under a certain nervous tension and therefore may not show so well at the very first test as their mental equipment should allow.  Hence it is decidedly desirable not to rely on the first test, but to repeat it.  If those various interferences are taken into account, the correspondence between efficiency and the results of the tests is fairly satisfactory.  It justified me in proposing that the experiments be continued and in regarding it as quite possible that later tests on the basis of this principle may be introduced at the employment of motormen.

A difficulty is presented by the valuation of the numerical results.  The mere number of omissions alone cannot be decisive, as it is clear that no intelligent man would make any omissions if he should give an unlimited amount of time to it; for instance, if he were to spend fifteen minutes on those 12 cards.  But this is the same thing as to say that a motorman would not run over any one if he were to drive his car one mile in an hour.  The practical problem is to combine the greatest possible speed with the smallest number of oversights and both factors must therefore be considered.  The subject who makes relatively many mistakes but uses a very short time must be acknowledged to be as good as the man who makes fewer mistakes but takes a longer time.  In the results which I have gathered in experiments with motormen, no one has gone through those 12 cards in a shorter time than 140 seconds, while the longest time was 427 seconds.  On the other hand, no one of the motormen made less than 4 omissions, while the worst ones made 28 omissions.  I abstract from one extreme case with 36 omissions.  On the whole, we may say that the time fluctuates between 180 and 420, the mistakes between 4 and 28.  The aim is to find a formula which gives full value to both factors and makes the material directly comparable in the form of one numerical value instead of the two.  If we were simply to add the number of seconds and the number of omissions, the omissions would count far too little, inasmuch as 10 additional omissions would then mean no more than 10 additional seconds.  On the other hand, if we were to multiply the two figures the omissions would mean by far too much, as the transition from 4 mistakes to 8 mistakes would then be as great a change as the transition from 200 to 400 seconds, that is, from the one extreme of time to the other.  Evidently we balance both factors if we multiply the number of omissions by 10 and add them to the number of seconds.  The variations between 4 and 28 omissions are 24 steps, which multiplied by 10 correspond to the 240 steps which lie between 180 and 420 seconds.  On that basis any additional 50 seconds would be equal to 5 additional omissions.  If of two men one takes 100 seconds less than his neighbor, he is equal to him in his ability to satisfy the demands of the service, if he makes 10 mistakes more.