Leibniz's criticism upon these statements of Locke is brief and to the point. He admits Locke's definition of knowledge, qualifying it, however, by the statement that in much of our knowledge, perhaps in all that is merely empirical, we do not know the reason and connection of things and hence cannot be said to _perceive_ the agreement or disagreement of ideas, but only to feel it confusedly. His most important remark, however, is to the effect that relation is not a special kind of knowledge, but that all Locke's four kinds are varieties of relation. Locke's "connection" of ideas which makes knowledge is nothing but relation. And there are two kinds of relation,--those of "comparison" and of "concourse." That of comparison states the identity or distinction of ideas, either in whole or in part. That of concourse contains Locke's two classes of co-existence and existence. "When we say that a thing really exists, this existence is the predicate,--that is to say, a notion connected with the idea which is the subject; and there is connection between these two notions. The existence of an object of an idea may be considered as the concourse of this object with me. Hence comparison, which marks identity or diversity, and concourse of an object with me (or with the _ego_) are the only forms of knowledge." Leibniz leaves the matter here; but he only needed to develop what is contained in this statement to anticipate Berkeley and Kant in some of the most important of their discoveries. The contradiction which lies concealed in Locke's account is between his definition of knowledge in general, and knowledge of real existence in particular. One is the agreement or disagreement of _ideas_; the other is the agreement of an idea _with an object_. Berkeley's work, in its simplest form, was to remove this inconsistency. He saw clearly that the "object" was an intruder here. If knowledge lies in the connection of _ideas_, it is impossible to get outside the ideas to find an object with which they agree. Either that object is entirely unknown, or it is an idea. It is impossible, therefore, to find the knowledge of reality in the comparison of an idea with an object. It must be in some property of the ideas themselves.