1 8 R. E. Mont:: 



In the preceding table the figures in the second and third 

 columns are reduced from Mr. Gerwig's tables, the fourth column 

 is made up from Miss Whiting's results: 



A^=the number of periods from which the averages are taken. 



P=the average number of predications per sentence. 



6"=the per cent of simple sentences. 



L=the average number of words per sentence. 



If we compare the figures under P with those under L, we see 

 that Macaulay in his history uses almost exactly ten words to 

 each finite verb employed, and so do More, Hooker, Sidney, and 

 Channing, while Chaucer, Spenser, Lyly, Bacon, Milton, Bun- 

 yan, De Quincey, and Newman come ■ approximately within ten 

 per cent of this amount. But there is a marked tendency to 

 depart from the ratio ten when the sentence-length approaches 

 extremes as in the case of Hakluyt, Emerson, or Bartol. Ob- 

 viously, when L is less than ten, the ratio L to P must be less 

 than ten. The data are, however, too limited to reveal much 

 more than the mere fact of interdependence. 



The fact of interrelation is perhaps more apparent from a 

 graphical representation as shown in fig. i. The values of L 

 have been used as abscissas, ten times the values of P as ordinates, 

 and the resulting points have been marked by letters correspond- 

 ing to those to the left of the names of the authors in the table 

 above. The graph shows that four points are approximately in 

 range with the line whose equation is 



L=io P, 'I) 



while all the points but four lie between the dotted lines, mak- 

 ing an angle of only six degrees with the former. 



A more striking, though less obvious, relation exists between 

 predication-averages and simple-sentence-percentages. An' in- 

 spection of the table convinces one that there is some sort of 

 reciprocal relation, the larger P going with the smaller 5" and 

 vice versa, but much more than this is not warranted by the 

 figures in our table. Nor should we expect the table to reveal 

 any very definite relation, when we consider the uncertainty of 

 the numbers involved. There is a large probable error in all 



246 



