ON A QUARTZ THIJKAD GRAVITY BALANCE. 



These values being so nearly the same, it is obvious that an approximate value of 

 g at any station is all that is necessary to enable all observations to be reduced to a 

 common temperature, which, if desired, may then be used to obtain a second approxi- 

 mation ; this, however, is not necessary with our present accuracy of observation. 



OBSERVATIONS. 



Supposing that the line of collimation of the microscope occupies a fixed position, 

 the variables which have to be observed at any one place are the temperature of the 

 interior of the instrument as given by the platinum thermometer, and the amount of 

 twist in the thread necessary at any moment to bring the image of the end of the 

 lever coincident with a cross wire in the eye-piece of the microscope. This latter 

 factor is given by the reading of the position of the vernier arm on the sextant arc. 

 We have found that the relation between these variables depends very greatly not 

 only on whether the temperature is rising or falling, and on the rate of the change, 

 but also, theoretically at all events, on all the previous variations of temperature 

 since the last steady state. In our instrument, the platinum wire, after a change of 

 temperature, assumes its new resistance appropriate to the final temperature more 

 quickly than the lever takes up its new position. 



Plot 1. 



* ' falling. 



In Plot 1 are given observations showing the relation between the temperature and 

 the position of the vernier arm when the temperature is first rising and afterwards 

 falling. The change of temperature on October 17th was due to a "southerly 

 burster" coming up after a hot westerly wind. On the 18th the change is the 



