352 



Mr. Carl Barus on the 



day of observation, and if N# be the corresponding twist on 

 the day of final calibration, both observed at the same torsion- 

 galvanometer, then Nfl/N/ is the reduction factor. 



To test this inference I adduce two independent calibra 

 tions, over a month apart. If N 

 t = 20°, the data are expressed in Table VI 



a(T-0+6(T 2 -a,and 



Table VI. — Behaviour of the Torsion-Galvanometer*. 

 = •0844 „;__ aail j a = -0829 



April 20th {« Jo000 717 May 2Uh { 6 = -0000694 



T. 



N*'. 



N„ 



Ne/Ne'. 



T. 



w. 



Ne. 



Ne/Ne'. 



°C. 





o 





°C. 



o 







50 



2-683 



2-632 



•9810 



250 



23-865 



23-377 



•9796 



100 



7440 



7-298 



•9809 



300 



30-057 



29-430 



•9792 



150 



12-557 



12-311 



•9804 



350 



36-611 



35-831 



•9787 



200 



18-031 



17-670 



•9800 



400 



43-520 



42-578 



•9784 



The change of 'Ne/^J is less than *3 per cent, for the whole 

 interval 50° to 400°, a satisfactory result. Small reciprocal 

 changes of a and b would wipe out this discrepancy without 

 appreciably slighting the observations. 



. 36. A more serious question, relative to the degree of truth 

 of the interpolation equation Ne = a(T — £) +b(T 2 — t 2 ) remains 

 for solution. This can only be decided by direct air-ther- 

 mometer comparisons. I used a re-entrant air-thermometer 

 bulbf, and determined constancy of volume by aid of an 

 electric contact. The bulb was heated in a closed cylindrical 

 paraffin bath, 7 inches in diameter and about 7 inches long, 

 heavily jacketed (1 inch thick) with asbestos. The disposition 

 met was such as placed the thermoelectric junction at the 

 centre of the bulb, which in its turn was at the centre of the 

 bath. Observations were made in time series during the 

 period of cooling down from 300°. I give an example of 

 one of the three series of comparisons made in Table VII., 

 where T 0> * is the temperature measured by the air-thermome- 

 ter, N0 the corresponding twist at the torsion-galvanometer. 

 T t , c is the temperature computed from ~N e , by aid of Table VI. 

 Again, V is the zero volume of the air-thermometer bulb, v' 

 and v" the volumes of the hot and cold parts of the stem and 

 capillaries, and h is the tension of the gas at zero Centigrade. 



* Computed from observations at 100° and 327°. 



t Bull. U.S. Geolog. Survey, No. 54, pp. 168 et seq., 188 et seq., 1889. 



