92 PROCEEDINGS OF THE BALTIMORE MEETING 



is sho\yn in the use of the norm as distinguished from the mode, or actual rock. 

 The use of the norm increases the definiteness of chemical classification, but 

 loses touch with the actual rock. The present paper is an attempt to test the 

 relative refinement of chemical characterization of correlative groups in the 

 old and new methods of classification to see whether or not the gains compen- 

 sate for the losses. 



The problem proposed is the testing of the relative homogeneity of the types 

 as determined by specific diverse systems of classification. The method em- 

 ploj^ed has been an arithmetical consideration of the eight constituents com- 

 monly present in rocks as represented in the superior analyses collected by 

 Washington. The grouping of these analyses was based on that proposed by 

 Iddings for the normative and by the names applied by different authors for 

 the modal classification. 



A preliminary study included the empirical examination of the effect of the 

 choice of few or many examples, taken at random, on the range of values of 

 the several constituents. This shows that in most instances 50 examples give 

 only approximate results and that more than 150 examples add little to the 

 observed ranges. The quantities of the several constituents were tested for the 

 range in actual amount and the ratio between oxides and groups of oxides in 

 the analogous groups developed by the normative and modal classifications re- 

 spectively. 



The method applied failed to show that the introduction of the norm, with 

 subsequent groupings as applied by Iddings, gives an improvement in chemical 

 characterization to commensurate groups sufiicient to compensate for the loss 

 in other characterizing factors. For chemical units smaller than those gev.- 

 erally recognized in the classification of rocks, the normative method is capa- 

 ble of progressive logical refinement impossible in the modal classification, 

 involving minerals of isomorphic series. When these smaller units are grouped 

 together the chemical characterization is more definitive verbally, but not actu- 

 ally, and probably inferior to results obtainable by a quantative modal classifi- 

 cation. 



As a method of dividing chemical complexes (rocks) into groups in which 

 some inter-relationships of constituents are fairly constant and others practi- 

 cally disregarded, the normative method is eminently successful. That this 

 success in the grouping of rocks implies not only that the particular chemical 

 relationships selected are more fundamental than others, but that they are also 

 more fundamental as characteristic of igneous rocks and more valuable than 

 the sum of all of the modal characteristics combined, since the normative classi- 

 fication disregards modal characteristics for the sake of bringing out with re- 

 finement the few chemical characteristics. E^en so, the actual chemical char- 

 acterization is not appreciably closer. 



Presented in abstract by the author from the manuscript. 

 Discussed by Prof. J. P. Iddings and Dr. Whitman Cross, with reply 

 bv the author. 



