IN THE MINERAL SERIES. 45 



of the author. If we make the macrodiagonal unity, these prisms 

 "give for the brachydiagonal, the following values : — 0"4fi60 (= V ; 

 prism-angle, 130°2') ;— 09484 (=F2; prism-angle, 93°3') ;— 1-397 

 (= F 3 ; prism-angle, Ti'^lO'} ;— 1-G84 (= F4 ; prism-angle, 6L°47'). 

 In Lievrite, the two commonly occurring prisms, give, respectively — 

 0-6840 (=F; prism-angle, nri2');— and 1-370 (= V 2 ; prism- 

 angle, 72"" 16')- To make these correspond with the chrysolite values, 

 the first prism must be considered to equal V a, and the second F 3. 

 The vertical axes of the two forms stand to each other very nearly ia 

 the ratio of 5 to 4. But we need not attempt to push these analogies 

 beyond their legitimate limits. The affinity, in general characters 

 and conditions of occurrence, of Lievrite to Fayalite (and through 

 this latter to Chrysolite^ cannot be overlooked if we take the entire 

 relations of these substances into consideration. Single characters, in 

 the determination of analogies, are necessarily useless. 



Now, the atomic constitution of the Chrysolites is perfectly 

 well established. The oxygen atoms in base and acid are equal, and 

 the base consists of simple oxides only, the general formula being 

 2 (RO), SiO^ ; or two atoms of monoxidized base -|- one atom of 

 Silica. In the normal Chrysolites the base consists of magnesia ; in 

 the Olivines, of magnesia and protoxide of iron ; and in Fayalite (at 

 least, essentially) of the latter oxide alone. In Lievrite, on the other 

 hand, as shewn above, both protoxide and sesquioxide of iron are 

 present ; and the oxygen atoms of the Silica do not correspond with 

 those of the bases. 



In attempting to reconcile these discrepancies, I called to recollection 

 a fact that came under my notice some time ago. In examining a 

 specimen of Lievrite that had been broken up into small pieces, I 

 found that certain fragments exerted a much more powerful influence 

 on the magnet, than others. Whilst some of the particles scarcely 

 shewed a trace of magnetism, others were magnetic in a marked de- 

 gree. These latter, tested by the blow -pipe, seemed almost free from 

 silica ; whilst the feebly-magnetic fragments gave very readily, with 

 microcosmic salt, the well-known reaction of that substance. It 

 occurred to me, therefore, that the analysed specimens of Lievrite 

 might have contained a certain portion of magnetic iron ore : a cii - 

 cumstance easily conceivable, if we call to mind the geological associa- 

 tions of our mineral. The nearly uniform proportions of the Fe O 

 and Fe^O^ found in the various analyses, seemed, it is true, opposed 



