132 PROCEEDINGS OF THE AMERICAN ACADEMY 



much more complicated pattern. The total capacity of the solder 

 used was only about ^fa of the total capacity including the water; 

 and if we should neglect the whole, and call it copper, the error would 

 be only about -r^^. Hence it was considered sufficient to weigh the 

 solder before and after use, being careful to weigh the scraps. The 

 * error in the weight of solder could not possibly have been as great as 

 ten per cent., which would affect the capacity only 1 part in 12,000. 



To determine the nickel used in plating, the calorimeter was weighed 

 before and after plating; but it weighed less after than before, owing 

 to the polishing of the copper. But I estimated the amount from the 

 thickness of a loose portion of the plating. I thus found the approxi- 

 mate weight of nickel, but as it was so small, I counted it as copper. 

 The following are the constituents of the calorimeter: — 



To determine the mean specific heat, the basket of a Regnault's 

 apparatus was filled with the scraps in the above proportion, allowing 

 the basket of brass gauze, which was very light, to count toward the 

 drawn brass. The specific heat was then determined between 20° 

 and 100°, and between about 10° and 40°. Between 20° and 100° 

 the ordinary steam apparatus was used, but between 10° and 40° a 

 special apparatus filled with water was used, the water being around 

 the tube containing the basket, in the same manner as the steam is in 

 the original apparatus. In the calorimeter a stirrer was used, so that 

 the basket and water should rapidly attain the same temperature. 

 The water was weighed before and after the experiment, to allow for 

 evaporation. A correction of about 1 part in 1,000 was made, on 

 account of the heat lost by the basket in passing from the apparatus 

 to the calorimeter^ in the 100° series, but no correction was made in 

 the other series. The thermometers in the calorimeter were Nos. G1G3 

 and C1GG in the different experiments. 



The principal difficulty in the determination is in the correction for 

 radiation, and for the heat which still remains in the basket after some 



