July 22, 1921] 



SCIENCE 



65 



Carbon ^ (3) 



Nitrogen 4 S (2) 



Oxygen (3) S 1 



Fluorine 1 



Silicon 4 (or electro-positive) 



Phosphorus ^ 3 



Sulfur 4 3 2 1 



Chlorine (4) (3) (2) 1 



In this table the numbers in italics give 

 the most common valences, while those in pa- 

 rentheses are only rarely found. It is clear 

 that a large kernel charge favors covalences 

 less than the negative valences while a small 

 kernel charge has the opposite efEect. A com- 

 parison of the elements of the second period 

 with those of the first, shows a slight tendency 

 for larger covalences among the heavier ele- 

 ments. This is to be explained as the effect 

 of the larger kernel and hence weaker forces. 

 There is also much more scattering among the 

 valences of the heavier elements. This is an- 

 other result of the weaker forces acting on the 

 electrons for the covalence of such atoms is 

 more dependent upon the electron affinity of 

 the other atoms with which they are combined. 



As an example of these relationships, let us 

 consider the compounds, CH^, NHj, H„0, 

 and HF. In each atom of these compounds 

 the covalence is equal to the negative valence 

 so that the residual charge is zero and the 

 tendencies of Postulates 1 and 3 are satis- 

 fied. If we mix the NHg and HF together the 

 larger kernel charge of the fluorine as com- 

 pared with the nitrogen, gives a tendency for 

 the fluorine atom to become negative at the 

 expense of the nitrogen. Thus the covalence 

 of the fluorine decreases to zero while that of 

 the nitrogen increases to four. This leads to 

 the formation of the compound NH^F which 

 consists of NH^+ ions and F" ions. The total 

 number of covalence bonds has not been 

 changed, they have merely been distributed 

 differently. But this causes the atoms to be- 

 come charged and makes the compound an 

 electrolyte. It should be noted that this theory 

 indicates definitely in what direction the 

 change of charge occurs. Thus we should not 



* See Langmuir, Jour. Amer. Chem. Soo., 41, 

 927 (1919). 



expect NH3 and HF to give a compound con- 

 sisting of ions NH,- and H„F+ although under 

 other conditions these ions might exist. 



Similarly NH3 and H,0 may react to form 

 NH^OH which will consist of ions NH^* and 

 OH". But the tendency to form a compound 

 such as this is much less than in the case we 

 have just considered, for the charge on the 

 kernel of the oxygen atom is less than that 

 of the fluorine atom so it has less tendency to 

 become negative. As a result NH, is much 

 less active towards H.,0 than towards HF. 

 Examples of this kind can be extended almost 

 indefinitely and can even be used to obtain 

 quantitative relationships between the heats 

 of formation of various substances. 



Since the sheaths of atoms of atomic num- 

 ber less than about 25 never contain more than 

 8 electrons, the covalence of these atoms can 

 not exceed 4. With heavier atoms, however, 

 we might expect in some eases larger co- 

 valences than 4. Large covalences are improb- 

 able in most cases for they imply equally large 

 negative valences which means that the num- 

 ber of electrons in the sheath must be very 

 much larger than the charge of the kerneL 

 There are a few compounds, however, which, 

 suggest that large covalences sometimes exist. 

 For example the compounds reCCO)^ and 

 Ni(CO)^ correspond to complete compounds 

 in which the central atoms have the covalences 

 10 and 8 respectively. Since e for iron is 8 

 and for nickel is 10 and the complete sheaths 

 for these atoms contain 18 electrons, the nega- 

 tive valences of iron and nickel are 10 and 8, 

 that is the same as the covalences needed to 

 account for the above compounds. Thus these 

 compounds are in accord with both Postulates 

 1 and 3, and are to be regarded as of a type 

 analogous to organic compounds in which the 

 covalence of every atoin is equal to its nega- 

 tive valence. It should be noted that both of 

 these compounds are liquids of low boiling 

 point (102° and 43°) and their molecular 

 weights have been determined. Their proper- 

 ties are about those to be expected if they have 

 the structure assumed above. Other com- 

 pounds of iron with carbon monoxide are 

 known, but they have only been obtained in 



