362 Prof. J. J. Thomson on the Passage 



rounded by steam ; and neither Maxwell nor Hittorff could 

 find any trace of conduction through hot mercury- vapour. I 

 have repeated all these experiments and verified the results. 

 Hittorff found, too, that the conductivity of a flame was very 

 much increased by putting volatile salts, such as the chlorides 

 of sodium and potassium, into it, and that the increase in the 

 conductivity depended on the nature of the salt. 



Methods of Eosperirnenting . 



I have used three different methods of investigating the 

 conductivity of hot gases, as the method which is the most 

 convenient for one gas is not always so for another. In all 

 these experiments the conductivity of the gas was sufficiently 

 large to enable the current sent through a centimetre or two of 

 the gas by a few Darnell's cells to be easily measured by a gal- 

 vanometer of about 4000 ohms resistance. The galvanometer 

 method, when available, is much more convenient and easy to 

 use than the much more delicate one in which a leak through 

 the gas is detected by an electrometer. Whatever method is 

 employed it is essential that the current measured really 

 passes through the gas, and is not due to the defective insula- 

 tion of the electrodes which carry the current into it. The 

 effective insulation of these electrodes requires considerable 

 care, because no substance is known wdnose insulation can be 

 trusted even at temperatures very much lower than those used 

 in these experiments. It is therefore necessary to keep the 

 places at which insulation is essential quite cool. The only 

 really effective way of doing this is to keep such places well 

 away from the hot gas and well screened from currents of hot 

 air. In the following experiments this was done by making 

 the upper part of the electrodes pass through vertical glass 

 tubes, to which two long horizontal pieces of glass tube were 

 fused; these pieces were about 8 inches in length, and were 

 supported at the ends away from the electrodes by passing 

 through holes drilled in solid pieces of ebonite, supported by 

 retort-holders. These pieces of ebonite remained cool even 

 when the gas had been kept at a white heat for some hours, 

 and though they were tested repeatedly they were never found 

 to leak. The electrodes were connected with the battery by 

 long overhead wires insulated by pieces of sealing-wax placed 

 so far from the hot gas that they never became appreciably 

 warm. 



In the greater number of experiments the gas to be investi- 

 gated was contained in a platinum tube made by Johnson and 

 Matthey. It was 7 inches long and 1 inch in diameter. Into this 



