200 
NATURE 
[ DECEMBER 31, 1896 
In those earlier experiments the air drawn away was 
replaced by air coming in from the laboratory at the open 
end of the tube. We found evidence of disturbance due to 
electrification of air of the laboratory by brush discharges 
from electrodes between the induction coil and Réntgen 
lamp, and perhaps from circuit-break spark of induction 
coil. These sources of disturbance are eliminated by 
our later arrangement of lead cylinder covered with card- 
board at both ends, as described above, and air drawn 
into it from open-air outside the laboratory. ; 
We have also founda very decided electrification of 
air—sometimes negative, sometimes positive—when the 
Réntgen rays are directed across a glass tube or an 
aluminium tube, through which air was drawn from the 
quadrangle outside the laboratory, to the filter. 
A primary object of our experiments was to test 
whether air electrified positively or negatively lost its 
charge by the passage of Réntgen rays through it. We 
soon obtained an affirmative answer to this question, both 
for negative and positive electricity. We found that 
positively electrified air lost its positive electricity, and 
in some cases acquired negative electricity, under the 
influence of Réntgen rays; and we were thus led to 
investigate the effect of Réntgen rays on air unelectrified 
to begin with. 
Note on Diagram.—For the sake of simplicity, the 
screening of the electrometer is not shown in the 
diagram. In carrying out the above experiments, how- 
ever, we have found it absolutely necessary not only to 
surround the electrometer with wire gauze in the usual 
manner, but we have had also to place a sheet of lead 
below it, and to screen also the side next the Rontgen 
lamp by a lead screen. In some cases it was even 
necessary to cover up the whole with paper to prevent 
the electrified air of the room from disturbing the 
instrument. KELVIN. 
jC. (BRADITE 
M. SMOLUCHOWSKI DE SMOLAN. 
Physical Laboratory, University, Glasgow, 
December 19. 
ON A NEW LAW CONNECTING THE PERIODS 
OF MOLECULAR VIBRATIONS. 
FTER the attempts to detect harmonic ratios in the 
wave-lengths of the light emitted by incandescent 
gases had failed, Balmer led the way ina line of research 
which promises to furnish a rational explanation of the 
different periods of a vibrating molecule. If 7, is a con- 
stant, the frequencies of all the hydrogen lines can be 
represented by Balmer’s formula : 
a 7 = 2) 
where for 7z we must put successively 3, 4, &c., and lines 
have been observed as far as #72 = 15. We may call t) 
the convergence frequency, as it is that to which the 
vibrations approach as 7 increases. We also call the 
slowest vibration of the series the “fundamental.” For 
other spectra the relationship is not quite so simple, but 
confining ourselves to the alkaline metals, the spectra of 
which have been most carefully examined and discussed 
by Kayser and Runge, these authors show that the 
vibration frequencies may be represented by means of 
three series of the form 
B C 
7=A — - 
mi m+ 
the fundamental in every case being given by m = 3. 
The first, or principal series, has lines which are single 
in the spectrum of lithium, but double in the other cases, 
the components separating further and further as the 
atomic weight increases. The two supplementary series 
consist of lines which easily widen and show the remark- 
No. 1418, VOL. 55] 
able property that the convergence frequency A has 
nearly the same value for both supplementary series. 
I must refer to Kayser and Runge’s paper for a dis- 
cussion as to how far the formula is approximate only, as 
well as to other details ; but as the majority of physicists 
have not hitherto paid much attention to this subject, I 
may add the remark, that the division of spectra into a 
number of series of the above nature is by no means 
arbitrary, but constitutes a most important step in the 
simplification of a very complex problem. It is known that 
Runge and Paschen have shown that the constituents of 
cleveite gas have spectra resembling those of the alkali 
metals ; and I hope in a future communication to show 
that the spectrum which I have called the compound line 
spectrum of oxygen also divides into three series of the 
same type. 
As regards my new law, it is so simple that it 1s 
astonishing how it could so long have remained un- 
noticed. It may be enunciated as follows : 
If we subtract the frequency of the fundamental 
vibration from the convergence frequency of the prin- 
cipal series, we obtain the convergence frequency of the 
supplementary series. 
The following table shows how far the law is accurate = 
Wave Number of 
A (Principal Wunidamertal A (Supplement= 
Difference 
Series) Vibration ary Séries} 
Lithium 43585 14907 28678 28667 
28587 
Sodium ... 41537 16960... 24577 24566: 
16977... 24560 24547 
24496: 
24476 
Potassium... 35087 12988 .. 22099 22077 
13045 ... 22042 22022 
22050 
21991 
Rubidium... 33762 ... 12579 21183 21179 
12802 20960 20939 
The numbers given in the table are proportional to the 
frequencies, being inverse wave-lengths in centimetres. 
The two numbers given in the second column refer to the 
two components of the double lines. As the lines of the 
supplementary series are double in the spectra of sodium 
and potassium, there are the four convergence frequencies 
in these cases which are all given. In comparing the 
two last columns it must be remembered that the quantity 
denoted by A, which is the convergence frequency, cannot 
be determined with the highest accuracy because Kayser 
and Runge’s formula is approximate only and fails to give 
accurate results for the case 7z=3. The only serious. 
differences between the third and fourth columns occur 
in the two cases in which Runge and Paschen used the 
case 72=3 in determining the constant. 
In the case of caesium, the fundamental vibration lies. 
in the infra-red, and has not been observed, but we may 
use the law to forecast its position,and obtain a wave-length 
of 8908 for the less refrangible, and of 8518 for the most: 
refrangible component, numbers not differing much from 
Kayser and Runge’s estimate for the same lines. The 
numbers given by Runge and Paschen for the gases from 
cleveite are sufficient to show that the law also holds in 
the case of both sets of the three series into which the 
spectrum divides itself. In the case of oxygen, I have 
not obtained sufficient data as yet to determine the 
position of the principal series, as it lies chiefly in the 
ultra-violet, and the lines measured so far belong nearly 
all to the supplementary sets. 
The supplementary series have not been observed in 
the spectrum of hydrogen, but the new law shows that 
if they exist they must lie in the infra-red, and it is with 
some confidence that I predict the existence of hydrogen 
lines in the infra-red, the convergence frequency being 
1218°51 (A = 82066). 
