JUNE 22, 1899] 
WATURE 
179 
the lateral components of the magnetic triplet should 
vary inversely as the square of the wave-length of the 
spectral line under consideration. Now, when we ex- 
amine this point by experiment, we find that this simple 
law is very far from being fulfilled. In fact,a very casual 
survey of the spectrum of any substance shows that the 
law does not hold even as a rough approximation ; for, 
while some spectral lines show a considerable resolution 
in the magnetic field, other lines of nearly the same 
wave-length, in the same substance, are scarcely affected 
at all. This deviation is most interesting to those who 
concern themselves with the ultimate structure of matter, 
for it shows that the mechanism which produces the 
spectral lines of any given substance is not of the 
simplicity postulated in the elementary theory of this 
magnetic effect. 
Grouping of Spectral Lines. 
Our previous knowledge of the line spectra of different 
substances might indeed have led us to suspect some 
such deviation as this from the results predicted by the 
simple theory. For if we view the line spectrum of a 
given substance we find that some of the lines are sharp 
while others are nebulous or diffuse, and that some are 
long while others are short—in fact, the lines exhibit 
characteristic differences which lead us to suspect that 
they are not all produced by the motion of a single un- 
constrained ion. On closer scrutiny, they are seen to 
throw themselves into natural groups. For example, in 
the case of the monad metals sodium, potassium, &c., 
the spectral lines of each metal form three series of 
natural pairs, and again, in the case of the diad group, 
cadmium, zinc, &c., the spectrum of each shows two 
series of natural triplets, and so on. 
Thus, speaking generally, the lines which form the 
spectrum of a given substance may be arranged in groups 
which possess similar characteristics as groups. Calling 
the lines of these groups Aj, B,,C,... As, B,C... ., 
Az, Bz, C, . . . we may regard the successive groups as 
repetitions of the first, so that the A’s—that is A,, A,, 
A,, &c.—are corresponding lines produced probably by 
the same ion ; while the B’s—namely, Bj, B,, B,, &c.— 
correspond to one another and are produced by 
another ion, and so on. This grouping of the spectral 
lines has been noticed in the case of several sub- 
stances, and it has been a subject of earnest inquiry 
amongst spectroscopists for some time past. All such 
grouping, however, up to the present, has had to depend 
on the judgment of the observer as to certain similar- 
ities in the general character and arrangement of the 
lines, and similarities which indeed may or may not have 
any specific relation to the mechanism by which the lines 
are produced. In fact, such grouping has been effected 
by guess-work, or by empirical formule, and we need not 
be surprised if it is found that the groups so far obtained 
are more or less imperfect. 
I introduce this grouping of the spectral lines to your 
notice in order that we may attack the problem of re- 
ducing to order the so far apparently lawless magnetic 
effect. As I have already mentioned, the lines in the 
spectrum of any given substance are notall resolved into 
triplets by the magnetic field, but some are resolved into 
triplets while others become sextets, &c. ; and further, 
the magnitude of this resolution, that is, the interval 6A 
between the lateral components, does not appear at first 
sight to obey any simple law. 
Complex Atoms. 
According to the prediction of the simple theory, the 
separation 6A should be proportional to A*, and although 
this law is not at all obeyed, if we take all the lines 
of the spectrum as a single group, yet we find that it is 
obeyed for the different groups it we divide the lines into 
NO. 1547, VOL. 60] 
a series of groups. In other words, the corresponding 
lines A,, Ay, A, &c., have the same value for the 
quantity e/7,* or, as we may say, they are produced by 
the motion of the same ion. The other corresponding 
lines, B,, B,, Bs, &c., have another common value for 
e/m, and are produced therefore by a different ion, and 
so on. We are thus led by this magnetic effect to- 
arrange the lines of a given spectrum into natural 
groups, and from the nature of the effect we are led 
to suspect that the corresponding lines of these groups. 
are produced by the same ion, and therefore that the 
atom of any given substance is really a complex consist- 
ing of several different ions, each of which gives rise 
to certain spectral lines, and these ions are associated to: 
form an atom in some peculiar way which stamps the 
substance with its own peculiar properties. 
In order to illustrate the meaning of this, let us con- 
sider the spectrum of some such metal as zinc. The 
bright lines forming the spectrum of this metal arrange: 
themselves to a large extent in sets of three—that is, they 
group themselves naturally in triplets. Denoting these 
triplets in ascending order of refrangibility by A,, B,, Cj, 
As, By, Cy, &e., we find that the lines A,, A,, &c., show 
the same magnetic effect in character, and have the 
same value of e/7, so that they form a series obeying 
the theoretical law deduced by Lorentz and Larmor. In 
the same way, the lines B,, B,, B,, &c., form another 
series, which also obeys the theoretical law, and 
possess a common value for the quantity e/7, similarly 
for the lines C,, C,, C3, &c. The value of e/7 for the 
A series differs from that possessed by the B series, or 
the C series, and this leads us to infer that the atom of 
zinc is built up of ions which differ from each other in 
the value of the quantity e/7z, that each of these different 
ions is effective in producing a certain series of lines in 
the spectrum of the metal. When we examine the 
spectrum of cadmium, or of magnesium—that is, when we 
examine the spectra of other metals of the same 
chemical group—we find that not only are the spectra 
homologous, not only do the lines group themselves in 
similar groups, but we find in addition that the corre- 
sponding lines of the different spectra are szvz/arly 
affected by the magnetic field. And further, not only is 
the character of the magnetic effect the same for the 
corresponding lines of the different metals of the same 
chemical group, but the actual magnitude of the resolu- 
tion as measured by the quantity e/7 is the same for the 
corresponding series of lines in the different spectra. 
This is illustrated in the following table, and leads us to 
Nonets or 
Magnetic effect complex Sextets Triplets 
triplets 
Cadmium... i 5086. 4800 4678 
ZinChes vn 4811 4722 4680 
Magnesium N= 5184 5173 5167 
P ional spin ee 1s iy ee 
pias m9 m 87 Mt oe 
{This table shows the effect for the three lines which form the first natural 
triplet in the spectrum of cadmium compared with the corresponding lines 
in the spectra of zinc and magnesium. It will be seen that the corre- 
sponding lines in the different spectra suffer the same magnetic effect both 
in character and magnitude. Thus the corresponding lines 4800, 4722, 5173 
are each resolved into sextets, and the rate at which the ionic orbit is caused 
to precess is the same for each (denoted by e/# = 87 in the table). Similarly, 
for the other corresponding lines. ] 
believe, or at least to suspect, that the ion which pro- 
duces the lines Aj, A., A;, &c., in the spectrum of zinc is. 
* The quantity e is the electric charge of the ion, and 7 is its inertia, and 
the ratio e/#: determines the precessional frequency, or spin, of the ionic. 
orbit round the lines of magnetic force in a given field. 
