276 
CHEMISTRY: HARKINS AND WILSON 
THE STRUCTURE OF COMPLEX ATOMS AND THE CHANGES 
OF MASS AND WEIGHT INVOLVED IN THEIR 
FORMATION 
By William D. Harkins and Ernest D. Wilson 
KENT CHEMICAL LABORATORY. UNIVERSITY OF CHICAGO 
Presented to the Academy. April 13, 1915 
A study of the atomic weights referred to hydrogen as unity shows 
that for the series of elements from helium to fluorine the average 
deviation from a whole number is 0.09 units, which is smaller than 
would be expected if the atomic weights were wholly the result of acci- 
dent. However, the average deviation for the first 27 elements (0.21) 
is almost as great as would be expected as the result of chance (0.25 
units). This, on first sight, does not seem to accord well with the 
numerical form of Trout's hypothesis, that these atomic weights should 
be whole numbers. If now the atomic weights referred to oxygen are 
considered, the remarkable relation is found that the numbers are ex- 
ceedingly close to whole numbers as a rule. In order to save space 
these relations may be considered in the form of the results of calcula- 
tions of probability. The chance that the sum of the deviations from 
whole numbers for the 27 lighter elements should be as small as it is 
when hydrogen is taken as 1 is found to be one chance in ten, when the 
method used for the calculation is of the form known as De Moivre's 
problem, and when the probable errors in the atomic weights are taken 
into accoimt. In contrast with this it is found that there is only about 
one chance in fifteen miUion that the sum of the deviations from whole 
numbers of the atomic weights on the oxygen basis should be as small 
as it is. 
This might seem to indicate that the atoms other than hydrogen are 
built up from units different from hydrogen and of a weight 1.000, 
while hydrogen has a weight L0078 (on the oxygen basis) ; but this would 
involve the existence of two kinds of particles with masses very nearly 
the same, and gives the additional difficulty that the atomic weights 
are not exact multiples of 1, which indicates that if the particle from 
which the atoms are built has a weight of nearly 1, this weight must 
vary somewhat from atom to atom. Attempts have been made to 
explain the discrepancy as the result of a variation in the number of 
negative electrons, but this method is neither in accord with recent 
theories of atomic structure, nor does it seem to agree with the fact 
that the atoms are neutral electrically. 
