5t Dr, WoLLASTON on the elementary Particles 



plate divisible into prismatic rods, the section of which is ar 

 rhomb of 70° 32' and 109° 28' nearly ; and if we again split these 

 rods transversely, we shall obtain a number of regular acute 

 rhomboids, all similar to each other, having their superficial 

 angles 60* and 120°, and presenting an appearance of primi- 

 tive molecule, from which all the other modifications of such 

 crystals might very simply be derived. And we find, more- 

 over, that the whole mass of fluor might be divided into, and 

 conceived to consist of, these acute rhomboids alone, which 

 may be put together so as to fit each other without any inter- 

 vening vacuity. 



But, since the solid thus obtained (as represented fig. 2.) 

 may be again split by natural fractures at right angles to its 

 axis (fig. 3.), so that a regular tetrahedron may be detached 

 from each extremity,, while the remaining portion assumes 

 the form of a regular octohedron ; and, since every rhomboid, 

 that can be obtained, must admit of the same division into one 

 octohedron and two tetrahedrons, the rhomboid can no longer 

 be regarded as the primitive form ; and since the parts inta 

 which it is divisible are dissimilar, we are left in doubt which 

 of them is to have precedence as primitive. 



In the examination of this question, whether we adopt the 

 octohedron or the tetrahedron as the primitive form, since 

 neither of them can fill space without leaving vacuities, there 

 is a difficulty in conceiving any arrangement in which the par- 

 ticles will remain at rest : for, whether we suppose, with the 

 Abb6 Hauy, that the particles are tetrahedral with octohedral 

 cavities, or, on the contrary, octohedral particles regularly 

 arranged with tetrahedral cavities, in each case the mutual 

 contact of adjacent particles is only at their edges ; and 



