684 Scientific Proceedings, Royal Dublin Society. 
kind filtering in one direction, another in the opposite direction at 
the same time, and the process continuing until the arrangement ~ 
throughout is of the closest-packed nature. If the constituents — 
dislodged are, before the change occurs, linked to other parts of — 
the same assemblage, the readiness with which the links break will 
do something towards determining which of these two alternatives 
prevails in any given case. 
As before, it is not obligatory to suppose the newly-formed 
assemblage to be continuous ; it may be produced in comparatively 
minute fragments as here and there the requisite symmetry is from 
time to time reached, and then stereotyped by the formation of 
links. 
The readiness of a compound assemblage to effect exchanges of 
this kind with other assemblages will depend on whether its con- 
stituents are loosely or closely packed, in other words on the extent 
to which some constituent of it is saturated with the others. 
This corresponds very well with the facts concerning double 
decomposition. Thus in speaking of certain cyclic compounds, 
Bischoff says, that the dimethylene. complex is very unstable 
and is broken by hydrogen bromide, bromine, and even by iodine. 
Trimethylene is alone decomposed by hydrogen bromide, not by 
bromine. Finally tetramethylene and hexamethylene are very 
difficult to break or not to be broken up.’ 
Inversion by exchange. 
Suppose that a certain collection of balls is so constituted as to 
find equilibrium in two slightly different homogeneous assemblages 
under some slight variation of the conditions, and that when these 
assemblages break up into groups as a change to the partially- 
linked or liquid condition® takes place, these groups are in both 
cases all of one kind and the composition the same in both cases ; 
and suppose further that the two assemblages of groups A and B are 
so related that the only material difference in the arrangement, apart 
Jrom the linking, is that one kind of ballin A occupies entirely different 
situations from those it occupies in B, and that the unoccupied points 
in B, which correspond to the places of these balls in A, and also the 
unoccupied points in A, corresponding to the places of the same balls in 
1See p. 680. * Bischoff’s ‘‘ Handbuch der Stereochemie,’’ p. 51. °% Comp. p. 583. 
