128 



SCIENCE 



[N. S. Vol. XLIX. No. 1258 



counted on one's fingers. Since then the more 

 assiduous observation of modern times has 

 raised the number into the himdreds, and the 

 application of mechanical methods of search 

 (that is, of photography) has swelled our list 

 into the thousands, with prospects of further 

 increase. 



It is now possible to classify almost all 

 variables into fairly definite — often very defi- 

 nite — ^natural groups. In each of these groups 

 more or less numerous empirical regularities 

 of behavior, or " laws " have been detected ; 

 and in some instances the number of these 

 relations is large, and the accuracy with 

 which they represent the characteristics of 

 individual stars is surprising. It has even 

 been possible to use these relations to deduce 

 information regarding the distribution of the 

 stars in space, for example, which has revolu- 

 tionized our previously existing ideas. Yet 

 the humiliating admission must still be made 

 that, in spite of these advances, we know ex- 

 tremely little of the real causes of stellar 

 variation. A satisfactory theory exists in the 

 ease of but one group — and this is based on 

 the fimdamental assumption that the stars of 

 this class are not really variable at all, but 

 owe their apparent changes to the geometrical 

 accident of eclipse! Of the causes, mech- 

 anism and physical relations of the intrinsic 

 variability of the stars we are still in dense 

 ignorance. Could we solve the riddle, there 

 is good reason to hope that the key to some of 

 the fundamental problems of astrophysics 

 would be found. 



In the study of variable stars, therefore, 

 , we have a series of problems which are at 

 once laborious, difficult, fascinating and of 

 great promise, and a summary survey of the 

 field may afford very appropriate material for 

 this address — considering first the methods of 

 observation, and then the known facts, to- 

 gether with such theoretical conclusions as 

 may be drawn from them. 



From the observational side, the study of 

 variable stars affords excellent examples of 

 the advantages of " scientific management " 

 and of cooperation. Until about a generation 



ago, the discovery of variables was made by 

 accident, in the course of other work — such 

 as the making of meridian catalogues or star 

 charts, or the search for asteroids, — and the 

 rate of discovery was naturally small. The 

 great superiority of photography for this pur- 

 pose was first effectively realized by Pickering 

 and his assistants at Harvard. By super- 

 position of a positive of a given star-field 

 upon a negative of the same field taken at 

 some other date any variables which may be 

 present can be picked out at once. More 

 than a thousand variables have been dis- 

 covered at Harvard in this way — a number 

 far exceeding the total which all the astron- 

 omers of the world had found by visual means 

 in the preceding three centuries. The process 

 is so easy that Miss Leavitt and her associates 

 when working up a new region, never trouble 

 to identify the previously known variables, but 

 simply rediscover them along with the rest. 

 From the percentage of known variables 

 which are missed (usually a small one) it is 

 possible to estimate how many unknown ones 

 have been passed over and await future dis- 

 covery. 



Similar photographic studies of globular 

 clusters led to Bailey's important discovery 

 of the presence of variables in them — the im- 

 portance of which is only now being fully 

 realized. Mention should also be made of the 

 successful work of Max Wolf and others with 

 the blink microscope. 



Another fruitful method of discovery is by 

 means of spectrum photographs with the ob- 

 jective prism. Certain types of spectra with 

 bright lines are practically certain to belong 

 to variable stars. Mrs. Fleming and Miss 

 Cannon have thus discovered some two hun- 

 dred variables of long period and about half 

 of the known galactic novse. 



Wlien however a variable star has been 

 discovered, the observer's work has just begun. 

 Its changes must be followed and their laws 

 determined. In many cases the variations are 

 foimd to be exceedingly regiilar, both in 

 period and amplitude, so that a very precise 

 mean light curve may be obtained. Such 



