674 PROCEEDINGS OF THE AMERICAN ACADEMY. 



nearly a one-thousaiulth of a degree. Further, the stirring of the mixture, 

 though lasting for only a minute, must have exerted a warming effect. 

 This also was found to amount to about one one-thousandth of a degree. 



These two cou>tant effects, the further possibility of the cold li(|uid 

 not being exactly at zero degrees, constant thermometric erroi-s, and other 

 hiddden errors, were completely eliminated by making parallel determi- 

 nations, using water in each pipette. Had it not been for these small 

 constant errors, we should of course have obtained relative specific heats 

 of the hot and cold water whose ratio corresponded to the ratio of the 

 known mean specilic heats of water between these temperatures, provided, 

 that is, that the knowMi value also is free from such errors. As a matter 

 of fact, we found that these ratios were not exactly equal, but that they 

 differed by a small constant amount, — less than 0.05 percent.* The 

 small magnitude of this difference is equivalent to proof that the appara- 

 tus just described gives about the same relation between the specific heats 

 of the warm and cold water as has the work of others, and forms an 

 excellent test of the efficiency of the present apparatus. 



The cum[)utation with water in both pipettes is very simple; but, as 

 has been said, the simple heat capacity of a solutioa is not determined 

 by this method, for the heat of dilution is superposed when the cold 

 solution is mixed with warm water. It is necessary, therefore, to make a 

 separate measurement of the heat of dilution. In spite of the ease with 

 whicii this latter measurement can be made, this might, under some con- 

 ditions, be considered a drawback, and for that reason it was originally 

 planned to have the liquids equalize their temperature without uniting. 

 For this purpose there was devised a cylindrical vessel divided into two 

 compartments by means of a thin, greatly convoluted partition. Since iu 

 the present work the heat of dilution was also of interest and value, as 

 will be shown, the determination of the heat of dilution at this time 

 involved no extra labor. For this reason the specific heats of solutions 

 were determined by actual mixing, and corrected for the heat of dilution. 



In passing, it may be noted that this apjiaratus could be used for 

 determining the heat of reaction of solutions upon one another, when 

 thermometric lag is to be excluded. One of the solutions could be placed 

 in the calorimeter at the temperature of the room, and the other in one of 



* Tlie standard of comparison was the talde clioscn by Ostwald and Lutlier as 

 representing tlie true values (Pliys. Cliem. Mess.,p 195 [1902]). Our value for the 

 ratio, 1 : 1.0059, was between the value derived from this table and the lower value 

 to be derived from the work of Callendar and Barnes (Zeitsciir. i)liys. Chcm.. 

 32, 161 [1900] ). 



