i6o 



POPULAR SCIENCE MONTHLY. 



do graduate. It is evident that a few students graduating in a class 

 above forty years of age — by no means an unheard-of state of affairs — 

 would unfairly raise the average age of that class, since it is mani- 

 festly impossible to graduate twenty years below the normal age. 

 Again, a class, or series of classes, may graduate a considerable number 

 of its members below twenty, while a still larger number graduates 

 above twenty-four or twenty-five. The curve of distribution of the 

 ages of graduation will then resemble the letter M. Manifestly, 

 in such a case, which occurs several times, the arithmetical average 

 tells us nothing of value. Finally, the median age gives us the exact 

 information that one half the students in question graduated at or 

 above the given age, and the other half at or below it. The curves 

 of distribution, moreover, given in the plates for all graduates and all 

 colleges for the years 1850-59 and 1890-99, show exactly what per- 

 centage graduated at each age. 



Table I. 

 Median Ages of Graduation by Decades. 



We now come to a consideration of Table I.* The most obvious 

 and surprising thing that strikes us at first sight is the fact that our 



*In Table I., decade '1770-79,' equals Dartmouth 1771-79; decade '180U- 

 09,' equals Middlebury 1803-09; decade '1830-39,' equals Alabama 1832-39, 

 New York University 1833-39, Oberlin 1837-39, Wesleyan 1833-39; decade 

 ' 1850-59,' equals in Syracuse 1852-1)9. In each case the corrected year 

 marks the date of the first gradvuiting class. In decade ' 1890-99 ' Adelbert in- 

 cludes only the years 1890-95; New York University, 1890-94; Syracuse, 

 1890-98. In Alabama University there were no graduates for the years 

 1806-71 inclusive. During several of these years the luiiversity was closed. 



The data for the decade '1900- ' are as follows: Dartmouth, Oberlin, 

 DePauvv, each, class of 1900 only; Wesleyan, Alabama and Vermont, classes 

 of 1900-01; Howdoin. 1900-02. The whole number of cases in this 'decade' 

 is 572. 



In reference to tiie degrees included in tlic investigation, I have attempted 

 to use only A.B., Ph.B. and B.S. In a few instances the last named degree 



