ABSOLUTE THERMAL CONDUCTIVITY OF NICKEL. 



365 



numbers found in the curve, using the formula 



d*6 , d (log. 6) J d (log. 6) d i d (log. 6) \ ) 

 dx 2 dx \ dx ^ d$\ dx ) ] 



The temperatures given in the tables are actual temperatures, not temperature excesses. 

 They have been got by adding 19 — about the average temperature of the air during the 

 experiments — to the numbers used in the curves which were differences of temperature 

 between the bar and the air. 



Temperatures. 



d (log. 0) 

 dx 



d?d 

 dx 2 



Temperatures. 



d (log, e) 



dx 



d 2 e 



dx 2 



40 



•0335 



•0239 



110 



•0370 



•138 



50 



•0340 



•0369 



120 



■0375 



•159 



60 



•0345 



•0509 



130 



•0380 



•181 



70 



•0350 



•0661 



140 



•0385 



•205 



80 



•0355 



•0823 



150 



•0390 



•230 



90 



•0360 



•0997 



160 



0395 



•256 



100 



•0365 



•1183 



170 



•0400 



•284 









180 



•0405 



•313 









190 



•0410 



•344 









200 



0415 



•376 



§ 4. The Cooling Experiment. — For this experiment a short bar turned down from 



a, left-over portion of Dr Knott's nickel bars was used. It was a piece of the same rod 



is the bar used in the statical experiment : it was turned down in the same way, at the 



same time, and to the same diameter, as was found by careful measurement. The length 



}f the cooling bar was 21 "55 cm., and as its diameter was 4 '67 cm., the surface exposed 



it the ends was 9f per cent, of the whole surface. This involves an increase in the 



•ates of cooling of about ten per cent. This is a serious drawback in these experiments. 



[t has been allowed for by diminishing the observed rates of cooling in the ratio of the 



vhole area of the cooling bar to the area of the curved portion only. It is possible that 



n air the emissivity of a vertical surface is, ceteris paribus, greater than the average emis- 



livity of a curved cylindrical surface of the same diameter. As there is heat lost from the 



;nds of the cooling bar, there must be some fall of temperature between the centre and the 



nds. I have given up all attempts at making allowance for this. The best way of 



neeting difficulties of that kind is to make the end correction negligible altogether. The 



»ar was heated over a row of bunsen burners, without the previous warming necessary to 



void " sweating." The bar was heated pretty rapidly, and turned round rapidly while 



eing heated, and very little moisture condensed upon it. A Kew mercury thermometer 



'as used to measure the rate of cooling; of the short bar which was heated to about 250° C. 



leadings were not taken until the bar had cooled for some time with the thermometer 



i position, since the distribution of temperature in the thermometer itself is at first 



regular. This is discussed very fully by Professor Tait in his paper already referred 



