the Tait, and the Amagat High-Pressure Gauges, 405 



7. Table III. substantiates the general inferences of § 4. 

 In the first series, above 300 atm. on, the data make up a 

 straight line. Between and 300 atin., on, the registration 

 of the Tait gauge is too high, being more in error as the 

 pressure-zero is approached. In the off series all the points 

 lie on a straight line excepting the zero, which is 10 atm. too 

 high on the Tait gauge. Thus the locus consists of two 

 straight lines converging at 1500 atm., and the cyclic cha- 

 racter in § 3 does not appear. In the second series these 

 results are sustained, though the off march returns in the 

 line of the on march. As a whole, therefore, the observa- 

 tions lie on a very flat curve, which above 300 atm. passes 

 into a straight line. It is difficult to conjecture the cause of 

 this curious low-pressure discrepancy, in consequence of the 

 number of minor errors wdiich may be encountered. Possibly 

 the rise of temperature of a liquid, per 100 atm. of adiabatic 

 compression, may decrease with increasing pressure, seeing 

 that at low pressures compressibility is relatively large. 

 Table III. shows, finally, that the mean factor derived from 

 all the data is '03596 cm. /atm. If this be increased *4 per 

 cent, to allow for the temperature of the mercury column of 

 the manometre, the more correct value is '0361 cm./atm. In 

 my former paper * the results were '0360 to '0369, showing 

 that my empiric standard atmosphere must have been less 

 than 2 per cent, in error. It is altogether probable that this 

 difference of factor is the result of the excessive use and abuse 

 to which the Tait gauge was put during the intervening nine 

 months. Indeed I believe the residual errors of Table III. to 

 be due rather to unavoidable fluctuations of the temperature 

 of the water-jacket than to thermal effects of compression or 

 viscosity of metal. For this reason I doubt whether any 

 permanent expansion of the metal for the range 1500 atm. 

 has been measured. 



8. Summary. — In the attempt to carry these comparisons 

 above 2000 atm., I ruptured the stout tube which coupled the 

 compressor and the manometre. The above work, however, 

 shows conclusively that the Tait gauge may be perfected so as 

 to be available for precise measurement. It is perhaps per- 

 missible to indicate the way in which I am endeavouring to 

 do this here, since many of the experimental contrivances 

 involved have already been tested. Thus far the most annoy- 

 ing difficulty is encountered in maintaining the temperature 

 of the water-jacket sufficiently constant. A point of progress 

 will be gained when the whole gauge (excepting the capillary 



* Phil. Mag. [5] xxx., Table IV., p. 345 (1890). In this table rend 

 10 2 X cm. atm. instead of cm./atm. 



