Thermoelectric Position of Carbon, 121 



arrangements. By using the precautions detailed above, I 

 ensured, however, that the place of any disturbances that 

 might come into play should be located with certainty in the 

 thermo-joints under examination. 



The .proper amount of resistance to be inserted in the 

 primary circuit having been determined beforehand, the 

 primary circuit was closed. The current was allowed to 

 flow during the whole time the experiment lasted. The 

 temperature of the hot bath was raised very slowly up to the 

 highest point reached; then as slowly lowered again. Fre- 

 quent readings of the thermometers in the hot and cold baths 

 were taken. Simultaneously were noted the positions of the 

 movable sliders on the graduated wire, for which no deflec- 

 tion was shown by the galvanometer on depressing the key in 

 the secondary circuit. 



When a temperature reading was intended to be taken, 

 the temperature was kept as steady as possible for ^ to ^ a 

 minute, accompanied with constant stirring of the hot bath. 

 In this way a series of corresponding readings of temperature 

 and of distances on the graduated wire were obtained. The 

 results were plotted out, taking centimetres of the graduated 

 ware as ordinates, and temperatures of the hot bath as ab- 

 scissae. A regular curve was then drawn amongst the points 

 thus obtained, so as to give as nearly as could be judged the 

 mean of the observations. The curves A, B, C (Plate VIII.) 

 were obtained in this way. The scales for these curves are 

 not quite the same. Details are given below. The curves A 

 and B were obtained on successive days. 



The want of regularity in the individual readings, after all 

 the precautions taken, I considered could only be due to the 

 properties of the carbon. Partly to satisfy myself that no 

 essential precaution had been overlooked, however, and partly 

 to test an idea I formed as to the cause of the irregularities, 

 the apparatus was set up again about a month later. Curve 

 C gives the results of the observations. The heating of the 

 hot joint gave the curve to the left hand ; the subsequent 

 cooling gave the right-hand curve. These curves have not a 

 point of intersection at their upper ends shown in the diagram 

 (Plate VIII.), because the temperature w r as intentionally raised 

 to a point higher than that for which a reading on the graduated 

 wire could be obtained, before cooling down was begun. 



Now, as 4-^r hours were given to the observations, the heat- 

 ing and cooling took place very slowly indeed. And as a 

 stirrer was kept almost continuously in operation in the hot 

 bath, the lagging of the thermometer behind the thermo- 

 joint must have been very small. I believe that the dif- 



