Knoivledge of the Specific Volumes of Steam. 17 



equal to the specific heat at about 20°. His experiments are 

 as concordant as those of Dieterici. 



Let us now compare Regnault with these other investigators. 

 The table below gives the fractional deviations of c Q _ l00 accord- 

 ing to them from 1*00358 c lb . 



(1) Experiments which are unreliable. 



Von Muenchausen , (mixtures) +0-0113 



Velten (mixtures) —00139 



Henrichsen (ice-calorimeter) +0*0214 



Yelten (ice-calorimeter) —0*0106 



Baumgartner (mixtures) +0*0124 



Joly (steam-calorimeter) — 0*00*79 



(2) Experiments in which evidences of unreliability are not 

 present. 



Rowland (mixtures) +0*0005 



Ludin (mixtures) +0*0018 



Dieterici (mech. equiv.) +00044 



Reynolds (mech. equiv.) —0*0050 



The first set are grouped so far as possible into pairs of equal 

 probability and nearly balance. Of the second set the last two 

 balance, and the deviations in the first two are very small. 

 As the numbers stand the positive deviations seem to rather 

 outweigh the negative ; some additional evidence on the nega- 

 tive side is afforded by the following: Sahulka* has found for 

 the mechanical equivalent at 58° the value 426*262 kilogram- 

 meters, or 41*81 megalergs, thus making c 58 equal to 0*9981 c lf> . 

 Assuming the specific heat to vary linearly with the tempera- 

 ture from 35° to 100°, this would make the mean specific heat 

 from 0° to 100° equal to 0'9992 c 1B , the fractional deviation 

 from the value we have assumed being thus —0*00436. This 

 evidence is not entitled to very much weight, however, because 

 Sahulka's experiments deviate from their mean over one-half 

 of one per cent. 



It is thus seen that the assumption that Reguault's thermo- 

 metric scale was that of the air-thermometer makes his value of 

 c _ ]00 agree well with that obtained by others. It cannot be 

 more than 0*5 per cent out, and is probably much more accu- 

 rate than that. Any error in it should be ascribed to the unit 

 in which it is expressed rather than to inaccuracies in the 

 experiments themselves. 



For temperatures below 100° the writer has formed for his 

 present purposes the following formula : 



h — 1-00449*— 0*0001 904 f + 0*000001813 t\ 



* Wied. Ann., xli, p. 748, 1890. 



Am. Jour. Sci. — Fourth Series, Vol. VII, No. SI. — January, 1899. 

 2 



