Positive Ionization from Hot Salts. 671 



that the thermions, which give values of m/H between 25 and 

 35 or thereabouts, in cases where the chemical composition 

 does not make such values probable a priori, arise from con- 

 taminations with the salts of the alkali metals. The view 

 has been recently advocated by several authors that these 

 ions are charged atoms or molecules of some gas, such as 

 carbon monoxide, which has access to the heated salt. I 

 believe this view to be a mistaken one. 



Magnesium Chloride. — The curves given by this substance 

 were irregular, and made the experiments somewhat unsatis- 

 factory. There is a great deal of difference between the 

 salts of the same metal in this respect. I am inclined to 

 attribute the irregularity to the salt redistributing itself on 

 the strip during the course of an experiment ; some salts 

 seem to have a tendency to collect into irregular lumps, 

 whilst others form a nice uniform layer all over the strip. 



The ionization from magnesium chloride was relatively 

 small. The only satisfactory experiment gave the values 

 e/m = 360, and ??i/H = 27. This may have been either Mg + 

 or Na + , but there was certainly no indication of Mg ++ . 

 The magnesium chloride used was Kahlbaum's " Zur 

 Analyse." 



Zinc Chloride. — The curves given by this salt were also 

 somewhat irregular. The ionization developed at a low 

 temperature and did not persist very long. The two most 

 satisfactory determinations gave the values ??i/H = 78, and 

 ???/H = M respectively. The mean of these is 61 as against 

 65 for Zn + . The zinc chloride used was Kahlbaum's 

 granulated. 



Zinc Bromide. — After some preliminary experiments, which 

 gave rather irregular curves, some very satisfactory results 

 were obtained with this substance. The deflected curves 

 were similar and symmetrical, and the undeflected curve 

 (H = 0) was similar and exactly halfway between them. 

 The mean of two concordant experiments gave £/??i = 194, 

 and ???/H = 49"4. This is almost halfway between the values 

 of mjS. for Zn + and Zn ++ , which are 65 and 32*5 respec- 

 tively. It may imply a mixture of these ions, or it may be 

 due to Zn + contaminated with alkalis. 



Zinc Iodide. — This substance behaved rather like zinc 

 bromide. The freshly heated substance gave irregular curves 

 which afterwards became more consistent. Two concordant 

 sets of curves, which were symmetrical about the undeflected 

 position, gave for the mean values e/m = 2'64: and m/H = 41'4. 

 These are much nearer the values for Zn ++ than for Zn + . 



Considering the results for the zinc haloid compounds 



2 Y2 



