204 Professor Joseph John Thomson [Jan. 19, 



lines were the first to appear, and were the last to disappear, when the 

 rate of work was reduced from a high value, at which both the 

 nitrogen and mercury spectra were bright down to a point where the 

 discharge ceased to be luminous. 



The preceding considerations have also an important application 

 to the difference between the arc and spark spectra. In the con- 

 tinuous arc discharge, although the average rate of work is much 

 higher than in the spark, the maximum rate is very much less ; in 

 the spark discharge we have exceedingly intense cun-ent density lasting 

 for a very short time, and while the spark is passing we have a very 

 much greater rate of work than in the arc. Hence the state of things 

 in the spark will be analogous to that represented in Fig. 5, and the 

 lines corresponding to systems of the type B will be enhanced re- 

 latively to those of type A ; we conclude then that the arc hnes 

 correspond to systems of the type A, the spark lines to those of type B. 



The work done in the discharge tube is probably ultimately con- 

 verted for the most part into heat, so that the rate at which work is 

 being done at any part of the tube is approximately proportional to 

 the rate at which heat is being produced in the tube. I do not, 

 however, regard temperature, i.e. the energy due to the translation of 

 the atoms as a whole, as having any direct connection with the pro- 

 duction of spectra. The work done by the electric field on the 

 corpuscles is, since the corpuscles can easily penetrate the atoms of 

 the gas, first converted into internal atomic energy. This energy may 

 ultimately be for the most part transformed into the energy of trans- 

 lation of the molecules of the gas and so appear as temperature, but 

 it by no means follows that, if we heated the molecules of the gas by 

 non-electrical means to the temperature to which even a few of its 

 molecules are raised by the electric discharge, that we should get a 

 luminous spectrum. The production of the spectrum depends upon 

 the internal energy of the atom. When we use the electric discharge, 

 all the work done by the corpuscles goes at first into the form of 

 internal atomic energy, while if we supplied the same amount of 

 energy to the gas by thermal, as distinguished from electrical means, 

 the energy would go first into increasing the energy of translation of 

 the atoms, and very little of it would ever get inside the atom. It 

 is probable, however, that some of the energy of translation would 

 get converted into internal energy, and that temperature is one way 

 of giving internal energy to the atom and so producing luminosity. 

 From one point of view, however, it is a very extravagant method, as 

 the fraction of the energy spent in lieating the gas which goes in 

 producing luminosity is small. 



The coefficient of absorption a of the systems will depend 

 upon the way in which the internal energy is given to the atom, as 

 well as upon the rate at which the electric field is doing work in the 

 neighbourhood of the atom. Thus, for example, if the internal work 

 is given by means of rapidly moving corpuscles, the coefficient of 



