98 THE ROMAN COMAGMATIC REGION. 
The two analyses I and II are very closely alike, especially in silica, ferrous 
oxide, and potash. The alumina in I is the higher by 2 per cent, and as this is 
accompanied by i per cent less of magnesia, the suspicion may present itself that 
some of the magnesia has been thrown down with the alumina. This analysis, 
however, is the later of the two, and especial attention was paid at the time to this 
very point. Furthermore, it will be observed that the general character of I in other 
respects is more salic than II, as shown by the slightly higher soda and lower 
ferric oxide, lime, titanium, and phosphoric oxide; and these chemical differences 
are in harmony with the rocks themselves. The Cotognola rock (I) is very notably 
lighter in color, and the sections show somewhat less alferric and femic minerals 
than that from Madonna del Riposo (II). 
The analyses in themselves call for no special comment. Silica is lower than 
in any rock so far described here, and while potash is high, it is distinctly less so 
than in any previously described types of vicose. The magmatic positions shown 
by the norm are in most cases well within the limits, except the ordinal position of 
II, which approaches rather near that of the lenfelic order, but neither rock can be 
justly regarded as transitional. 
On the basis of the analysis shown in III the rock from Madonna del Riposo 
has been assigned* to the subrang braccianose, II. 7. 2. 2. This was correct as 
based on the analytical data then available, but the new analysis shows that the true 
position is rather in vicose, II. 6. 2. 2, though, as we have seen above, somewhat 
near the border of the division to which it was first assigned. The change is brought 
about by several factors, each of slight moment in itself, but all acting in the same 
direction, the increase in the figures for silica and magnesia, the decrease in alumina 
and lime, and the additional determinations of titanium and phosphoric oxides. 
Mode. The mode of the Cotognola rock was calculated from the norm by 
readjustment of the mineral molecules, in the usual way. It was somewhat difficult 
to distinguish the colorless groundmass minerals with sufficient accuracy, so that 
the Rosiwal method was not applied, though a rough estimate of the leucite present 
was undertaken to use in the calculation. This amounted to 40 per cent by volume. 
As the Riposo rock is not holocrystaUine, containing about 5 per cent of glass, its 
mode was not calculated, but it will resemble the other closely. 
Orthoclase, OrjAbj 17.5 
Labradorite, Ab z An 2 16.8 
Leucite 37.1 
Nephelite 4.6 
Augite 18.6 
Magnetite 4.4 
Apatite i . o 
IOO.O 
This mode shows the distinctly more alferric character of the type, indicated in 
the analysis by the lower silica and the higher bivalent metal oxides. The amounts 
* Washington, Prof. Paper U. S. Geol. Surv. No. 14, 1003, p. 305. 
