within reasonable measure of the values 983 '004 and 983 003 found for the Cape 

 Evans station. 



The values found by Bernacchi with the three pendulums used by him were 082-970, 

 982-979 and 983-025. From the published figures, it is clear that little reliance ran 

 be placed on the value of the adopted mean, but it is interesting to note that Bernacchi's 

 mean value at Melbourne was about 0-03 low, and that his Christchurch value, 980-512, 

 is also somewhat lower than ours. 



Attention should be drawn to the fact that the observations at Cape Evans in 

 the first year have been neglected. It is quite impossible to determine the probable 

 error of the observations in this first year and the conditions of observation were very 

 unsatisfactory. The clock was hung in a very unfavourable position, the time observa- 

 tions were unsatisfactory and the coincidence apparatus was working very badly. In 

 addition, the low temperature in the pendulum cave did not improve the observational 

 conditions. 



Of the two sets of observations made in the first winter, the first gives values for 

 " g " not far removed from those obtained in the second year, but the times of swing in 

 the second set are such as would correspond to an error in the daily clock rate of 0- 8 sec. 

 an error which it is difficult to believe could have occurred. The agreement of the 

 first set with the third and fourth (apparently within the probable error) is therefore a 

 matter for satisfaction, but no explanation for the disagreement of the second set can 

 be offered. 



GRAVITY AT CHRISTCHURCH. 



Observations were made by Bernacchi on the Discovery Expedition in 1901 and 

 1904, but little agreement is shown between the values for " g " in the different years 

 or with the different pendulums. Unfortunately, the observations of the present 

 expedition were not much more successful. Neither for the observations in 1910, nor 

 for those in 1913, is it possible to calculate the probable error. On the first occasion, 

 the observational conditions were wholly unsatisfactory and the values for time of 

 swing may be in error by20xlO~ 7 sec., while the value obtained with pendulum No. 5 

 is obviously wrong and must be discarded. In the second year, no two pendulums 

 gave satisfactory observations. The whole of the observations should therefore be 

 discarded, but as the previous observations by Bernacchi all appear even more 

 unsatisfactory, the value of " g " has been evaluated. 



The corrected values were as follows : 



No. 5. No. 7. No. 21. 



Christchurch, 1910 .. .. (0-5085396) sec. 0-5085121 sec. 0-5099429 sec. 

 1913.. .. (0-5085259) (0-5085052) (0-5099405),, 



The best one can hope for is that pendulum No. 21 was working satisfactorily in 

 1913, but no reliance can be placed on this. In all probability, the values obtained in 

 1910 and 1913 with No. 21 are hi agreement within the limits of probable error. 



93 



