124 C. Bar us — Counter-twisted Curl Aneroid. 



9. — To obtain preliminary evidence, I took curl IV, just dis- 

 cussed (§ 8), though it had become somewhat leaky from hard 

 usage. The counter-twisting was done by a long steel wire 

 with the following results : 



Turns uncoiled from j Pressure per degree, per turn, 13*4 cm , Hg. 

 4*5 to 3*7. ( Pressure per degree, per linear cm., 167 cm . 



Turns uncoiled from j Pressure per degree, per turn, 5'l cm , Hg. 

 4*5 to 3*0. ( Pressure per degree, per linear cm., V9 cm . 



In all these cases the deflections were instantaneous, and 

 there was no apparent viscosity. The curl itself has been 

 improved from one of low sensitiveness (330 cm ) to nearly the 

 same range of high sensitiveness (77 cm ) ? actually obtained in 

 § 6 for extremely thin copper tubes. 



This result deserves special study; but it already appears 

 that so long as the coil is sharp-edged in section — so long as the 

 strain is a case of nearly pure bending — the coil increases in 

 sensitiveness as its spindle-shaped section is more highly 

 arched. The smaller the medial radius of curvature of a right 

 section of the tube, the greater proportionately is its variation 

 for the same pressure-difference ; and the greater proportion- 

 ately must the corresponding variations of the longitudinal 

 radius of curvature (coil radius) also be, since the product of 

 two radii is to remain constant. Hence a rotundly arched^ 

 spindle-shaped section, maintained in an excessively thin-walled 

 tube, is compatible with the greatest amount of rotation at the 

 registering index. 



10. — A copper tube having walls # 01 cm thick was flattened 

 and rolled down as usual. The curl (No. VII) contained 9 

 turns and it was 2'8 cm in diameter and about 12*5 cm long, 

 each turn in section being 0*8 cm high, with blunt edges about 

 0*05 cm wide. So constructed, the sensitiveness (conformably 

 with the data in § 8) was low, for the pressure in cm. Hg per 

 degree per turn was 50, and the pressure per degree per linear 

 cm. of the curl 440 cm . — in spite of the thin copper walls 

 stated. This tube was now hammered flat and sharp-edged on 

 the steel mandrel. A part, VII & , free from flaws was then cut 

 from this tube and used in the following tests, Table 2. For 

 want of better material a helical spring of brass spring wire 

 was used in the counter-twisting and the system was weighted 

 to prevent knotting (cf. § 8). 



These results are given graphically in the chart figure 3, 

 where the ordinates are pressures in cm. of mercury, corre- 

 sponding to a deflection of 1° for a length of l cm measured on 

 the turns of the curl. The abscissas show the amount of 

 counter-twisting at the lower end of the system in degrees. 

 The sensitiveness therefore increases in marked degree with 



