fl1s 



844 SCIENCE. [N. s. vol. xxiii. no. 596. 



Table VIII. marks of an individual's separate examina- 



Difference between First and Last Trials of an tions in Latin, for instance, was over 30 



Entrance Examination. Class of 1906. j^ tj^g gggg Qf iy^q glass of 1901, the dis- 



I a ■§ i§ tribution of these differences being given 



^1 ^ |s 3a |g I'S in Table VII. (the scale of marking being 



always the common one from 100 down). 

 In 140 cases of repeated examinations 

 chosen at random the average difference 

 between a candidate's first and his second 

 trial of precisely the same examination 

 was over 22, the distribution of these dif- 

 ferences being given in Table VIII. 



It should be noted that this defect (not 

 measuring capacity) of the examinations 

 of the Middle States Board probably exists 

 to a greater extent with the examinations 

 given independently by single colleges, for 

 in the former case the examination ques- 

 tions are planned and the papers rated 

 with great care. The eccentricities of in- 

 dividual examiners play a relatively minor 

 role, and the nature of the examination 

 can probably not be prophesied in advance 

 so accurately as in the case of the examina- 

 tions arranged by some one college. 



The inaccuracy of prophecy of achieve- 

 ment in college from achievement in en- 

 trance examinations becomes intolerable in 

 individual cases. For instance, there were 

 10 men out of the 130 who in their junior 

 year got A (the highest mark given) in at 

 least five studies. Their average marks at 

 entrance were in some eases in the lowest 

 tenth of the 130, barely above the passing 

 mark. Had the passing mark been set the 

 least bit higher, one of the very best stu- 

 dents of the three college classes would 

 have been debarred from entrance. There 

 is every reason to believe that of those stu- 

 dents who did yet worse in the entrance 

 examinations and so were shut out, a fairly 

 large percentage would have done better in 

 college than a third of those who were ad- 

 mitted. Sooner or later some one will be 

 barred out M'ho would have been the best 

 man of his class. 



