I may now^ quote from the paper my conclusions :* "It is interesting
to note that u> [here] plays the part of a control in refraction, and that
the form of the dispersion curve is really due to the distribution law
and not to radiation. The value of o>/2ir obtained must be about three
times the frequency for the Na lines and hence w corresponds to a
frequency far beyond the violet. I see no grave objection to this
result, as the same result would arise in any theory in which a single
term is supposed to represent the facts. The controlling period must
l>e less than the imposed period to explain ordinary dispersion in the
spectral range.
" The high value of u> precludes the possibility of explaining spectral lines by the molecules of this theory. I am inclined to think that spectral lines are in some way connected with dissociation, but not necessarily of simple molecules.
" I return to a suggestion which I made earlier, that these are due to multiple molecules. "\Ve cannot avoid the conclusion that there are such in a gas. They need not, and indeed do not, have a permanent existence. It is only necessary that there should be a small proportion on average. The frequencies of these small planetary systems are probably less than the frequency of the molecule par excellence. Further it is probable that their frequency will not be greatly affected by pressure, but the number would probably increase with pressure. This would agree with Professor Sir Norman Lockyer's observations that the period of the lines does not alter much with pressure, but the intensity does increase considerably with increase of pressure."
Earlier in the paper I point out the importance of multiple molecules in connection with the theory of the ratio of specific heats. " The ratio of specific heats for chlorine indicates more nearly six degrees of freedom, and one must either suppose that there is a con- siderable proportion of the set B present, or else a proportion of multiple molecules. The first supposition could hardly be supported quantitatively on electrical grounds ; while the second is more probable in view of the comparative proximity of chlorine, under ordinary con- ditions, to its critical state."
Note. In the Thesis presented to the University of Gottingen, for his doctorate, Herr Karl Baedecker has measured the effect of tem- perature on the dielectric constant for several gases. The thesis was published in pamphlet form in August, 1900, and in the ' Zeitschrift Phys. Chem.,' vol. 36, pp. 305-335. The bearing of his results on the formula above given is analysed in the following table, 6 being taken as temperature centigrade plus 273 0< 3.
- These views were first expressed in n Fellowship Dissertation presented to
Trinity College, Cambridge, in August, 1900.
