128 C. Barus — Counter-twisted Curl Aneroid. 



Proceeding thus, I found a = 0*001 ; and with this value I 

 reduced the observations by placing (1) under the form 



s=— - — -a— s =Bc — s (3) 



1 + at v J 



Hence s is a linear function of B; and this postulate can be 

 tested graphically without computing c and s . B is given in 

 the last column of Table 3. 



The conditions under which these data were obtained were 

 very unfavorable. As a result of the delicate suspension, the 

 jar of wagons and cars passing the laboratory often made it 

 impossible to read the scale millimeters in the telescope. 

 Moreover I was unprepared for so large a temperature coeffi- 

 cient, «, as has just been adduced ; and I did not therefore 

 take such special precautions against currents of air which 

 should have been taken, seeing that the thin metallic helix 

 adjusted itself at once to temperature, whereas the thermome- 

 ter follows sluggishly after. Nor was the latter placed near 

 enough to the coil. Hence since a single degree of tempera- 

 ture corresponds to nearly 0'l cm of the barometer, discrepancies* 

 of half a millimeter of pressure must have been incident to the 

 work, and the table and chart bear this assertion out. Points 

 which lie out of position are isolated in each group and the 

 discrepancy is the result of a current of air. 



In spite of the unfavorable choice of metals (copper and 

 brass) the behavior of the copper curl is thus seen to be satis- 

 factory so far as the present purposes go : for a development 

 of principles of construction has only been aimed at. The tel- 

 escope reading for this small curl is 9'8 cm per cm. of the mer- 

 curial barometer. 



12. Conclusion. — Having reached this stage of progress, it 

 seemed expedient to break off the work, for it would not be 

 worth while to proceed with fine measurement without making 

 the system of more highly viscous material at the outset. In 

 brief : tubing preferably of resilient brass 3-4 millimeters in 

 diameter or less, with walls 0'01 cm thick or less (by solution) is 

 desirable. The tube should be heated in steam to remove 

 excesses of drawn strain and thereafter flattened and coiled 

 until the walls all but touch. The counter-twisting spring 

 is to be either helical or watchspring-shaped, and of steel 

 annealed at 400° C, or a suitable long quartz fiber. A long 

 curl would not be self-sustaining ; but it could be made so 

 without interfering with its free action, by a series of equidis- 



* In the face of the large thermal effect discussed in the text it would be use- 

 less to endeavor to evaluate the effect of viscosity. Nor would a metal of low 

 viscosity, like a spring of drawn brass, be chosen in tests of a final character. 



