of Glass at different Temperatures. 227 



rial. I believe that some such law as this will be found to hold 

 for all dielectrics when the effects of the permanent changes 

 produced in the glass by heating are eliminated. These 

 changes are, I find, of considerable magnitude in the case of 

 glass ; and it is to them especially that I desire to call atten- 

 tion in the present paper. 



With the exception of one experiment, in which a Thomson's 

 quadrant-electrometer jar was used, the whole of the experi- 

 ments, the results of which are detailed below, were made on 

 ordinary test-tubes, which had lain in our store-room for 

 several years, but which had not been used for any other pur- 

 pose. The resistances were measured by the direct deflection 

 of the needle of a very sensitive astatic galvanometer, pro- 

 duced by the current passing through the glass. The resist- 

 ance of this galvanometer was 10,000 ohms; and one Daniell's 

 element produced a deflection of one millimetre on the scale 

 when a resistance of 9*96 x 10 9 ohms was interposed in the 

 circuit. 



The battery most commonly used consisted of 10 Daniell's 

 elements ; but at low temperatures a battery of 110 Daniell's 

 cells was sometimes used. I was thus enabled to measure 

 with a somewhat close approach to accuracy a resistance equal 

 to 2 x 10 14 ohms per cubic centimetre in my test-tube experi- 

 ments, and equal to 14 x 10 14 in my electrometer-jar experi- 

 ment. The test-tube glass had generally a resistance consi- 

 derably below the former figure even at low temperatures ; 

 but the electrometer jar reached the latter figure at a tempe- 

 rature of about 100° Cent. 



The advantage in these experiments of using this galvano- 

 metric method of measuring the resistances, rather than the 

 method with electrometer and condenser of great capacity, 

 was the ease with which the temperature and resistance could 

 be simultaneously observed. 



In determining the resistances two methods of observation 

 were employed. One consisted in passing the current through 

 the glass first in one direction and then in the opposite direc- 

 tion, and taking the deflection in each case half a minute after 

 the battery had been applied. The mean of these deflections 

 was then taken as the deflection corresponding to the resist- 

 ance to be measured; and the mean temperature was taken as 

 the temperature of that particular determination. 



The second method consisted in passing the current conti- 

 nuously through the glass during the whole experiment, and 

 taking deflections by short-circuiting the galvanometer. 



The first method gives the resistance before any very great 

 amount of polarization has taken place, at least before more 



