510 Her. T.G. Botvney — Pitchstones and Felsites of Arran. 
traction, and will liave a tendency to break up into fragments. Tbe 
great facility witlr wliicb spheres are formed under contractile strain 
will render a perlitic structure a very common result , 1 but seeing 
tbat obviously space cannot be entirely filled by spherical surfaces, 
fracture into regulär polyhedrons is likely to result, the form of 
wkich, in accordance with the principle of least action, will be 
rhombic dodecabedrons. 
Consider now tbe case of a single one of tbese, and suppose it to go 
on very gradually losing beat uniformly from its exterior, tben, just 
as I liave shown happens in tbe case of spberoidal structure on alarger 
scale, tbe surfaces of equal tension will be approximately spherical. 
As the mass contracts, all tbe particles in any one of tbese surfaces 
will be sitnultaneously acted upon by an equal central force: if tben 
the temperature, pressure, etc., be such tbat microlithic crystalliza- 
tion is set up, the crystals will grow towards the centre, because botli 
tbat is tbe direction of the normal force, and tbe surface of crystal- 
lizing temperature is approaching tbe centre. There is thus a certain 
analogy with the growth of radiatiug columns in a spberoidal mass 
of igneous rock, like the spindle at St. Andrews. Tbe nucleus, wben 
present, is therefore more often likely to be a residuum tlian an 
included foreign body. As tbe mass is in a state of normal strain, 
it may once or twice rupture along a concentric surface during the 
formation of the spberule, and this may account for tbe concentric 
layers in tbe spherules. Certainly tbese are not seldom surfaces of 
weak cobesion, because tbe spherule exfoliates easily along tliem, 
and tbey are often covered with a sort of white dust. There is 
tlius a common cause in tbe development of perlitic and spherulitic 
structure, namely, a contractile force, but this in the latter case is 
acting on a crystallizing body. I think, liowever, tbat tbe same 
explanation will apply to tbe spherules wbicb do not show a crystal- 
line structure, and probably to some of the larger felsitic balls in 
igneous rocks, tbe outer surface denoting a surface of equal force, 
wbicb bas caused tbe part within to assume a structure (not neces- 
sarily crystalline) different from tbat without. Further, it will be 
obvious that the process of formation of a line or band of spherules 
will be analogous to the formation of spberoidal structure in a 
colurnn of an igneous rock, especially to tbat case wbere the spheroids 
are formed without external rupture of the colurnn. 
Some of tbe above evidence, such as tbe banding of pitchstones 
towards tbe exterior, and one at least of tbe banded felsites, also 
seems to show tbat this latter structure may be tbe result of squeez- 
ing. not of flowing; tbe same result, as I liave pointed out in 
another case , 2 being produced by diverse forces. The former cause 
is likely to produce tbe more regulär structure ; for it will be readily 
seen that if a number of particles, not bomogeneous, are mixed 
togetber, and tben squeezed, so as to be forced to pack into smaller 
space, there will be a kind of rliytbmic arrangement, like getting 
1 See Quart Journ. Geol. Soc. vol. xxsi. p. 152. 
2 loc. eit. p. 153. 
