Washington — Kaersutite from Linosa and Greenland. 209 



ness of the analyses made by Stanley under the direction of 

 so preeminent an analyst as the late Professor Penfield can- 

 not be doubted, and it is not probable that Schneider's figures 

 are seriously in error, especially as he states* that special atten- 

 tion was paid to the determination of this constituent, though 

 he does not mention the method employed. 



In passing, a few remarks may be made in regard to Nos. 

 IV and V. The exact localities of these hornblendes are not 

 given by either author, but they both are said to come from 

 Bohemia, and the extremely close similarity in the figures for 

 Si0 2 , A1 2 3 , FeO, MgO, CaO, Na 2 0, and K 2 makes it proba- 

 ble that they were made on identical amphibole from the same 

 locality. In view of this remarkable agreement of the con- 

 stituents mentioned, the discrepancies observed in the figures 

 for Ti0 2 and Fe 2 3 are noteworthy. In IV Ti0 2 is higher 

 and Fe 2 3 lower, while in V the reverse is true. The sum of 

 the two in each case is, however, exactly the same, 13'26. The 

 exact agreement is, of course, a coincidence, but taken in con- 

 nection with the close concordance in the other constituents, 

 and having regard to the methods of analysis, it suggests the 

 explanation that in IV the ferric oxide was determined by 

 reducing the iron by II 2 S, which does not act upon Ti0 2 , while 

 in V the iron was reduced by zinc, which would reduce the 

 Ti0 2 to Ti 2 3 , and the latter would appear as Fe 2 3 , after sub- 

 traction of the FeO. On this supposition the lower Ti0 2 of V, 

 in conjunction with the close agreement between the figures 

 for Al 2 3 , would be explained by the use of the method of 

 boiling an acid solution with S0 2 , which is as apt to give too 

 low as too high results, or even possibly by the assumption 

 that the residue left on evaporation of the silica with HF is 

 wholly Ti0 2 , a not uncommon error, especially in former days. 

 The presumption therefore is that IV represents the true com- 

 position more accurately than V. The matter, after all, is of 

 little importance, but serves to illustrate the need of a critical 

 study of analytical data in the light of the methods of analysis. 



The blue arfvedsonitic amphiboles (IX) and the triclinic 

 aenigmatite (X and XI) resemble ours in the figures for Si0 2 

 and K 2 0, and especially in the high Ti0 2 of the Greenland 

 aenigmatite. Ti0 2 was not determined in the analysis of arfved- 

 sonite and the Pantelleria cossyrite, but it is presumably present 

 in both, and almost certainly very abundantly in the cossyrite, 

 as Brogger has shown. Additional reasons for this belief will 

 be given in a forthcoming paper on the rocks of Pantelleria. 

 These hornblendes differ, however, widely from ours in the 

 lower A1 2 3 , the very much higher FeO and Na 2 0, and in the 

 very low MgO and CaO. The recently discovered triclinic 

 *C. Schneider, Zeitschr. Kryst., vol. xviii, p. 579, 1890. 



