Some ideas had a connection traceable by reason; others were only connected by 'chance and custom.' Association does not explain reasoning, only the deviations from reasoning. But with Hume and Hartley the relation is inverted. The principle, instead of being an exceptional case, is simply the universal rule from which logical connection may be deduced as a special case. The facts upon which Mill relied, and the account of them which he gave, require notice and embodiment in any sound psychology. In some shape or other they form the starting-point of all later systems. Mill's vigorous application of his principle, worked out with imperfect appreciation and with many oversights, had therefrom, at least, the merit of preparing the ground for a more scientific method. In any case, however, his conclusions, so far as sound, must be placed in a different framework of theory. It becomes necessary to dwell chiefly upon the curious defects of his theory, if taken as he wished it to be taken, for an ultimate scientific statement. The fact that there is a synthesis and an analysis is expressed by 'association.' But what more can we say? What are the 'laws' of association? Unless some rule can be given, we shall get nothing that can be called a theory. One idea is not suggested by the other through any logical process. They are still 'conjoined' but not 'connected.' The connection, therefore, must be given by something different from the ideas themselves. Now the order of the original 'sensations' depends upon the 'objects of nature,' and is therefore left to 'physical philosophy.'[516] They occur, however, either in 'synchronous' or in 'successive' order. Then 'ideas' spring up in the order of 'sensations,' and this is the 'general law of association of ideas.'[517] The synchronous sensations produce synchronous ideas and the successive sensations successive ideas. Finally, the strength of the association between the ideas depends upon 'the vividness of the associated feelings, and the frequency of the association.'[518] Hume had said that association depended upon three principles, 'contiguity in time and place,' 'causation,' and 'resemblance.' Contiguity in time corresponds to the successive, and contiguity in place to the synchronous, order. Causation, as Brown had finally proved,[519] means simply antecedence and consequence. 'Resemblance' remains and is, as Mill afterwards says,[520] a most important principle; but in an unlucky moment he is half inclined to reduce even 'resemblance' to 'contiguity.'[521] Resemblance is, he even suggests, merely 'a case of frequency,' because we generally see like things together. When we see one tree or sheep, we generally see several trees or sheep. J. S. Mill mildly remarks upon this quaint suggestion as the 'least successful simplification' in the book. He argues the point gravely. Sheep, it is clear, are not seen to be like because they often compose a flock, but