652 HENR V S. WA SHING TON 



AI3O3, which is almost a straight line but slightly convex,^ all 

 the curves are concave' (toward the bottom). 



The curves divide themselves naturally into two groups, 

 according to general direction. Those of SiOg, Al^Og, NagO 

 and K3O ascend toward the right (the periphery), the three last 

 almost arithmetically. It will be noticed that the K3O curve 

 ascends more rapidly than that of NagO to a point a little to the 

 right of IV, when it drops a trifle more rapidly. This is expressed 

 in the series of the ratios of these two oxides, already given in 

 Table I. The second group is that of (FeO), MgO and CaO, 

 which descend toward the rig-ht, and at a g'reater rate than those 

 of the other group rise, SiOg excepted. 



It may be noted, by the way, that if TiOa be plotted with 

 SiOa, the curve of the sums of the two becomes rather more flat 

 (especially about IV), and the inflexion at the left is almost 

 overcome. For facilitating this observation I have put the 

 molecular ratios of TiOg next to those of SiO^ in the table, 

 though it would complicate the diagram unnecessarily to put in 

 this joint curve. This seems to confirm the general belief that 

 TiOg plays the part of an acid radical, like SiOg, in rock magmas. 



The general results can be concisely shown by plotting the 

 sums respectively of the "ascending" and the "descending" 

 oxides, except silica. We then get the two curves F (that of 

 AI2O3, Na^O, KgO) and ^(FeO, MgO, CaO), on the line of 

 which all the determining points fall very exactly. They are 

 both decidedly concave, /^ascending and v^descending, and their 

 smoothness and regularity are very striking. 



These results are in strong harmony with those of Pirsson, 

 and the general characters of the curves in each diagram are very 

 similar, though there are some differences in detail. Thus all 

 his curves are much flatter, the SiOg does not show signs of 

 inflexion, and the AlgOg is concave and KgO convex, while in 

 mine these two are reversed. But these are small matters, pos- 

 sibly due to the abscissal distances at Yogo Peak not being as 



' The terms convex and concave will be understood as referring always to the X 

 axis, the bottom of the diagram. 



