4 HISTORY AND INTRODUCTION Ch. 1 



begun to study the effects of sampling errors on conclusions drawn 

 from samples. 



By the end of the nineteenth century the Staatenkunde had ceased 

 to exist, and "Political Arithmetic" had died in name but had devel- 

 oped into a science of statistical analysis, with emphasis on socio- 

 logical and economic applications. The theory of mathematical 

 probability had grown extensively as a branch of pure mathematics, 

 and also was beginning to be associated with applied statistics. Thus 

 the groundwork was laid for the present phase in the evolution of 

 statistical theory and methods. 



In 1908 William Seely Gosset, who wrote under the pseudonym 

 "Student," published an article in the journal Biometrika which was 

 later to typify the opening of a new era in the statistical analysis and 

 interpretation of sampling data. From 1899 until his death in 1937, 

 Gosset worked for the brewing firm, Messrs. Guiness. His associa- 

 tions with this firm led him into a variety of experiences and sug- 

 gested uses for statistical methods which are typical of several of the 

 present-day applications of statistics. 



Messrs. Guiness were interested in barley, not just any barley, but 

 in those varieties, growing conditions, and practices which would 

 produce the best barley for breweries to use. These circumstances 

 brought Gosset into contact with agricultural experimentation aimed 

 at the improvement of crops and of agricultural practices. Moreover, 

 Messrs. Guiness did not wish to subsidize the raising of large crops 

 purely for the sake of scientific experimentation; they were a com- 

 mercial firm which wanted to show a profit from their enterprises. 

 That fact, plus the shortage of tillable land in Ireland and England, 

 made Gosset well aware of the importance of small samples and of 

 methods for deriving reliable information from such samples. 



Furthermore, a large brewery conducts many chemical analyses, 

 and hence needs to take proper account of errors of measurement. 

 And, finally, the firm with which Student was associated was con- 

 fronted with problems concerning industrial statistics: production and 

 marketing analyses, price analyses, and methods for controlling the 

 quality of the products which it manufactures for market. Thus Stu- 

 dent came into contact with a wide variety of agricultural, economic, 

 and industrial problems which would require some form of statistical 

 study. Moreover, those problems had to be solved for a commercial 

 firm, a situation which demanded efficiency and reliability with a 

 minimum cost consistent with these qualities. The twentieth cen- 

 tury renaissance of statistical theory and methods appears to be 



