18 EXPERIMENTS WITH THE DISPLACEMENT INTERFEROMETER. 



adequate indication of increased sensitiveness due to larger period T which 

 is in question. It will thus be necessary to endeavor to account for the anoma- 

 lous apparent variability of a observed in these experiments with the float. 



11. Effect of temperature on the float, etc. It will be necessary at the 

 outset to obtain the changes of buoyancy due to corresponding changes of 

 temperature of the water in which the float is submerged. To obtain the 

 change of buoyancy with temperature, the following table of increments may 

 be consulted, where the normal temperature is taken as 20 and V=8o2 cm.: 



15 + .706 



20 .0 



25 - .914 



30 -I.Q97 



The total effect of temperature between 10 and 30 would thus be but 

 slightly over 3 grams. This may be estimated to act upon a lever arm not 

 exceeding i cm., the endeavor having been made to keep the float axial. 

 Thus we may assume that a moment of 3X981 dyne cm. would not be ex- 

 ceeded in any variation of temperature, the moment being 0.15X981 per 

 degree centigrade. 



Direct preliminary experiments on the effect in terms of the deflection x of 

 definite moments around a horizontal axis were made by placing 10 gram 

 weights on the pan (i, fig. 7) carrying the float, at a distance of nearly 5 cm. 

 on either side of the vertical axis of rotation. The effective moment is thus 

 95X981 dyne cm. The successive deflections were (differences due to devia- 

 tion of pendulum during observation) 



10 grams in front, # = 3.4 cm. ro grams in rear, = 35.4 cm. 



6-7 33-9 



This is equivalent to # = 29.6 cm. for the given torque, or 



of deflection per dyne-centimeter of torque. Hence, using the preceding esti- 

 mate, the deflection should be 



o T.I 



-~- Xo 15X981 = 0-46 cm. 



901 



per degree centigrade per centimeter of eccentricity, which would be equivalent 

 to about x= i cm. for a range of temperature from 10 to 30. It does not 

 appear, therefore, even if the eccentricity of the float is greater than was 

 assumed, that the temperature decrement can be a menace. 



The endeavors to measure the temperature discrepancy directly were all 

 failures, inasmuch as, during the long time of cooling, the deviations of the 



