530 SCIENCE PROGRESS. 



obvious exceptions which exist at the present time and which 

 may be relegated to the position required only after very 

 careful and accurate research as to their true atomic weights. 

 With respect to this law as with Prout's, one must not for- 

 get that it must express the results of experiments and not 

 override them. To take an example, we must not assign 

 to tellurium an atomic weight less than that of iodine simply 

 because that of oxygen is less than that of fluorine, and that 

 of sulphur less than that of chlorine, nor say that it must be 

 a mixture when all experimental evidence points to the con- 

 trary. What we call tellurium may be a mixture and true 

 tellurium may have an atomic weight less than that of iodine. 

 When this is conclusively proved, but not till then, let such 

 an atomic weight be assigned to it. 



It is obvious that if we are to have the atomic weights 

 accurately represented by whole numbers this must depend 

 to a very great extent on the unit selected, if it be possible 

 at all. Berzelius adopted oxygen as his standard, making 

 its atomic weight equal to ioo. He did this because his 

 values were so largely derived from the oxidation of the 

 elements or the reduction of their oxides so that he obtained 

 his results directly in terms of the atomic weight of oxygen. 

 This standard gives numbers too large for convenient use, 

 and therefore those chemists who believe in referring 

 atomic weights to an oxygen standard have adopted six- 

 teen as its atomic weight, primarily because the atomic 

 weight of oxygen is approximately sixteen times that of 

 hydrogen and for a long time it was supposed to be exactly 

 so. Chemists are still divided in opinion as to what is the 

 most scientific standard for atomic weights. Many, for the 

 same reasons which influenced Berzelius, hold that sixteen 

 for the atomic weight of oxygen is better than one for the 

 atomic weight of hydrogen. To refer very many to the 

 hydrogen standard simply means correcting those on the 

 scale for oxygen by multiplying by a factor, but in not a few 

 cases this cannot be done so simply, especially when we 

 have obtained the number partly by direct reference to 

 hydrogen as well as to oxygen and have taken a mean value. 

 Further, at the present time it can hardly be said to be 



