In partial or mixed association we have all along supposed the interesting portion of the disappearing thought to be of considerable extent, and to be sufficiently complex to constitute by itself a concrete object. Sir William Hamilton relates, for instance, that after thinking of Ben Lomond he found himself thinking of the Prussian system of education, and discovered that the links of association were a German gentleman whom he had met on Ben Lomond, Germany, etc. The interesting part of Ben Lomond, as he had experienced it, the part operative in determining the train of his ideas was the complex image of a particular man. But now let us suppose that that selective agency of interested attention, which may thus convert impartial redintegration into partial association—let us suppose that it refines itself still further and accentuates a portion of the passing thought, so small as to be no longer the image of a concrete thing, but only of an abstract quality or property. Let us moreover suppose that the part thus accentuated persists in consciousness (or, in cerebral terms, has its brain-process continue) after the other portions of the thought have faded. This small surviving portion will then surround itself with its own associates after the fashion we have already seen, and the relation between the new thought's object and the object of the faded thought will be a relation of similarity. The pair of thoughts will form an instance of what is called 'Association by Similarity.'[484]
The similars which are here associated, or of which the first is followed by the second in the mind, are seen to be compounds. Experience proves that this is always the case. There is no tendency on the part of simple 'ideas,' attributes, or qualities to remind us of their like. The thought of one shade of blue does not remind us of that of another shade of blue, etc., unless indeed we have in mind some general purpose like naming the tint, when we should naturally think of other blues of the scale, through 'mixed association' of purpose, names, and tints, together. But there is no elementary tendency of pure qualities to awaken their similars in the mind.
We saw in the chapter on Discrimination that two compound things are similar when some one quality or group of qualities is shared alike by both, although as regards their other qualities they may have nothing in common. The moon is similar to a gas-jet, it is also similar to a foot-ball; but a gas-jet and a foot-ball are not similar to each other. When we affirm the similarity of two compound things, we should always say in what respect it obtains. Moon and gas-jet are similar in respect of luminosity, and nothing else; moon and foot-ball in respect of rotundity, and nothing else. Foot-ball and gas-jet are in no respect similar—that is, they possess no common point, no identical attribute. Similarity, in compounds, is partial identity. When the same attribute appears in two phenomena, though it be their only common property, the two phenomena are similar in so far forth. To return now to our associated representations. If the thought of the moon is succeeded by the thought of a foot-ball, and that by the thought of one of Mr. X's railroads, it is because the attribute rotundity in the moon broke away from all the rest and surrounded itself with an entirely new set of companions—elasticity, leathery integument, swift mobility in obedience to human caprice, etc.; and because the last-named attribute in the foot-ball in turn broke away from its companions, and, itself persisting, surrounded itself with such new attributes as make up the notions of a 'railroad king,' of a rising and falling stock-market, and the like.
The gradual passage from impartial redintegration to similar association through what we have called ordinary mixed association may be symbolized by diagrams. Fig. 41 is impartial redintegration, Fig. 42 is mixed, and Fig. 43 similar association. A in each is the passing, B the coming thought. In 'impartial,' all parts of A are equally operative in calling up B. In 'mixed,' most parts of A are inert. The part M alone breaks out and awakens B. In 'similar,' the focalized part M is much smaller than in the previous case, and after awakening its new set of associates, instead of fading out itself, it continues persistently active along with them, forming an identical part in the two ideas, and making these, pro tanto, resemble each other.
Fig. 41. 
Fig. 42. 
Fig. 43. Why a single portion of the passing thought should break out from its concert with the rest and act, as we say, on its own hook, why the other parts should become inert, are mysteries which we can ascertain but not explain. Possibly a minuter insight into the laws of neural action will some day clear the matter up; possibly neural laws will not suffice, and we shall need to invoke a dynamic reaction of the form of consciousness upon its content. But into this we cannot enter now.
To sum up, then, we see that the difference between the three kinds of association reduces itself to a simple difference in the amount of that portion of the nerve-tract supporting the going thought which is operative in calling up the thought which comes. But the modus operandi of this active part is the same, be it large or be it small. The items constituting the coming object waken in every instance because their nerve-tracts once were excited continuously with those of the going object or its operative part. This ultimate physiological law of habit among the neural elements is what runs the train. The direction of its course and the form of its transitions, whether redintegrative, associative, or similar, are due to unknown regulative or determinative conditions which accomplish their effect by opening this switch and closing that, setting the engine sometimes at half-speed, and coupling or uncoupling cars.
This last figure of speech, into which I have glided unwittingly, affords itself an excellent instance of association by similarity. I was thinking of the deflections of the course of ideas. Now, from Hobbes's time downward, English writers have been fond of speaking of the train of our representations. This word happened to stand out in the midst of my complex thought with peculiarly sharp accentuation, and to surround itself with numerous details of railroad imagery. Only such details became clear, however, as had their nerve-tracts besieged by a double set of influences—those from train on the one hand, and those from the movement of thought on the other. It may possibly be that the prepotency of the suggestions of the word train at this moment were due to the recent excitation of the railroad brain-tract by the instance chosen a few pages back of ii railroad king playing foot-ball with the stock-market.
It is apparent from such an example how inextricably complex are all the contributory factors whose resultant is the line of our reverie. It would be folly in most cases to attempt to trace them out. From an instance like the above, where the pivot of the Similar Association was formed by a definite concrete word, train, to those where it is so subtile as utterly to elude our analysis, the passage is unbroken. We can form a series of examples. When Mr. Bagehot says that the mind of the savage, so far from being in a state of nature, is tattooed all over with monstrous superstitions, the case is very like the one we have just been considering. When Sir James Stephen compares our belief in the uniformity of nature, the congruity of the future with the past, to a man rowing one way and looking another, and steering his boat by keeping her stern in a line with an object behind him, the operative link becomes harder to dissect out. It is subtler still in Dr. Holmes's phrase, that stories in passing from mouth to mouth make a great deal of lee-way in proportion to their headway; or in Mr. Lowell's description of German sentences, that they have a way of yawing and going stern-foremost and not minding the helm for several minutes after it has been put down. And finally, it is a real puzzle when the color pale-blue is said to have feminine and blood-red masculine affinities. And if I hear a friend describe a certain family as having blotting-paper voices, the image, though immediately felt to be apposite, baffles the utmost powers of analysis. The higher poets all use abrupt epithets, which are alike intimate and remote, and, as Emerson says, sweetly torment us with invitations to their inaccessible homes.
In these latter instances we must suppose that there is an identical portion in the similar objects, and that its brain-tract is energetically operative, without, however, being sufficiently isolable in its activity as to stand out per se, and form the condition of a distinctly discriminated 'abstract idea.' We cannot even by careful search see the bridge over which we passed from the heart of one representation to that of the next. In some brains, however, this mode of transition is extremely common. It would be one of the most important of physiological discoveries could we assign the mechanical or chemical difference which makes the thoughts of one brain cling close to impartial redintegration, while those of another shoot about in all the lawless revelry of similarity. Why, in these latter brains, action should tend to focalize itself in small spots, while in the others it fills patiently its broad bed, it seems impossible to guess. Whatever the difference may be, it is what separates the man of genius from the prosaic creature of habit and routine thinking. In Chapter XXII we shall need to recur again to this point.