of Tourmaline for Heat. 431 



before the contact, no change of temperature of the junction 

 (a) was observable. 



The experiment was conducted in the following way : — I 

 first observed the position of the needle at rest with a closed 

 circuit ; I then short-circuited the current (excluding the three 

 thermoelectric junctions) to convince myself of the absence 

 of every other electromotive force ; the temperature of the 

 turpentine was now read, and after removing the short circuit 

 the deviation of the magnetic needle observed. The piece of 

 ice was then placed on the glass plate of the test-tube, the stop- 

 cock opened, and with the help of a chronometer the moment 

 of contact between the ice and the tourmaline slice observed 

 through a telescope. In order to keep the readings for the 

 different times expressed in divisions of the scale, the junction 

 (a) was then immediately placed in ice and the deviation noted. 

 The results were controlled at the end of every experiment by 

 observing whether the temperature of the turpentine and the 

 deviation of the needle, by open and by short circuit, were 

 the same as before. 



In this way a number of experiments were made for both 

 tourmaline slices, sometimes the analogous, sometimes the 

 antilogous, end of the crystal being brought into contact with 

 the ice. The results I represented graphically by means of 

 curves, of which the number of seconds, reckoned from the 

 first contact of the tourmaline with the ice, form the abscissae, 

 and the corresponding temperatures of the junction (a), ex- 

 pressed in scale-divisions, the ordinate. 5 *. The galvanometer 

 was so arranged that about sixty divisions on the scale corre- 

 sponded to 1° C. difference of temperature. 



Of the curves thus obtained only the middle one has been 

 used. The first shows, naturally, a less regular course ; while 

 the last is also useless, because the changes of temperature 

 towards the end of the experiment took place rather slowly, 

 making the errors of observation of course greater. For 

 the thicker slice I have used, in the construction of the curve, 

 the part between the abscissae t — 20 seconds and t x = 200 

 seconds ; for the thinner, in which the rise of temperature 

 takes place with more rapidity, from £ = 20 to t x = 150 seconds. 



In consequence of the unavoidable variations in the tempe- 

 rature of the room in which the observations w r ere made, the 

 first few ordinates of the temperature-curves obtained in the 

 different experiments are not quite equal ; the greatest differ- 

 ence which I have allowed amounts to 3 percent. In order to 

 obtain from all the experiments the probable change of tem- 

 perature for one and the same position of one of the slices, I 

 have increased or diminished the curve in such a way that for 



