tion of the hydroxyls as well as of the imide hydrogen atoms; in short, quite a number of stereo-isomers may be conceived. I have thus far sought in vain for indications of the existence of such forms.

EXPERIMENTAL PART.

Analytical methods.-Salts of trimetaphosphimic acid and derived bodies are easily decomposed by fusion with sodium carbonate. As the process involves oxidation it is essential to keep the mass in a state of fusion for some time with access of air. Salts of silver of other easily reducible metals and those in which it is desired to determine alkali or nitrogen must be decomposed by strong sulphuric acid. The substance, contained in a platinum crucible, is first moistened with dilute, and then covered with 4 or 5 cubic centimeters strong sulphuric acid, and the covered crucible heated to fuming for an hour in a radiator, care being taken that the fumes do not escape. In determining alkali metal, the resulting phosphoric acid may be conveniently removed by the ferric chloride-acetate method. As the water of crystallization can not be completely expelled by heat without partial decomposition, it is best determined, together with the hydrogen of the salt proper, by combustion. This is conveniently carried out in a short combustion tube with a 10-centimeter spiral of oxidized copper gauze, the substance, contained in a boat, being covered with fused potassium bichromate. As nitrous fumes are formed in abundance, the usual precautions must be taken. In the following, where more than one determination is given, the numbers always refer to different preparations.

Preparation of trimetaphosphimic acid.-The sodium salt affords the starting material for all preparations of trimetaphosphimic acid. It may be obtained by saponifying triphosphonitrilic chloride by alcoholic soda, but with much loss, owing to the formation of ethers and other substances of unknown nature. The chloride, as observed by Liebig and by Gladstone, is practically unacted on either by water or anhydrous ether, but, as I have already pointed out elsewhere, its decompo sition may readily be effected by shaking its ethereal solution protractedly with water, whereby intimate contact is secured. Decomposition by ether and water alone, however, results in much loss, as the liberated hydrochloric acid rapidly decomposes trimetaphosphimic acid. The following method is perfectly satisfactory and gives practically the theoretical yield.

Thirty grams of the chloride are dissolved in 150 cubic centimeters ether free from alcohol, and the solution gently agitated with a solution of 110 grams crystallized sodium acetate in 200 cubic centimeters water, the agitation being conveniently effected by slowly rotating with a small turbine. After about fifteen hours, well-formed crystals of sodium salt begin to ap pear, and about seventy or eighty hours are required for complete decomposition. This point is best observed by evaporating a few drops of the ether and taking up the residue with water, any undecomposed chloride

remaining undissolved. It is best to continue the agitation for a short time longer in order to decompose the chlorhydrines, which are always formed as intermediate products. At the end of the operation nearly all the sodium salt has crystallized out, being almost insoluble in the strong salt solution; a further small amount can be recovered by mixing the solution with alcohol. After washing with 50 per cent alcohol, the salt is pure enough for most purposes, but may be redissolved in water and precipitated by gradual addition of alcohol.

The same salt also results by decomposing triphosphonitrilic tetrachlorhydrine with soda. It exists in two forms, apparently differing only in the amount of crystal water, the a-salt being formed at ordinary temperatures, the -salt only above 80°.

a-Sodium trimetaphosphimate, P ̧Ñ‚О¿H ̧Ñа ̧ + 4H2O.—The air-dried salt, prepared as above, gave the following figures:

It loses weight slowly in vacuo, and 3 molecules of water are given off on long heating at 100°. After five hours at 100° the loss was:

Loss

Calculated for
3H20.

Found.

14.38

14.05

This is one of the most characteristic salts of trimetaphosphimic acid. It forms brittle orthorhombic prisms, usually about 1-2 millimeters across, of which fig. 1 represents the most common form.

The angle a, measured roughly under the microscope, is about 130°. No other faces are ever observed, and D is frequently lacking, giving the form shown in fig. 2, which is often seen to be elongated in the direction of any one of the axes. They are especially well developed when slowly thrown down by acetic acid and alcohol from alkaline solution. When formed by rapid addition of alcohol in large excess they tend to elongation in the direction of the principal axis, being often acicular, often long, flat prisms with imperfectly developed ends. At 200, 100 parts water dissolve 18.3 parts; in hot water it is much more soluble, and from its hot solution it crystallizes but slowly; the larger crystals dissolve in hot water with a crepitating sound. The reaction toward litmus is neutral. On rapidly heating the dry salt it gives off water and later ammonia, and fuses to a clear glass. It is but slowly decomposed by boiling with water; after three hours heating traces of phosphoric and imidodiphosphoric acid were shown by magnesia mixture. Alkalies do not cause the evolution of any appreciable quantity of ammonia, even when hot; apparently, however, their very long-continued action is attended with decomposition.

B-Sodium trimetaphosphimate, P3N3O6H ̧Ñа3+H2O.-This form is deposited when the solution has a temperature of 80° or higher. It was obtained

(1) By boiling the solid a-salt under a strong solution of sodium nitrite (analysis 1).

(2) By slowly adding boiling alcohol to a boiling solution of the a-salt (analysis 2).

(3) By pouring a boiling solution of the a-salt into a boiling 25 per cent solution of sodium acetate.

Doubtless other sodium salts would serve equally well, provided their hot solutions are sufficiently strong to throw it out. The salt lost nothing at 100° and gave:

This modification of the sodium salt forms needles, which, under the microscope, are seen to consist of flat prisms terminating in points; the terminal angles are of two kinds, one rather more, the other rather less than 90°; frequently both of these may be observed at opposite ends of the same crystal. Often, too, the crystals are cut off obliquely,

owing to the development of but one terminal plane, in which case the terminal angles are respectively rather more and rather less than 45°. In general properties it resembles the a-salt, which it gives when reprecipitated from cold solution; whether the difference consists merely in the amount of crystal water, or whether it is a chemically distinct body can not be decided at present.

Tetra-sodium salt (sodium amido-diimidotriphosphate), P3N2O,H,Na+ H2O.-As pointed out in the introduction, this is best regarded as the neutral salt of an open chain acid,

of which trimetaphosphimic acid is the inner anhydride. As positive proof of this can not be adduced at present, I have preferred to place it among the trimetaphosphimates.

On dissolving sodium trimetaphosphimate in an excess of caustic soda, concentrating and allowing to cool, the salt crystallizes in long brittle needles (analysis 1). On adding alcohol to a caustic soda solution of sodium trimetaphosphimate (which need not have been heated), it is thrown out as a sirup which crystallizes, slowly if left to itself, at once on adding a fragment of previously prepared salt, to a mass of delicate colorless needles (analysis 2). These must be washed with alcohol containing a little caustic soda in solution, strongly pressed out and dried out of contact with carbon dioxide. It contains a large amount of crystal water, which it loses, with the exception of the last molecule, on drying in vacuo; the exact amount could not be determined. The substance dried in vacuo lost nothing at 100° and gave:

This salt is very unstable, being decomposed even in the solid state by carbon dioxide. From its cold aqueous solution alcohol precipitates a mixture of unchanged salt and a-sodium trimetaphosphimate, a few reprecipitations converting it completely into the latter and free alkali. Its behavior toward silver nitrate is described under the silver trimetaphosphimates.

Salts with less than 3 or more than 4 atoms of sodium could not be obtained.

Potassium trimetaphosphimate, P3N2OH3K3.-A solution of this salt was obtained by decomposing the barium salt with potassium sulphate; on evaporation a transparent, glassy mass resulted. Alcohol precipitated it as a sirup, which on repeated treatment with absolute alcohol was converted into indistinct crystals. It was not further examined. Ammonium trimetaphosphimate, P3N3OH (NH)3 + H2O.—A solution of this salt was obtained by double decomposition from the barium and silver salts and by neutralizing the free acid with ammonia. It loses ammonia on evaporation and leaves a transparent amorphous residue. Its solution, treated with alcohol, gives a deposit of large and beautiful pearly scales, which are characteristic. As these form even in the presence of a large excess of ammonia it appears that a body analogous to the tetra-sodium salt can not be thus formed.

Magnesium trimetaphosphimate, (P3NO,H3),Mg(?).-Strong solutions of sodium trimetaphosphimate and magnesium chloride give no precip itate, even on boiling. A solution of the magnesium salt may be obtained by double decomposition. From this alcohol throws down an amorphous flocculent precipitate and it leaves a soluble transparent residue on evaporating. On boiling the solution is slowly decomposed. A mixture of a not too dilute solution of sodium salt with an excess of a strong solution of magnesium acetate remains clear, but is precipitated by acetic acid; on adding a little water this precipitate redissolves, reappears on heating, and redissolves on cooling. The nature of these precipitates has not been investigated. With more water the solution remains clear on short boiling, but on continued heating a crystalline precipitate forms which examination shows to consist of magnesium pyrophosphate and imidodiphosphate, orthophosphoric acid remaining in solution.

The fact that trimetaphosphimic acid is not precipitated in the cold by ammoniacal magnesia mixture affords a valuable means of separating it from some of its decomposition products.

Barium trimetaphosphimates.-Much time was spent in studying these salts, until it was found that they are devoid of characteristic properties, and that they are of exceedingly varying composition. Both neutral and basic salts exist, and there is a strong tendency to form double salts. The tendency to form complex salts is so strong that when barium nitrate, chloride, or acetate is used as the precipitant, the product is invariably contaminated by these. It has proved impos