commonly styled limestones, being in fact a dolomite; and the resemblance, except in hardness, of some of these rocks to compact grey gypsums or even quartzites makes it necessary to emphasise this caution.

It is constantly of service to examine the compact or glassy groundmass of an igneous rock for potassium, by the method described on p. 85, which has the advantage of giving roughly quantitative results.

The treatment of a rock with acid is frequently important as revealing an insoluble residue, which should always be examined further. The division, however, of every rock into a soluble and insoluble portion, prior to analysis, is now regarded as of little value, and the ordinary plan pursued is to make a thorough fusion of a weighed quantity of the powder with carbonate of potash and carbonate of soda. The powder must be obtained by breaking up little fragments of the rock still further upon an anvil. An enlarged form of the steel mortar used in blowpipe analysis (p. 40) will serve well. The material is ground and reground, a portion at a time, in a fair-sized agate mortar until the powder is practically impalpable between the fingers. Too much care cannot be given to this simple preparation of the material used in the analysis, since imperfect fusion may result if the particles are not sufficiently fine, and the silica ultimately separated will contain gritty undecomposed matter. Although the precautions and details of the methods employed must be left to chemical works and to personal practice, it may be of service to remind the reader of the successive operations performed during a simple rock-analysis, such as would suffice for ordinary determinative purposes. Naturally the list of substances that might be looked for and separately estimated in an elaborate analysis of material from the earth's crust is as long as that of the known chemical elements; but the proportions in which the below-mentioned oxides occur are often of fundamental geological importance. Unless, however, such substances as manganese, titanium, barium, &c., are separately determined, the analysis must be regarded as only approximate, and as serving for classificatory purposes rather than for refined discussion. This is clear from the detailed papers by Messrs. Clarke and Hillebrand, which should be in the hands of all who would analyse silicates (Bull. U.S. Geol. Survey, Nos. 167 and, especially, 176, price 15 c. The latter is a revised edition by Hillebrand in 1900 of No. 148, issued three years previously).

SUMMARY OF DETERMINATIVE CHEMICAL ANALYSIS OF A ROCK.

1. Loss on Ignition.-Dry the powdered rock in an air-bath at 110° C.; transfer about 1 gramme to a platinum crucible, and determine the weight of the quantity thus used. Then ignite strongly over a gas blowpipe, cool in a desiccator, and weigh again. Ignite a second time and weigh, repeating this until the weight is constant. The difference thus found is due to loss on ignition, which generally represents water. Where it is necessary to determine carbon dioxide, a sample of the powder must be decomposed by acid in an apparatus in which either the gas evolved is allowed to escape and is determined by loss, or in which it is collected in an absorption-tube by soda-lime and weighed. (See Hillebrand, op. cit., p. 101.)

2. Silica. Prepare a fusion-mixture by minutely mixing 13 parts by weight of potassium carbonate with 10 parts sodium carbonate. Add to the ignited powder in the crucible, or to a fresh sample if the heating has caused it to fuse or frit together, about four times its weight of fusion-mixture, mixing carefully and very thoroughly with a rod or platinum spatula. Fuse at first over a Bunsen-burner, the lid of the crucible being kept on, and avoiding too great heat at the outset. Then apply the blowpipe until the whole mass runs freely together and ebullition ceases. The flame should be directed obliquely, and should not envelope the whole crucible.

Remove and stand the crucible on a cool surface, such as an iron plate, so that the fused mass may crack away from the wall of the crucible. Place in a porcelain or platinum dish with hydrochloric acid and water, covering quickly with a clock-glass to avoid loss by effervescence of the carbonates. Warm, and allow to stand until decomposition is complete. Evaporate to approximate dryness in a water-bath (Clarke and Hillebrand). Moisten again with strong hydrochloric acid, add water, and warm. The silica should now float about lightly in the liquid when stirred, while all the bases are in solution. Filter off the silica; evaporate the filtrate, treat as before, and add the small quantity of silica thus obtained to that already in the filter. Ignite for about twenty minutes, and weigh. If gritty matter occurs amid the silica, the fusion has not been satisfactory, and the process must be begun again.

3. Alumina and Ferric Oxide.-Add to the filtrate a few drops of nitric acid, in order to ensure the conversion of ferrous to ferric salts. Then add ammonium chloride, and ammonia in

very slight excess, and boil. Filter off the precipitate of alumina and ferric oxide, obtaining the filtrate a. When thoroughly washed, re-dissolve the precipitate into another vessel, and divide the subsidiary filtrate thus obtained into two measured quantities. Thus it may be made up to half a litre by dilution in a marked flask, and 250 cc. may be drawn off with a pipette. In this portion precipitate alumina and ferric oxide as before; filter, ignite, and weigh. Draw off 100 cc. from the portion remaining in the flask, and determine the iron in this volumetrically by means of bichromate or permanganate of potash. Make a check-determination by drawing off another 50 or 100 cc. Divide the weight of iron found by 7, which will give the weight of ferric oxide. Deduct this from the joint oxides, the alumina being thus found by difference. Ferrous and ferric oxides must be separately determined in all exact analyses. (See especially Hillebrand, op. cit., p. 88.)

4. Lime. To the original filtrate a, which must contain ammonia in excess, add excess of ammonium oxalate. Allow to stand for 12 hours. Filter, and ignite strongly; weigh, and repeat till the weight is constant. The precipitate is thus converted into lime.

5. Magnesia.-Ammonia being in excess, add hydric disodic phosphate to the filtrate, stirring very carefully with a rod, since the precipitate clings to any parts of the beaker that may have been in the least degree abraded by touching. Stand for 12 hours and filter cold. Wash the precipitate with a mixture of 1 part ammonia and 3 water, and ignite, the filter being burnt separately in the lid of the crucible. Where a large quantity of magnesia is expected, a porcelain crucible should be used, to avoid injury to the platinum. The ignited precipitate is the pyrophosphate (Mg, P2 O). To estimate as magnesia, multiply by ·36036.

6. Potash and Soda.-These alkalies are best determined by the Lawrence-Smith method. Mix intimately 1 part of the powdered rock (about half a gramme) with one part of ammonium chloride and 8 parts of pure calcium carbonate. Heat for about an hour in a deep platinum crucible, which is best supported almost horizontally over a flat-sided Bunsen-flame, and under a conical iron shield. The flame must be applied very gradually at first to avoid rapid volatilisation of the ammonium chloride, and the temperature should at no time rise above dull redness. The decomposition is effected without complete fusion. Dissolve out the fritted mass in water in a dish and filter. The filtrate contains the metals of the alkalies in the form of chlorides, with some portion of the materials used in decomposition.

Precipitate the lime from the filtrate by ammonium carbonate; filter and evaporate down, testing the filtrate as it becomes more concentrated with a drop or two of ammonium carbonate solution. If lime is still present, precipitate it and filter again.

Evaporate to dryness in a small dish, and gently drive off by further heating the ammonium chloride and ammonium carbonate. A dark stain may appear, which is due to impurities in the ammonium carbonate, and may be neglected. Excessive heat must be avoided, lest a portion of the chlorides of the alkalimetals should be lost. Weigh the joint chlorides in the dish while the latter is still slightly warm.

Dissolve up in water, add platinic chloride, and evaporate almost to dryness on a water-bath. Add alcohol, and allow to stand for some hours, the precipitate of potassium platinic chloride being insoluble in alcohol. Filter on to a weighed filter, wash with alcohol, and dry at 100°. Weigh with the filter without ignition.

To calculate this precipitate as potash, multiply by 19272. Divide this result by 63173, which gives the weight of the potassium chloride in the joint chlorides. Deduct this from the joint weight and multiply the remainder by 53022. This gives the weight of soda.

CHAPTER XIV.

THE ISOLATION OF THE CONSTITUENTS OF ROCKS.

In the case of a coarse-grained rock, clearly composed of heterogeneous materials, it is not difficult to break out with the hammer or the pliers fragments or crystals of individual constituents, which can then be submitted to special tests. It is, however, highly desirable that a microscopic section should have been previously prepared, in order that the purity of the crystals which are to be examined, and their freedom from enclosures, may be satisfactorily ascertained. This precaution is especially necessary when chemical or microchemical tests are about to be applied. Where the selected grains are small and translucent, examination of them when mounted in water under the microscope will often assure the observer of their purity or the reverse.

Many sedimentary rocks, such as sandstones, can be broken up with the pliers or even with the fingers, and the grains spread out on paper for identification. Other rocks, such as clays, may be broken up after prolonged treatment in water, the materials of varying fineness being successively washed off into separate vessels, and an often valuable residue of larger grains, small fossils, &c., being finally left behind.

When a rock is, however, compact and coherent, its constituents can be isolated only with difficulty; and at the beginning of the present century a large number of masses were classed as homogeneous, or even as mineral species, which were in reality fine-grained rocks in which it seemed impossible to determine the constituents. To the French geologist Cordier we owe a series of researches that shed a vast amount of light on the constitution of the igneous varieties of such rocks. Since the task he set himself was so eminently one of mineral-isolation, a summary of his work may be given appropriately here, before we discuss the modern methods by which such investigations have been facilitated. For if modern petrology appears to owe little to the men of 1800, it is not because these early researches were less accurate or in any way less laborious than our own, but because they were for a time forgotten by geologists amid the excitement of paleontological discovery.

In the autumn of 1815, P. Louis Cordier read to the Academy of Sciences at Paris his Mémoire sur les substances minérales dites en masse, qui entrent dans la composition des Roches Volcaniques