1888.] S. A. Hill — Psychrometer and Condensing Hygrometer. 



379 



in which the pressure was above 24'89 inches, the average value of all 

 the pressure observations, we find the mean value of A to he '00091, and 

 all the remaining observations give a mean of '00097. 



From these figures it is abundantly evident that it is not the com- 

 bined factor Ah oi the second term of the formula which is constant, as 

 Hazen supposes, and as was tacitly assumed by Grlaisher in constructing 

 his table of empirical factors, but only A that is so ; and, wherever the 

 pressures are considerably less than at sea level, Glaisher's factors, or 

 any table constructed on the assumption that variations of pressure are 

 of no account, must lead to erroneous results. 



The factor A is thus nearly if not quite independent of pressure, 

 but varies with the amount of ventilation up to a certain moderate 

 velocity of wind, after which it appears to remain constant, except per- 

 haps when the humidity is very low, as during the hot winds. With a 

 view to testing Regnault's opinion that the formula adapted for ordi- 

 nary open air conditions gives too low results when the air is very dry, 

 we may tabulate the values of A, given in Table II., according to the re- 

 lative humidities deduced from the temperature of the air and the dew 

 point. This is here done, the observations being divided into two sets by 

 the limit of 40 per cent, humidity, supposed by Regnault to be that 

 below which his formula was inapplicable. 



Series. 



No. 



Relative 

 Humidity. 



A. 



Series. 



No. 

 4 



Relative 

 Humidity. 



A. 



II. 



4 



7o 



8 



•00077 



1 



7o 



43 



•00072 





3 



9 



•00073 



}} 



3 



53 



•00084 





7 



14 



•00087 





5 



55 



•00092 





1 



15 



•00106 



II 



19 



56 



•00090 





2 



16 



00105 





20 



56 



•00118 



I. 



2 



19 



•00098 





10 



58 



•00117 



II. 



6 



19 



•00121 



jj 



18 



58 



•00119 





15 



22 



•00077 



) J 



9 



59 



•00077 





5 



23 



•00124 





8 



65 



•00105 





14 



24 



•00112 



,, 



21 



87 



•00100 



>> 

 )) 

 if 



17 

 16 

 11 

 13 



24 

 26 



27 

 29 



•00089 

 00097 

 •00107 

 •00080 



)> 



22 



92 



•00075 





12 



30 



•00082 











I. 



1 



32 



•00083 











Mean for observation with R. 

 H. below 40 % 



j -00095 



Mean for obscrvati 

 H. above 40 % 



3ns with R. 



1 -00095 



The large and small values of A are not distributed in this table 

 according to any regular law, and the means of the two columns are as 

 nearly as possible identical. It seems probable therefore that, when 



