The psychologist, now, stands before a like problem. The mental world, no less than the material, comes to us in the gross; mental phenomena are complex, often highly complex; we must reduce them to their elements, we must keep analysing till we can analyse no further, if we are to describe them in a scientific way. And here too =synthesis= goes hand in hand with analysis. Psychology, to be sure, does not write graphic formulas; but psychology has to show how its elements go together, to discover the laws of their connection; we shall find that tones and colours go together in very different ways. All the while that we are tearing a bit of our world apart, and finding its elements, we are trying to put those elements back again in their places and to reconstruct the original experience. Synthesis, unfortunately, is often very difficult; and you must notice that a _failure to reconstruct does not necessarily mean that the preceding analysis was wrong_. A chemist may analyse a given substance into a certain number of elements, each one represented by a certain number of atoms; yet if he puts these elements together again, in the right proportions, he may—perhaps because he is now working at a different temperature—come out with another substance of different properties. His analysis was not therefore wrong; but his attempt at synthesis is a failure because he has not taken account of all the relevant circumstances. It may happen similarly in psychology that we do not know all the relevant circumstances; or it may happen that we know them but cannot control them; in such cases we cannot reconstruct. The only thing to do is then to make analysis its own test; we analyse again and again; and if the result is always the same, we are satisfied to let it stand. Children who do not know how to prove an example in arithmetic follow the same plan; if they get the same answer several times over, and if their schoolmate gets that answer too, they are satisfied; and when the work has been honestly done, the agreement is pretty good evidence that they are right. Notice one other point: that if you sit down to describe, there is simply no escape from analysis. _To begin a description is to be analysing._ Well-meaning people sometimes shake their heads at scientific psychology; all this dissecting work, they say, misses the real issue; it kills mind; it destroys the living, breathing reality of experience, and offers in its place a catalogue of dead facts. The mannikin again! Of course, if mind is a little man inside you, you must kill him to dissect him,—though he nevertheless crops up again, alive and well, after the autopsy. The mannikin, as we have seen, cannot face cold logic. No, the task of science is to describe; if you are to describe you must analyse; and the results are every bit as real as the unanalysed experience. Dead facts? But a fact is the most live thing possible; it will survive any number of theories, and will still give birth to more.