XVIIL] SOLUTION AND THE CHEMICAL PROCESS. 451 



unity, in which we can no more assert the existence of the 

 chloride or of water, than of chlorine, hydrochloric acid, or a 

 metallic oxide, although these and many others are conceiv- 

 able results of its differentiation. If the solution be one of 

 chloride of potassium, evaporation resolves it into water and 

 the chloride, but if chloride of aluminum, it is decomposed by 

 boiling into water, hydrochloric acid, and alumina, or in the 

 case of the corresponding magnesiaii salt, into hydrochloric acid 

 and an oxychloride. 



The precipitation of the sulphates of cerium, lanthanum, and 

 calcium from their solutions by heat, and of most other salts 

 by cold, is chemical decomposition or differentiation. Dilution 

 may also effect decomposition in solutions ; we have already 

 said that the combination of terchloride of arsenic, AsCl 3 , with 

 36 HO is stable at ordinary temperatures, but a further addi- 

 tion of water causes the solution to divide into aqueous hy- 

 drochloric acid and crystalline oxide of arsenic. The precipi- 

 tation of chloride of antimony, and of many salts of bismuth 

 and mercury by water, is an analogous process. This decom- 

 position of the solution of chloride of arsenic is an example 

 of what is called double elective affinity (attractio electiva du- 

 plex), and is generally explained by saying that the attraction 

 of arsenic for oxygen, and that of chlorine for hydrogen, en- 

 .able the chloride and water to decompose each other. But 

 these elemental species do not exist in the solution, although 

 they are possible results of its decomposition, and to explain 

 the process in this manner is to ascribe it to the affinities of 

 yet unformed species. 



I have elsewhere asserted that double decomposition always 

 involves union followed by division (ante, page 428), although 

 we cannot in every case arrest the process at the first stage. 

 Under some changed conditions of temperature and pressure 

 the decomposition may be the counterpart of the previous 

 union, and thus reproduce the original species, as in the case 

 of mercuric oxide, which is decomposed into mercury and oxy- 

 gen at a temperature a little above that at which it was formed. 

 When the division takes place in a sense different from the 



