432 On the Conductivity of Tourmaline for Heat. 



£ = 20 seconds the ordinates of all the curves have one and the 

 same value. For the later times the different curves will be 

 Fig. 2. Crystal from Brazil. 





■'..:' 





: :L. 



40 so 60 70 so so ioo 110 120 130 uo 150 igo no iso mo 200 seconds 



Fig. 3. Crystal from Ceylon. 





:* 



20 30 40 so 60 70 so 90 wo itoi2o 130140 150 seconds. 



somewhat differently shaped ; and the most probable tempe- 

 rature-curve for any particular temperature at commencing 

 will be obtained if, for its ordinates, the mean values of the 

 ordinates of the new curves be chosen. This method of pro- 

 ceeding is not strictly correct. As, however, experiment shows 

 that the different temperatures at commencing of corresponding 

 curves approach one another more and more as one proceeds to 

 greater abscissa?, and the variation among the individual curves 

 being moreover small, the error can be of little importance. 



Concerning the curves, which have been reduced so as to 

 have the same temperature at commencing, I only repeat the 

 statement that the deviations amount at most to 3 per cent. 

 In figs. 2 and 3 I give the most probable temperature-curves, 

 obtained in the way given above, for the two slices : here A 

 denotes the line obtained when the flow of heat is from the 

 analogous to the antilogous pole, and the corresponding one 

 (B) in the reverse direction. 



