26 American Statistical Association. 



exacting from all that which should be exacted only from 

 the mean, could be rationally adapted to the special needs 

 of the exceptionally weak and the excei)tionally strong. 

 These beneficent reforms, it is at present believed, must 

 await the slow collection of data by the individualizing 

 method. If it is indeed true that the laws of growth deter- 

 mined for the mean cannot be nsed for the individual until 

 the individualizing method has established the probability of 

 each individual deviation remaining constant throughout the 

 period of growth, then a generation must elapse — so slow is 

 the gathering of data by this method — before the necessary 

 knowledge is in our hands. I hope to show that this long 

 waiting is unnecessary, and that the data collected by the 

 generalizing method maybe used, in a wa}^ hitherto unrecog- 

 nized, for the making of standards b}^ which the deviation of 

 an individual from the mean of his age can be seen to be 

 normal or abnormal. 



Let the problem be clearly understood. The question is : 

 This boy or girl is above or below the mean height, or weight, 

 etc. of his or her age, — how shall it be known that this devia- 

 tion is normal or abnormalT "^niere has been hitherto no 

 satisfactory reply to this question. A vague and partial 

 answer is possible after two measurements separated by at 

 least a year's interval. If the deviation is the same, or very 

 nearly the same, at both measurements, the probability is that 

 the child is growing normally. This probability is greater 

 than the general probability that a normal deviation is more 

 likely to occur than an abnormal one, but its numerical value 

 is wholly unknown. If, on the other hand, the two devia- 

 tions are unequal, the probability is that the greater of them 

 is abnormal, but the numerical value is here also unknowai. 

 How dehnitelv the individualizino- metliod could answer this 

 question is difficult of conjecture, in the present lack of data, 

 but certainly no answer whatever covdd be expected until 

 after two measurements separated b}' a year's interval, — a 

 year in which the unchecked cause of an abnormal deviation 



