de tous le âges.* Struck with the probability that the compact or more glassy matrix of volcanic rocks consisted of determinable mineral substances, he put before himself the following question:-" Are the undetermined volcanic groundmasses mechanically constituted, and, in the event of it being possible to determine their mechanical constitution, what are the mineralogical units that compose them?"

After many unsatisfactory experiments, Cordier had recourse to two modes of discrimination, the microscope and blowpipeexamination after the method practised by de Saussure. † He considered it probable that the particles produced on the breaking up of his volcanic groundmasses would belong to the common crystallised minerals that occurred in coarser specimens of such rocks. Hence he began by reducing well-known minerals to powder by pressure under a pestle, until he obtained samples in which the grains were from 05 to 01 millimetre in diameter. He then examined these under a microscope magnifying 13 or 20 diameters, rotating the object-carrier so as to view each grain in several positions with regard to the light falling on it. The clearness of the characters displayed by the various minerals came to him as a welcome surprise.

He examined in this way pyroxene, felspar, olivine, titanic iron (much magnetite was probably included by him under this term), amphibole, mica, leucite, and hæmatite. He also tested the hardness of the grains, and their fusibility on de Saussure's kyanite splinter, and observed on the same support the action of one mineral upon another, placing the two grains under examination in contact with one another and fusing them in the same flame.

In his comparisons of these types with the constituent granules of the groundmasses, he felt that local variations must be eliminated; hence he was careful to employ, as far as possible, the larger and identifiable crystals of a rock as guides in the determination of the particles of its matrix. We now know that such porphyritic crystals are apt to differ very widely from those of the second consolidation;" but even this precaution of Cordier is an illustration of the general refinement of his work.

Under the head of volcanic rocks he included the products of active volcanos, of denuded cones, and of ancient centres which had been covered by marine deposits of the remotest age. He powdered up these rocks, microcrystalline or glassy, washed the

Journal de Physique, t. lxxxiii. (1816), pp. 135, 285, and 352. Abstract by A. Brongniart in same journal, t. lxxxii., p. 261. +See p. 46 of this volume.

particles so as to procure samples of suitable fineness, and submitted them to the same tests as his typical mineral series. He extracted the opaque black granules with a bar magnet,* and referred them to his "fer titané," which contained only a very small proportion of titanium, and which corresponds in these rocks to our magnetite and titaniferous iron oxide. Finding a small portion of the opaque grains not thus attracted by the magnet, he compared them first with chromite and melanite, and finally classed them as ilmenite. His preliminary assumptions were now fully justified, and he distinguished as the constituents of his "pâtes lithoïdes" the minerals felspar, pyroxene, amphibole, mica, "fer titané," leucite, olivine, and hæmatite. He also put an end to the idea that amphibole rather than pyroxene was the dominant black silicate in basalt and in the allied darkly coloured rocks. He finally divided his lithoidal lavas into "leucostines," in which felspar predominated, and "basalts,' which fused to a black glass and in which pyroxene was

abundant.

Cordier then compared in detail the granular and often schistose rocks known as petrosilex, cornéenne, and trap, with his ancient and modern lavas, and concluded that the two groups. had nothing in common, beyond a few familiar crystallised minerals. In the former group he notes as distinctive constituents quartz, diallage, talc, chlorite, magnetite, iron pyrites, and pyrrhotine.

He next turned his attention to volcanic scoriæ, and proved their composite character with similar success, showing that the microscopic crystals in them were often embedded in a glassy matrix. His researches on truly vitreous lavas ought similarly to have gone far towards a rational treatment of such rocks; many glasses, however, have been regarded as minerals rather than mineral aggregates from the middle ages to the present day. Cordier traces admirably the passage from the compactest basalt to the black glass (tachylyte) for which he reserves the old name of "gallinace." The more felspathic glasses that fused. to a white or lightly coloured product he classes in distinction as obsidian.

After an elaborate examination of the altered matrix (wacke) of many tuffs and amygdaloidal lavas, he sums up practically as follows:

That the mineral substances styled massive, forming the

Fleuriau de Bellevue also used the magnet, and may have determined his minerals partly with the microscope. "Mémoire sur les cristaux microscopiques." Journ. de Physique, t. li. (1800), p. 442.

groundmass of volcanic rocks, are, with the rarest exceptions, composite in character.

That their constituents are microscopic crystals and glass. That the crystals belong to the common species above given. That the vitreous groundmasses probably contain material the further development of which would have produced the lithoidal types.

That in many altered volcanic rocks the materials are held together by foreign matter interposed in very minute particles.

That the sixteen types of rock which he establishes are connected one with another by a sufficiently complete series of intermediate types.

That the volcanic rocks have no analogy with those called petrosilex and trap.

That the differences alleged to exist between ancient and modern lavas are entirely superficial, and consist only in minute modifications of texture; vacuoles are thus always present between the constituents of modern lavas, while in the older examples they have become rare or completely absent through infiltration.

Although Cordier probably exaggerated the points of difference between some truly igneous "traps" and his volcanic series, yet his general conclusions were in the highest degree philosophic; and it is indeed pleasant to refer to his classic memoir as an example of what may be done in determinative geology by the union of scientific method with the simplest means. (See also p. 132.)

The crushing of crystalline rocks, with a view to the microscopic examination or isolation of their constituents, may be performed between folds of smooth cloth or even paper, to avoid the introduction of extraneous metallic or mineral material; but a hard steel anvil and hammer generally suffice.

The powder of the rock, which must be fairly coarse, is passed through sieves of various mesh, until a sample is procured, as coarse as possible, in which each grain consists of only one mineral species. For this purpose the sieves used in chemical laboratories are convenient, several fitting one above the other; the crushed mineral is placed in the topmost, which has the widest mesh, and, the whole being shaken, each sieve selects a sample increasing in fineness till we reach the lowest pan.

The objection to the use of sieves lies in the fact that some of the constituents may be much more friable than others, and hence for quantitative purposes no one sample may be satisfactory. The contents of each sieve must be examined in order to determine if any mineral has become eliminated from this cause.

The sample, when selected after examination with the lens, may be picked over by the aid of that instrument, or upon the stage of a microscope with a low power. A fine brush should be moistened with water (Dr. Sorby recommends glycerine) and brought in contact with the grain to be picked out. It is then dipped just below the surface of distilled water in a watch-glass, and the grain is at once detached and sinks.

In this way, by care and patience, a quantity of any one constituent can be accumulated, sufficient even for a chemical analysis. But for merely qualitative tests a very few grains will be sufficient, and excellent material can be quickly obtained to which microchemical reagents may be applied.

a

The removal of light material, such as clay, fine dust, &c., from heavier or coarser constituents may be performed by washing, as in an apparatus described by M. Thoulet* (fig. 12). A large tube, a, terminating in a tap below, is fitted with a rubber cork through which a finer tube, b, passes. A tube, c, opens through the side of a. The powdered material is placed in a and water is introduced through b. This rises in a and flows over at c, carrying with it, if the operation is sufficiently prolonged, all the light substances thus washed out of the material.

In separating minerals of different specific gravities, water is introduced at c and flows out up b when a has become full. This current keeps the powder well disturbed, and by regulating it none of the material escapes up b. Check the flow gradually, and the grains of different characters will descend successively, forming distinct layers at the bottom. These can be drawn off by the tap, and a fairly pure amount of any particular constituent can be collected. Plate-like minerals, such as mica, will probably appear among the upper layers. It is clear that simple forms of such an apparatus can be constructed with glass tubes, corks, rubber tubing, and a clip to act as a stop-cock.

Fig. 12.

Prof. Derby, in 1891, showed how an ordinary miner's pan will suffice to separate a good quantity of the heavier minerals from a powdered rock. Dr. G. P. Grimsley ("Granites of Cecil "Séparation mécanique." Bull. Soc. Min. de France t. ii. (1879), p. 22.

Co., Md.," Journ. Cincinnati Soc. Nat. Hist., 1894) used a pan 12 inches in diameter for his work on granite soils. The powder is well stirred with water, and the earthy portion poured off; the granular residue is stirred with more water, "by a combination of a spherical and elliptical movement of the pan." Then, by a quick side-movement, the water is thrown off, carrying the lighter minerals and the floating mica. Repeat until only a small residue is left, which contains the heavy minerals.

The various methods of decantation and separation by washing in moving water, which have been adopted for the mechanical analysis of soils, are well discussed by Wiley (Principles of Agric. Analysis, vol. i., 1894, pp. 171-247). Simple but systematic methods of decantation in ordinary beakers appear to be as reliable quantitatively as far more elaborate methods.

Cordier extracted some constituents by the use of the magnet (p. 112) after washing. Composite grains, containing only minute particles of magnetite, may be taken up, but can be picked out if the iron oxide itself is required to be pure. It is useful to cover the end of the magnet with a sliding cap of tissuepaper. This is kept in contact with the end while passing over the powdered rock, and the magnetic particles adhere to it. On withdrawing the magnet to the collecting-vessel, the cap is thrust forward and the material falls off into the vessel.

M. Fouqué * uses an electro-magnet, connected, if necessary, with six Bunsen-cells. By graduating the strength of the current, the constituents of a rock can be fairly sorted; first the magnetite, then the pyroxene, the olivine, and the felspars and allied minerals which contain traces of magnetic substances. A residue of felspars and "felspathoids" remains. The glassy matrix of igneous rocks is the most common source of error; if it is pyroxenic, it may, by inclusion in the felspar, cause the removal of a large quantity of the latter, leaving only the purer quality; but in many cases it is highly silicated and scarcely ferriferous, and cannot be separated from the felspars that are to be tested by Szabó's or other reactions. Microscopic examination, then, must decide on the suitability of such selected material for refined determinative tests.

In practice with Fouqué's method, the ends of the electromagnet may be covered with thin paper, to prevent the adhesion of non-magnetic particles to any moisture on the surface of the iron. The powder is placed on a large card and jerked about close under the poles. When a certain amount of material has * "Nouveaux procédés d'analyse médiate des roches." Mémoires présentés par diver savants à l'Acad. des Sciences, t. xxxii., No. 11. See also Minéralogie Micrographique.