102 PROCEEDINGS OF THE AMERICAN ACADEMY 



sion, and unreasonably hasty work. To diminish this error further, 

 it is necessary to decrease the bore of the glass tube tt, Figure 4, 

 thus making the thermometer feature of the gauge of smaller mo- 

 ment. § 18. In the above apparatus this tube was 1.15 cm. in 

 internal diameter, corresponding, therefore, to a liquid interstratum 

 .07 cm. thick. Clearly, this admits of further reduction. It is also 

 feasible to decrease the strength of the steel tube, making the ap- 

 paratus more delicate. 



16. A second point of view is obtained by comparing the data of 

 the pressure " on " and the pressure " off " phases of the above experi- 

 ments. For convenience, I have constructed the results of Table I. 

 in the chart. Figure 6. The degree of accordance in the "on" phase 

 is satisfactory, since the errors are not above 10 atm., and the loci, 

 apart from slight circumflexure attributable to the Bourdon mechan- 

 ism, are straight lines. In the " off " series, this good uniformity is 

 lost, and the data lie on consecutive broken lines. The " off " data 

 do not return in the lines of the "on" data. Indeed, the two curves, 

 "on" and "off," enclose a band the maximum width of which is 1.5 

 cm., corresponding to 90 atm. The interpi-etation of this curious and 

 persistent phenomenon is therefore of importance. 



17. With this end in view, I again compared the gauges after 

 several months' use. The data are contained in Table II. At least 

 five minutes were allowed per observation. Two complete series are 

 given. 



The second part of the table gives some data on the constancy of 

 the zero after very high pressure. The slight shifting of the zero 

 observed (equivalent respectively to 10 and to 5 atmosplieres) is due 

 to a mechanical imperfection which I subsequently remedied. Hence 

 the gauge is vouched for within 2,000 atmospheres. 



It is seen at once that the mean data of Tables I. and II. are 

 identical. Hence, to account for the apparent hysteresis observed, 

 it is necessary to construct other Tait gauges. I did this with 

 much care. 



18. New Tait Gauges. — The new gauges are numbered 1 and 4. 

 No. 1 is nearly of the same length as No. ; No. 4 is considerably 

 shorter. The glass tubes {tt, Fig. 4) are as close fitting as practica- 

 ble, and the capillary of No. 1 is of small bore, to insure special accu- 

 racy in reading. Of the many series of observations made, two for 

 No. 1 are given in Table III. Table IV. contains a comparison of 

 Nos. 1 and 4. Five minutes were allowed per observation. The 

 capillary tubes were carefully calibrated. lu the second part of 



