138 PROCEEDINGS OF THE AMERICAN ACADEMY. 



The values of K given above take no account of the variation in the 

 Bpecific heal <>f water between 28°. 2 and 58°.3. The temperature coelli- 

 cient of K obtained from the trials made July 31 and August 2 after a 

 change in the arrangement of the spirals, which change was supposed to 

 make for greater accuracy of results, is decidedly greater than that ob- 

 tained from the trials which preceded this change; but the considerable 

 difference between the values of K at low temperature found July 31 

 and August l' makes it unsafe to give especial weight to the value of the 

 temperature coefficient calculated from the trials of these two day-. 

 Tin- beet course appears to be to take the mean of all the results at low 

 temperature and compare it with the mean of all those at high tempera- 

 ture ; and this has beeu done, with the result stated at the beginning of 

 tin- paper. 



Measurement of the Electrical Resistance. 



A -ati-factory determination of the mean electrical resistance of the 

 iron was a work of considerable difficulty. I attempted it at first by use 

 of several little cylinders, each taken singly. Buch as had been used in 

 testing the thermo-electric quality of the iron (see p. 128). I found, 

 however, that the last milling which these cylinders had been Bubjected 

 to had left them somewhat irregular in diameter, so that it was impossible 

 to measure this dimension accurately, even when calipers with jaws 

 meeting along a narrow line were used. Accordingly I used the some- 

 what larger cylinders, already mentioned, which had suffered much less 

 from the imperfections of the milling process. These were about 0.23 cm. 

 in diameter, and could be measured with satisfactory accuracy. 



The straight thick wire of a Carey Foster bridge having been replaced 

 by two .-tout lira-- rods lying iii line, with a gap between their ends, 



which were amalgamated, one of the iron cylinders was placed end to end 

 between these rods and firmly held there under considerable pressure. Then 

 the resistance of a certain measured length, 1.472 cm., of the iron bar was 

 rmined in the usual way by means of the bridge and the accompany- 

 ing low resistance coils. Four cylinders were tested in this way, and 



their mean Specific resistance was found to be 12-1."><>, ('. < '•■ S.. at L'"J C. 



Later the resistance of the same length of each of these bars and of 

 f I %•• other similar bars, held u jual described, was measured by a poten- 

 tiometer method, the potentiometer wire being oi copper drawn especially 

 for this U I The mean specific resistance of the nine bars, a> found bj 

 latter method, was 12240, C G. 8., a1 18 C, the value given in the 



ginning of this paper. 



