REPORT 1893. 



returning quickly to the telescope, found there had been an apparent 

 tilting of about one-sixth of a second to the east. Standing for the same 

 time on the east side of the pendulum, there was an apparent tilting to 

 the west. At a distance from the pendulum these etfects were much less 

 marked. After sitting for a short time at the telescope-seat in the 

 passage outside the cellar a somewhat similar, but very slight, eflPect is 

 noticeable. For about a minute the image remains perfectly steady. 

 It then starts slowly, but rather suddenly, to the right, indicating a 

 movement to the east, the total deflection in eight minutes being about 

 ^Lth of a second. I attributed these movements at first to my weight 

 bending the cellar floor, but some experiments made afterwards at Mr. 

 Darwin's suggestion showed that they were due rather to changes of 

 temperature, and that the deflections caused by my weight were in reality 

 almost imperceptible. 



Supposing the foundation to remain fixed, the effects of a change in 



Fig. fi. 



soo 



700 



eoo 



i)00\ 



3 00 



zoo 



.'iajn.K 



the distribution of temperature might be manifested in two ways — either 

 by the unequal expansion of diS"erent parts of the instrument or by con- 

 vection curi'ents in the paraffin oil. Since the gas-jet stands west of 

 the pendulum, the expansion due to its heat would produce an apparent 

 tilting to the east. On the other hand, since the vertical axis about 

 which the mirror turns lies to the south of its centre (see fig. 1) the 

 •eSect of convection currents on the oil would be to cause an apparent 

 tilting to the west. Thus the initial deflection of the mirror in the pre- 

 ceding experiments appears to be due to expansion. 



If, however, the heating be continued for some time a retrograde 

 movement sets in. On August 20 the gas was kept alight for six hours, 



