440 Dr. F. W. Aston on the 



were obtained which contained both sodium and potassium 

 lines. The latter consisted of a bright component which, 

 when measured against the sodium line, corresponded to 39, 

 and a very faint component at 41. 



Rubidium chloride was now added to a little of the mixture 

 used in the potassium experiments and spectra containing 

 the potassium and rubidium lines were obtained. Rubidium 

 is very definitely double. Its components are more nearly 

 equal in intensity than those of lithium or potassium. 

 Measured against the potassium line 39 its stronger com- 

 ponent is 85 and the weaker 87. The intensity relation agrees 

 reasonably well with the accepted atomic weight 85 45. 



When a mixture of rubidium chloride aud caesium 

 chloride was used evidence of a line at 133, measured against 

 the two rubidium lines, was soon obtained. Pure caesium 

 chloride was then substituted and the utmost possible 

 exposure given to ascertain if a lighter component, expected 

 from the chemical atomic weight 132'81, could be detected. 

 Although by this means the intensity of the line 133 was 

 increased to a satisfactory pitch no other neighbouring line 

 was found. If, therefore, a lighter isotope of caesium exists 

 and does not differ from 133 by many units, which seems 

 very unlikely, it cannot be present in proportion sufficient to 

 account for the fractional atomic weight obtained by chemical 

 means. 



No special accuracy of mass measurement was aimed at in 

 these experiments. The calibration curve previously plotted 

 from the results obtained by the ordinary discharge-tube 

 method was made use of, but it was known that the new 

 strains set up by the change in the apparatus might be 

 sufficient to impair its reliability, and the lines produced by 

 the hot-anode method were neither suitable nor sufficiently 

 numerous to enable accurate recalibration. It is enough to 

 state that the differences between the masses of isotopes 

 of the same metal were whole numbers to a very high degree 

 of accuracy, and that no observation was made which would 

 lead to the supposition that the weights themselves did not 

 follow the whole number rule as closely as those of the non- 

 metallic elements. 



Table of Results. 



Atomic Atomic Minimum number Mass of isotopes 



Element. number. weight. of isotopes. in order of intensity. 



Li 3 694 2 7,6 



Na 11 2300 1 23 



K 19 39-10 2 39, 41 



Kb 37 65-45 2 85,87 



Cs 55 132-81 1 133 



