I'fiijcli roiiirl rlc Fornndii. 159 



When this methnd of reduction was first recoj^nised as more 

 (•orret-t. it was not known that any otlier experimenter had used it. 

 lUit on investigation, the same method, with a small alteration, the 

 reason for which I oould not discover, was found given without 

 remark in Fen-el's paper on '' Psychrometrical Tables,"! though 

 there are some obvious misprints. The method used by Regnault and 

 most others is not stated in their papers : on the other hand, Angot2, 

 Pernter' and Svensson'' certainly used arithmetic mean values. For 

 this reason I have thought it well to call attention to the discrepancy 

 between the two methods, though the actual results may not be 

 much different in a good series of observations. For investigating 

 the effects of wind-velocity and other circumstances, the same method 

 is appropriate, and was used in all further study of my own observa- 

 tions. 



Results. 



The value of A derived from the 103 observations recorded in this 

 paper is 0.0007232 ± 0.0000048. Taking the mean of individual 

 values, the result is higher — viz., 0.01M>7330 ; but the probable errors 

 of a single observation of x in the two cases are respectively 0.229 

 mm. and 0.231 mm., so that the difference in the value of A is of 

 small practical moment. The value given in the previous paper 

 was 0.0007228 : using the more correct method of reduction it would 

 become 0.0007167. From the whole set of 166 observations taken 

 together the resulting value is 0.0007227 ± 0.0000043. Thege 

 various values all agree when only two significant figures are taken, 

 and that is all that can be regarded as really valuable. The final 

 result is then that A = 0.00072, with a probable error of about 

 half a unit in the last place, i.e., (72 ± .1) x 10—5. The equation 

 thus becomes 



x = f-0.00072}i(f-O- 



Applying the two-constant formula, the values of r; and of A are 

 found to be 0.9877 and 0.0006967 respectively. For the €3 observa- 

 tions of the earlier series. // had the value 0.9974. The lower value 

 now obtained might be regarded as due to incomplete saturation of 

 the air leaving the wet-bulb, or some similar failure in the action 

 which is assumed in theory to occur. But it seemed scarcely likely 

 that this would be more noticeable in the present «eries than in the 

 other one, since the later observations were distinctively superior in 

 other respects. In order to determine whether the lessened value 

 might be due to the observations at low humidity, the series was 



1 Kerrel, Report of Secretary of War, V.S , 1886, vol iv., p 233 



2. Aiijrot, J de Physicpie, 1, l!SS-2, p 119 



3. Fernter, Sitzunu'sber. Wiener .\kad. 87, 188;i 



4. Sveiisson, Meteor. Zeitschr. 1896, p. 201. 



