1903.] Specific Heats, especially at Low Temperatures. 191 



standing a few inches above a fixed calorimeter, the latter protected 

 from the direct radiation of the former (except during the short time 

 occupied in transferring the hot body to the calorimeter) by a screen 

 of metal and wood forming the base of the heater. The hot body 

 could be very rapidly lowered into the calorimeter, probably without 

 appreciable cooling. A chief cause of uncertainty appeared to lie in 

 the temperature of the hot body ; this was read on a thermometer 

 with its bulb very near the body, but there was no doubt that 

 the thermometer acquired the temperature of the surrounding hot 

 air much more rapidly than did the comparatively massive piece of 

 metal. To eliminate this source of error, the heating process was 

 generally continued for 3 or 4 hours, and in certain cases for 6 hours. 



The thermometer used in the steam-heater was graduated in tenths 

 of a degree from 95 to 105, the length of a degree on the scale being 

 approximately 24 mm. It had been examined at Kew, but the cor- 

 rections actually applied were obtained by the frequent use of the 

 hypsometer, which showed during the determinations a progressive rise 

 in the steam-point amounting altogether to J. 



In the calorimeter a Beckmann thermometer was used. It was con- 

 venient in many of the experiments to use a different instrument from 

 that used in the low temperature determinations. This second instru- 

 ment was carefully standardised in the same manner as the first, but 

 unfortunately proved to have a bore far from uniform. The results 

 obtained with it are those for copper (all), thallium (both), aluminium 

 approx. pure (first result), aluminium commercial (both), nickel (both), 

 zinc approx. pure (both). 



The quantity of water used was 200, 300 or 400 grammes, and the 

 rise of temperature 1 0< 1 (thallium) to 5 -6 (nickel). 



Table VIII gives the results of all the experiments made with the 

 exception of the following : (1) preliminary tests of the method, made 

 with copper, thallium, lead commercial, zinc commercial ; (2) one 

 experiment with lead pure, result 0'0301, rejected on the ground of 

 probably insufficient time of heating; (3) one experiment with tin 

 commercial, result 0*0551, rejected on the ground of an unduly long 

 transference. 



Dulong and Petifs Law. 



In Table IX the specific heats of Tables IV and VIII are compared. 

 The metals are here arranged in order of atomic weight. The atomic 

 heats, calculated from the international atomic weights (oxygen 16) 

 for 1903, are added. So far as the metals examined are concerned, 

 the table shows that the proportional alteration of specific heat is 

 relatively large for metals of low atomic weight. Hence the approximate 

 constancy of atomic heat for the higher temperature range is not 

 preserved for the lower temperature range. In fact for the lower 



