sible to obtain a single satisfactory analysis, but the following data may be mentioned:

A salt obtained by decomposing an ethereal solution of phosphorus chloronitride with excess of baryta water consisted of a pulverulent, indistinctly crystalline substance for which the ratio P: Ba 6:4.48 was found, while the calculated ratio for neutral salt is 6 : 3..

=

The potassium salt, precipitated by ammoniacal bariu chloride, gave a preparation in which P: Ba= 6:3.68.

Triphosphonitrilic tetrachlorhydrine gave a clear solution with ammonia and barium chloride; on boiling a granular precipitate resulted (analysis 1).

Ethereal chloronitride and barium acetate solution gave a preparation from which the figures under analysis 2 were obtained.

A precipitate from the potassium salt and neutral barium chloride. gave the results recorded under analysis 3.

The neutral salt is slightly soluble in water and easily in a solution. of ammonium chloride; on boiling the latter it is partially reprecipitated. It also dissolves readily in a strong solution of sodium chloride, and, on warming, the barium sodium salt is precipitated.

Barium sodium trimetaphosphimate, P3N3OH,NaBa + 13(?) H2O.— This salt is almost insoluble in water, and forms a heavy, crystalline powder, which is seen under the microscope to consist of spherical aggregates with many rhombic facets, or occasionally of single minute crystals, apparently rhombohedra. The ordinary amorphous barium salt is invariably formed by precipitating any soluble trimetaphosphimate with a barium salt in the cold, but if the chloride or other salts of sodium be present in sufficient quantity, this passes into the crystalline barium sodium salt, at once on heating, or in the cold after a period varying from a few minutes to several hours. The latter also forms at once on precipitating in hot solutions. It was also obtained by decomposing an ethereal solution of chloronitride (1 molecular weight) with a solution of barium acetate (1 molecular weight) and sodium acetate (9 molecular weights). As is so common with salts of barium, this tends to "carry down" a considerable amount of the precipitant. A preparation made by the last method mentioned contained 0.49 per cent chlorine,

corresponding to a contaminatio.of 1.44 per cent barium chloride. After making correction.for this the figures under analysis 3 were obtained, which approximate closely to the theoretical. The excess of barium in analyses 1 and 2 is accounted for in the same manner. Possibly the difficulty is due to the presence of half dissociated barium ions (e. g. BaCl) in the solution.

The substance lost but little at 1000 and gave:

Similar salts containing potassium or ammonium could not be obtained. A barium silver salt consisting of microscopic spherules was obtained by adding silver nitrate to a solution of the neutral barium salt in ammonium nitrate; a qualitative examination showed the presence of barium and silver.

Tri-silver trimetaphosphimate, P3N3O6Agз.—This is the most important salt obtained, on account of its highly characteristic crystalline form, and because it is anhydrous and easily prepared pure, thereby establishing the empirical composition of trimetaphosphimic acid. As those salts which contain crystal water do not lose it completely at a temperature short of decompositon, we should otherwise be unable to decide between the formulas PзN3OH, and P3N3OH. For comparison the theoretical figures for a silver salt of both of these are given.

6

The salt may be prepared by precipitating a cold dilute solution (1-50) of the sodium salt by silver nitrate, when it is thrown down at once in the form of colorless elongated microscopic plates, terminated by an angle of 78° (analyses 1-4). Much better crystals are obtained if the solution of the sodium salt be acidified with several equivalents of nitric acid and the silver nitrate added slowly, finally in large excess. In this way the crystallization is made to extend over a considerable time, but the precipitation is never complete (analysis 5). Thus formed the crys tals are usually very well developed, though seldom over 2 millimeters long, and consist of thick, monoclinic plates of the forms represented in figs. 3 and 4, the latter resulting when the crystallization extends

Richards has pointed out the necessity of making a similar correction in determining sulphuric acid as barium sulphate. Zeitschr. anorg. Chem., Vol. VIII, p. 413.

over several hours. No other faces are ever observed, and the angle a (between the edges) measured under the microscope is very nearly 78°.

The air-dried salt lost nothing at 100° and gave:

The salt is insoluble in water, easily soluble in ammonia, and somewhat difficultly in dilute nitric acid. It is not affected by light, but on boiling under water it becomes somewhat yellow, owing to the formation of a superficial coating of the 6 atom salt and the liberation of free acid, the decomposition products of which can be detected in the liquid. If free from silver nitrate, caustic soda does not discolor it, even on boiling. On heating it swells up, gives off ammonia, and fuses to a mass which contains metallic silver. In order to ascertain if it is actually precipitated in the anhydrous form, or if it contains water which is given off on standing, a freshly precipitated sample was washed with alcohol and ether, pressed out, and at once weighed; it lost nothing at 100°, indicating that even at the moment of formation it is anhydrous. Precipitation from boiling solution seems to have no influence on its composition; if, however, the solution of the sodium salt

be added to a large excess of a solution of silver nitrate, the resulting precipitate, though white, is amorphous and contains much too high a percentage of silver (analysis 6), probably indicating an admixture of a more basic salt.

If tri-silver trimetaphosphimate be heated in a current of dry air at a temperature rising gradually from 2000 to 260°, very nearly one-third of the nitrogen is given off as ammonia:

Loss

Calculated
for NH3.

Found.

3.05

2.86

The product is no longer completely soluble in dilute acid. Alternate treatment with sodium chloride and ammonia extracts the silver and an acid which gives an amorphous precipitate with magnesium nitrate mixture. A considerable white residue remains in the form of pseudomorphs of the original substance, which is insoluble in ammonia, and soluble in hydrochloric acid only on boiling.

2

4

4

Salt with 4 atoms of silver, P3N3OH2Ag, or P3N3O,H,Ag1.-If silver nitrate be added slowly to a solution of the 4-atom sodium salt (sodium amidodiimidotriphosphate), each drop produces a white precipitate, which, on stirring, turns yellowish, in presence of an excess of sodium salt; when the latter is used up and an excess of silver nitrate has been added, it again becomes white. If, on the contrary, the sodium salt be added to the silver nitrate, each drop gives a yellow precipitate, turning white on stirring, and remaining so as long as an excess of silver is present. This white amorphous salt is decomposed by washing with pure water, turning yellow, while trimetaphosphimic acid goes into solution. The change may be represented thus:

3P3N3O7H4Ag1 = 2P3N3O7H2Ag1 + P3N3O»H¿+H2O.

Silver nitrate prevents this change; the precipitate must therefore be washed with 1 per cent aqueous solution of silver nitrate, then with 1 per cent alcoholic solution of the same, and finally with ether. On pressing and drying, it then remains colorless. Preparations made in this manner gave the following results after drying at 1000, at which temperature the salt remains white:

Unfortunately, the figures do not admit of a positive decision as to the formula, but the extreme instability of the salt, as compared with the others, may indicate an acid of different nature from trimetaphosphimic acid.

If this salt be allowed to stand for a day or two under pure water, it is converted into a mixture of tertiary silver trimetaphosphimate and a yellow crystalline salt with 6 atoms silver.

3P3N307H4Ag1=P3N3O7H2Ag1+2P3N3O6HзAgз+2H2O.

The same change occurs rapidly on boiling. Whatever may be the nature of the salt, this change is doubtless due to the tendency of the silver to wander to the nitrogen; silver nitrate, by opposing dissociation, prevents alteration.

Salts with 6 atoms of silver, PзNОAg and perhaps P,N,O,H2Ag1 also. Equally unsatisfactory analytical results are given by the salts with 6 atoms of silver. The figures indicate the tendency to formation of salts with the ratio P: Ag=3:6, but the actual ratio only approximates to this, and it is impossible on the basis of the analyses alone to decide which of the formulas P3N3O.Ag, and PзN,O,H2Age should be adopted in most cases. After many experiments I have been compelled to abandon for the present the attempt to obtain these substances in entirely satisfactory condition, and give the figures as actually obtained.1

If sodium trimetaphosphimate be added to an excess of an ammoniacal solution of silver nitrate, or of a solution of silver oxide in ammonium nitrate, a portion of the trimetaphosphimic acid is precipitated as a pure white voluminous and amorphous body, which, as analysis shows, contains approximately 6 atoms of silver. The precipitation is very incomplete, especially when ammoniacal silver nitrate. is used, but in this case the substance contains very nearly 6 atoms of silver, while if the silver oxide solution be employed the percentage of silver is considerably too low. Heated under the mother liquor it becomes yellowish, and becomes white again on cooling. After washing out (whereby it becomes yellowish) it dissolves quite readily in strong ammonium nitrate solution, but if boiled with less of the same, or less rapidly if heated with water alone, it is quickly converted into a heavy pulverulent substance. This, if highly magnified, is seen to consist of spherules or indistinct prisms, and has a color varying from orange to deep orange-red, usually the latter. This salt likewise contains 6 atoms of silver. The transformation is prevented by any silver nitrate in the solution, unless the effect of this is neutralized by a large quantity of ammonium nitrate.

1See p. 146 of this bulletin for an explanation of these difficulties.

* Prepared by adding ammonia to silver nitrate solution till the brown precipitate just redissolves. Three grams silver oxide, 10 grams ammonium nitrate, and 50 cubic centimeters water.

4 The filtrate, on evaporation over sulphuric acid, deposits an imperfectly crystalline substance, at first white, then yellow, which is probably identical with the yellow salt referred to below.