40 ROYAL SOCIETY OF CANADA 
take place whenever possible, and proceed until one at least of the react- 
ing phases totally disappears. Which of the four will be the first to 
disappear, depends obviously on the direction of the reaction and on 
the relative amounts of the different phases present. When the number 
of phases has been reduced to three, the possibility of a reaction of this 
nature has vanished. 
The occurrence of reactions of this type, then, is the mechanism by 
means of which the number of phases in a three-component system is 
kept down to three. If it fails, the state of the system is no longer 
defined solely by temperature, pressure, and the masses of the compo- 
nents. Experimental evidence of such definition, must therefore 
be regarded as proof that not more than three phases are present in the 
(three-component) system. 
The result predicted by the second theorem is effected by means 
of reactions of a second type, which can take place in a three-component 
system of three phases only if one of the three is capable of continuous 
variation in composition (a solution). 
By dissolving more or less of the solid phases in the solution, it 
would be possible to construct a whole series of systems—formed of 
the same amounts of the same components—but differing in the com- 
position of the solutions and in the relative quantities of the solid 
phases. If only one of these is found to exist in fact, the disappearance 
of the others must be ascribed to the occurrence of reactions of this 
second type, which fix the composition of the solution irrespective of 
the relative quantities of the three phases of which the system is com- 
posed. 
Extension of the Method. 
The two theorems which serve to interpret the experimental results, 
distinguishing between mixtures and single substances, and identifying 
the chemical individuals in the precipitates, are thus dependent on the 
occurrence of reactions of two types:—the first (which does not involve 
change of composition of any of the phases) resulting in the disappear- 
ance of all phases in excess of three; the second (which involves change 
of composition of the solution) keeping the composition of the solution 
constant so long as it is in contact with the same pair of basic salts. 
The possibility of applying the same method to the study of preci- 
pitates formed by potash in solutions of metallic salts, depends on the 
discovery of conditions under which a fourth component may be added 
to the system without interfering with the occurrence of these two 
classes of reactions. 
A 
fi 
