Tan prral are. The temperature was measured by thermometer No. 29110 in the 

 dummy pendulum resting on its metal base, and the appropriate corrections applied. 

 The temperature gradient from floor to roof was not measured. 



Barometer. Barometer Hicks No. C 895 was read three times during each swing 

 by each pendulum, and the hygrometric correction was calculated from the readings of 

 the standard wet and dry bulb thermometers of the Observatory. 



The rate of the sidereal clock " S.C." was obtained directly from the time observa- 

 tions, and there is strong internal evidence that the rate did not remain constant during 

 any period of 24 hours. This anomalous behaviour may have been caused by a 

 new suspension strip for the pendulum, which was made by a local jeweller to replace 

 that broken on the journey from Potsdam. 



The results of the observations are given in Table XVII. 



The final figures, it will be seen, are very discordant. Even after application of 

 the correction for temperature lag in Table XVIII, there is little improvement. The 

 differences from the mean of each pendulum are also given in this Table and the squares 

 of these differences. 



TABLE XVIII. 



From this, we get the probable errors of a single result : 



e b ~ 43-OxlO-' sec. 



e- = 32-4 



2i = 32-7 



e m = 33-3x10-' sec. 



25 



