xxi.] RESULTS WITH POTASSIUM. 301 



represent a complex something which cannot withstand the 

 temperature of the sun, and that the yellow line represents a 

 something finer which can withstand the temperature of the 

 sun, what happens when we try to drive off the vapour of this 

 potassium at the lowest temperature at which we can get it to 

 volatilise at all, is that if the experiment is carefully performed 

 it gives precisely those lines which are reversed in the solar 

 spectrum alone ; and of that line which is the strongest line at 

 the temperature of the Bunsen burner we see absolutely nothing 

 at all. Referring to the spectrum which we get in the lilac and 

 yellow-green part of the tube, two out of the three lines visible 

 at all events are seen in the sun, whereas the other lines which 

 we get in the flame and some of them which we get with an 

 induction coil are not represented in the fine vapour which was 

 produced at the lowest possible temperature. 



While heated with the Bunsen burner some very exquisite 

 colour-effects are seen in the tube, and especially a beautiful 

 blood-red colour which might be imagined to be the product of 

 that molecule which gives the red line seen in the Bunsen flame ; 

 but that is not the fact. The line seen in the flame of the 

 Bunsen burner is not visible as a rule in the vapour when heated 

 in this way, the lines actually seen being more refrangible. 



The experiment then comes to this. If we assume potas- 

 sium to be a compound body and that its finer constituent 

 molecules are those which resist the solar temperature, then it 

 behaves exactly like a mixture of hydrocarbons is known to do, 

 that is, the finer vapours come off in greatest quantity at the 

 lowest temperature, and the more complex ones as the tempera- 

 ture is raised. 



The result of the application of this new method indicates 

 that in the case of a considerable number of chemical sub- 

 stances not only is the line spectrum compound in its origin, 

 as I suggested many years ago, but that a large number of the 

 lines is due to molecular groupings of considerable complexity, 



