CHEMISTRY: RICHARDS AND HALL 
343 
complete separation could not be attained (if at all) in less than eighty 
thousand fractionations. 
Estimation of Radioactivity. — The /3-ray activity of the lead used indi- 
cates that it contains in about 2 parts so-called 'radium G' and 1 part 
lead, an amount of 'radium D' of the order of 10~^ parts. This is be- 
trayed by the steady growth of a penetrating /3-ray product (RaE) 
which comes to practical equiHbrium with its parent in about a 
month after they are separated. Evidently the ionization — if the 
a-rays from the polonium present are cut off — caused by a weighed 
amount of the material under constant conditions, is a measure of the 
concentration of radium D relative to its isotopes 'radium G' and lead. 
Thus the determination of the activity of the end fractions gives infor- 
mation as to the relation between the solubility of the nitrate of 'radium 
and the mean solubility of the nitrates of 'radium G' and ordinary 
lead, a relation of great interest because it cannot be tested by atomic 
weight determinations. This case should afford an especially favor- 
able test of the theory of complete identity. 
The method was as follows: To eliminate differences in moisture 
content, and consequently in absorbtive power, the two samples were 
kept in the same exhausted desiccator after purification. When sixty- 
six days had elapsed 10.25 =t 0.03 gram of each sample was weighed 
into a marked metal dish and tamped with a platinum spatula. The 
material formed a layer of thickness defined by the expression 0.88 g/cm.^ 
This is far too thin to give a maximum ionization due to the complete 
absorption of the /3-rays from the lowest layers, but is thick enough so 
that no great accidental changes in thickness needed to be feared, and 
the effect of small changes was annulled as described below. The dishes 
were covered in turn with the same thin sheet of paper and introduced 
into a grounded aluminum-leaf electroscope which was very kindly 
loaned by Prof. William Duane. The rate of fall of the leaf was then 
observed with a stop-watch in the usual manner. All errors arising 
from differences of potential, differences of stiffness of the leaf in different 
positions, and inhomogeneity of the scale were eliminated in the usual 
way. The two dishes were introduced alternately into the electroscope 
to eliminate the effect of changes in the natural leak, and were thor- 
oughly stirred and repacked between every two measurements to allow 
for small accidental differences in thickness, and other precautions to cut 
out extraneous effects and secure exact comparabiUty were employed. 
It is beHeved that the table below, which includes all the observations 
affected by no known source of error, represents about as good repro- 
ducibility as this electroscope can be made to give. 
