Speight. — On an Occurrence of Olivine-andesite. 369 



I. II. III. IV. 



Loss on ignition ... 2-97 3-16 3-13 3-13 



Silica, SiO.> 55-70 54-99 55-10 55-12 



Alumina, A1,0 3 ... 21-10 20-96 ... 2041 



Ferrous and ferric oxide 8-47 8-75 ... 7*74 



Lime, CaO 5-39 5-38 5-38 535 



Magnesia, MgO ... 1-29 1-45 ... 2-75 



Soda, Na.,0 3-72 3-94 4-03 3-80 



Potash, K 2 2-66 2-66 2-17 2-50 



101-30 101-29 ... 100-80 



It will be noticed that the loss on ignition in No. I. is less 

 than in the other cases. This was due to the fact that the 

 mineral was heated in an open crucible, and so the iron pre- 

 sent in the ferrous state was oxidized to the ferric state ; in 

 the succeeding analysis the powdered rock was heated in a 

 closed crucible, so that there might be no oxidization. After 

 determination of the loss on ignition, the mineral was fused 

 with fusion-mixture, and on extraction with HC1 the silica was 

 determined. The result obtained was high in the first case, 

 owing to an error in weighing, care not being taken to pre- 

 vent absorption of water during the process ; but in the 

 succeeding cases this was provided for. 



The alumina and iron-oxide were determined by precipita- 

 tion with amnionic hydrate, and reckoned as A1 2 3 and Fe 2 3 , 

 and the amount of iron present was determined from another 

 part of the solution by titration with permanganate of potash. 

 This was reckoned as ferric oxide, and subtracted from the 

 combined weight of the AL0 3 and Fe.,0 3 . This gave the A1 2 3 , 

 and the iron was estimated as ferrous oxide, since on testing 

 with sulphocyanate of ammonia only a very faint coloration 

 was obtained in a part of the original solution. Some Fe„0 3 

 was undoubtedly present, as microscopic examination showed. 

 It will be noticed that in analysis No. I. the amount of the 

 A1 2 3 and FeO is high. This was due to the fact that insuffi- 

 cient amnionic chloride was added, and the magnesia came 

 down with it. This made the reading for magnesia low. It 

 will be noticed in No. IV. that the alumina is lower and 

 the magnesia higher. In estimating the FeO by titration in 

 the first cases the ordinary decinormal solution of permanga- 

 nate was used, and, on account of the small quantity of iron 

 present in the solution tested, the readings were not sufficiently 

 accurate. In the last determination the permanganate was 

 diluted to a known amount, and greater accuracy obtained. 

 The lime was precipitated with amnionic oxalate, and esti- 

 mated as oxide. 



The magnesia was precipitated with sodic phosphate and 

 24 



