LIQUID HYDROGEN. 139 



would theoretically become a perfect conductor. The effect of cooling- 

 platinum from the boiling point of liquid oxygen to that of liquid 

 hydrogen is to diminish its resistance to one-eleventh. 



The difficulties in liquefying hydrogen caused by the presence of air 

 in the gas have been referred to/ and later experiments had for their 

 object the removal of this fruitful source of trouble. This is by no 

 means an easy task, as quantities amounting to only a fraction of 1 

 per cent accumulate in the solid state and eventually choke the nozzle 

 of the apparatus, necessitating the abandonment of the operation. 

 Later experiments enabled me to procure a larger supply of liquid 

 hydrogen with which the determination of certain physical constants 

 has been continued. The first observations made with a pure platinum- 

 resistance thermometer had given —238° as the boiling point. A new 

 thermometer, constructed of platinum from a different source, gave 

 practically the same value. As these results might be affected by some 

 constant error, the determination was checked by employing a ther- 

 mometer constructed from an alloy of rhodium and platinum, contain- 

 ing 10 per cent of the former. Alloys had been shown by Professor 

 Fleming and the author to differ from pure metals in showing no sign 

 of becoming perfect conductors at the absolute zero of temperature, 

 and a study of the rhodium-platinum alloy had shown that the change 

 in conductivity produced by cooling from 0° to the boiling point of 

 liquid air is regular and may be represented by a straight line. As 

 determined by the rhodium-platinum thermometer, the boiling point 

 of hydrogen was found to be —216°, or some 8° lower than the plati- 

 num thermometer gave. Two ways of explaining the discrepancy 

 between the two values suggested themselves. Pure platinum, although 

 its resistance may be represented by a straight line almost down to the 

 solidifying point of air, shows signs of a departure from regularity at 

 about this point, and the curve may become asymptotic at lower tem- 

 peratures. On the other hand, the resistance of the rhodium-platinum 

 alloy diminishes less rapidly at these lower temperatures and is nuich 

 higher than that of pure platinum under similar conditions. It follows 

 that its resistance curve in all probability deviates less from a straight 

 line than is the case with platinum. Either cause would explain the 

 differences observed, but the lower boiling point (—246°, or 27° abso- 

 lute) seemed to be the more probable, as it agreed very fairly with 

 the value for the boiling point calculated by the author from Wro- 

 blewski's results. As the use of other pure metals or alloys was not 

 likely to lead to more satisfactory results, the problem had to be 

 attacked in a different way, namely, by means of an "air" thermometer 

 containing hydrogen under diminished pressure. 



A first attempt has been made at determining the boiling point 

 by a constant-volume hydrogen thermometer working under dimin- 



1 Proceedings, 1898, 14, 130. 



