DR. W. M. HICKS : A CRITICAL STUDY OF SPECTRAL SERIES. 
4-25 
required by known spectrum laws. It must be remembered that the appearance of 
the Em lines varies very much in relative intensity with different observers {cf., for 
instance 4604, 4460 above) that some appear early and then disappear, that others 
come in after the emanation has stood for a few days, and further that the copper 
electrodes, which extended the useful duration of the tubes, would prol^ahiy have some 
effect on the nature of the emitting sources in the gas. To account for this, the 
suggestion might be thrown out that the activity of the emanation would by itself 
ionize the molecules of the gas, and that especially the a-rays would ionize in a 
different and more drastic way than the ordinary cathode or vacuum tube ionization. 
That with time the y-rays from the active deposit might ionize in again a different 
way and produce again new lines. One would expect that the self-effect—as it may 
be called—is so drastic that it destroys those configurations which should give the red 
spectrum analogous to that in the other gases. It is a fact, as I hope to show, that 
the spectrum, so much as there is of it, is decidedly of the jar, or blue kind. 
The degree of accuracy of the observations is not of the best. Royd claims an 
accuracy of OTA. The spectrum was obtained by a concave grating of 1 metre 
radius and extended from 5084 to 3005, with some additional lines by a prism 
spectrograph, subject to errors of ’5A. Watson’s lines extended from 7057 to 3867 
vdth several new lines. His degree of accuracy is probably about the same as that of 
Royd. In the following we shall treat the maximum errors as ■ 2 A except where lines 
are only gixen to the nearest unit. 
The extent of the spectrum observed is too restricted to expect to find more than 
the S ( 2 ) and I) (l) lines, and even in the case of S ( 2 ) the 82 ( 2 ) and 83 ( 2 ) may be in 
the violet where only glass apparatus was used. Further, there is the added 
disadvantage that the links are so large that they can stretch from the unobserved 
ultra-red to the unobserved ultra-violet, and consequently can only act as sounders 
for lines so far in the idtra-violet that an e link lands within the visible red. The F 
lines should be expected to lie wholly in the observed region, and this must be the 
chief guide in the unravelling of the series relations. 
As a preliminary and definite starting point, we have the value of the oun as 
calculated from the atomic weight. But here also there is some uncertainty. The 
value of Honigschmidts’ determination of the atomic weight of Ra, 225'97 is now 
generally accepted as close to the real value, as against the earlier value of.226‘4. 
This makes the atomic weight of the emanation to be 222 to 222 '4. These two give 
values of ^ = 361‘80tc^ as between 1783T and 178 9'5, with the probability that it is 
close to 1783. The uncertainty in the value of the constant 361‘80 will not affect 
this. 
An examination of the spectrum for constant separations shows a large number of 
triplets with i/j in the region 5371 to 5383 and ^’2 at 2671 and less. Further, the 
higher values appear in sets which show inverted order of intensities. This suggests 
that the lines belong to D satellite systems and that the separations about 5383, 2671 
3 M 2 
