THERMAL CONDUCTIVITY. 29 



ing the quantity of ice which is frozen in a given time, say 

 an hour, and then the quantity frozen is a measure of the 

 quantity of heat passed through. But these experiments 

 can hardly be considered as extremely accurate any more 

 than can those made by several other philosophers. There 

 is a great difficulty in keeping the temperature of the two 

 surfaces of the substance absolutely constant and exactly 

 the same as what you are measuring. For example, if I 

 separate two vessels of water by a wall of metal and let 

 one vessel be at a temperature of, say 80 C. and the other at 

 a temperature of 10 C., it would be impossible to maintain 

 that relation between the water close in contact with the 

 sides of the tin vessels. The metal is such a good con- 

 ductor that it would be an uniform temperature throughout, 

 and there would be a thin layer of water on each side of it, 

 and these layers would be of exactly the same temperature. 

 What we require is to have the temperature of the water 

 in contact with the substance differing by measured 

 quantity from the water in contact with the other surface 

 of the substance. This has been the great difficulty, and 

 consequently our knowledge of the conductivity of bad 

 conductors has not been very certain hitherto. A large 

 number of results have been collected, and they have 

 considerable value, as you can judge for yourselves. The 

 agreement between different experimenters, if you consult 

 this excellent book of Professor Everett's an illustration of 

 the centimeter, gramme, second system of unit you will 

 find a large number of results on conductivity classed 

 together, and if you care to examine it you will see some 

 results I have mentioned to-day spoken of. 



Some experiments made quite lately under the auspices 

 of the British Association by Professor Herschel, not only 

 on heat, but also on measuring the conductivity in the 

 laboratory, it is expected will give better results than we 

 have had hitherto. You see, then, that a great deal of 

 labour has been spent both on theory and in the experi- 

 mental determination of conductivity, but a great deal 

 has still to be done, and it is very desirable that some 

 simple and easy method should be found for measuring the 

 conductivity of different substances. It is of immense 

 practical importance, now-a days, when non-conducting 

 cements for roofing materials are being constructed, and it 



