OP ARTS AND SCIENCES. 



151 



weights stood to each other in the relation of whole numbers within 

 the limits of accuracy of the most refined experimental work. The 

 number of these cases was so large that it seemed highly improbable 

 that the coincidences should be the result of chance. 



This idea was prominently set forth by Professor Mallet of the 

 University of Virginia, in his admirable paper on the Atomic Weight of 

 Aluminum,* which was a striking illustration in point; and the same 

 feature was also made prominent by Professor F. W. Clarke of 

 Washington, after a careful review of all the determinations of atomic 

 weights.! 



The coincidences appear more striking if the values of the weights 

 referred to are given in values of the oxygen atom assumed to be 16, 

 as has been done by Professor George F. Becker in his digest of 

 atomic weight determinations.^ The following table from the writer's 

 work on Chemical Philosophy will make clear the point in question. 



ATOMIC WEIGHTS 



MOST ACCURATELY DETERMINED. 



Hydrogen 1.002 



Lithium 7.01 



Carbon 12.00 



Nitrogen 14.04 



Oxygen 16.00 



Aluminum 27.02 



Sodium 23.05 



Magnesium 24.00 



Phosphorus 31 05 



Sulphur 32.07 



Chlorine . 

 Potassium 

 Calcium . 

 Bromine . 



35.46 



39.14 



40.00 



79.94 



Silver 107.93 



Antimony 119.92 



Iodine 126.85 



Barium ......... 137.14 



Thallium 204.11 



Lead 20G.91 



This table includes all the atomic weights which up to 1882 could 

 be regarded as known within one thousandth of their value, and with 

 one or two notable exceptions there is no instance iu which the value 

 differs from a whole number by a quantity greater than the possible 

 error, thougli not always the " probable error," of the processes em- 

 ployed in their determination. 



Were these numbers wholly independent of each other and distrib- 

 uted by no law, we should expect to find every possible intermediate 

 value, and the fact that they so nearly approach whole numbers can- 



* Phil. Trans., 1880, p. 1003. 



t Smithson. Misc. Coll.; Constants of Nature, Part V. p. 270. 



i Smithson. Misc. Coll. ; Constants of Nature, Part IV. 



