084 Scientific Proceedings, Royal Dublin Society. 
for the present,’ and devoting our attention to the groups, we notice 
that the presence of a single detached group of any kind will 
involve the existence of a number of groups similar to it dispersed 
through space, and without going outside our data, we have, there- 
_fore, means for the manufacture of a number of groups of precisely 
the same pattern, such as we have already seen can be packed 
together in various ways. 
An assemblage which by the rupture of some of the ties in a 
homogeneous manner has been partitioned into uniform groups, or 
groups of a limited number of kinds, may display as orderly an 
arrangement of its parts as prevailed before the ties were broken, 
but we can, on the other hand, conceive of it as disturbed by some 
external influence, so that the arrangement and orientation of the 
groups with respect to one another becomes more or less irregular. 
Yet, and this is important, notwithstandiny any disturbance from 
without, the interaction of its parts, to which the tendency to pack 
closely is traceable, may continually operate to diminish irregularity, 
and although the work done may be continually undone, may perpe- 
tually re-arrange the groups, ever striving after that orderly arrange- 
ment which gives the closest-packing possible. 
A pure chemical compound in a liquid state will evidently be 
comparable to an assemblage in which the ties have been symmetri- 
cally broken in such a way as to give but a single kind of group 
of linked spheres, the arrangement and orientation of the groups 
having simultaneously become more or less irregular. 
2. As to the nature of the isolated groups, we may regard these 
as small similar fragments of a solid assemblage, mobile as to one 
another, but of stereotyped pattern. The component parts of such 
a group will be arranged with as high symmetry with respect to 
one another, as they displayed in the homogeneous assemblage in 
which the grouping was produced ; in some cases they may display 
greater symmetry, namely, if the balls of a group, when the links 
connecting them with the balls of other groups are broken, are 
able to take up more symmetrical relative situations consistently 
with the variety of balls found in the group. 
1 See pp. 640, 676, and 679. The case whose consideration is thus deferred must 
not be confounded with the spongelike structure produced by intercalation subsequently 
suggested to be that of crystalloids, see p. 666. 
