300 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1909. 
silica and high in magnesia, though it has been observed by Kemp 
in gabbros, which were presumably connected genetically with peri- 
dotitic rocks. Recent developments point to a somewhat wider dis- 
tribution than was formerly thought to be the case, and indicate that 
platinum not infrequently is associated with copper ores. 
The true relations of such elements as arsenic, antimony, bis- 
muth, selenium, and tellurium to igneous magmas are quite un- 
known. It is possible that arsenic and selenium are most at home 
in the basic rocks, while antimony, bismuth, and tellurium are more 
apt to occur in siliceous ones. 
We may summarize the observations recorded above as follows: 
Of the rarer elements whose distribution is better known, lithium, 
beryllium, cerium, and yttrium, zirconium, uranium, thorium, sul- 
phur (as trioxide), fluorine, chlorine, and possibly tin occur most 
abundantly in sodic magmas; barium in potassic magmas; titanium, 
vanadium, manganese, nickel, and cobalt in iron-rich magmas; chro- 
*mium and platinum in magnesian magmas; and phosphorus (?) and 
chromium (7?) in calcic magmas. 
Of the other elements it can only be said that boron and molyb- 
denum are certainly, and zinc, cadmium, lead, antimony, bismuth, 
and tellurium are possibly, connected with magmas high in silica; 
sulphur and copper almost certainly, and arsenic and selenium pos- 
sibly, with those low in silica; while the relations of gold, silver, and 
mercury are very uncertain, but they are probaby most at home in 
acid rocks. 
This statement, it will be seen, differs from that of Vogt, in that, 
in the best-established cases, silica plays a less determinative role 
than some of the other major constituents. At the same time, the 
influence of the general law of the association of the most abundant 
oxides comes into play, and in a general way the potassic and sodic 
magmas are most apt to be highly siliceous (though the facts of dis- 
tribution are shown in them even when silica is low); while those 
which are high in iron, magnesia, and lime are most apt to be low 
in silica. 
Possibly the most striking feature of the distribution as thus shown 
is the great number of elements which are prone to occur in highly 
sodic magmas. As is well known, such magmas are those which show 
most tendency to differentiation and the formation of a great variety 
of rocks, many of them characterized by the presence of rare or other- 
wise unusual and interesting minerals, and there may probably be 
some connection between the two features of these magmas. 
It will be noted that some of these elements, as fluorine, chlorine, 
sulphur (as trioxide), and boron, are among those to which is usually 
attributed the role of so-called “ mineralizing agents,” they being 
supposed to be present as dissolved vapors in the magma and to exert 
