606 
LETTERS TO THE EDITOR. 
(The Editor does not hold himself responsible for opinions 
expressed by his correspondents. Neither can he undertake 
to return, or to correspond with the writers of, rejected 
manuscripts intended for this or any other part of NATURE. 
No notice is taken of anonymous communications.] 
The Complex Nature of Thorium. 
With regard to several letters on thorium and its complex 
nature that appeared in Nature of March 24 and 31, April 
7 and 14, and in which my name is mentioned, I take the 
liberty of adding a few remarks, having had ten years’ 
experience in working with thorium. 
In 1897, at a meeting of the British Association in Toronto 
(Canada), I read a paper in which I pointed out that 
spectrum evidence proves the complex nature of thorium. 
In 1898 (Chem. Soc. Trans., p. 953) 1 isolated from some 
thorium fractions an earth with an atomic weight of 225-8 
(tetrad). Knowing the difficulties of the separation of 
rare earths (I have been engaged in this kind of work 
since 1878), and not wishing to publish a premature con- 
clusion, I did not declare this to be a novel constituent of 
thorium, but said that foreign earths were present, in spite 
of the fact that the reaction used ought to have separated 
them. 
In 1901 I published another short paper (Proc. Chem. 
Soc., March 21, 1901, pp. 67-68), in which I said that 
“my experiments may be regarded as proving the complex 
nature of thorium.’’ Thorium was split up into Tha and 
Ths. With ThB I obtained so low an atomic weight as 
Riv =220. The fractions Tha gave by the analysis of the 
oxalate, though it was prepared by pouring the thorium 
salt solution into an excess of oxalic acid, in order to avoid 
the formation of a basic salt, the high atomic weight 
Riv = 236-3. But I stated expressly, and I feel obliged to 
repeat it, that these fractions show a great tendency to 
form basic salts. Assuming these to be normal, a higher 
atomic weight than the true one is obtained. This is true 
especially in regard to the oxalate. 
The splitting up of thorium into Tha and Thf was, of 
course, not so sensational an event as the announcement 
from America of the splitting up of thorium into “ caro- 
linium ’’ and ‘‘ berzelium.”’ Bonustav BRAUNER. 
Bohemian University, Prague, April 18. 
Radio-activity and the Law of Conservation of Mass. 
Mr. Soppy in the Wilde lecture on the ‘‘ Evolution of 
Matter as Revealed by the Radio-active Elements ’’ (Proc 
Manchester Phil. Soc., vol. xlviii., part ii., p. 29) gives 
two methods of deducing the average life of a radium atom. 
The results become concordant if we assume that the 
complete disintegration of an atom of radium involves the 
emission of four a particles. Now the atomic mass of 
radium is 225, and that of an a particle about 2; the question 
therefore arises as to what has become of the rest of the 
mass, 
There appear to be three possible answers to this 
question. In the first place Mr. Soddy’s estimate may be 
wrong by a factor of ten, although it is hardly likely that 
the data are uncertain to this extent; secondly, the various 
stages of the disintegration may involve the liberation of 
non-radio-active by-products which would necessarily be in- 
capable of detection by the methods of investigation 
employed ; and, finally, there may be a decrease in the total 
mass of the system owing to the decrease in the velocities 
of some of the constituent electrons. 
A priori the second hypothesis appears to have the balance 
of probability in its favour, as it agrees best with our 
present ideas; but I think the third solution should not 
be dismissed too hastily. In discussing the matter recently 
with Mr. G. A. Schott, I found that he also had been led 
to consider the tenability of this view in connection with 
some theoretical work on the structure of the atom which 
will probably soon be published. O. W. RicHarpson. 
Trinity College, Cambridge, April 109. 
NO. 1800, VOL 69] 
NA OTs 
[ApRIL 28, 1904 
The Atomic Weight of Radium. 
In the Philosophical Magazine for April, 1903, Runge 
and Precht work out the atomic weight of radium from 
its spectrum to be 258, instead of the 225 found by Madame 
Curie. I should like to point out that the spectrum data 
of radium support the value given by Madame Curie, if 
handled according to Section v. of *‘ The Cause of the 
Structure of Spectra’’ (Phil. Mag., September, 1901). 
There is an important question of principle involved in 
the distinction between the two methods of using the spec- 
trum of an element for determining its place in the periodic 
classification. 
The method adopted by Runge and Precht is founded on 
a slight alteration of the approximate empirical law dis- 
covered by Rydberg, that, when similar elements have 
corresponding pairs or triplets of lines in their spectra, the 
difference of the frequencies of vibration of the two lines 
in a pair belonging to any one element is proportional to 
the square of its atomic weight. Runge and Precht alter 
this law by substituting the words ** some power of its 
atomic weight ’”’ for ‘‘ the square of its atomic weight.” 
Thus in the alkaline earth family they give the power as 
1/0-5997- The frequency differences of corresponding pairs 
of lines in Mg, Ca, Sr and Ba, namely, 91-7, 223, 801 and 
1691 being denoted by x, the logarithms of the atomic 
weights M of these elements are given by the formula 
log M=o0.2005+0-5997 log x. As x for radium is 4858-5, 
the value 258 is found by Runge and Precht for the atomic 
weight of radium. But the frequency differences are con- 
nected with one another by numerical relations, and not 
directly by their atomic weights. 
The clearest instance of this purely numerical relation- 
ship is shown by the spectra of Zn, Cd, and Hg. A 
characteristic set of corresponding frequency differences for 
these elements is 386.4 for Zn, 1159-4 lor Cd, and 4633-3 for 
Hg. The number for Zn multiplied by 3 gives 1159 2, which 
is indistinguishable from the value for Cd, and 12 times 
the number for Zn gives 46368, which is within 1 part in 
1000 of the value for Hg. 
In the paper mentioned I have shown how this numerical 
series, I, 3, 12, appears elsewhere in the frequency differ- 
ences for other families of elements, with a fourth member, 
namely, 28. Now these numbers are the first four terms 
of the series the general term of which is 1—3n/2+7n"/2. 
This series, and not atomic weights, controls the relations 
between the frequency differences for the spectra of allied 
elements. 
Perhaps the most striking evidence in support of this 
assertion is afforded by some characteristic frequency differ- 
ences discovered by Runge and Paschen in the complicated 
spectra of O, S, and Se. These are for O 3-7 and 2-08, 
for S 18-15 and 11-13, and for Se 103-7 and 44-07. Of 
these six numbers four are the first four terms of the series 
the general term of which is 3-7(1—3n/2+7n"/2), namely, 
3-7, 11-1, 44-4, and 103-6. The numerical law applies, then, 
to the non-metals as well as to the metals. The distinction 
between a purely numerical and an atomic weight relation 
between the frequency differences of allied spectra is funda- 
mental, the one implying a kinematical, the other a 
dynamical, origin for the structure of spectra. 
As to the bearing of these considerations on the deter- 
mination of the atomic weight of radium from its spectrum 
we can represent the differences given above for the alkaline 
earth elements by the formula 63 8(1, 3, 12, 26)+31-6, in 
which 26 for Ba takes the place of 28 in the standard series. 
There are other instances in which a serial number like 
12 or 28 is reduced by 2 or 4, a phenomenon probably of 
kinematic origin. For the next three elements of this 
family with higher atomic weights than Ba, the frequency 
differences corresponding to those just given should be the 
values of 63-8(1—3n/2+7n*/2)+31-6, given when n=4, 5 
and 6. The coefficients of 63-8 are 51, 81 and 118. Now the 
4858-5 of Runge and Precht for radium is 63-8 75-7+31-6. 
Thus the 75-7 of radium corresponds to the 81 of the 
regular series in the same way that the 26 of Ba does 
to the 28 of the regular series. So the numerical law for 
the frequency differences places radium two main rows 
lower than Ba in the table of elements, and gives to it, 
therefore, an atomic weight exceeding that of Ba, namely, 
