334 SECTIONAL TRANSACTIONS.—A. 
1,000. This instrument definitely proved the isotopic nature of neon, chlorine and 
many other elements and established the Whole Number Rule. The measurements 
with this apparatus were always referred to the line O'6. In the absence of any 
evidence to the contrary oxygen was assumed to be a simple element, so that this 
system of measurement made the new physical scale of the weights of atoms identical 
with the chemical scale of atomic weights. With but few exceptions the atomic 
weights deduced from mass spectra agreed well with those generally accepted, and since 
it was decided to retain the chemical meaning of the words, the numerous new species 
of atom were distinguished by the chemical symbol and their whole number weight, 
called the mass number. 
The accuracy of 1 in 1,000 was high enough to indicate that variations from the 
whole number rule, expected theoretically from the nuclear packing, were present, 
notably in the exceptional case of hydrogen, but that higher accuracy would be 
necessary to measure them. In 1925 the second mass-spectrograph was built. This 
had a resolving power of 1 in 500, and an accuracy, in favourable cases, of 1 in 10,000. 
The first results obtained were described in the Bakerian Lecture of 1927. The per- 
centage deviation of the weight of an atom from a whole number on the scale of 
O0'%=16 was expressed in parts per 10,000 and called its packing fraction. These 
packing fractions when plotted against mass numbers, lie roughly on a hyperbolic 
curve. 
During the last three years a photometric method of calculating the relative 
abundance of isotopes from the intensity of their lines on mass spectra has been used 
to determine the mean atomic weights of complex elements. In most cases these 
can be fixed to a few hundredths of a unit of atomic weight. The method is rapid 
(some 26 elements have already been done), and being in general independent of purity 
gives a valuable check on chemical determinations. The agreement in the majority 
of cases is quite satisfactory. The two most glaring discrepancies, those of Kr and 
Xe, have now been removed by density redeterminations. Further chemical work 
appears to be desirable on Se, Te, Cs and Os. In all, 62 of the non-radioactive elements 
have been examined for isotopes. In addition to the 62 species of atoms inferred to 
occur in these there have been discovered 101 by the mass-spectrograph, eight by 
other methods of positive ray analysis, and four by optical observations on band 
spectra, making a total of 175 atomic species in sixty-two elements. 
In 1929 Giauque and Johnston made their surprising announcement that observa- 
tions on oxygen band spectra indicated the presence of O17 and O18. The quantitative 
relations to O!6 are small, and of the order 10-4 and 10-° respectively. Their measure- 
ment is still a matter of great difficulty. Naudé concludes that the combined effect 
of the isotopes makes a difference between the scale 0%=16 and the chemical scale 
of 1-25 x 10-4 ; Mecke gives the higher figure 2-2 x 10-4. In the calculations of chemical 
atomio weights from photometry of mass spectra I have used the former figure. The 
complexity of oxygen accentuates the fact that a unit founded on a chemical standard 
can never be really suitable for physics. We can never be quite certain that any 
element is absolutely simple, although it seems highly probable that this is the 
case with hydrogen and helium. For physical measurement it is essential to take 
some definite atom as our standard. 
Of those proposed the most hopeful are: the Proton or the Hydrogen Atom ; 
the Alpha Particle or the Helium Atom; the Oxygen Atom of mass number 16. 
Some idea of the numerical relations between-scales derived from these and the present 
chemical scale can be obtained from the following table. The figures for the masses 
are purely illustrative and are based on the results of the mass-spectrograph. The 
mass of an electron is about 0-00054 on such scales. 
te = Chemical i 
H=1 He=4 aula O4=16 
Ht Be = 1-00000 1-00723 1-00765 1-00778 
He! Hic 5s 3:97127 4-00000 4-00166 4-00216 
Qe os .. 15-8765 15-9914 15-9980 16-0000 
Oxygen .. ..  15:8785 15-9933 16-0000 16-0020 
Chlorine .. <3 | BOrLSo 35-443 35-458 35-462 
ge 00 as .. 198-392 199-908 199-991 200-016 
It is obvious that the scale of H=1, although it would have the advantage of making 
all the packing fractions of the same sign, is quite inadmissible, since it throws the 
masses of heavy atoms right off their proper mass numbers. He=4 avoids this 
i ted be et 
ae 
