previous state of mind in which they were brought together; and endeavours to explain various cases (as, for example, association by 'contrast' as well as by 'likeness' or 'continuity') by a more 'subtile' analysis.[491] He then works out an elaborate theory of 'simple' and 'relative' suggestion. Simple 'suggestion'[492] corresponds mainly to ordinary association, as when a friend's name or his book calls up the thought of the man himself. 'Relative suggestion' arises when two or more objects are perceived and suggest various relations of likeness and so forth.[493] This provides a scheme for working out the whole doctrine of the sequences of ideas so far as the sequences depend upon the mind itself and not upon external causes. It thus leads to problems of abstraction and generalisation and to his whole theory of what he calls the 'intellectual states.' He again closely coincides with the French ideologists. He starts by examining Locke and Condillac. He of course professes to hold that Condillac's version of Locke is illegitimate, and ridicules the famous formula _penser c'est sentir_. He is, however, equally unwilling to admit Reid's 'variety of powers.'[494] In fact, his criticism of Condillac shows more affinity than contrast. Condillac erred, he says, in holding that thoughts are 'transformed sensations.' This was a false simplification into which he considers Condillac to have been led partly by the ambiguity of the word _sentir_.[495] Condillac applied to the mind the theory, true in 'the chemistry of the material chemists,' that the 'compounds are the elements themselves.'[496] He errs when he infers from the analogy that a feeling which arises out of others can be resolved into them. 'Love and hate' and other emotions are fundamentally different from the sensations by which they are occasioned, not mere 'transformations' of those sensations. We, on the other hand (that is to say, Reid and Stewart), have erred by excessive amplification. Instead of identifying different things, we have admitted a superfluous number of 'ultimate principles.'