ON SEISMOLOGICAL INVESTIGATIONS. 



61 



Mr. Shaw specially designed his instrument (with a thin metal cover, 

 &c.) so that it might take up the temperature quickly supports this 

 view. But the magnitude of the lag in M.-B., viz. 6 hours, is somewhat 

 surprising, and to test its reality the numbers of Table II. were treated 

 in a different way. Subtracting each constituent from that which follows 

 it, we get the movements for each two hours, and can study the corre- 

 spondence. It wall be convenient to reverse M.-B. for this purpose. 



Now, if we try the effect of associating the movement of M.-S., 

 (a) with the contemporary movement of M.-B., 

 (6) with the reading of M.-B. two hours later, 

 (c) with the reading of M.-B. four hours later, 

 and so on, we get, on grouping the results, the means shown in Table V. 



Table V. 



-Effect of Associating M.-S. with 31. -B. of Various 

 Times Later. 



Inspection of these figures indicates that the best correspondence 

 is somewhere between 4 h. and 6 h. later, and nearer 6 h. than 4 h. But 

 it was clearly desirable to have more light on the matter, and, as the 

 next step, a thermograph was installed in the stable which forms the 

 Observatory. It was not a very good thermograph, having been rather 

 roaghly treated in some mining experiments ; i)ut it gave a pretty fair 

 indication of the temperature for several weeks. The readings May 4-10 

 and May 31- June 10 were selected for discussion. The traces were 



