516 SPECIAL METHODS OF DISCOVERY. 



and observations; and, as either our hypotheses are in 

 many cases false, or the tests we employ are unsuccessful, 

 and as every new combfnation of matter or its forces pro- 

 duces new effects, it not unfrequently happens that, when 

 employing this method, we search for one thing and find 

 another ; i.e., we fail to confirm our hypothesis, but discover 

 an unexpected new truth. 



The discovery of the laws of double stars is an instance 

 of this kind. ' Among the stars there are some which are 

 called double stars, and which consist of two stars, so 

 near to each other that the telescope alone can separate 

 them. The elder Herschel diligently observed and mea- 

 sured the relative positions of the two stars in such pairs, 

 and as has so often happened in astronomical history- 

 pursuing one object, he fell in with another. Supposing 

 such pairs to be really unconnected, he wished to learn 

 from their phenomena something respecting the annual 

 parallax of the earth's orbit. But in the course of twenty 

 years' observations he made the discovery (in 1803) that 

 some of these couples were turning round each other with 

 various angular velocities. These revolutions were, for 

 the most part, so slow that he was obliged to leave their 

 completed determination as an inheritance to the next 

 generation. His son was not careless of the bequest, and, 

 after having added an enormous mass of observations to 

 those of his father, he applied himself to determine the 

 laws of these revolutions. A problem so obvious and so 

 tempting was attacked also by others, as Savary and Encke, 

 in 1830 and 1832, with the resources of analysis. But a 

 problem in which the data are so minute and inevitably 

 imperfect required the mathematician to employ much 

 judgment as well as skill in using and combining these 

 data ; and Sir John Herschel, by employing positions 

 only of the line joining the pair of stars (which can be 



