148 CARNEGIE INSTITUTION 



Pickering and Hale that it need not be considered in the present 

 one. The following remarks therefore refer mainly to the branch of 

 celestial measurement sometimes termed asiromelry. 



It must not be inferred that these two branches are really distinct. 

 The forces of both have to be combined in working out the great 

 problems of astronomy. 



At the outset it is necessary to consider certain features in which 

 astronomy is radically different from other branches of science, and 

 which render it inadvisable to apply to it^ promotion any general 

 rules derived from experience in the case of other sciences. 



I. Generai, Character of Astronomicai. Research. 



Astronomy differs from other physical sciences in being founded 

 entirely on observations, while the others rest mainly upon experi- 

 ment. 



Its laws and phenomena are of slow development, generally in 

 cycles of years, centuries, or ages. 



Their complete investigation involves difScult and complex mathe- 

 matical problems, the nu merical computations connected with which 

 are long and arduous. 



For all these reasons important astronomical results generally re- 

 quire a combination and comparison of observations continued 

 through long periods of time. 



In its relation to the welfare of mankind astronomy is also pecu- 

 liar. In addition to its immediate connection with that welfare, as- 

 tronomical research may bring to light cosmical processes pregnant 

 with the destiny of our race. 



These same circumstances render a different kind of genius neces- 

 sary in the case of astronomy and of the sciences of experiment. 

 The skill required in an astronomical investigator is not that of the 

 experimentalist, but of one having such a grasp of the mathematical 

 relations involved in the subject which he is treating that he can 

 devise methods which most economically and surely lead to the de- 

 sired results. 



In working out these results a great amount of computation or ob- 

 servation is generally required. This part of the work can generally 

 be done by routine computers, who are not necessarily masters of 

 the great problems of astronomy, but 'A'ho simply follow the formulae 

 supplied them. 



