24 R. E. M or its 



which coincides nearly with the vahie 2.30, determined by the 

 laborious process of actually counting the finite verbs in over 

 40,000 periods. 



I now offer the following as a tentative statement of the Sher- 

 man principle : 



1. Good zin'itcrs manifest their individuality by the uncon- 

 scious use of certain average sentence-proportions which remain 

 nearly constant throughout a given form of composition. 



2. These sentence-proportions vary undely, not only in varions 

 authors, hut also for different forms of composition employed by 

 the same author. 



3. These proportions are interrelated, and in some instances 

 at least their lazvs of dependence may be mathematically ex- 

 pressed. 



The principle as now stated opens an almost unexplored field 

 for investigation. The present inquiry has been limited to the 

 constants L, P, and S, but there are many other sentence-pro- 

 portions, such as the percentages of various kinds of ^rlauses and 

 phrases, the relative abundance of adjectives, conjunctions, and 

 other parts of speech, the ratio of Anglo-Saxon derivatives to 

 those derived from other languages, the average length of the 

 words employed by an author, etc. Are these proportions more 

 or less variable than those already considered? Which of these 

 are independent, and which are interrelated, and by what laws? 

 Which of these are most persistent and hence most character- 

 istic of a given author? These and many other questions press 

 for answers. 



Concerning the constants L, P, and 6" many questions remain 

 open. We have seen that L depends at least upon two factors, 

 the author's sentence-sense and the form of composition em- 

 ployed; but are these the only factors upon which it depends? 

 If so, can the element of variation in L due to the form of com- 

 position be determined and the residue, due to the author's 

 sentence-sense alone, be set free? Can the relation between sen- 

 tence-length, sentence-instinct, and sentence-form be as defi- 

 nitely expressed as we expressed the relation between P and S? 



252 



