Variation of Sentence-Constants in Literature 15 



suits, I find that the normal sentence-lengths of writers before 

 the seventeenth century, during the seventeenth century, the 

 eighteenth century, and the nineteenth century, are represented 

 respectively by the numbers 48, 42, 36, and 27, if the original 

 punctuation is used, and by 42, 38, 34, and 26, respectively 

 if the works examined are repunctuated. But in the selection of 

 the works examined, no thought seems to have been given to 

 uniformity in the form of composition ; at any rate we find works 

 so widely divergent in structure as Chaucer's Tale of Mcliheus, 

 Bacon's Essays, Milton's Areopagitica, Bunyan's Pilgrim's Prog- 

 ress, Locke's Essay on the Human Understanding, Taylor's 

 Sermons, Hume's History of England, Goldsmith's Vicar of 

 Wakeiield, Irving's Life of Washington, Spencer's Data of 

 Ethics, and Howell's Rise of Silas Lapham enjoying equal 

 suffrage in the comparison. 



Professor Sherman brought to light the remarkable fact that 

 an author's sentence-length remains practically constant through- 

 out a given work, an extensive test having been made on the 

 40,000 sentences of Macaulay's History of England, but here 

 we have the incontestable fact that the sentence-length of one 

 and the same author may vary by almost insensible degrees be- 

 tween limits so widely divergent as five and thirty-five. The 

 conclusion from which there seems to be no escape is that the 

 sentence-length of a work depends both upon the zvriter's sen- 

 tence-instinct and upon the particular form of composition into 

 which his thought is cast. That is to say, sentence-rhythm, inas- 

 much as it manifests itself in constant sentence-length, is a func- 

 tion of at least two variables .r and y, where x signifies the 

 author's sentence-sense and 3; the form into which he moulds 

 his thought,^ 



^ I tru'^t that this statement will not be interpreted as contradicting or 

 de-util zing; th:- Sherman pinciple. All that it insists upon is the necessity 

 of so mod fying the principle as to recognize the facts of variability in the 

 sentence constHnts. 



The results ob ained from Macaulay's History are in perfect harmony 

 with the-e conductions. Professor Sherm in found the average .'entence- 

 length of the History (5 volumes containing 41,500 periods) to be 23.43. 

 This is practically the same as the sentence- ength 23.H5 or Mach aveiii, 24 

 for PAtf and 23.L0 for the Essay on History by the same author. Now the 



243 



