A.—MATHEMATICAL AND PHYSICAL SCIENCES 29 
element mercury seems specially provided by nature to help in the work. 
Not only do its nine isotopes provide a most valuable scale of abundance 
but it usually occurs in the discharge, to which its presence is advantageous 
from the point of view of smooth running, and it is unique in its property 
of forming multiply charged ions. Mercury lines up to the fifth and sixth 
orders can be detected so that it provides a perfect natural scale, a link 
between light and heavy atoms absolutely necessary to extend accurate 
measurements to the latter. ‘The packing fraction of mercury, which is 
practically zero, was determined by means of its third order line °*8Hgt+ ++. 
Of the recent episodes in the story I relate certainly the most sensational 
is the discovery of deuterium, the heavy isotope of hydrogen. The events 
leading up to this and following it form a most remarkable sequence. 
In them the elements of nature seem to have joined in an impish, but 
fortunately benign, conspiracy to delude the observer and to turn his 
most sober researches into a sort of blind man’s buff. 
The first comparison of the masses, now termed ‘ isotopic weights,’ of 
the atoms 1H, 1##C, 14N, and 160 to a high degree of accuracy were made 
with my second mass-spectrograph and published in 1927. Various 
methods were used which cannot be given in detail here, but since the 
comparison of H with O could only be done through the intermediate 
4He, and even then the ratios measured were very large, little reliance 
could have been placed on the figure for H unless it could be checked in 
some quite independent manner. It was possible to do this by means of 
the close doublet O—CH,, and, when measurements of this appeared 
to support my values for C and H, I had no reason to doubt their sub- 
stantial accuracy. In this I have been justified to some extent for the 
figures have stood for seven years, and a direct determination of the 
He,H ratio, made later by Bainbridge agreed exactly with mine. A 
further support was afforded by the fact that the figures for the four 
elements, all then supposed to be simple, agreed within 1 or 2 parts in 
, 
10,000 with the accepted chemical atomic weights. 
This satisfactory agreement was completely upset in 1929 by the startling 
discovery of the heavy isotopes of oxygen 17 and 18 which, present in 
small quantity, had naturally been overlooked on mass-spectra of that 
element owing to the technical difficulty of ensuring the absence of the 
isobaric compound lines OH and OH,. The discovery was made by 
Giauque and Johnson by observations on band spectra, which are free from 
this confusing disability, and the careful quantitative work of Mecke, 
made later, showed that, owing to the presence of these isotopes, the 
chemical standard of atomic weight O = 16 was about 2 parts in 
10,000 heavier than the physical one *O = 16. Examination of com- 
pounds of carbon and of nitrogen by the same method showed not only 
that these elements also contained heavy isotopes #8C and 1°N but that 
their apparent abundance, by a most incredible coincidence, was just 
about enough to bring their mean weights into line with that of oxygen. 
Birge pointed out that to satisfy my low estimate of 1H hydrogen must 
also contain at least one heavy isotope. Urey took up the problem and, 
happily unaware of the real uncertainty in the figures concerned, with 
the collaboration of Brickwedde and Murphy fractionated liquid hydrogen 
and proved by examination of the Balmer lines that "=H was present. 
