14 L. A. Sherjfian, 



two principal articles in the second number of our University 

 Studies, I put together certain chief facts and findings from 

 the analysis in the paper " Some Observations upon the 

 Sentence-length in English Prose," hoping to attract other 

 hands to the work. But, though I outlined with some clear- 

 ness the course of investigation to be followed, no one of 

 those communicating with me concerning it seemed drawn to 

 the task, or, as I thought, to realize the promise it held out of 

 solving some or many of the mysteries of literature. Clearly, 

 by study of individual styles the course of evolution in 

 modern prose English might be traced. Moreover, if it were 

 true that each author writes always in a consistent numerical 

 sentence average, it would follow that he must be constant 

 in other peculiarities, as proportion of verbs, substitutes for 

 verbs, conjunctions, etc., if a sufficiently large number of 

 sentences were taken as the basis. Meanwhile, in a series 

 of communications to Science (beginning with the issue for 

 March 22, 1889) upon a kindred topic, it had been seriously 

 questioned whether there could be any such thing as con- 

 sistency in such cases, — except perhaps on the basis of 

 many thousand sentences. The first thing therefore to be 

 done was to demonstrate undeniably the fact of a constant 

 numerical average. For this I chose Macaulay's History of 

 England. The style of this was noticeably less stereotyped 

 and regular than of the Essays, there was much curt dia- 

 logue, there were long descriptions. If the findings for the 

 Essays were confirmed in the History as a whole, the case 

 would be closed, at least for Macaulay. I had devised a 

 plan of accurately registering the results in counting, and 

 had reached such facility with the method that I no longer 

 dreaded the drudgery of such a task. In about three weeks 

 of the summer of 1889 I finished the five volumes. The fol- 

 lowing were the results obtained. Each entry is the average 

 of one hundred periods in consecutive order throughout. 



350 



