494 HISTORICAL INTRODUCTION TO CHEMISTRY CHAP. 



less for hydrogen, and almost exactly a unit greater for Li, 

 Be, B, F, Na, Al, P. The nature of this secondary factor 

 is not yet known, but if it exists, there is no reason why 

 the atomic weights should not occasionally overlap, as in 

 the case of Ar and K, Co and N, Te and I ; it need 

 not therefore be assumed that the atomic weights now 

 given for these elements are incorrect, as Mendeleeff 

 asserted. 



SUMMARY AND SUPPLEMENT. 

 A. METALS AND NON-METALS. 



The idea that all bodies were composed of four "elements" 

 or three " principles " was opposed by Boyle, who urged that 

 the number of elements could only be determined by experi- 

 ment (Sceptical Chymist, 1661). Lavoisier, in 1789, gave a list 

 of 30 elements ; Berzelius, in 1819, described 50 ; the number 

 now known exceeds 80. 



Lavoisier classified the elements in three chief groups : (a) 

 Oxygen, with light and heat, azote and hydrogen, (b) Metals, 

 forming basic oxides, and (c) Non-metals, forming acid oxides. 

 Davy classified chlorine with oxygen as a " supporter of com- 

 bustion," and also described the elements as " positive " or 

 " negative " according as they were attracted to the negative or 

 to the positive pole in electrolysis. Berzelius classed fluorine 

 and iodine with chlorine, and sulphur and selenium with oxygen. 

 Finally the oxygen-group ceased to be distinguished from the 

 other non-metals, and the elements were divided into two groups, 

 as metals and non-metals. 



B. NUMERICAL RELATIONSHIPS AND THE PERIODIC LAW. 



Dobereiner, in 1829, showed that the atomic weight .of 

 bromine was approximately the mean of the atomic weights 

 of chlorine and iodine. Other triads were S, Se, Te ; Ca, Sr, 

 Ba ; Li, Na, K ; also Ni, Cu, Zn and Pt, Ir, Os, where the 

 atomic weights differed very little. Dumas, in 1859, discovered 

 regular increments of atomic weights in families of 4 or 5 



