202 Washington — Kaersutite from Linosa and Greenland. 



In both cases the ratio of (Si, Ti)0 2 to (R' 2 , R'^O is greater 

 than unity, being 1*22 in the Linosa, and P06 in the Kaersut 

 hornblende. Such relations differ widely from those presented 

 by the hornblendes analyzed by Penfield and Stanley, in which, 

 where the ratio differs notably from unity, it is always on the 

 side of an excess of RO over Si0 2 . 



Assuming first that all the titanium is present as the dioxide, 

 if we calculate the composition of the molecules in terms of 

 the bivalent radicals suggested by Penfield and Stanley, we 

 obtain residues of Si0 2 , (Mg, Fe)0, and CaO, which differ 

 widely from the metasilicate ratio Si0 2 : RO = l : 1, in that the 

 original excess of silica is here greatly accentuated. This is 

 shown in the table below. 



Linosa. Kaersut. 



(Al. Fe) 2 0(F, OH)J SiO,_ -017 '033 



(Al, Fe) 2 4 RNa 2 ] Si0 3 ..'. -039 -059 



( Al, Fe) 2 3 R] Si0 3 -097 '026 



(Mg,Fe)0..__ ■2'34)._ 1 '370 



•234 ) 

 •217 ] ' 



451 



CaO -217 f x '195 



Residual Si0 2 _ . -634 -670 



If, on the other hand, we assume that some of the titanium 

 is present as sesquioxide, by making the necessary calculations 

 and readjustments (the results of which it is needless to give 

 here), we find that this will increase still more the ratio 

 Si0 2 :RO, in spite of the reduction in the amount of R0 2 , 

 through the diminution in the amount of FeO and the taking 

 up of MgO to form one of the complex radicals. We may 

 therefore assume that all the titanium is present as Ti0 2 , as this 

 shows less divergence from metasilicate ratios. 



It is therefore evident that, for our hornblendes at least, the 

 presence of some other bases or radicals must be assumed, 

 which will take up this seeming excess of silica and at the same 

 time conform to the metasilicate ratio Si0 2 :RO = l : 1. Such 

 may be found in the molecule Na(Fe, Al)Si 2 6 , which exists 

 as the ferric or alumina extreme respectively in riebeckite and 

 glaucophane, and whose presence in many amphiboles was 

 suggested by Tschermak, though not found essential to the 

 interpretation of the hornblendes studied by Penfield and Stan- 

 ley. This, and the analogous (Mg, Fe) (Fe, Al) 2 Si 4 12 , con- 

 form to the normal metasilicate ratio,- owing to the trivalence 

 of the basic iron and aluminum ; and the molecules may be 

 expressed graphically as ring formulas, quite analogous to those 

 suggested by Penfield, as is shown by that of riebeckite, as 

 follows : 



