This, of course, raises difficulties when we try to describe the facts in words, since words stand for abstractions and carry their logical implications. All descriptions in words of what is non-logical are bound to be a mass of contradictions, for, having applied any word it is necessary immediately to guard against its logical implications by adding another which contradicts them. Thus we say our experience is of facts, and must then hastily add that nevertheless they are not plural, and we must further qualify this statement by adding that neither are they singular. A description of what is non-logical can only convey its meaning if we discount all the logical implications of the words which, for want of a better medium of expression, we are driven to employ. Our whole intellectual bias urges us towards describing everything that comes within our experience, even if the description is only for our own private benefit Unfortunately the language in which these descriptions have to be expressed is so full of logical implications that, unless we are constantly on our guard, we are liable to be carried away by them, and then, at once, we lose contact with the actual facts. In order to get round this almost universal tendency to confuse abstractions with facts Bergson sometimes tries to get us to see the facts as they actually are by using metaphor instead of description in terms of abstract general notions. He has been much criticised for this but there is really a good deal to be said for attempting to convey facts by substituting metaphors for them rather than by using the ordinary intellectual method of substituting abstractions reached by analysis. Those who have criticised the use of metaphor have for the most part not realized how little removed such description is from the ordinary intellectual method of analysis. They have supposed that in analysis we stick to the fact itself, whereas in using metaphor we substitute for the fact to be described some quite different fact which is only connected with it by a more or less remote analogy. If Bergson’s view of the intellectual method is right, however, when we describe in abstract terms arrived at by analysis we are not sticking to the facts at all, we are substituting something else for them just as much as if we were using an out and out metaphor. Qualities and all abstract general notions are, indeed, nothing but marks of analogies between a given fact and all the other facts belonging to the same class: they may mark rather closer analogies than those brought out by an ordinary metaphor, but on the other hand in a frank metaphor we at least stick to the concrete, we substitute fact for fact and we are in no danger of confusing the fact introduced by the metaphor with the actual fact to which the metaphor applies. In description in terms of abstract general notions such as common qualities we substitute for fact something which is not fact at all, we lose touch with the