In Table LXVIIT, the mean temperature and barometer readings are put down for all 

 the stations, together with the differences A from the mean of the Potsdam observa- 

 tions. 



TABLE LXVIIT. 



From pages 8 and 9, it is seen that the mean probable error of the determination 

 of the temperature constant is 0-07xlO~ 7 sec. and of the pressure constant 

 6-8 XlO^ 7 sec. 



From these values, we may calculate for the different stations 



For temperature and for barometer constants. 



-7 



sec. 



Cape Evans " C " probable error 

 " D " 



;j 



Wellington ,, 



Melbourne ,, 



Christchurch, 1910 ,, 

 1913 



(4) Probable error of lay correction. It may with confidence, I think, be assumed 

 that no correction for lag needs to be applied in the case of the observations made 

 subsequent to the replacement of the brass disc, upon which the dummy pendulum 

 rested, by a disc of cork. 



The brass disc was only used for the observations at Potsdam I, Christchurch, 1910, 

 and Cape Evans (A). The error in assuming a lag correction of 25 x 10~ 7 sec. per degree 

 per hour for the first is inappreciable, but may be significant in the case of Cape Evans 

 (A) and particularly Christchurch. 1910. Of these, the former series is neglected in 

 forming the final mean, while the errors of the latter cannot be estimated, though, in 

 any case, the probable errors due to this cause must be insignificant in comparison with 

 errors connected with the rating of the clock. 



(5) The determination of the probable error of the correction for clock rate. At 

 Potsdam, the probable error of the rate may be taken as not greater than +0-01 

 sec., which corresponds to a probable error in the correction to time of swing 

 = 0-6xlQ- 7 sec. 



86 



