Chemical Composition of Amphibole. 29 



was selected which floated and sank within narrow limits. 

 The heavy solutions used were potassium-mercuric iodide for 

 the lighter and barium-mercuric iodide for the heavier varieties 

 of amphibole. In every case the powders were washed with 

 great care to remove every trace of the heavy solutions ; they 

 were also examined under the microscope to note the presence 

 of any possible impurities. It may be stated with confidence 

 that all of the materials which have been analyzed were of the 

 utmost possible purity. 



Method of Analysis. — The methods employed in making 

 the analyses were like those for a long time in use in the Shef- 

 field Laboratory, and essentially the suggestions as outlined by 

 Clarke and Hillebrand* and by Wasbingtonf were followed. 

 The analyses were made by Stanley, and before starting on 

 them considerable time was spent in gaining familiarity with 

 the methods of separation, for which purpose numerous arti- 

 ficial mixtures were made and analyzed. In cases where much 

 iron and alumina were present the separation from magnesia was 

 made by means of a double basic acetate precipitation. Fluorine 

 was determined by the Berzelius method as outlined by Penfield 

 and Minor,;}; and from experiments made with artificial mix- 

 tures containing the same constituents in about the same pro- 

 portions as "in amphibole it is probable that the results obtained 

 for fluorine are all too low by about O10 to 0*15 per cent. Water 

 was in all cases determined by the closed-tube method de- 

 scribed by one of the present writers.§ In several cases, especi- 

 ally in the early part of the investigation, analyses of the same 

 material were repeated four or six times in order to gain the 

 greatest possible accuracy in the determinations. With few 

 exceptions the determinations were always made in duplicate. 



New Analyses and discussion of residts : — Amphibole 

 occurs in several distinct varieties, some of which are conveni- 

 ently designated by names, and it seems best to take up the 

 analyses in groups, commencing with the simplest types. The 

 discussion of the analyses is based upon the ratios derived by 

 dividing the percentages of the several constituents by their 

 molecular weights. In the case of fluorine the percentages are 

 divided by twice the atomic weight, in order to make the 

 quotients comparable to those obtained from the percentages 

 of H 2 ; two fluorine atoms being equivalent to two hydroxyl 

 radicals, rejjresented in the analyses by the hydrogen atoms of 

 H 2 0. Hydrogen unquestionably plays a double role in 

 amphibole and numerous other minerals having a complex 



"Analyses of Kocks and Analytical Methods, U. S. G. S. Bull. 148. 

 fThe Chemical Analysis of Rocks. 

 % This Journal, xlvii, 387, 1894. 

 § This Journal, xlviii, 31, 1894. 



