Numerical tests may, no doubt, be used to compare the results of processes which are in themselves nonmechanical and perhaps inscrutable. Thus of two salesmen spending the same time on similar routes selling the same goods at the same prices, one will sell twice as many as the other; it is often impossible to say why. “Personality” does it; that is, a complex of influences beyond the sphere of precise analysis. But you can measure the results of its operation and be fairly sure they will be repeated. It is the same with authors. When a new writer submits the manuscript of a novel the publisher can make only a very uncertain guess as to how many copies of it will sell; but when several novels have been published and shown their power to interest the public a reasonably safe prediction is possible. The statistical method does not require that the process we are testing be understood, but only that it be uniform. In that case its future working may be predicted from its past. There is a large and legitimate field for ingenuity in thus standardizing human function. Formalism is apt to come in, however, by our taking a mechanical view of the function itself, of the end to be sought, in order to make it more easily measurable. This objection may be made, for example, to rating and rewarding salesmen according to the amount of their sales. It seems that this is not, in practice, a good plan, because their behavior counts in many ways not covered by such a calculation. A merchant says: “If you have five hundred sales-people working on a straight commission basis, you have five hundred individuals who are, in principle, each one in business for himself.... This means that there is no group spirit, no sense of unity in the organization, no co-operative spirit present. It works out very badly.” He suggests a modified test, also numerical, which is inadequate in theory, however it may work. The complete function of personality is never measurable. We have the same fallacy in the attempts to measure the value of a professor by the number of students electing his courses, or the number of hours he spends in his classroom. It seems to be a general truth that the higher a social or mental function the less capable it is of numerical measurement; the reason being that the higher functions are acts of creative organization that can be appreciated only by a judgment of the same order. The work of a lawyer, a teacher, a clergyman, a man of science, even of an artistic craftsman, can be measured only by expert opinion. Our tests of the mental capacity of children should be mechanical only in so far as they relate to mechanical processes, like verbal memory or calculation. When it comes to higher capacities, like the understanding of complex ideas or sentiments, such as honor, the test, if it is to be of any value, must be applied, not mechanically, but by some one of imagination to understand what the child means by his answers.