Variation of Sentence-Constants in Literature H 



LONGFELLOW LONGFELLOW 



{The Spanish Student) ' {Hyperion) 



First hundred periods 12.0 First hundred periods 20.5 



Second " " 8.8 Second '• " 25.4 



Third '• " 11.1 Third " " 27.5 



Fourth " " 8.8 Fourth " " 24.3 



Fifth " " 10.1 Fifth " " 21.3 



Average for 500 periods 10.2 Average for 500 periods 23.8 



The list could be indefinitely prolonged, but it was not neces- 

 sary. The evidence seemed to me to show that sentence-lengths, 

 and presumably also predication-averages and simple-sentence- 

 percentages, are quite as much dependent upon the nature of the 

 composition employed as upon -the author's sentence instinct. In 

 fact, I surmised that the variety of sentence-lengths which an 

 author employs is limited only by his versatility as a writer. 

 Acting on this surmise, I decided to make a test. Goethe was 

 the author selected for this purpose, for he seemed to meet most 

 nearly ideal conditions. His style is unquestioned, his punctu- 

 ation is consistent and scientific. Most important of all, he is 

 the most universal writer of the last century. He ranks high 

 as a writer in the fields of poetry, drama, fiction, biography, art, 

 literary criticism, travel, and science. The results^ fully cor- 

 roborated my conjecture. 



^In securing these as well as the preceding results, the sentences were 

 actually counted, but where the nature of the text permitted it, the number 

 of words has been obtained by counting the lines and multiplying by the 

 average number of words per line. In this way the drudgery of the work 

 may be greatly shortened without impairing the accuracy of the results, 

 since counting, like every other arithmetical operation which is not carefully 

 checked, involves unavoidable accidental errors. To test the accuracj' of my 

 method, I determined by means of it the number of words in Chaucer's 7 ale 

 of Melibeiis and obtained 16,633 words as against 16,659 which was obtained 

 by an actual count. In less than two hours I determined the number of 

 words in Macaulay's History of JLnp;land to be 979,6*58 as against 974,195 

 obtained by actual count. Assuming that this last number is absolutely 

 correct, my estimate involves an error of J^ per cent, which may be safely 

 neglected since it will be shown that the average sentence lengths based 

 upon 500 sentences involve an average error of over 2 per cent. 



239 



