"Suppose that the general intellectual ability of the dullest men who are able to support and look after themselves (men who though temperate and strong earn say $400 a year in good times in New York City) be represented by a and that of Aristotle or Goethe by _a_+_b_, the difference, _b_, being 100. Then the amount of such ability assigned by the tests alone would not, on the average, vary from the individual's true amount by more than 5; and would not vary therefrom by more than 14 in one case out of a hundred. The 5 and 14 are very cautious estimates, 4 and 11 being probably nearer what such an experiment would in fact reveal." He further remarks, "There is excellent reason to believe that it is literally true that the result of two hours' tests properly chosen from those already tested gives a better diagnosis of an educated adult's general intellectual ability than the result of the judgments of two teachers or friends who have observed him in the ordinary course of life each for a thousand hours."[14] Interesting applications of tests of this general character have been reported by Scott. Workers of various kinds, such as salesmen and clerks, were graded by their employers or supervisors on the basis of their actual ability at their task. It was possible in some cases to get very accurate objective measures of ability to sell goods, etc., by keeping records of achievement over a considerable period of time. These objective measures have been compared with the results of psychological tests administered at the time the men were employed. Positive correlations ranging in several instances as high as .80 to .90 were secured. This means that ability in the performance of the particular mental tests used was a very reliable sign of ability in the field. Various instances similar to these have already been described in Chapter 5.