344 
CHEMISTRY: RICHARDS AND HALL 
TABLE 2 
/8-RAY Activities 
Sample A 
11.30; 11.43; 11.69; 11.56; 11.61; 11.47; 11.49; 11.50; 11.41; 11.58 Mean 11.504. 
Sample B 
11.32; 11.62; 12.06; 11.43; 11.70; 11.34; 11.32; 11.43; 11.47; 11.36; Mean 11.505. 
The ionization is expressed in arbitrary units (scale divisions per 
minute.) The natural leak (which must have been the same for both 
as they were taken alternately) was found to be 0.50 =b 0.01. The agree- 
ment of the averages to one part in ten thousand must be regarded as 
fortuitous, since their mean probable error is about 0.11 unit. This 
error, divided by the mean value from which has been subtracted the 
natural leak may be taken, roughly, as a measure of the accuracy of 
the determination. Its value is = This work definitely 
shows, then, that no change in the relative concentration of radium 
D and its isotopes as great as one per cent has occurred in nine hundred 
fractionations, and certainly gives no indication that radium D could 
be completely separated from its isotopes by this method even by crys- 
tallizing as many as 90,000 times. 
If ordinary lead is a mixture of isotopes, this mixture must have been 
made very long ago while the earth was still in a highly mobile condition 
(since all the ordinary lead throughout the world seems to have the same 
atomic weight^) . Could the composite nature of ordinary lead be proved, 
the identity of the several samples through geological aeons would form 
another argument in favor of the inseparability of the constituents. 
To the Carnegie Institution of Washington we are greatly indebted 
for some of the apparatus and material used in this research, which is 
to be continued in the near future. 
Summary. — ^Lead from AustraHan carnotite (beheved to contain about 
one part of ordinary lead to two parts of radium G, with a mere trace 
of radium D) has been fractionally crystallized over nine hundred 
times as nitrate, and the end-fractions purified. 
The atomic weights of the samples so obtained from the crystal and 
the mother liquor ends of the series respectively agreed within the ex- 
perimental error of 6 parts in 100,000. 
The j8-ray activities agreed within the experimental error of 1%. 
These observations indicate that the nitrates of Radium D and lead 
on the one hand and radium G and lead on the other hand could hardly 
be separated, if at all, by less than 80,000 or 100,000 crystallizations. 
