Bartow—A Mechanical Cause of Homogeneity of Crystals. 688 
As in the case of assemblages combining, it is not necessary to 
suppose that all the parts of any considerable portion of the 
changing mass reach a condition of equilibrium simultaneously ; 
the effect will still take place if the passing disturbances are such 
as to prevent any great continuity in either time or space of the 
new symmetrical arrangements produced. And as was just now 
pointed out, if any particular homogeneous arrangement momen- 
tarily and locally brought about is one at which linking occurs, 
and this change of state does not occur where the regularity falls 
short of this, the mass may at least partially, notwithstanding the 
disturbances, pass by degrees, as here and there the requisite 
arrangements are produced, into new permanent combinations, of 
definite symmetry." 
We find a resemblance to the process just described in the 
orderly separation of different kinds of atoms which follows a 
chemical decomposition, the mass breaking up into less complex 
substances of a definite composition. 
Also, in the orderly arrangement arrived at by the atoms of a 
body as it crystallizes out of a liquid or semi-liquid mixture. 
VII.—The exchange of some of the constituents of two or more as- 
semblages so as to constitute fresh assemblages; an effect 
which resembles the redistribution of the atoms which occurs 
in a chemical double decomposition. 
If two assemblages, each of which is composed of one or more 
different kinds of balls} whose arrangement is the closest-packed 
possible, are when brought together capable of closer-packing if a 
redistribution of the balls is made by which two fresh assemblages 
are formed, it is evident that, unless the linking in the original 
assemblages prevents, the action of the principle of closest-packing 
will produce such a redistribution. 
In some cases the effect may be produced by a combination of 
all the constituents to form a single assemblage, followed by a 
separation along fresh lines, while in other cases an actual inter- 
change takes place at all points where the parts in juxtaposition 
have not the closest-packed arrangement possible for them; one 
1 Comp, p. 674. 
