96 ALBERT JOHANNSEN 



therefore, to receive at once answers to the following questions as well 

 as further comments from all who are interested. 



QUESTIONS 



i. Classes. — Should there be a fifth class for rocks having 

 approximately equal amounts of light and dark constituents ? The 

 limits would then be 0-5-35-65-95-100 instead of 0-5-50-95-100, 

 as here proposed. The introduction of an extra class would add 

 104 families. 



2. Orders. — Should Order 4 (Fig. 23), in which there are very 

 few rocks, be combined with Order 3 ? Order 3 would then contain 

 all rocks with plagioclase from labradorite to anorthite inclusive. 

 This would make the subdivisions from Ab to An at 0-5-50-100, 

 and would reduce the number of families by 72. Of course, if the 

 fourth order is retained the pigeonholes need not be named until 

 rocks occupying them have been found. 



3. The line separating the granites, adamellites, etc., from the 

 corresponding quartz-rich varieties is here taken at 50 per cent 

 quartz. Should there be a dividing line here, or should granite, for 

 example, include all rocks having from 5 to 95 per cent of quartz ? 

 As suggested above, the division line might be made at 65, making 

 the lines 0-5-65-95-100. 



4. In the older classifications albite is united with orthoclase 

 for the alkali rocks. This would throw out Order 1, but in the 

 older systems, with the introduction of lime, the soda molecules are 

 divided into two parts, and orthoclase plus albite is contrasted with 

 the lime-soda plagioclases. This division is not logical, but is it 

 desirable ? If such a division were made, Order 1 (Fig. 20) would be 

 dropped and the alkali rocks would form Families 1, 6, n, 16, 21, 

 and 26 of the triangles now representing Order 2 (Fig. 21), and 

 soda- and potash-rocks would have to be separated in the sub- 

 families. The double triangle would then have orthoclase-}- albite 

 +microperthite+anorthoclase for the left angle of the base, while 

 the right corner would be CaiVaf, NaCaf, or Caf, depending upon 

 the orders. Such a combination would simplify the placing of 

 rocks containing microperthite, which is worth careful consideration, 

 but the grouping is not so correct theoretically. All of the rocks 



