218 INTEGRA.L ATOMIC WEIGHTS. DODD. 



each single element. Xevertiieless, the heights of these lines 



persisted in approximating to multiples of -i or 8 with 



a few exceptions. Where it is a multiple of 3, that will crop 



up repeatedlY in the same group of elements. Similarly when 



it is a multiple of 5. This happens, by the "^ay, in the pentad 



group, which is decidedly interesting. 



super-ordinate 



He afterwards found that the ratio -— approx- 



1.8 



imated more closely to the specific gravity if the heights of the 

 horizontal lines above the base were measured with 1.008 

 (atomic weight H) as a unit instead of with 1 as a unit. 



The number of unit? in the ordinates running from the 

 base line right up to the horizontal lines may be called the 

 ••'Atomic Multiples" ; and the Atomic Multiples x 1.008 are 

 what have been called above, the "Integral Atomic Weights." 



A list is given in Tig. 2, of these Atomic Multiples, 

 arranged after the manner of Mendeleeff's table. 



It will be found that, 



. ^ ^ . IntegTal At. Wt.— actual At. Wt. 

 Specific Gravity= — r 



with an average error of 0.5.5. A large number work out with 

 great aecurae3\ 



Manganese is not very amenable to the system, but it should 

 be remarked that it stands very much alone in its group any 

 way. 



It is well to keep the two factors of the Integral Atomic 

 Weight — namely the Atomic Multiple and the unit of 1.008 

 — separately in mind, as it may be found desirable to alter the 

 the Atomic Mutltiple for any element, or the unit for all the 

 elements in order to get more precise results later on. 



MeUing Points. 



These Atomic Multiples have a very distinct relation to 

 the melting point curve of the elements. As the list of the 



