PERIODIC ARRANGEMENT OF ELEMENTS 169 



and sodium, 16'09; and so on. In other groups the 

 divergences are still more striking. 



The cause of this Jrregularity has, therefore, to be 

 sought. In seeking for a clue, the first question is : Are 

 the atomic weights invariable ? A further question is : Is 

 weight invariable ? Does a body always possess the same 

 weight under all conditions ? For example, would the 

 weight of a body remain the same if it were to be weighed 

 at different temperatures? Or, if electrically charged, 

 would its weight remain unaltered ? 



It is a very difficult problem to weigh an object at a 

 high temperature. If the balance, as is usual, contains 

 air, convection currents are produced by the ascent of air 

 heated by the warm body, and the body apparently weighs 

 too little. If the whole balance were uniformly heated, 

 the weights would be at the same temperature as the 

 substance weighed ; and it is to be presumed that both 

 they and the substance would alter in weight equally, and 

 still remain in counterpoise. And if the balance case be 

 pumped empty of air, as was done by Crookes in deter- 

 mining the atomic weight of thallium, other phenomena 

 intervene, which, however interesting in themselves (they 

 led Crookes to the invention of the radiometer), are very 

 disconcerting ; for attractions and repulsions, which com- 

 pletely disturb equilibrium, are produced by the slightest 

 variations of temperature. However, some curious calcula- 

 tions have been made by Hicks in dealing with Baily's 

 experiments on the attraction of leaden balls by masses 

 of lead experiments which afford data for calculating the 

 density of the earth. At a high temperature the attraction 

 appeared to be less than at a low one ; and as the attrac- 

 tion of the earth is the cause of weight, supposing these 

 experiments to be correct, and the deductions legitimate, 

 it would follow that weight is altered by temperature. 

 The subject is well worthy of further experiment. 



