Strength-Duration Curves and the Theory of Electrical Excitation.

regardless of how carefully the particular prediction-basis is computed, it indicates that some major factor or factors are omitted, or are improperly weighted or coordinated, in the particular prediction-basis being tested for accuracy. In comparing two different prediction-bases against the same subsequently determined set of actual distribution-values, the coefficient of predictionaccuracy tests the relative merit of these two prediction-bases as to their relative correctness in the inclusion of their own postulated constituent factors, and in the relative weights and co6rdinations given such factors. In any case in which one has in hand a prediction-basis in terms of a probability-distribution, and has also in hand the actual subsequent or other independently secured measuring and counting of the same quality in the same population, the coefficient of, prediction-accuracy (CPA) presents a sound formula for judging the given probability-distributionproduction-basis as to its degree of success-no excuses taken. The next step in good research is, of course, to find out why the particular prediction-basis which was actually used was not more successful.