102 BRIDGMAN. 



greater lag than he did in his later experiments. He was not able to 

 repeat the early experiments, and was able to explain them only by 

 supposing that the particular specimen of machine oil which he had 

 used to transmit pressure had some peculiar property which the later 

 specimens did not have. 



A comparison of the numerical values of Tammann's later paper and 

 the values found above gives the following results. Tammann found 

 the change of volume I-II between 1000 kgm. and the triple point to 

 be constant within the limits of error at 0.0095, the probable error 

 being 5%. I found above that AT' increases on the transition line 

 from 0.0086 at atmospheric pressure to 0.00102 at 3000. Tammann's 

 triple point is 2940 kgm. and 100°, against 2810 and 99.4° above. 

 Tammann found the I-II line straight within the limits of error, with 

 a slope of 0.01473; I found the slope to increase in absolute value from 

 0.0146 at 1 kgm. to 0.0182 at 3000 kgm. Tammann calculates the 

 slope of the I-III line at the triple point to be 0.0195; I find above 

 0.0140. The slope of the II-III line Tammann gives as —0.75, and 

 I find it to vary from —.415 to —.273. At the triple point Tammann 

 gives for the change of volume (uncorrected) of II-III, 0.0205, and 

 for I-III, 0.0115, against 0.0241 and 0.0140 above (corrected). Tam- 

 mann, however, had to leave an unexplained discrepancy of 7-10% 

 in the AV values for II-III, the change of volume being different 

 for the two directions of reaction. 



For the values of Aa, A/S, and ACp, two methods of estimation are 

 available. At the triple point the various quantities may be com- 

 puted, and there are also experimental determinations of the difference 

 of compressibility. Neither of these methods gives satisfactory 

 results, however. The calculations at the triple point are uncertain 

 because the variation of AV along the II-III line is so slight that even 

 its sign is uncertain. The experimental values of the difl'erence of 

 compressibility are irregular and uncertain. We can, however, make 

 the following rough statements at the triple point. I is more com- 

 pressible than II, the difference being of the order of 0.065. The 

 difference of expansion between I and II is very small and is uncertain 

 as to sign. The specific heat of I is greater than that of II, and the 

 difference is of the order of 0.3 kgm. cm. per gm. Ill is more compressi- 

 ble than I, and the difference is fairly large, of the order of O.O55, 

 but there is a good deal of uncertainty as to the numerical value of 

 this difference. Ill is less expansible than I, the difference being of 

 the order of O.O44 and III has a smaller specific heat than I, of the 

 order of 0.3. From these values, of course, the differences between II 

 and III may be found immediately by a subtraction. 



