vom Rath10 gives the values 2.512 and 2.592 as the densities of the glassy crusts of Vesuvian bombs of 1872, and Lagorio11 2.319 as that of a Vesuvius obsidian. We may therefore provisionally accept a density of about 2.5 for the Vesuvius glass. As to the liquid lava, we have no data; but, assuming a density of 2.8 for the solid lava, we may conclude from the discussion of Daly12 based on Mellard Reade's and Barus' data as to expansion, that the molten lava would have a density of about 2.35. This does not take into account the presence of dissolved gases, which would unquestionably lower the density, and at the same time decrease the viscosity, very materially. It is clearly evident, therefore, that augite crystals, placed in such a leucite tephrite magma, would be much more dense than the magma, and would tend to sink, though the actual sinking of many of them would be more or less prevented by movements in the liquid, and by the possible presence of attached gas bubbles in the upper portions of the mass. Anyone who has made mineral separations with heavy liquids will appreciate the possibilities of disturbance of a "theoretically" perfect separation, adherent particles of the lighter minerals here replacing the gas bubbles of the magma.

But the occurrence of masses of rock of granitic texture, without euhedral crystals, or crystals formed freely in the body of the liquid, of the same composition as such crystals formed in what must have been a very similar magma and at the same volcano, seems to demand the recognition of some other factor than gravity, or at least one in addition to that of gravity.

This is not the place to enter into a discussion of the various kinds or causes of differentiation that have been suggested, but I must recall the case of Shonkin Sag and the explanation of its differentiation by fractional crystallization advanced by the late Prof. Pirsson, which, it seems to me, Daly has not adequately met by an appeal to gravitative differentiation. Pirsson and I held much the same views on these matters, and I feel inclined to revive a theory put forward many years ago,1 chiefly 10 Vom Rath, Z. deutsch geol. Ges., 25, 240, 1873. "Lagorio, Tscherm. Min. petr. Mitth., 8, 475, 1887.

1 R. A. Daly, this Journal, 15, 277, 1903.

13 L. V. Pirsson, this Journal, 11, 12, 1901; U. S. Geol. Surv., Bull. 237, 188, 1905. Cf. Daly, Igneous Rocks and their Origin, 223, 238, 1914.

14

Washington, Bull. Geol. Soc. Amer., 11, 409, 414, 1900; Jour. Geol., 9. 663, 1901.

to account for the different types of laccolithic differentiation. This is based on fractional crystallization, perhaps aided by convection currents, as Pirsson supposed, the crystallization beginning at the rough walls, and the crystals of this portion (in the present case monomineralic) interfering with each other so as to produce a granitoid textured rock. Crystallization of free floating crystals (therefore euhedral) in the magma could, and probably would, also go on simultaneously. The process is analogous to the slow freezing of a bottle of salt solution, which begins at the walls, so that clear ice forms above, at the sides, and at the bottom, leaving finally a central core of highly concentrated solution. With the more complex rock magmas the process would be conceivably more complex than this, but the same general principles would seem to apply. Unquestionably, the influence of gravity might or would be felt, especially on the loose floating crystals, but this would probably have less or no effect on the wall accumulations. The process is analogous to Daly's chilled border concept, but differs from this in that, according to Pirsson's and my hypothesis, the border crystallization product does not represent the original magma, as conceived by Daly, but would be an "extreme pole of differentation."

On such a hypothesis the formation, either simultaneous or successive, of a granular pyroxenite, composed of closely packed and adherent, anhedral crystals, and free-floating, euhedral crystals, is readily understandable; more readily thus, it seems to me, than on a hypothesis based purely on the influence of gravity. It also serves to explain, as that of gravity does not, such examples of vertical, tubular differentiation as those of Magnet Cove and Mount Johnston, Quebec, which are impliedly regarded by Suess1 as analogous to "piping" in a steel ingot.

But we are getting far away from our little augite crystals. Let us pass on to those of a near-by volcano, Etna.

AUGITE OF MONTI ROSSI, ETNA.

The loose crystals of augite that are found in abundance in the ashes and tuffs of Monti Rossi, formed by the eruption of 1669, and elsewhere around Etna, would

15

Suess, The Face of the Earth, 4, 559, 1909. Cf. H. S. Washington, Jour. Geol., 9, 607, 1901; F. D. Adams, Jour. Geol., 11, 254, 1903.

seem to have been among the first augites to be studied. They were described as early as 1783 by Romé de L'Isle, a few years later by Dolomieu (1788), and by Spallanzani about 1792. It may be of historic interest to cite here. Spallanzani's analysis, which seems to have been the first, or among the first, of the analyses of these augites.16 He found: "free silica 34.5, lime 18.7, iron 7.6, alumina 12.4, magnesia 11.0, sum 84.2" It will be seen that, imperfect as the analysis is from the modern standpoint, Spallanzani determined the presence of all the most essential constituents, and approximately in their relative order of abundance.

Physical characters.-The augite crystals examined were obtained in July, 1914, by Dr. Day and me in the ashes of the western summit of Monti Rossi, near Nicolosi. Though they do not appear to be as abundant as they were in Spallanzani's day, yet a handful was readily collected in half an hour.

The habit is the usual one, like that of the crystals of Vesuvius and Stromboli, though they are, on the whole, somewhat smaller, and with a decidedly greater tendency to prismatic development, some of them being three times as long as thick. They are bounded by the planes a(100), b(010), m(110), and s(111). The ordinary contact twins (twinning plane a(100)) seem to be rarer than at Vesuvius or Stromboli. They are jet black, and the faces are lustrous, much brighter than those of the Vesuvius crystals, though close examination shows that they are not flat planes, but are as if the crystals were cracked, so that they do not give good reflections for the goniometer. For this reason no crystallographic measurements were made.

In powder or particles under the microscope they are of a greenish gray color, without pronounced pleochroism. The refractive index B varied from 1.710 to 1.715; the highest value of y was 1.735, and the lowest of a was 1.702. Thus the chemical analysis is probably very closely represented by the values: a=1.704, B-1.711, y=1.732. The extinction angle is high, but was not determined, as the cleavage is poor and it was not thought worth while to grind a section parallel to b(010).

16 Spallanzani, Viaggi alle due Sicilie, (1) Chap. 7, (page 172 of Milan edition, 1825). In a (somewhat pathetic) note referring to the low summation he says: "It must be noted that, apart from the almost unavoidable loss in manipulation, and that of the moisture present in the schorls, the lime is here deprived of the acid with which it was originally provided (combined)."

The specific gravity of the crystal fragments used for the analysis, determined with a pycnometer, was 3.373 referred to water at 22°.

Chemical composition.-The loose crystals appear to be very pure, except for patches of a siliceous material (which was readily removed by dilute hydrofluoric acid). Small grains of yellow olivine project from the surfaces, but they are not found in the interior of the crystals. After crushing a batch of crystals to small fragments for analysis, all these olivine particles were carefully removed by repeated search under a binocular, and the material used for analysis is confidently believed to have been free from them. The crystals, however, contain a small amount of minute inclusions of magnetite, which it was impracticable to remove entirely. Treatment with a magnet of about 0.6 g. of the powder analyzed showed that this was present to the extent of 3.94 per cent, and in column 2 of Table II the analysis is corrected for 4 per cent of magnetite.

1. Augite, Monti Rossi, Etna. Washington analyst.

2. Same, corrected for 4 per cent of magnetite.

3. Augite, Monti Rossi, Etna. S. van Waltershausen analyst. Der Etna,

2, 490, 1880. (First published in 1853).

4. Augite, Monti Rossi, Etna. Rammelsberg analyst. Pogg. Ann., 103, 436, 1858.

Only five or six published analyses of these augites are to be found, and all suffer from the same defects that were pointed out in the case of the Vesuvian augite, that is, non-determination of titanium and alkalies, and, in many, non-separation of the iron oxides.

Discussion. Any extended discussion of my analysis is unnecessary here, and will be reserved for a future

occasion, in connection with those of other Italian augites. It may, however, be of interest to give the composition of the Etna augite in terms of the usual molecules, which is as follows (No. 1):

It will be seen that the Etna augite (1) is composed very largely of diopside and hedenbergite molecules, with a little wollastonite and acmite, and a small amount of the aluminous Tschermak molecule. In general it much resembles the augite (2) of Stromboli,17 though this carries somewhat less of the diopside molecule, and considerable hypersthene instead of wollastonite. The respective refractive indices are as follows:

a

Augite, Etna 1.704, 6-1.711, y=1.735 Augite, Stromboli, a 1.693, B-1.699, y=1.719 This is not the place to discuss the differences, but it may be mentioned that the higher indices of the Etna augite are to be connected with its higher wollastonite and titanium content, which seem to more than counterbalance the higher ferrous oxide of the Stromboli augite. The Stromboli augite, furthermore, is slightly higher in magnesia, which would tend to lower the refractive index. On the whole, it may be said that the chemical and optical data for both of these augites are quite in accord, and bear out observations made on the pyroxene molecules generally.

"Kozu and Washington, this Journal, 45, 468, 1918.