30 



II. DISCUSSION OF PENDULUM RESULTS 



BY 



C. CHREE, Sc.D., LL.D., F.R.S. 

 (FROM THE NATIONAL PHYSICAL LABORATORY). 



1. IN drawing conclusions from the pendulum observations made during the British Antarctic Expedition 

 of 1902-1904, due allowance must be made for the conditions under which the work was done. Those 

 responsible for the Expedition found themselves shortly before its departure without a physical observer. 

 At the last moment, Mr. L. C. BERNACCHI consented to fill the breach, and in the very short time that 

 remained he did all that was possible to obtain familiarity with the instruments. He had fortunately had 

 a good deal of previous experience in observing times of vibration in connection with magnetic observa- 

 tions, and the observational results obtained by him and Mr. SKELTON during the Expedition appear as 

 consistent as could be expected under the conditions of observation. The apparatus had arrived at Ken- 

 some time before Mr. BERNACCHI joined the Expedition, and the pendulums had been swung by Mr. E. G. 

 CONSTABLE, senior assistant in the Observatory Department, in order to obtain their periods. Even then 

 some difficulty was experienced in getting the cylinder containing the pendulums to remain air-tight 

 during the observations, which were taken at a pressure of about 60 millims. of mercury. Greater 

 difficulty was experienced during Mr. BERNACCHI'S introduction to the instruments, but this was 

 attributed to the fact that meantime the apparatus had been dismounted and had been somewhat hurriedly 

 re-erected. The defect, however, as explained in Mr. BERNACCHI'S introduction, proved even more 

 troublesome in the Antarctic. 



When observing, Mr. BERNACCHI'S usual practice was to set the pendulum swinging, and then shortly 

 after take two sets of observations of the time answering to 50 coincidences, one set with the pendulum 

 moving in the one direction, the other with it moving in the opposite direction. The pendulum was then left 

 swinging unobserved for about an hour and a half, and thereafter two other similar sets of 50 coincidences 

 were taken. The mean time of the two sets of observations differed by about two hours, and the leakage 

 was such that in this interval the pressure inside the receiver rose on an average from about 60 millims. to 

 110 millims. The leakage was not conspicuously worse during any one set of experiments than during 

 the others. 



During each set of coincidences four readings were taken of the pressure, and the arithmetic mean of 

 these was accepted as the pressure of the observation. 



The " pressure correction " is, within the limits of accuracy of its determination, a linear function of the 

 pressure, and the rate of leak would normally be nearly uniform during the time occupied by a set of 

 coincidences. Thus the fact that the cylinder was leaky will presumably have made little if any difference 

 in the accuracy of the mean final values ; but it is probably in part accountable for the somewhat large 

 discrepancies occasionally apparent between the results of the different sets of coincidence observations 

 with the same pendulum on the same day. 



Difficulty was also experienced in connection with the temperature. At Winter Quarters, whilst the 

 regular diurnal inequality of temperature was small, large sudden changes were not unusual. The 

 room in which the pendulums were swung usually varied very perceptibly in temperature during the 

 observations, and the temperature in different parts of the room (e.g., beside the pendulums and beside 

 the barometer) sometimes differed rather largely. The change of temperature in progress during the 

 observations was sometimes a rise, sometimes a fall. A change of 1 C. in the temperature of the pendulums 

 means an alteration of 46 x 10~" second in their time of swing. Thus a very little error in the temperature 

 assigned has an appreciable effect on the period. Here, again, there was probably little, if any, effect on 

 the mean final values, but there was unquestionably in the temperature variations an active source of 

 irregularity between the different individual results. 



