62S Physical Conditions and Absorption of Light. 



theoretical conclusions derived agree with the experimental 

 results given by Wood. 



This theory also provides for the broadening of the lines 

 in Humphrey's experiments on the " pressure shift/'' There 

 is, however, a still more obvious reason for taking a range 

 of values of the constant a as above. The density of the gas 

 varies considerably from the cooler parts in the surrounding 

 air to the very hot portions in the centre of the arc. 



It thus appears that by taking advantage of the latitude 

 allowed by the theoretical uncertainty in the value of the 

 constant a, we can explain a large number of the compli- 

 cations present in ordinary spectra. We have merely to 

 avoid mutually inconsistent hypotheses in explaining the 

 var ous types of the phenomena. Any consistent and rea- 

 sonable set of properties for the constant a would appear to 

 provide a theoretically effective explanation of the phenomena 

 under discussion. We might regard the experimental results 

 as providing the necessary laws for the variation of the 

 constant a with the physical conditions of the material. It 

 is, however, important to notice that the more extensive list 

 of physical properties would involve only the mean value of 

 the constant a taken for all the electrons of the substance. 

 Such values are deducible from known properties of the 

 substance in bulk. Spectrum analysis alone can separate out 

 the possibly different values of a with which we are concerned 

 when discussing the individual electrons. 



Dr. Havelock has given a discussion of the dependence of 

 the mean value of the constant a on the physical conditions 

 of the substance. 



The theory here given is not, however, complete. It 

 appears to fail in one very important respect. Although in 

 the majority of cases the actual facts are as stated, yet there 

 are some important exceptions. For instance, the dis- 

 placement of certain absorption and emission lines in the 

 spectra of not uncommon substances is not towards the red 

 end of the spectrum, but towards the violet end. The asso- 

 ciated phenomena of asymmetrical broadening also frequently 

 takes place in this direction. The theory here developed 

 appears, at present, quite unable to admit of an explanation 

 of these divergences from what appears, after all, to be the 

 general rule which it does so well explain. Further investi- 

 gation may, however, confirm the results already deduced : 

 and we must then look for the loophole which will enable us 

 to extend the theory to the above anomalies. 



