ammonium oxalate and thiosulphate methods. The weighed thoria was pure white, the ceria of a pale salmon color.

The condition of the uranium was shown by its precipitability as tetrafluoride on dissolving the mineral in hydrofluoric acid. This reaction afforded a ready means of estimating with exactness the ferrous iron by permanganate after rapidly filtering off the fluorides precipitated in an atmosphere of carbon dioxide. Solution of the mineral in a sealed tube with dilute sulphuric acid allowed of finding the oxygen value of both UO, and FeO by permanganate. The result thus found for OU, agreed marvelously well with that calculated from the UO, found gravimetrically in a separate portion.

Nitrogen (?) and helium (?) were obtained quantitatively by fusing the mineral with sodium-potassium carbonate in a current of carbon dioxide and collecting the gases in a nitrometer over potassium hydroxide. The volume was between one and two cubic centimeters per gram of yttrialite. The gases were not further examined.

The analysis is given above, together with that of Mackintosh for comparison.

The ratios of Mackintosh's analysis as calculated above by me are certainly wrong in so far as they are affected by the value for iron, which he assumed to be wholly ferrous. If corrected in accordance with the statement of my analysis, or, what amounts to the same thing for the purpose of illustration and is simpler, if my ratio is altered to conform to his statement for RO and R,O, bases it becomes 0644: 1882, a very close agreement.

It is altogether probable that Mackintosh's separations of the earths were not so far reaching as mine, and this belief is borne out by the differences in the experimental molecular weights. for the lanthanum and yttrium groups, mine being more in accordance with what might be expected and, moreover, agreeing almost exactly with those which were found by me for rowlandite in 1893, namely, 336-8 and 266-2 respectively.*

It is of course impossible to say what disposition should be made of the small amounts of firmly held water, phosphorus, carbon dioxide, fluorine, and alkalies. The ratios of my analysis are, therefore, to a slight extent incorrect, but probably not enough to influence any conclusions that may be drawn. One thing is apparent, that the preliminary purification by acid has had no pronounced effect on the composition of the mass acted on, otherwise Mackintosh's and my analyses should show far greater differences in the main constituents.

The crude empirical formulas deducible from the ratios of the two analyses are nearly

*The three minerals gadolinite, yttrialite, aud rowlandite occur in Llano County in most intimate association, suggestive of close community of origin, a suggestion which is emphasized by the marvelous agreement for gadolinite and yttrialite, not only in the relative proportions of the trivalent earth metals but in their absolute amounts as well.

This concordant testimony of three analysts may be regarded as strong evidence of the correctness of the earth separations made by them in these cases. Nearly the same relation is shown by the trivalent earth-metals of rowlandite, as seen in the table above.

there being a slight deficiency of oxygen atoms in each case for the radical Si,O,, which is increased by allowing for the CO, and P,O,.

In so far then as the character of the acid radical is concerned the results of Mackintosh's analysis are fully confirmed and there is absolutely no ground for accepting Benedicks' basic formula, which as I have already shown (p. 145) is based on a palpable error. But the ratios are not at all such as to lend themselves to ready resolution into isomorphous salts of the acid H.Si,O,. By doing quite unwarranted violence to the analytical data the above formulas might be reduced to

R" R"" """ (Si2O;)4
6

which can be readily represented structurally as a single complex molecule or as a mixture of molecules like 3R"",Si2O,+ R"R""Si,O,. In neither case, however, is the type of the rowlandite molecule approached, which requires an altogether different ratio of monoxide, dioxide, and trioxide bases, nor, if the second be accepted, is it at all clear that the two molecules would be mineralogically equivalent, that is, isomorphous.

An alternative hypothesis is to regard the mineral as a mixture containing the anhydrous thorite molecule. Proceeding on this assumption and deducting all thorium and uranium and the proper amounts of silicon and oxygen, the crude empirical formulas become

which may be interpreted as basic salts of metasilicic acid :

this last being easily susceptible of symmetrical representation in graphic form.

On the whole I prefer to leave the constitution of yttrialite unsettled until further evidence can be gathered, either from analyses of allied minerals or from yttrialite itself of more certain purity than any that has yet been discovered.

It must not be forgotten that the gases other than CO, contained in the mineral may be the cause of the inability to arrive at satisfactory conclusions in the case of this and all other minerals which contain them, as I have already pointed ont on p. 147.

My excuse for such a lengthy publication on a matter still unsettled is the desire to prevent general acquiescence in the grouping under one type of minerals which can by no means be regarded as proven to belong to that type, and to which yttrialite certainly does not belong.

Summary.

The empirical formula of Hidden and Mackintosh for yttrialite is confirmed, and it is shown that the basic formula of Benedicks rests on error and has no standing.

The formula of Hidden and Mackintosh is not, however, susceptible of representation as a simple salt of the acid H.Si,O.. On the purely hypothetical assumption of admixture of anhydrous thorite, the remaining constituents afford ratios conform ing quite closely to those of a basic metasilicate R''R''',(R'''O)', (Sio,)..

It is shown for two reasons fully discussed that the formula proposed by Benedicks for thalénite is to be regarded as doubtful.

Laboratory U. S. Geol. Survey,

Washington, D. C., Nov. 19th, 1901.

SCIENTIFIC INTELLIGENCE.

I. CHEMISTRY AND PHYSICS.

1. Apparatus for Determining the Density of Liquids.—A very simple means for taking the specific gravity of liquids has been devised by F. GIRARDET. It consists of two perpendicular glass tubes, resembling barometer-tubes, which are open at the bottom, and are connected at the top so that suction can be applied to both tubes at once by means of a rubber tube. Each tube has behind it a paper millimeter scale extending from an arbitrary zero point a few centimeters above its bottom to 500mm or more, near the top. To take the specific gravity of a liquid, the latter is placed in a small, cylindrical, glass vessel so that one of the upright tubes dips into it, while distilled water is placed at the mouth of the other tube and in a dish of the same size. Suction is now applied until the liquid to be tested remains, when a screw-clamp on the rubber tube is closed, at a convenient point high up in the tube, say 500mm, and the height of the water is also read. Then air is admitted until the columns have fallen to points near the bottoms of the scales, for example, until the liquid to be tested is at zero, when the positions of the two columns are again recorded. It is evident that the columns thus measured are inversely as the specific gravities of the liquids; hence, since one of them is water, it is only necessary to divide the fall of the water column by that of the other liquid in order to obtain the specific gravity of the latter. The effect of capillarity is eliminated by the two readings if the tubes are of uniform caliber. In the apparatus described the tubes were of about 8 exterior diameter, so that the quantity of liquid needed was very small. Determinations were made with various liquids, and results were obtained which agreed closely with those obtained by a Mohr's balance. Slight errors were encountered when volatile liquids were used, but satisfactory results were obtained with ethyl alcohol and with ammonia of 985 specific gravity. The apparatus is easily constructed at slight expense; it is much more accurate than ordinary specific gravity spindles, and it may be used to replace a series of these, as it is applicable to liquids which are heavier or lighter than water. Where readings are made to mm, with columns of about 400mm, the probable error is calculated to be of the absolute value. The time required for the determination is only about one minute.-Bulletin, xxv,

936.

H. L. W.

2. The supposed Existence of an Oxide of Hydrogen higher than the Dioxide.-The possible existence of an oxide of hydrogen higher than H,O, was suggested by Bach a year or two ago, on the ground that when hydrogen peroxide is acidified with sulphuric acid the amount of oxygen evolved by potassium permanganate is greater that the amount calculated from the permanganate used. RAMSAY has now confirmed the fact just stated, but