the Tait, and the Amagat High-Pressure Gauges. 403 



parallel to the line of motion of the free end of the Bourdon 

 tube, which line was found to be about *4 centim. long, and 

 inclined at an angle of nearly 45° with the tangent at the free 

 end. The long slender conoidal plug, at the end of which 

 these measurements were made, was about 10 centim. long. 

 Of the two series given in Table II., the first shows some 

 convexity upward in the on march, where the mean factor is 

 •000362 cm./atm. The data of the off march show dis- 

 tinct convexity downward. Thus the curves form a cycle 

 whose mean breadth is 40 atmospheres, agreeing substantially 

 with § 4. The data of the second series, which show rather 

 better agreement among themselves, in other respects cor- 

 roborate the first series throughout. The factor is -000361 

 cm./atm. Many other measurements both within 500 

 atmospheres and 1000 atmospheres were made, from which 

 the absence of cycles in case of the smaller interval was 

 again manifest. 



The bow-shaped cycles obtained in the present and the 

 earlier paper* are therefore cases of actual hysteresis. I 

 venture to regard the phenomenon as depending on the 

 occurrence of volume-lag in those parts of the tube which 

 are directly influenced by pressure. In other words, the 

 molecules of the metal near the inside of the Bourdon tube 

 pass from the original to a second molecular state, in pro- 

 portion as a certain pressure P is approached and exceeded ; 

 whereas these molecules pass from the second state back to 

 the first again, in proportion as pressure falls below a certain 

 other datum p, where P>/>. Tables I. and II. show that 

 P must lie somewhere between 500 atm. and 1000 atm., 

 whereas p may lie below 300 atmospheres. Here, therefore, 

 is a phenomenon very similar to the effect of pressure on an 

 under-cooled liquid |. 



* Phil. Mag. [5] xxx. p. 344 et seg. (1890). See plate x. 



t Two other possible explanations of the cyclic changes may be noted. 

 In the first place it is conceivable that the Bourdon tube is hotter during 

 the off march than during the on march. To test this directly it would 

 be necessary to submerge the tube in a water-bath, which is scarcely 

 feasible without impairing the mechanism. I hold this explanation im- 

 probable; for if it were true, the comparison within 500 atmospheres 

 should show a cycle proportional in magnitude to the cycle for the interval 

 1000 atmospheres, which it does not. Again, cycles should quite vanish 

 on long waiting (five minutes were allowed per observation in my earlier 

 work), which they do not. The other explanation seems equally im- 

 probable. There may be two figures of equilibrium of the Bourdon tube, 

 one corresponding to low pressure, the other to high pressure ; and the 

 figure of labile equilibrium through which the tube passes from the first 

 to the second may be maintained over a relatively large interval of 

 pressure. 



2G2 



