Multiple Proportions of the A tomic Weights. J I 



The absolute parallelism of the positive and negative 

 series of elements Hn and H272 in their numerical, chemical 

 and physical relations, together with their close resemblance, 

 to homologous series, will be at once apparent to philo- 

 sophical chemists. The small differences observable 

 between the experimental and a few of the theoretic atomic 

 weights, when distributed among the twenty-four members 

 of the four series, only amount to 0*0036, or less than half 

 of 1 per cent of the actual determinations. Excluding 

 these differences, the atomic weights are a simple multipli- 

 cation table, the significance of which will be understood by 

 students in other departments of science quite as well as 

 by the most able chemists. 



The theoretic atomic weights are also in much closer 

 agreement with experimental results than the fundamental 

 law of atomic heats formulated by Dulong and Petit for 

 the same series. No one doubts the general accuracy of 

 this law because it does not hold good for boron and 

 silicon, or from its inexactitude to fractional quantities 

 throughout the whole number of the elements. 



I would take this opportunity of remarking as a prin- 

 ciple of scientific reasoning, that when the number of 

 recurring facts is sufficient to establish the relation of cause 

 and effect, or, in other words, the general accuracy of a law, 

 the road to further discovery lies rather in the direction of 

 explaining the causes of anomalous departures from it, 

 than in challenging the truth of the law itself. I would 

 also emphasise the fact, the importance of which is hardly 

 yet realised by chemists, that as the received atomic 

 weights are all expressed in units of hydrogen and are 

 equivalents of this element, the multiple relations sub- 

 sisting among the higher atomic weights, as shown in the 

 above table, have an immensely greater validity in 

 determining the question of their being whole numbers of 

 hydrogen, than when the atomic weights were compared 



