TRANSACTIONS OF SECTION B. 713 



Moreover, the diimide has been heated to full redness in a gas combustion 

 furnace while dry hydrogen was still passed over it ; even under these conditions 

 little charring occurred, but some nitrogen and a phenyl radical were eliminated, 

 and the purified residue was found to approximate in composition to SiNPh, which 

 would represent the body as phenylsilicocyanide or a polymer of it. Even careful 

 heating of the diimide in ammonia gas has not enabled me to remove all the phenyl 

 from the compound, but rather to retain nitrogen, as the best residue obtained 

 from such treatment consisted of SioNjPh, or the phenylic derivative of one of the 

 substances produced by Schutzenberger and Colson from the ammonia reaction. 

 It may be that both these substances are compounds of silicocyanogen with an 

 imide group of the kind indicated below — 



SiN. SiN. 



>NH : >NPh 



sin/ SiN/ 



Further investigation must decide whether this is a real relationship ; if it be, 

 we should be able to remove the imidic group and obtain silicocyanogen in the free 

 state. One other point only need be noticed, namely, that when the above silicon 

 compounds are heated in oxygen they are slowly converted into SiO.,; but the last 

 traces of nitrogen are removed with great difficulty, unless water-vapour is present, 

 when ammonia and silica are quickly formed. 



Much remains to be done in this department of comparative chemistry, but we 

 may fairly claim to have established the fact that silicon, like carbon, can be made 

 to form perfectly well-defined compounds in which it is exclusively united with the 

 triad nitrogen of amidic and imidic groups. 



Now, having proved the capacity of silicon for the formation of compounds of 

 this order with a triad element, Nature very distinctively lets us understand that 

 nitrogen is not the particular element which is best adapted to play the triad role 

 towards silicon in its high-temperature changes, which are ultimately dominated 

 by oxygen. We are not acquainted with any natural compounds which include 

 silicon and nitrogen ; but large numbers of the most important minerals contain 

 the pseudo-triad element aluminium combined with silicon, and few include any 

 other triad. Phosphorus follows silicon in the periodic system of the elements as 

 nitrogen does carbon, but silicates containing more than traces of phosphorus are 

 rare ; on the other hand, several silicates are known containing boron, the lower 

 liomologue of aluminium ; for example, axinite, datholite, and tourmaline. 



^Moreover, it is well known that silicon dissolves freely in molten aluminium, 

 though much of the former separates on cooling. Winkler has analysed the gangue 

 of aluminium saturated with silicon, and found that its composition is approxi- 

 mately represented by the formula SiAl, or, perhaps, SioAU, if we are to regard 

 this as analogous to OjNo or cyanogen. Here aluminium at least resembles nitrogen 

 in directly forming a compound with silicon at moderately high temperature. It 

 would appear, then, that while silicon can combine with both the triads nitrogen 

 and aluminium, the marked positive characters of the latter, and its extremely 

 low volatility, suit it best for the production of permanent silicon compounds 

 similar to those which nitrogen can afford. 



With these facts in mind we may carry our thoughts back to that period in the 

 earth's history when our planet was at a higher temperature than the dissociation 

 point of oxygen compounds. Under such conditions the least volatile elements were 

 probably liquids, while silicides and carbides of various metals were formed in 

 the fluid globe. We can imagine that the attraction of aluminium for the large excess 

 of silicon would assert itself, and that, as the temperature fell below the point at 

 which oxidation became possible, these silicides and carbides underwent some 

 degree of oxidation, the carbides suffering most owing to the volatility of the 

 oxides of carbon, while the fixity of the products of oxidation of silicides rendered 

 the latter process a more gradual one. The oxidation of silicides of metals which 

 had little attraction for silicon would lead to the formation of simple metallic 

 silicates and to the separation of the large quantities of free silica we meet with in 

 the solid crust of the earth, whereas oxidation of silicides of aluminium would not 



