DETERMINATIONS OF HEAT CONDUCTIVITY OF SELENIUM. 359 



hours before any reading could be taken. When this steady 

 temperature was reached, two sets of readings were taken, 

 separated by an interval of half an hour. Thus in Table I 

 we have, at every temperature, two values of the conduc- 

 tivity, each of these being calculated from the mean of a 

 number of readings. 



Table I also shows the general order adopted for any disc. 

 The conductivity was measured first at room temperature, 

 then at several higher temperatures, and finally the read- 

 ings at room temperature were repeated. In most cases 

 this procedure has been repeated at intervals which extend 

 up to one year from the date of preparation. 

 Sources of Error. 



The errors involved were necessarily large. They natur- 

 ally fall into two divisions, those due to the imperfections 

 of apparatus and method, and those due to such changes 

 in the selenium as were beyond control. 



With the instruments used, the errors of the first class 

 should not have exceeded 1%, these probably being small 

 in comparison with the errors of the second class. 



The latter appear to be due chiefly to heating effects, and 

 are most marked in the case of newly prepared crystalline 

 discs. Two types of irregularity were of frequent occur- 

 rence : — 



(i.) In many cases the first reading taken for any 

 erystalline disc at a temperature above that of the room 

 was abnormally high. 



(ii.) The conductivity at room temperature was almost 

 invariably increased after the measurements at higher 

 temperatures had been made, but the actual increase thus 

 produced varied greatly, even in the case of discs which 

 had apparently been treated in exactly the same manner. 



Results. 



Eleven discs have been prepared, five vitreous and six 

 crystalline, the nature of the results obtained being illus- 

 trated in Tables I, II and V. 



