614 HENRY S. WASHINGTON 



As the original specimens were small, these differences cannot 

 be ascribed to differences of composition in different parts of 

 the mass, but are more probably due to the different analytical 

 methods used. This applies especially to the alkalis,- in the 

 determination of which the usual German method of treatment 

 of the rock with HF and H 3 S0 4 , and subsequent separation of 

 Al, Fe, Mg, and Ca, instead of the much more expeditious and 

 accurate method of J. Lawrence Smith, tends to erroneous and 

 discrepant results largely through contamination with alkalis 

 derived from the glass vessels and reagents employed. 



Having thus obtained an insight into the chemical and min- 

 eralogical composition of these rocks, it will be well to deter- 

 mine their place in the present scheme of classification. 



In the first place, the very close correspondence of both my 

 analyses, except in the minor constituents Ti0 2 and P 3 5 , show 

 that, in any scheme based on this character alone, they are to be 

 classed together. Indeed, this is true of all the rocks of the 

 Kula region examined by me, as shown by the analyses of 

 Rohrig given in the original paper. 



From a purely chemical standpoint it is abundantly evident 

 that they are not to be referred to the basalts as this name is at 

 present understood, since the alkalis are far too high for this 

 group of rocks. The alumina is also too high, at least in com- 

 parison with the basalt analyses in which this constituent has 

 been determined with proper care. Nor are they to be put with 

 the subgroup of hornblende-basalts, as is shown by comparison 

 with some typical analyses of these given in V, VI, and VIL 

 The Kula rocks differ from these in having higher Si0 2 , Al 2 O s 

 and alkalis, and lower iron oxides, MgO and CaO. In other 

 words, they are more properly leucocratic, while the hornblende- 

 basalts are melanocratic. At the same time the hornblende- 

 basalt from Kosk Creek (VII) , described by Diller, occupies a 

 rather intermediate position between the Kula rocks and the 

 typical hornblende-basalts of the Rhone and other regions. 



The closest analogues of the Kula rocks are to be found 

 among the nepheline-tephrites and nepheline-basanites. It is 



