The acid gneisses are divisible into three well-marked groups:—

(a) Biotite and sillimanite-gneiss;

(b) Garnetiferous leptynite;

(c) Granulitic microcline-gneiss.

The first includes some remarkable accidental rocks (andalusite- and sillimanite-bearing exceptional rocks).

In the second group are varieties sometimes containing pyroxene, and marking a passage into the more basic rocks; they recall, in a general way, the leptynites of the central plateau of France, but with a larger proportion of garnet.

Finally, the third group includes, very probably, granulites (granulites) that a mere laboratory examination does not suffice to distinguish with certainty from granulitic gneiss.

Whilst the rocks of the two first groups are especially abundant at Colombo, those of the [third] are particularly frequent at Kandy, and only occur accidentally in the neighbourhood of Salem (probably in the form of granulites).

These rocks are pierced by veins of coarsely crystalline pegmatite, so frequently found in all gneissose regions. By the great development of minerals, and the nature of the constituent felspar, they recall the pegmatites of the south-east coast of Norway. The basic gneissose series is much more varied. It includes

Pyroxenic and hornblendic gneisses (a).

Pyroxenic and hornblendic gneisses (b).
Anorthite-gneiss.
Cipolins.

The groups a and b differ from each other in the nature of their pyroxene, and b admits of a four-fold division in accordance with the variety of hornblende which accompanies or replaces the pyroxene.

In the same way the anorthite-bearing rocks can be divided into several sub-groups founded on the presence or the predominance of scapolite, garnet, pyroxene, or hornblende.

Some rocks of exceptional types are closely related to them, such as that composed of wollastonite, wernerite, and garnet.

Looked at stratigraphically, it is easy to see the analogy between all the acid gneisses and those of the series distinguished by the letter 1 in the geological map of France. The pyroxenic and hornblendic gneisses a should be included in the same series.

In the group b certain types are closely connected with those of the first, whilst most of them are higher in position. Their age is apparently that of the pyroxenic gneiss of Brittany, the position of which, in the upper part of the gneissose series, is well established.

The rocks described further on, having anorthite, pyroxene, and scapolite as constituents, are closely equivalent to the scapolite-bearing pyroxenic gneisses of Loire-Inférieure and of Waldviertel. The same may be said of the cipolins which are associated with them.

Above these gneisses are found mica-schists of variable composition, closely comparable to the French and Alpine mica-schists of 2; as the chloritic and sericiteschists are analogous to those of x in the same region.

K

The succession of these different rocks may therefore be represented as in the

following table, which indicates, (at the same time, the order adopted in describing

These gneisses are identical in Ceylon and in Salem. They are rich in black mica, and possess a freely rubanée1 structure.

The black mica, garnet, and oligoclase form thin layers, separated by seams of white translucent quartz, often mixed with oligoclase, and sometimes attaining several centimetres in thickness. In addition to these constituents, sillimanite frequently (Colombo) occurs in the micaceous layers. The crystals of sillimanite lie parallel to the foliation of the rock. These gneisses are, at times, much crumpled. In some cases the felspar is very deficient, and the rock is composed almost exclusively of black mica, garnet, and sillimanite.

Microscopical examination discloses the presence of the following minerals: apatite, magnetite, zircon, sillimanite, biotite, almandine garnet, oligoclase, [orthoclase], quartz, and occasionally hornblende [and graphite] (Salem).

The apatite, zircon, and magnetite only play a subordinate part The sillimanite occurs in crystals elongated in the direction of the zone oo P∞ ∞ Poo (100), (010) with extremely numerous cleavage-fractures parallel to oP (001). The crystals are always distinct, and sometimes attain a length of one centimetre, with a breadth of a quarter of a millimetre.

They do not form a tangle of little needles, as in the granulitic gneisses of the central plateau of France, and of most gneissose regions. They possess all the characteristic properties of sillimanite.

The garnet belongs to the almandine variety; currant-red macroscopically, it is but faintly tinged with bright pink in thin sections.

The biotite occurs in scales, without definite form, about 2 millimetres in diameter. It is sensibly uniaxial. Pleochroic aureoles around inclusions of zircon are not uncommon. The pleochroism is most intense, with

c and b, dark black-brown.

a, bright yellow.

Often the absorption is so great, parallel to c and b, that the plate is quite opaque to the rays vibrating in those directions.

"non-schistose rocks in whichsome of the constituents are disposed

The felspar is oligoclase in large hemitropic lamella on the albite- and periclinetypes. It is very pure, free from inclusions of mica, but often filled with granules of quartz de corrosion1 less than the th of a millimetre in size and distributed in great numbers through one and the same crystal.

The sillimanite and biotite are intimately associated; [graphite, in hexagonal spangles, accompanies the mica]. The first of these two minerals has crystallized between the leaves of the second in such a way that on fractures perpendicular to the foliation of the rock the crystals of sillimanite may be seen to alternate with the scales of mica.

It was during the crystallization of these two minerals, and before it had finished, that the garnet was formed. In fact, numerous patches (plages) of sillimanite and of biotite are noticeable completely, or partially, enclosed in the garnet, and, on the other hand, there are crystals of garnet evidently formed prior to the biotite, which has been moulded on them (qui les a moulés)3.

Facts of a similar kind will be often described in the course of the present memoir. Whilst in the eruptive rocks most of the constituent minerals have a relatively fixed period of crystallization, and the order of succession is sensibly the same in a given rock, in the gneissose series under consideration, we shall see continual oscillations in the order of crystallization of the minerals, and that, not only in the different types, but also in one and the same rock.

The dark green hornblende is only found accidentally and in very small quantity; there is never a passage into hornblendic gneiss, such as occurs amongst the gneisses described further on.

After the crystallization of the felspar that of the quartz took place, which is moulded on all the other constituents of the rock in the same way as the quartz in granite.

When the rock is altered, the felspar becomes charged with damourite, and the mica changes to chlorite.

Exceptional Rocks.

Amongst the accidental mineralogical varieties that the gneisses of Ceylon, or of Salem, present, a special place should be given to the andalusite- and sillimanitebearing rocks, and to those containing sillimanite and corundum, which are represented by large specimens in the collections of the Collège de France and of the Museum.

' [By quartz de corrosion French petrographers mean the quartz that is met with in divers minerals (and especially in the felspars) in the form of tears with curved and vermicular outlines, slightly resembling the quartz of graphic pegmatites, but without the regularity of the latter. The name of quarts de corrosion has been given to it on account of its probable secondary origin The minerals that enclose it are believed to have been corroded by the solutions that have deposited the quartz.] a continuous portion of mineral in a rock, having uniform optical properties over a certain area."]

2

[Plage

=

$ [F'emplois 'mouler' dans le sens d'entourer, le minéral moulé étant antérieur au minéral moulant.] The equivalent of mouler in this sense is "to be moulded on or around."F. R. M.

The first comes from King's Fort, about 50 miles (20 lieues) west1 from Colombo, as well as from the neighbourhood of Salem.

With the naked eye one may recognize flesh-red andalusite, in long striated prisms amongst which felspar is developed, quartz, sillimanite, and damourite. In other instances sillimanite predominates, owing to which the rock is extremely tough; there are specimens of this series, formed exclusively of sillimanite, which have served for the description of fibrolite given for the first time by de Bournon2.

When microscopic sections of these specimens are examined, one immediately remarks the curious association of andalusite and sillimanite that I have recently described, and which has been since noticed by MM. Michel-Lévy and Termier1; an association in which the two minerals have their crystallographic axes parallel, and in which they are easily distinguished by their double refraction and the different positions of their principal indices

Fig. 1. Association, with parallel axes, of andalusite and sillimanite (Ceylon).5

There are, in addition, groupings of sillimanite at angles of 60° and 90°. Figure 46 illustrates this association.

The andalusite is intensely pleochroic, with

;}greenish white, almost colourless.

a, bright flesh-red.

This pleochroism is variable in intensity, and sometimes only exists in the state of faculæ. The cleavages oo P. ∞ P. (110) (110) are well marked.

1 Sic; probably a slip of the pen for east. It may be remarked here that the localities, bearings and distances, given by M. Lacroix, are those on the labels attached to the specimens.-F. R. M.

2 Description of the corundum stone and its varieties.-Phil. Trans., 1802,p. 289.

3 Bull. Soc. Minér. XI (1888), 150 and XII (1889), 59.

4 Idem, XII (1889), 56.

5 In the above illustration ng = C, Nm =

As previously remarked, the sillimanite is associated with andalusite: in some cases it forms crystals measuring 0 30 millimetres and well adapted for optical examination; in others little needles form closely felted aggregates, with irregular extinction; these are found in groups (bandes) or flakes (flammèches) both in the andalusite and in the felspar. [Some specimens that I have examined include corundum and rutile. The latter forms rounded grains, or crystals greatly elongated in the direction of the vertical axes, and orientated in the sillimanite and andalusite in such a way that the vertical axes of the two minerals are parallel.]

From one specimen I obtained a fragment, about 3 centimetres long by 2 broad, composed of crystals, measuring 3 or 4 millimetres, lying side by side in the direction ∞ P∞ (100), with a degree of parallelism sufficient to give a section o P. (001) homogeneous from an optical point of view; this has served for measuring the principal indices by the method of total reflexion (sodium light).

The plane of the optic axes is parallel to oo P oo (100): the bisectrix is positive c and normal to o P. (001)

This sillimanite encloses only a little biotite and corundum. A quantity of pure material, sufficient for an analysis (a), was obtained by treatment with boiling hydrochloric acid, hydrofluoric acid, and liquids of high specific gravity, successively. The analysis of sillimanite (fibrolite) from the same rocks, executed by Chenevix, is given in column (b)

The white mica is a damourite, the divergence between the axes of which is about:

It occurs in lamine on the crystals of andalusite, following the asperities and depressions of the latter; further, thin sections of the rock show them pierced with holes like a skimmer (écumoir). Frequently the white mica is developed around the jagged patches of biotite, the crystallographic orientation being common to the two micas.

The most interesting fact to notice with reference to this rock is the mutual relation of the minerals which compose it.

The andalusite, associated with sillimanite, occurs in large crystals often eaten away, corroded, and surrounded by a great number of little fragments orientated upon the principal patch: they seem to indicate corrosion in situ [or an impeded crystallization]. In other cases it is riddled with holes filled with quartz.