ABSTRACT
In trying to discuss the problem of universals, one immediately runs into the initial difficulty, that we are using a number of technical terms, or pairs of correlative technical terms, which tend to be defined in terms of each other. The difficulty of avoiding such circularities has suggested to some philosophers that the reason why the problem of universals has not been satisfactorily solved even after two thousand years is that there is really no problem to solve. There are three cases of universals, which appear to have no actual instances, but with which we are perfectly familiar. Similarity is held to be a universal in the traditional sense because, although we can define other universals in terms of it, we cannot define it in terms of itself, and therefore we cannot avoid treating it as a universal of which there are instances, i.e. as a realist universal either in the Aristotelian or in the Platonic sense.
