272 Prof. W. J. Sollas. 



germ of a great truth, which subsequent refinements in the geo- 

 metrical treatment of the subject have to some extent obscured. 



Recent work by Lord Kelvin, Mr. Tutton, and others, has given 

 great encouragement to those who cherish a hope of ultimately 

 arriving at some sure representation of crystalline structure ; but 

 the remarkable observations of Penfield* on the several species of the 

 mineral chondrodite are of an especially suggestive character. The 

 chondrodites present us with a homologous series of chemical com- 

 pounds, having the following constitution : 



No. of atoms of Mg. 



(a) Mg 3 [Mg(F.OH)] 2 (Si0 4 ) 2 5 



(6) Mg 6 [Mg(F.OH)] 8 (Si0 4 )s 7 



(c) M g7 [Mg(F.OH)] 2 (Si0 4 ) 4 9 



As Professor Miers tells us, "the three minerals have almost the 

 same form, but differ only in the length of one crystallographic axis 

 (parameter). The lengths of this . . in the three minerals are in 

 the ratio of 5 : 7 : 9. Thus the addition of the olifine radicle, 

 Mg 2 Si0 4 , exerts a so-called ' morphotropic ' action along one definite 

 axis, and causes a certain constant increase in the length of this axis 

 (parameter)." " It is further a curious fact that the length of this 

 (parameter) varies directly as the number of magnesium atoms in 

 the compound." The inference that the magnesium atoms are dis- 

 posed along this axis would seem to follow naturally. 



Mr. Barlow has lately completed an investigation into the differ- 

 ent ways in which spherical bodies of different volumes may be 

 packed together so as to produce different crystal forms. His 

 method of treatment is general, and many of the arrangements of 

 atoms, which I am myself led to imagine as existing in certain 

 crystals, will, no doubt, be found to fall as special cases under his 

 general laws. But I do not propose in this communication to merely 

 suggest some way in which atoms regarded as spheres may be built 

 up into geometrical forms ; my purpose is different, it is to consider 

 csrtain particular substances, and to endeavour to discover the par- 

 ticular structure which must most rationally be assigned to them. 

 Further, to avoid misconception, it should be added that the "law of: 

 closest packing," which is so admirably developed by Mr. Barlow, 

 does not seem to be obeyed by the greater part of the crystals I have 

 investigated ; so far as the cubic system is concerned, there is but 

 one simple snbstance in which the atoms are as closely packed as 

 possible, that is the diamond. 



It is a singular fact that crystals of the cubic system, when com- 

 posed of salts which are diatomic, such as NaCl,KCl, are character- 

 ised by a cubic cleavage, while those composed of triatomic salts 

 * ' Amer. Journ. Science,' vol. 47, p. 188. 



I 



