cubic capacity of the cave, it was not possible to obtain M steady temperature by arti- 

 ficial heating. The first difficulty met was due to the frosting of all the mirror and 

 lens surfaces by the breath of the observer. Thus, the operations of levelling the 

 apparatus, determining the flexure correction and placing the pendulums in position, 

 so covered the prisms, agate planes, and the pendulums themselves with hoar-frost, 

 that one could get no proper reflection of the light from the pendulum mirrors. Another 

 possible source of error was due to changes in the period of the pendulums due to the 

 hoar-frost accumulated on them. 



The operations finally adopted were, in order : 



( 1 ) Levelling. 



(2) Determination of flexure correction. 



(3) Placing pendulums in position with CaCl 2 , or P 2 5 , in the case. 



(4) Wait of about two weeks while the mirrors cleared. 



In adopting this procedure, it was necessary to assume that, during the two weeks 

 wait, no serious change of level or change in the flexure correction occurred, and that, 

 by the time the mirrors were clear of hoar-frost, the pendulums themselves were also 

 clear. Observations were then made in the usual way and the flexure correction re- 

 determined at the end of the series. 



A minor trouble consisted in the frosting of the telescope object glass and eyepiece, 

 and of the windows in the cover of the pendulum stand, so that these had to be frequently 

 cleaned during the course of the observations. 



A much more serious difficulty lay in the large temperature changes as the observer 

 entered and left the cave, due to the smallness of its size. For this reason, immediately 

 the first readings had been made, the observer left the cave and only re-entered it as 

 the second set of readings became due. The uncertainty in the observed tempera- 

 ture is moreover intensified by the position of the dummy pendulum, which is not hung 

 as are the pendulums it is supposed to represent, but rests on a brass disc in direct 

 connection with the large metal base of the instrument. It is clear that, in these 

 unfavourable' circumstances, the march of temperature in the pendulums is probably 

 riot at all well represented by the reading of the thermometer in the pendulum 

 stand. 



Equally serious was the large temperature gradient in the cave due to gradual 

 conduction from the surface through the ice. At the time the cave was first built, the 

 temperature (after the warmth of the summer) cannot have been many degrees below 

 freezing-point, but, on the inception of winter conditions, a considerable temperature 

 gradient was produced. Even at the time of the first measurements this amounted 

 to 3 C. per metre. 



A further irregularity in the readings was unfortunately furnished by the behaviour 

 of the coincidence apparatus here and subsequently. This was shown in the following 

 way : In observing the course of successive flashes across the cross wire of the telescope, 

 under normal conditions, the flashes in the vicinity of the cross wire should be equally 



29 



