unseaworthy.'

And I think our discussion exhibiting the impossibility of relations in such a theory, conclusively demonstrates that the theory is unseaworthy. We find in such a theory that relations cannot directly relate either universals or particulars, nor can they be brought to relate these entities indirectly by any manipulation of the transaction of participation. Relations cannot be brought into the system by any means. Yet as insistent facts of experience they cannot be overlooked.

At the beginning of this paper I asserted that the theory we have been considering could be met in three ways: by developing another

"This is virtually W. E. Johnson's procedure (Logic, Pt. I, ch. XIII). A "characterising tie" alone is straight participation, a “characterising tie" plus a coupling tie" is a forked participation. These two "kinds" (?) of 'ties" are of course, not themselves relations (cf. p. 212).

theory of similarity, by finding this theory internally inconsistent, and by finding that this theory leads to consequences incompatible with experience. I am willing to grant provisionally that this theory is or could be made internally consistent. But the exclusion of relations as perceptible elements of experience is a consequence of this theory surely incompatible with a fact of experience at least as insistent as that of the perception of similarity. My conclusion is that another theory of similarity, even one not so neat and inviting as this one, is much in demand and heralded by the inadequacy of this one.

UNIVERSITY OF CALIFORNIA.

STEPHEN C. Pepper.

REVIEWS OF BOOKS

Der Eleatische Satz vom Widerspruch. Von SVEND RANULF. Christiania, Gyldendalske Boghandel, 1924.-pp. 222.

=

The thesis presented in this work is that all pre-Aristotelian thinking, including that of Socrates and Plato, was dominated by what the author terms the "logic of absolute ambiguity." According to this logic, a general concept is constructed, not by taking the common characteristics of the species subsumed under it, but by adding together the whole of the characteristics possessed by all the subordinate classes. Thus if A1 = abcd, A2 abef, As abgh, the concept A will connote the characteristics abcdefgh, and not ab alone as it would according to the Aristotelian logic. It is the use of this logic which accounts for the extraordinarily large number of fallacies —or arguments that we regard as fallacious according to Aristotelian logic-in pre-Aristotelian thought. The famous antinomies of Zeno, and the paradox of the Liar, are only better known and more striking illustrations of fallacies arising from the employment of this logical principle. In seven of the earlier Platonic dialogues which the author examines minutely, "there is hardly to be found a single conclusion which does not involve a fallacy," from the standpoint of Aristotle. Even the Platonic ideas are constructed according to the logic of absolute ambiguity. Finally the author appeals to M. Lévy-Bruhl's researches in primitive mentality, and finds that this principle of pre-Aristotelian logic corresponds to what Professor Lévy-Bruhl calls the "law of participation" which dominates the thinking of primitives.

There is no space here to criticize this, in parts suggestive, but on the whole over-strained interpretation of Greek thought. The prevalence of fallacious arguments in pre-Aristotelian thought can be explained, it would seem, as due either to the recognition and use of a different logic, or to the incorrect employment, owing to lurking ambiguities of language, of recognized logical principles. All the arguments that the author so painstakingly analyzes can be explained on the second hypothesis, which is inherently the more probable. The contribution of Aristotle to Greek logic consists, not in substituting a new logic for an older and different one, but in stating clearly, and