278 PROCEEDINGS OF THE AMERICAN ACADEMY. 



very point at which the jump occurs ; in fact, he had to renew many of 

 its parts, and to watch it continually thereafter, so that his conditions 

 may well have been somewhat changed. This discontinuity in his 

 curve has been noticed by many writers, one of whom attributes it to 

 a leak in his distributing valve, remedied at this point ; but this is not 

 definitely mentioned in the memoir. 



The recent publication by Henning of his measurements of L between 

 100° and 180° gives a valuable test of the new formula. All his 

 values in both papers are collected in the second column of Table II. 

 They are expressed in terms of a calorie of 4.188 19 international 

 Wattseconds. It is probable that the mean calorie (0° to 100°) is 

 about 4.184(2) international Wattseconds, for the fine work of Rey- 

 nolds and Moorby 20 by a mechanical method, leads, according to 

 Smith, 21 to the value 4.1836, while the equally good work of Barnes 22 

 by an electrical method must now 23 be regarded as leading to the 

 value 4.1849. Each of Henning's numbers should therefore be multi- 

 plied by 4.188 / 4.1842 = 1.00091. The results are given in the third 

 column of Table II. They lead to the values of H in the fifth column. 

 In the sixth and seventh columns are given the values at the cor- 

 responding temperatures of 



H - ff 100 = 0.3745 (t - 100) - 0.000990 (t - 100) 2 . 



and of #100 itself. The latter is practically constant as it should be if 

 the new formula is true. It will be noticed that the probable error of 

 the mean value of H is only one thirteenth of one per cent of that 

 mean, and that this agreement is within the one tenth of one per cent 

 which Henning claims for his observations. It will further be noticed 

 that practically all of the discrepancy is in two of the last three values. 

 If all three of these values are omitted, so that the range of the test is 

 cut down to that between 65° and 121°, the probable error of the new 

 mean is only ± 0.19 calories, or one thirty-fourth of one per cent. 



In estimating the significance of the comparatively great disagree- 

 ment between Henning's value at 180° and the new formula it should 

 be remembered that Henning himself says, "Bei der hochsten Tem- 

 peratur von 180° konnten nur an zwei tagen Versuchen angestellt 

 werden " (instead of on four days as at most of the other temperatures). 



19 This is Jager and von Steinwehr's value for the 15° calorie. The justi- 

 fication for it has not yet been published. 



20 Phil. Trans.. 1897, 190 A, 301. 



21 Monthly Weather Review, 1907, 35, 458. 



22 Phil. Trans., 1902, 199 A, 149. 



23 Proc. Roy. Soc, 1909, 82 A, 390. 



