soMF. co\TF.MPOR.iKy .iPr.-ixcFS IX rnvsics If 4«5 



einlu oU'iiUMits and llu- prfci-din^c <mv (IinkI) each in a (-oluniii <if its 

 i)\vn inarkiHl with its atomic mnnhiT (ami for i(k'ntitication thi- iiaiiie 

 of soiiu- I'lt'iiU'iU ill the s;inu' niiunin of the periodic table), while the 

 mass of each kind of atom is given by its elevation above the bottom 

 of the hgiire (the values are written along the vertical axis).'^ 



In these tabulations of Fig. 10 anil I"ig. II, all the numbers rep- 

 resenting atomic masses are written as integers. The conspicuous 

 post-decimal figures occurring in the sequence of combining weights 

 are absent; the notorious '.ir>Ar> of chlorine, the 24.32 of magnesium, 

 the 10.83 of lx>ron have vanished from the scene. Are then the 

 masses of all atoms really integer multiples of I'n of the mass of the 

 oxygen atom, using "really" in its only significant sense of "within 

 the uncertainty of observation? " Or do some of ihcm deviate ap- 

 preciabK' from the rule? The triaLcan be made most exactly upon the 

 lightest elements, as for these a given deviation from an integer value 

 would bulk as a larger percentage of the totJil mass, which is the 

 measured quantity, than it would for the hca\y elements. It is 

 fjcrformed by mingling the ions under test with ions of oxygen, or of 

 some other element, preferably one which has previously been com- 

 pared with oxygen; the locations of the traces of the two pencils of 

 ions upon the photographic plate are compared. Mingling lithium 

 ions with carbon ions, Aston finds that the masses of the two kinds 

 of lithium atoms stand to the mass of the carbon atom as 



(7.00(5 i.OOo): 12.000 and (6.008 ±.00.5): 12.000 



and if the mass of the carbon atom is exactly \e. that of the oxygen 

 atom, then the masses of the lithium atoms are very slightly distinct 

 from IP. and \\; of the oxygen mass (for, little as the difference exceetis 

 the uncertainty of experiment, .Aston regards it as real). Beryllium, 

 however, yielded the values 9.003 and 9.001 — indistinguishable 

 experimentally from 9.000 — in two separate experiments, in terms of 

 the same assumed value 12.000 for carbon. Farther along in the 



"The .itomic masses of these different kinds of atoms arc largely hy|X)thetical. 

 They have been measured for four single isotopes liclonging to four distinct elements; 

 radium (number 88, mass 226), radium emanation or niton (number 86, mass 222), 

 thorium (nunilxr 90, mass 232i, and uranium (number 92, mass 2.?8). Measure- 

 ments have also been made upon samples of lead Ijelicved to be composed almost 

 entirely of a single isr)to(ic, giving 2()6 for one kind of atom and 208 for another). 

 Each of the other atoms is a ilescendant of one or two of the four first-nametl atoms, 

 and its atomic mass is calculated by subtracting, from the atomic mass of its ancestor, 

 the masses of all the fragments which dropix'd away from the earlier atom during 

 its evolution. This procedure is confirmed by comparing the nteasured values 

 for uranium, radium, radium emanation, and one sample of lead, which all belong 

 to the same line of evolution: and the measured values for thorium and for another 

 sample of lead which is descended from thorium. 



