AND PLANTS 87 



whole angle : but beyond a certain limit, we cannot 

 do so. 



For most purposes, the people who want to know the 

 latitude of the Radcliffe telescope are content to take 

 a value somewhere within these very narrow limits, and 

 to neglect the observed discrepancies. The value chosen 

 for the latitude published in the General Catalogue of 

 Stars (1890) is about fifteen yards North of the lowest 

 observed latitude, about nineteen yards South of the 

 highest. This value was selected by a well-known 

 process, but I want you to realize clearly what we do 

 when we adopt it. We reject a carefully made record 

 of a variable experience, and substitute for it an ideal 

 record of a constant experience which we have not 

 attained. Why do we dare to do this ? If you ask an 

 astronomer this question, I suppose he will answer you 

 by saying that he wants to use the position of the 

 Radcliffe telescope in calculations which enable him to 

 predict some future experience, say of the relative posi- 

 tions of stars. But he will not be able to measure that 

 experience accurately, and the differences in his predictions, 

 involved by taking into account all the various observed 

 values of this latitude, would be so small that he could 

 not verify or disprove them by his subsequent measure- 

 ments, and therefore it is not worth while, as a practical 

 matter, to take them into account. This is a perfectly 

 reasonable statement, and its justification lies in the 

 accuracy with which future experience can be predicted 

 by the process adopted. But if you ask your astronomer 

 whether he believes that the constant value he has chosen 

 to adopt represents a real constancy in the angle of 

 latitude itself, his answer will be more difficult to follow. 

 Some of the discrepancies between the observations he 

 will certainly attribute to the observers ; and if you think 



