32 LIGHT SCIENCE FOR LEISURE HOURS. 



we select to regulate every other time-piece ? If a 

 man sets his watch every morning by the clock at 

 Westminster, it is clearly impossible for him to test 

 the accuracy of that clock by the motions of his 

 watch. It would, indeed, be possible to detect any 

 gross change of rate; but for the purpose of illus- 

 tration I assume, what is indeed the case, that the 

 clock is very accurate, and therefore that minute 

 errors only are to be looked for even in long intervals 

 of time. And just as the watch set by a clock cannot 

 be made use of to test the clock for small errors, so 

 our best time-pieces cannot be employed to detect slow 

 variations, if any such exist, in the earth's rotation- 

 period. 



Sir William Herschel, who early saw the importance 

 of the subject, suggested another method. Some of 

 the planets rotate in such a manner, and bear such 

 distinct marks upon their surface, that it is possible, by 

 a series of observations extending over a long interval 

 of time, to determine the length of their rotation-period 

 within a second or two. Supposing their rotation 

 uniform, we at once obtain an accurate measure of 

 time. Supposing their rotation not uniform, we obtain 

 (1) a hint of the kind of change we are looking 

 for; and (2), by the comparison of two or more 

 planets, the means of guessing how the variation is to 

 be distributed between the observed planets and our 

 earth. 



Unfortunately, it turned out that Jupiter, one of 

 the planets from which Herschel expected most, does 



