LEWIS. — AUTOCATALYTIC DECOMPOSITION OF SILVER OXIDE. 731 



rises and falls as the crystallization continues ; rises as the surface of the 

 crystal increases, falls as the degree of supersaturation decreases. 



Striking as the analogy is, it must be borne in mind that we are dealing 

 iu this case with the action of one solid upon another, and the great regu- 

 larity of the reaction seems hard to reconcile with all the other facts 

 known concerning reactions in solid phases. 



As a result of other studies I have been inclined to entertain an entirely 

 different explanation of the phenomenon. In the course of researches, 

 some of which will be shortly published, I have been led to believe iu 

 the great importance to chemistry of the two following reactions : 



20 ^ 0.. 



These two reactions have frequently been discussed in connection with 

 the theory of the nascent state, but little attention seems to have been 

 paid to the immense importance of the velocity of these reactions in a 

 large number of processes, such as the union of oxygen and hydrogen, 

 combustions in general, reduction by hydrogen, spontaneous decomposi- 

 tion of oxidizing and reducing agents, and especially iu the phenomenon 

 of galvanic polarization. Here, however, this larger problem concerns 

 us only as it offers a possible explanation of the autocatalysis of silver 

 oxide. When we consider the probable mechanism of the decomposition, 

 it is evident that instead of the one reaction 2AgoO = 4:Ag + 0-2, we may 

 write with at least equal probability the two reactions AgoO = 2Ag + O 

 and 20 = Oo. If in fact the decomposition takes place in these two stages 

 the important question is, Which is the reaction of which we are measur- 

 ing the velocity ? that is. Which of the two is the slower ? 



We have many reasons to believe that the reaction O^ = 20 is an ex- 

 traordinarily slow one. For example may be cited the inactivity of 

 oxygen and the preponderating tendency in slow oxidization at ordinary 

 temperatures for the oxygen to enter into the resulting compounds as the 

 radical (Oo) forming the unstable peroxides. We may assume that to an 

 extremely small extent oxgen gas is dissociated, that in ordinary Oo there 

 is a small concentration of O in equilibrium with it. If when the equi- 

 librium is destroyed by removing some monatomic oxygen it is restored 

 with great slowness, then it will be true also that if the etiuilibrinm is 

 destroyed in the opposite direction the recovery, according to the reaction 

 20 = Oo, will likewise be slow. 



Therefore, it is not unlikely that this latter reaction is the one whose 



