of the peroxide should be approximately measured by titration with permanganate on opening a fresh bottle, and again after a few weeks, otherwise very serious error may arise through its deterioration.

Mere traces of hydrofluoric acid, either in the peroxide or the titanium solution, render this method inexact,' hence care should be exercised as to the character of the peroxide, which, as sold in the market, often contains fluorine.

Dunnington has pointed out the necessity for the presence of at least 5 per cent of sulphuric acid in solutions which are to be thus tested for titanium, in order, as he concludes, to prevent partial rever sion to metatitanic acid, which does not give a color with hydrogen peroxide. The standard solution of titanium sulphate, holding conveniently about 1 centigram Ti, in 10 cm.3, equivalent to 1 per cent of TiO, in 1 gram of rock, contains, therefore, 5 per cent or more of sulphuric acid. Of this, 10 cm. are mixed with a sufficiency of hydrogen peroxide (2 cm.3 of most commercial brands is ample) and diluted to 100 cm. in a measuring flask.

Titanium can be estimated, as a rule, most conveniently in the sclution which has served for the titration of total iron (p. 57). This, having evaporated, if necessary, to less than 100 cm.", is to be fully oxidized with hydrogen peroxide, and if the color is less intense than that of the standard, is made up to 100 cm.3 with dilute sulphuric acid in a measuring flask, and mixed; otherwise, in a flask of sufficient size to insure that its color shall be less intense. One of the rectangular glasses described below being filled with the solution to be tested, 10 cm.3 of the diluted standard are run into the other from a burette, and water is added from a second burette until there is no distinction as to color. A second and a third portion of the standard can be run in and diluted and the mean of several determinations struck, when a simple calculation gives the percentage of TiO, in the rock.

If the convenient but expensive Soleil-Duboscq colorimeter is used, or the simple Nessler tubes, it is of course unnecessary to dilute the rock solution to the extent above required, should it be stronger than the standard. Experience has shown, however, that differences can not be sharply estimated in strongly colored solutions, and that the results are much more satisfactory when the color intensity is not much, if any, greater than that given by a standard of the above concentration. For the percentages of titanium found in rocks, clays, and soils, usually under 1 per cent, but rising to 2 or even 3 per cent or more occasionally, the colorimeter method gives results which are fully equal to those of the best gravimetric method, besides being a great time

1 Hillebrand: Jour. Am. Chem. Soc., Vol. XVII, p. 718, 1895; Chemical News, Vol. LXXII, p. 158, 1996; Bull. U. S. Geol. Survey No. 167, p. 56.

Jour. Am. Chem. Soc., Vol. XIII, p, 210, 1891.

saver. The error introduced by iron, in consequence of the yellowish color of its sulphate solution, is practically negligible unless its percentage is very high; then either the iron must be removed prior to making the color test, or correction should be applied for known amounts of ferric sulphate in solutions of the requisite dilution.

3

The exact correction to be applied in such cases is difficult of determination because of the impossibility of matching the colors of titanium peroxide solutions with those of ferric sulphate; but tests made go to show that the coloring effect of 0.1 gram of Fe,O, in 100 cm.3 5 per cent sulphuric-acid solution is about equal to 0.2 milligram of TiO, in 100 cm.3 when oxidized by hydrogen peroxide. This amounts to a correction of only 0.02 per cent on 1 gram of rock containing the unusual amount of 10 per cent Fe,O,. It will be more satisfactory, when much iron is present, to remove this as described on page 71 and to colorimetrically estimate the titanium thus freed from iron.

ALTERNATIVE MODE OF PREPARING THE TEST SOLUTION.

As said above (pp. 57 and 68), the solution that has been used for volumetric estimation of total iron can most conveniently be used for the colorimetric determination of titanium, but if desired this can, of course, be made on some other portion of rock powder. At one time it was the practice in this laboratory to combine it with the determination of barium, as described in Bulletin 148 of the United States Geological Survey, by decomposing the powder by sulphuric and hydrofluoric acids, expelling the latter by repeated evaporations with sulphuric acid, taking up with dilute sulphuric acid, filtering from barium sulphate, etc., and estimating the titanium colorimetrically in the filtrate. The expulsion of fluorine must be thorough, or else the titanium result will be low, as already stated (p. 68), and it is not always easy to effect this complete removal, though the time required to do so seems to be in no slight degree dependent on the nature of the fluorides to be decomposed. Long after every trace of fluorine seems to be gone, the formation of a crust on the evaporating solution sometimes allows an accumulation of enough hydrofluorie-acid gas to become plainly manifest to the smell on breaking the crust.

1 It is to be borne in mind that evaporation with hydrofluoric acid alone results in loss of titanium by volatilization, but that there is no loss if excess of sulphuric acid is also present.

With acid rocks solution is very complete, and it can be made nearly so with the most basic by transference to a small beaker and gentle boiling. The residue thus obtained may contain, besides barium sulphate, a little calcium sulphate, zircon, andalusite, topaz, and possibly a trace of titanium in some form. It is therefore to be thoroughly fused with sodium carbonate, leached with water, fused with potassium bisulphate, dissolved in dilute sulphuric acid, filtered, and the filtrate added to the main one. The insoluble matter will now be chiefly barium sulphate, for the further treatment of which see page 74.

THE COLORIMETRIC APPARATUS AND ITS USE.

The glasses G (fig. 12) may be of square or rectangular section, 8 to 12 cm. high and 3 to 34 cm. inside measurement between those sides through which the liquid is to be observed. These sides should, of course, be exactly parallel; the others need not be, but should be blackened externally. In order to further exclude the effect of side

FIG. 12.-Apparatus for colorimetric determinations, in different aspects. G, one of two glasses of square or rectangular section, 8 to 12 cm. high and 3 to 34 cm. inside measurement between those sides through which the liquid is to be observed. The other sides are blackened on the outside. B, rectangular box about 35 cm. long and 12 cm. square, stained black inside and out, one end closed by a ground-glass window, W, the other open, and a portion of the top removed. P, blackened partition, with openings corresponding to the interior dimensions of the glasses when in position. S. blackened cardboard shutter sliding stiffly up and down between partition and glasses, so as to shut off all light above the lowest surface of the liquid in the glasses.

light in this and other similar methods (chromium, for instance, p. 80), it is very convenient to have a simple light box (B, fig. 12) that can be easily held in one hand, about 35 cm. long and 12 cm. square, stained black inside and out and with one end closed by a piece of ground glass, W, the other open. For a space equal to the width of the glasses the

1 The allowable error in distance between the corresponding pairs of sides of the two glasses should not in any case exceed 1 per cent. Unfortunately there seems to be a disinclination or inability on the part of dealers in this country to furnish glasses fulfilling this requirement, and held together by a durable cement which shall be proof against dilute sulphuric acid. Canada balsam answers well for a time, but sooner or later it cracks, leaks then appear, and the sides soon drop off. It is, however, but a simple matter to cement them on again.

A pair of entirely satisfactory glasses can be made from a couple of square or rectangular 3 to 4 ounce bottles by cutting off one pair of sides from cach and grinding down till the calipers show that agreement is perfect. The tops are then to be sawed off and pieces of plate glass cemented on the

cover is removed at the top next the glass end to permit of the insertion of the glasses side by side in such a way that no light shall penetrate around their sides or between them. Immediately back of the glasses is a partition P, with openings of appropriate size cut in it. A stiffly sliding black cardboard shutter S is movable up and down immediately back of the partition, so that all light can be cut off except that which comes through the liquid.

Precautions of this kind are necessary if accurate results are to be counted on. Except for mere traces, this combination of glasses and darkened box insures greater accuracy and rapidity of work than Nessler tubes, and is preferable likewise, so far as the writer's experience goes, to expensive instruments like the colorimeter of Soleil-Duboscq, etc. In making the color comparisons the box is best held close to a window, so as to get a full, strong light. Daylight is far preferable to artificial light.

GOOCH'S GRAVIMETRIC METHOD.

When titanium is present in excess of 4 to 5 per cent and whenever for any reason it is desired to employ a gravimetric method, among the few that have been thoroughly tested that of Dr. Gooch' is unequaled. With one or two minor modifications introduced by Dr. T. M. Chatard, it is as follows:

Any solution of the rock freed from silica can be used, and the first step is to remove the iron. This is best done, after adding tartaric acid and reducing the iron by means of hydrogen sulphide to the ferrous condition, by rendering the solution ammoniacal and introducing more hydrogen sulphide. If the iron is not thus reduced before precipitation, titanium will be in part thrown down also. The amount of tartaric acid is to be gauged according to the combined weights of the oxides to be held by it in solution, and three times this weight is ample. After removing the iron sulphide by filtration--little washing suffices, because of the relatively small amount of titanium commonly present-the tartaric acid is destroyed as follows:

Potassium permanganate to the extent of two and one-half times the weight of the tartaric acid used is made into a strong solution, and to the ammoniacal filtrate from the iron sulphide enough sulphuric acid is introduced to leave some excess after all the permanganate has been reduced. After expulsion of hydrogen sulphide by boiling, the manganate is added gradually to the hot solution contained in a large beaker or flask. A vigorous reaction ensues. When a permanent brown precipitate of manganic hydrate appears the tartaric acid has

per

Proc. Am. Acad. Arts Sci., n.s., Vol. XII, p. 435; Bull, U. S. Geol. Survey No. 27, p. 16, 1886, Chemical News, Vol. LII, pp. 55 and 68, 1885.

Am. Chem. Jour., Vol. XIII, p. 106, 1891, Bull. U.S. Geol. Survey No. 78, p. 87, Chemical News Vol. LXIII, p. 267, 1891.

8 Cathrein: Zeitschr. fur Kryst., Vol. VI, p. 243, 1882; Vol. VII, p. 250, 1883.

been fully broken up, and the precipitated manganese is to be redis solved by a few drops of ammonium bisulphite or of sulphurous acid solution.

Ammonia is then added in slight excess, followed at once by acetie acid in considerable excess, and the boiling is continued for a few minutes. Thereby the titanium is freed from most of the alumina, and from lime and magnesia if they had not been earlier removed, also from most of the manganese introduced. The precipitate is filtered and washed with water containing acetic and sulphurous acids, then ignited, fused thoroughly with sodium carbonate, and leached with water to remove phosphoric acid and most of the remaining alumina. The residue is again ignited and fused with sodium carbonate. To the cooled melt in the crucible strong sulphuric acid is to be added. wherein it dissolves readily by aid of gentle heat. This solution is to be poured into a small volume of cold water and the platinum it contains precipitated by hydrogen sulphide at or near boiling temperature. After filtering and cooling, ammonia is added till the titanium is just precipitated, and a measured volume, containing a known weight of absolute sulphuric acid, is then added-just enough to redissolve the precipitate. The solution is then made up with acetic acid in such amount that the final bulk shall contain from 7 to 11 per cent of absolute acid, and then enough solid sodium acetate is stirred in to more than take up the sulphuric acid introduced. Upon rapidly bringing the liquid to ebullition the titanium is precipitated in flocculent and easily filterable condition, and the precipitation is complete after a minute's boiling, provided all the prescribed conditions have been followed and zirconium is absent.

The precipitate is washed first with acetic acid of 7 per cent strength and then with hot water. After 15 to 20 minutes' ignition over a good burner it is in condition for weighing and will lose no more weight over the blast lamp. For large amounts of titanium a repetition of the sodium carbonate fusion, etc., should be made. The actual carrying out of all these operations, when once the method is understood, requires much less time than the detailed description would indicate.

GOOCH'S METHOD NOT DIRECTLY APPLICABLE TO ROCKS CONTAINING ZIRCONIUM.

Prior to the adoption of the colorimetric method, Dr. Gooch's was invariably used in this laboratory. Occasional inability to secure clean and complete precipitation by it was experienced, especially with a certain series of rocks rather poor in titanium. Long research showed the difficulty to be due to the presence of zirconium, which acts as a marked preventive of the precipitation of titanium by boiling in an acetic acid solution under the conditions of the Gooch method.

The above rocks were found to contain up to 0.2 per cent of ZrO., and this amount was able to prevent precipitation of 0.3 per cent of