134 
Transacfi07is of the BoyaJ Society of South Africa. 
6. Rhombic Dodecahedra. 
Bultfonteiii rhombic dodecahedra merge on the one hand into occasional 
tetrahexahedra, white, yellow and green, and on the other hand by impercep- 
tible gradations through rounded octahedra, to some glassies. Those of the 
midway class are of special interest. For purposes of description they may 
be regarded either as octahedra whose edges have disappeared in solution, 
or as dodecahedra whose corners have been truncated b}^ octahedral faces, 
The octahedral faces are usually much indented with triangles, the dodeca- 
hedral and tetrahexahedral faces nevei'. When these two are indented at 
all it is with small round pits. It is curious that it is only in this midway 
class of Bultfontein diamonds that triakis — and hexakis — octahedral affinities 
are seen, namely when the dodecaliedral or tetrahexahedral edges invade the 
faces of the octahedron. The typical Wesseltoti. octahedron, for example, 
shows no such affinities.* The prevailing striated surfaces and rounded 
forms of Bultfontein diamonds create refraction troubles which make it 
difficult sometimes to see whether the interiors are flawed or spotted. 
7. Twins. 
Interpenetrating twins are common enough at Bultfontein and relatively 
rare elsewhere. The two niembei's of a twin combination are usually of the 
same habit, i.e. they may be both octahedra, or both dodecahedra, or both 
macles. But there is no hard and fast rule, and any possible form may 
occur twinned to any other. As to their origin, Crookes suggests that " two 
drops, joining after incipient crystallisation, might assume the not uncommon 
form of interpenetrating twin crystals." Some of these twins certainly 
look as if they might have been formed in some such way — say two octahedra 
or two dodecahedra joined point to point by the slightest of bonds, or a 
made joined to almost any isometric. Yet outside appearances on the 
whole point to consecutive rather than to concurrent growth. For there is 
unbroken gradation all the way from simple contact between two complete 
crystals down to absolute inclusion of one by another. It would be quite 
easy to arrange a collection of Bultfontein twins showing progressively one 
of a pair more and more deeply embedded in the other, i. e. a point buried, 
a quarter buried, a half, three-quarters, and so on, then only a point stick- 
ing out, and lastly complete immersion. 
It is not antecedently improbable, of course, that in each case the pair 
might have started together, and that one outgrew and in some cases finally, 
so to speak, overflowed the other. But it is just as likely that one might 
have started first, run its single course to the end, and then served as a 
nucleus for the other. And this view is supported by the venerable aspect 
of some inclusions. 
The careen octahedra from the Rand banket show such affinities to perfection. 
