788 Mr. S. G. Lusby on the 



and 1450°, Thus it appears that at high temperatures the 

 mass of the positive ion remains unaltered, and is equal to 

 that of a hydrogen atom. Ilelow 1400° the mean size of the 

 ion increases rapidly. Let us assume that </>(V) and ^r(ij) 

 are equal to unity, ?'. e. that the ion is one molecule. We 

 find, on solution, that an ion of specific mass 36 has a 

 mobility of 12 2, which is the value obtained at 1150°. 

 Within the limits of error this might represent a molecule of 

 either nitrogen or oxygen — both of which are present in the 

 flame, or else a molecule of carbon dioxide, or two molecules 

 of water vapour — these being the two main products of 

 combustion. Probably ions of all these types are present. 

 The most obvious way of accounting for the decrease in 

 mobility is to suppose that clustering commences round the 

 high temperature ion as nucleus. At moderately high tem- 

 peratures only a few such clusters exist, on the average, at 

 any instant ; but as the temperature falls the number of 

 clusters increases until we reach the stage noted at 1150° 

 when every ion is a single molecule. At any intermediate 

 temperature we should have some simple ions and some 

 clusters ; e. g. at 1300° the mobility is only 42, which means 

 that approximately six ions out of every seven consist of a 

 single molecule (see Table If., columns V. and VI.). Thus 

 at any such temperature we must imagine a constant combi- 

 nation and dissociation between the high temperature ion and 

 a molecule. If the simple ion is actually a charged atom of 

 hydrogen, this may be conceivable ; if not, we have to 

 explain how the positive charge can leave one body and 

 transfer itself to another. The question will be touched on 

 later in more detail* 



As the temperature falls still lower, we know from the 

 experiments of Moreau * and McClelland t that the size of 

 the ionic cluster increases very rapidly — in fact, as the fifth 

 power of the temperature. If we take McClelland's highest 

 result— a mobility of 23 at 500° — and the author's lowest 

 result- — a mobility of 12 at 1150° — this fifth-power law fits in 

 very well. This law of temperature change is very different 

 from that found by Phillips [loc. cit.) for ions produced by 

 Rontgen rays in gases, in which case the mobility varies 

 directly as the temperature. The tendency to cluster is far 

 stronger in flame gases than in ordinary pure gases. This is 

 probably due to the products of combustion; for de Broglie % 

 has shown that if a flame 'produces condensable products of 



* Morean, Ann. de CJiim. et Phi/s. 1906. 

 t McClelland, Phil. Mag. July 1898. 

 | de Broglie, 'journ. de i%s.Dee. 19.09. 



