SECTIONAL TRANSACTIONS.— C. 357 



In the polymorphic and polyploid genus Rosa, specific sets are composed of seven 

 chromosomes (septets). Five differential septets have been identified, and their 

 various paired combinations make thirty-one regular species possible (five diploids 

 and twenty-six polyploids) ; twenty-three of these have been found, leaving eight to 

 be discovered. 



Irregular polyploids with unpaired septets have the same mechanism. Thus, a 

 species is a precise taxonomic unit subject to experimental verification. Genetical 

 species are homozygous for specific characters and heterozygous for sub-specific 

 and varietal characters, being taxonomically equivalent to Linnean species. 



Afternoon. 

 Excursion to Headington and Shotover. 



Wednesday, August 11. 



18. Prof. W. L. Bragg, F.R.S. — The Relationship betiveen Orthosilicates 



and Metasilicates. 



Several types of the less complex natural silicates have recently been analysed. 

 Amongst them are included Zircon ZrSiOj, Garnet CaiAl.,Si,Oio, Olivine Mg,Si04. 

 Monticellite MgCaSiOj, Beryl Be.iAlaSioO.s, Phenacite Be.^SiO^, Topaz (AlF)oSi04, 

 Chondrodite Mg^ (MgOH).2 (SiOa)^. To this list should perhaps be added the 

 various forms of silica SiOj, analysed by W. H. Bragg and Wyckoff, because of the 

 relationship between them and the silicates. 



The number of structures yf>t known is small, and generalisation premature. It is 

 interesting, however, to examine a few features of these structures on account of the 

 importance of this group of compounds. 



There is justification for the view that the silicon plays the part rather of a metal 

 in a complex oxide than of the nucleus of an acid radicle. The structure of the olivine 

 group (MgoSiO.), &c.) is identical with that of chrysoberyl BeAl.204. The latter is a 

 typical metallic oxide, closely related to spinel MgAloOj, and corundum Al^O.^. In 

 all these structures the oxygen atoms are in approximate close-packing, as though the 

 repulsions between these atoms decided the shape and size of the structure, and the 

 very much smaller metal ions are inserted into the interstices of the tightly packed 

 oxygen atoms. In fact, the dimensions of the close-packed oxygen structure can be 

 traced throughout the following series of compounds, in order of complexity— BeO, 

 ALO.H, BeAl.Oj, MgoSiOj, MgCaSiOj, ( AlF).Si04, Mg, (MgOH), (8104)2. The replacement 

 of a metal by silicon does not break the continuity of the series. 



Beryl, BeaALSi^OiN, has the composition of a metasilicate. Examination shows 

 that there are no groups of formSiO.H= inthecrystal, like the very definite CO;j= groups 

 in a carbonate. Instead, each silicon is surrounded by four oxygen atoms in beryl 

 as it is in every other silicate as yet analysed, and the correct proportion of oxygen to 

 silicon is obtained by a sharing of oxygen atoms between neighbouring Si04 groups. 



The structures of garnet and phenacite are again of the ' co-ordination ' type, like 

 spinel and chrysoberyl. In all these silicates the oxygen atoms take up the greater 

 part of the volume of the structure. They are bound together by atoms of the metals 

 and silicon. The silicon is at the centre of a tetrahedral group of four oxygen atoms, 

 and the metal at the centre of a similar tetrahedral group or of six oxygen atoms at 

 the corners of an octahedron. All are typical co-ordination structures. 



19. Prof. G. HiCKLiNG. — Graphical Studies of ike Relations of Peat, Coa 



and Anthracite. 



The author has made an attempt to ascertain the relation between the various 

 types of coal by graphic representation of the distribution of the chief elementary 

 constituents (C, H, 0, N, S) in some 1,500 coals of all types. The principal constituents 

 have been plotted one against another in pairs (C : H, C : 0, &c.), the ' dry, ash-free ' 

 analyses being employed in all cases. The analyses of the U.S. Bureau of Mines are 

 particularly adapted to the purpose, and have provided the greater part of the material. 



By this method it is reasonable to suppose that any groups of coal derived from 

 parent-material of exceptional composition, or which have undergone different lines 

 of chemical alteration, will be distinguishable as separate groups on the graphs. 



