4

The felspar, which is oligoclase, has crystallized in large patches in the spaces left by the silicates of alumina: the quartz plays a similar part, occurring in rounded grains, generally of small size.

The sillimanite in capillary aggregates is also met with in the felspar; nontwinned patches much resembling cordierite are of frequent occurrence: although · such an association is extremely probable, I have not evidence sufficient to clearly establish the presence of this mineral.

Sometimes the andalusite is embedded in a crystal of felspar; their vertical axes being coincident. By lowering the condenser of the microscope, the andalusite may be clearly observed to show out more strongly refractive.

The second interesting rock is a mixture of corundum and sillimanite.

The most remarkable specimen that I have had the opportunity of examining is composed equally of violet-pink coloured, granular corundum, and of reddish yellow sillimanite; the rock possesses a high specific gravity; with the naked eye the sillimanite may be observed distributed through the corundum in crystalline tufts, converging towards a centre. The pearly cleavage ooPoo. (100) is very brilliant.

It is almost impossible at first to recognize the sillimanite. The rock, when examined in microscopic sections, shows the following structure: the corundum is made up of little rounded grains without distinct form, irregularly disseminated through large patches of sillimanite remarkable for the eminence of the cleavage (∞P∞(100). There are none of those tangles of fine needles so common in the preceding rock; the crystals are always largely developed. Often, however, moirées' extinctions indicate multiple groupings. The plane of the optic axes is parallel to coPoo.(100); the bisectrix is positive and normal to OP. (201) with:

2E = about 35°

and a strong dispersion with p > v.

[In one specimen the corundum, which is associated with large crystals of sillimanite, occurs in crystals which are flattened parallel to oP. (0001), and are rich in gaseous inclusions, and in inclusions of rutile disposed in parallelism to the base OP.(0001). The rock includes large patches of rutile surrounding the crystals of sillimanite. Further, tourmaline occurs, which is dark black-brown, and strongly dichroic in microscopic sections, and grouped with the corundum in such a way as to produce a fine ophitic structure.]

One difficulty in the way of diagnosis lies in the impossibility of obtaining sections of equal thickness; the presence of corundum in little particles, in fact, renders work on these rocks very difficult. Thanks to the well-known skill of M. Yvan Werlein, I have been enabled to obtain some good slides.

The double refraction is:

7 a = 0'020

There is, however, no doubt as to the identity of this mineral with sillimanite. About one gramme has been separated, by means of liquids of high specific gravity, from corundum, and from a small quantity of rutile which accompanies it, washed with boiling hydrochloric acid, and submitted to analysis. It is then of a violet-gray colour and transparent. The mineral is infusible before the blow-pipe, and insoluble in hydrofluoric acid.

Analysis gave practically the same results as those which have been recorded

[Since first writing this memoir, I have discovered in the Pyrenees of Ariège, in the rock-mass of St. Barthelemy, andalusite-bearing granulites (pegmatites) in thin veins piercing the gneiss. Corundum is there also associated with the andalusite.' It is probable that the exceptional rocks, just described, present themselves under similar geological conditions, instead of occurring, as I supposed at first, as accidental rocks in the gneisses themselves.]

De Bournon describes, in connection with indianite (anorthite), the occurrence of sillimanite (fibrolite) in association with corundum. It seems evident that the author confounds the two rocks (acid gneiss and anorthite gneiss) in which, as Newbold also attests, corundum is found.

The gneisses that occur [with] the sillimanite-gneiss present great difficulties in their delimitation. They are very abundant both in Ceylon and Salem. In the latter, however, more basic rocks (pyroxenic and hornblendic gneisses, anorthite bearing rocks) predominate.

Granulites composed of large crystals (pegmatite) occur in these regions, but independently of these rocks, eruptive granulites are found which it is not possible to distinguish, in the laboratory, from the granulitic gneisses which accompany them and into which there is a gradual passage. I have met with this difficulty in more than two hundred specimens of these rocks that I have examined.

They present the most varied aspects: sometimes relatively coarsely crystallized, sometimes nearly compact, in some cases they possess a well-marked stratified structure, in others they are granular. Some are composed almost exclusively of quartz, others contain scarcely any. They may, however, be classified in two groups.

One of these is very rich in garnet, and contains a little biotite; the dominant felspar is oligoclase.

The other is poor in, or without, garnet and biotite: microcline is the principal or only felspar.

The first, which is much the more schistose, approaches towards biotite- and sillimanite-gneiss; the second towards the granulites, and it is in this group that the granulites must be included that the imperfect means at my disposal do not enable me to separate with precision.

These are the groups (b) and (c) tabulated above.

(b) Garnetiferous leptynites.

The garnetiferous leptynites are especially largely developed in Ceylon, both at Colombo and Kandy.

They are of a bright colour, white or pink, and particularly rich in currant-red garnets, varying from the size of a pea to that of the head of a pin.

Although very variable in texture, the rock is extremely uniform in composition. There are present rutile, zircon, garnet, biotite, oligoclase, orthoclase, and quartz. The felspars are often clouded (troublès) and rich in micaceous inclusions; the garnet occurs in rounded forms, hollowed with sinuous cavities filled with the white minerals of the rock.

1 Contributions à l'étude des roches métamorphiques et éruptives de l'Ariège.-Bulletin des services de la carte géologique de la France, No. 11, C. II, 17, 1890.

2 Op. cit., p. 289.

3 Jour. Roy. As. Soc., VIII, 153.

The biotite, in little elongated laminæ, is very irregularly distributed, and is frequently wanting.

Some varieties of the rock are nearly compact, others, on the contrary, are somewhat largely crystallized. The schistose structure is marked by the orientation of the laminæ of mica and that of the garnet.

The structure is similar to that of the leptynites of the gneiss of the central plateau of France, and does not present anything special.

Pyroxenic leptynite.

At about three miles to the north of Colombo, on the cliffs bordering the sea, a variety of the preceding rock is found, which, owing to the presence of green strongly pleochroic, pyroxene, passes into the first of the pyroxenic gneisses which will be described further on.

Sometimes the rock is extremely compact, with a greenish-yellow colour, and slightly resinous appearance (it is then formed exclusively of quartz and of oligoclase, with very little garnet and pyroxene); sometimes it is largely crystallized, and composed of quartz, oligoclase, pyroxene, and felspar.

It is generally singularly free from weathering. Petrographical examination discloses, in addition to the minerals just enumerated, dark green spinel in large patches, often mixed with magnetite, as well as zircon in rounded crystals.

The pyroxene belongs to the very ferruginous and pleochroic variety which will be described some pages further on; that alluded to here, however, is more feebly doubly-refracting; the crystals are often broken, and in the fissures there is developed, after the manner of chrysotile in olivine, a greenish-yellow pleochroic mineral, with a maximum of absorption in the yellow tints parallel to c and to the elongation. The extinction is longitudinal: a = about o'025.

1, Quartz; 6, Oligoclase; 29, Magnetite with border of quartz; 52, Product due to the alteration of [cordierite] with vermicular quartz.

[Sometimes the pyroxene is completely transformed into this substance. In many specimens I have observed a yellowish secondary mineral, containing fine vermiculations of quartz, representing an elegant micro-pegmatite: although this

mineral possesses much more feeble double refraction than the alteration-product of the pyroxene, I had been led to refer it to the same substance, and it is so called in the explanation to Fig. 47. Since this memoir was written, I have had numerous microscopic sections of this rock cut, and I have satisfied myself that the yellow, but slightly doubly refractive, substance represented in the figure, is different from the pyroxene alteration product, and ought to be attributed to the decomposition of cordierite. I have found specimens in which the cordierite is absolutely unaltered. It is markedly differentiated from the felspars by all its optical properties. Inclusions of zircon are abundant in it, which are surrounded by pleochroic aureoles of a fine golden yellow. I believe that pegmatoidal associations of cordierite and quartz have not been described hitherto.]

(c) Granulitic microcline-gneiss.

This last group is more artificial than the preceding one, and probably includes some eruptive granulites that cannot be diagnosed with precision by petrographical analysis alone. The schistose structure is less accentuated than in the preceding rocks. These rocks are especially abundant in the vicinity of Kandy, and are likewise met with in the neighbourhood of Salem.

They are of a bright, often pink, colour; generally poor in oligoclase and in black mica, and very rich in orthoclase and, above all, in microcline.

The quartz is more individualized than in the preceding gneisses, and is often present in the granulitic state (à l'état granulitique).

The microcline is very abundant: it has been in general the last of the felspars to consolidate. It is of remarkable purity, being nearly always free from inclusions of albite or oligoclase.

All the varieties in the appearance of the cross-hatched twinning characteristic of this felspar, described by M. des Cloizeaux1, may be observed. The extinctionangles are about 15° on oP.(001) and about 7° on oooo. (010). ∞∞.(010).

Orthoclase sometimes occurs alone in large, bright, pink-coloured plates. An interesting fact that should be noticed with reference to these rocks, lies in the existence of much quartz de corrosion in the felspars, accompanied or replaced, by elongated inclusions, having their principal axes parallel to the vertical axis of the felspar.

In the sections ooPoo.(010) the inclusions have the form of long spindles more strongly refractive, and with greater double-refraction, than the felspar. In the sections oP.(001) these inclusions (cut transversely), are rounded, and of small diameter. It is clear, therefore, that they are not distributed in relation to determinate faces of the prismatic zone, but that they are irregularly disseminated (Fig. 3).

Frequently they assume vermicular forms, and are so abundant that in parallel polarized light it is difficult to measure the extinction-angle of their host.

[Their extinction is different from that of the felspar which encloses them. I have ascertained, beyond doubt, that their refraction (contrary to what I thought at first) is the same as that of the quartz of the rock, and they should probably be regarded as composed of quartz. They constitute, then, a particular form of the quartz de corrosion described on p. 8].

In some specimens all the felspars without exception contain them. They are

.1 Ann. Chimie et Physique 5me Sér., Vol. IX (1876).

1, Quartz; 6, Oligoclase with fusiform inclusions; 14, Sphene; 29, Magnetite surrounded by quartz.

equally to be found in the felspars of the pegmatites, and especially in the variety of orthoclase known as moonstone. [ *

Numerous crystals of felspar are filled with quartz de corrosion. The quartz is generally developed on the surface of the crystals, aud penetrates thence to the interior, where it forms fine vermiculations in connection with the quartzose peripheral zone. In certain cases these little quartzose canals increase in size, and the crystal of felspar is then formed only of débris included in a preponderating amount of quartz (noyés dans le quartz) (Fig. 4).

In other, and rarer, cases, the quartzification does not reach the centre of the crystal, and does not begin by a quartzose border.