may be, in the case of the chlorides, if methyl be present, methyl chloride is usually split off.

Collie and Schryver, after studying the effect of heat on a number of mixed ammonium compounds, came to the conclusion, however, that when a mixed quaternary ammonium chloride (made from trimethylamine or triethylamine) is heated, a mixed tertiary amine is always produced. But the actual amount formed varies very considerably in the case of different compounds, the only product from trimethylphenylammonium chloride, for example, being dimethylphenylamine, whilst trimethylamine is chiefly formed when allyl- and isopropyltrimethylammonium compounds are decomposed. The behaviour of the chlorides and hydroxides is also sometimes diverse: thus, whilst trimethylbenzylammonium chloride yields dimethylbenzylamine and methyl chloride, its hydroxide is resolved into trimethylamine and benzyl alcohol; but methyltriethylammonium chloride yields triethylamine, whilst the hydroxide is converted into methyldiethylamine.

The belief has gradually grown up, however, that nitrogen acts as a pentad in the ammonium compounds, and that no distinction is to be drawn such as is implied in Hofmann's reference to a supplemental position, although the evidence that this is the case is still incomplete. An important series of experiments has been made by V. Meyer and Lecco directly with the object of ascertaining whether isomeric compounds can be formed by combining, on the one hand, dimethylethylamine with ethyl iodide, and, on the other, diethylmethylamine with methyl iodide; finding the products to be identical, they came to the conclusion that the ammonium compounds are to be regarded as atomic, and not as molecular, compounds (Liebig's Annalen, 1876, 180, 173).

But Lossen has pointed out that their results do not necessarily establish the correctness of their conclusion. It is possible, he thinks, that an exchange of radicles may take place during the formation of the ammonium compound. Moreover, he does not regard the fact that the products are crystallographically identical as proof that they are actually identical in structure, his argument being that two compounds such as NMe.Et Etl and NEt,Me MeI would be so similar that they would probably crystallise in isomorphous forms (Liebig's Annalen, 1876, 181, 364).

On mathematical grounds, the five valencies of a nitrogen atom cannot be all similar. In the case of compounds of the type NRR',CI it may be assumed, however, that the five radicles are distributed upon a sphere, and either that their mutual position is changeable, and therefore no isomerism is possible; or that the positions of the five groups are fixed, and that at least two isomerides are possible. In the case of the simpler derivatives of ammonium chloride contain

ing hydrogen or methyl, the former view seems correct; but, in the case of derivatives containing more complicated radicles, it might be expected that, owing to their mass and size, these would not change their mutual positions so readily as do hydrogen or methyl. Le Bel is of opinion that isomerism of this kind is recognisable (Compt. rend., 1890, 110, 145). On examining a whole series of cubic mixed ammonium platinichlorides, and determining the stage at which cubic symmetry vanished, he found that whilst the trimethylpropylammonium platinichloride was still cubic, the homologous isobutyl compound was dimorphous-at first crystallising in long, highly doubly refracting forms, which change into feebly doubly refracting octahedra closely resembling cubic crystals. The chlorides from these platinichlorides were also different (ibid., 1893, 116, 513). Collie and Schryver came to somewhat similar conclusions in the case of methyldiethylisoamylplatinichloride.

Moreover, Le Bel has stated that it is possible to prepare optically active isobutylpropylethylammonium chloride (Compt. rend., 1891, 112, 724).

It is clear, therefore, that the last word has not yet been said on the subject.

About eight years after the publication of his researches on the ammonium compounds, Hofmann, in conjunction with Cahours, brought under the notice of the Royal Society an account of their researches on the phosphorus bases (Phil. Trans., June, 1857, 575), in which they described the now familiar method of preparing trimethyl- and triethyl-phosphine by means of Frankland's zinc methyl and zinc ethyl and phosphorus trichloride, as well as a number of phosphine derivatives, and also several phosphonium compounds corresponding in the closest manner possible to the ammonium compounds.

The history of this investigation may be traced by reference to a paper presented to the French Academy by Cahours in Hofmann's and his own names in November, 1855 (Compt. rend., 41, 831).

Hofmann, at the close of his great memoir on the amines, had not failed to draw attention to the compound prepared by Thénard in 1845 from calcium phosphide and methyl chloride as being probably the analogue of trimethylamine. Cahours, after referring to Thénard's work, points out that the subsequent discovery of stibethyl by Löwig and Schweitzer, and that of the compound ammonias by Hofmann, afforded proof that the hydrogen in ammonia and its analogues may be displaced by methyl, &c., without the loss by these

* The writer has been told by Dr. Frankland that at Hofmann's request he prepared the zinc ethyl used in the first experiments made to test the method. VOL. LXIX. 2 z

hydrides of their basic properties; nay, more, that in the case of arsenic and antimony the basic properties become actually enhanced. The discovery of the tetramethyl bases of Hofmann, followed by that of the corresponding arsenic and antimony bases made simultaneously by Cahours and Riche on the one hand and Landolt on the other, had shown, moreover, that it was possible to go further and displace the four hydrogen atoms in ammonium by alcohol radicles, and that corresponding compounds of arsenic and antimony could exist. There remained but the blank between the compounds of nitrogen and those of arsenic to be filled up; and to supply this want, therefore, during a stay of a month in London, he had undertaken the study of the phosphorus compounds with Hofmann.

The monograph published by the Royal Society was the outcome of the inquiry thus initiated, which was carried out with the assistance of Dr. Leibius and Messrs. Perkin and C. Hofmann. Trimethyl- and triethyl-phosphine are fully described in it, and their perfect analogy with triethylamine demonstrated by their behaviour with the iodides of methyl, ethyl and amyl. Tetrethylphosphonium iodide is shown to resemble the corresponding ammonium iodide, and to be extremely soluble in water, but precipitable by potash. The corresponding hydroxide, whilst behaving exactly like caustic potash, and therefore resembling tetrethylammonium hydroxide, differs-it is pointed out-from the latter in its behaviour on heating, owing to the affinity of phosphorus for oxygen coming into play, yielding ethane and triethylphosphine oxide.* Lastly, attention is drawn to the extraordinary affinity of triethylphosphine to sulphur and selenium.

The analogies presented by the four elements-nitrogen, phosphorus, arsenic, antimony-as well as their peculiarities are pointed out by Hofmann and Cahours in this memoir with a clearness and precision before unknown, and in a manner which leaves no word to be said even at the present day.

The analogy between the four elements, they remark,

"Is particularly manifest in the compounds belonging to the ammonium type. In these remarkable bodies, nitrogen, phosphorus, arsenic, and antimony appear to play absolutely the same part. It is more especially in the oxides of these compound metals that analogy of composition induces a perfect identity in properties, and, indeed, of very salient properties which may be traced in almost every direction. If we were satisfied with the study of the reactions of these bodies, we should never suspect, in compounds exhibiting such a close similarity of properties, the presence of elements so

* The manner in which a number of phosphonium derivatives decompose when heated has been since carefully studied by Letts and Collie and by Collie (Chem. Soc. Trans., 1888, 714).

dissimilar as nitrogen, phosphorus, arsenic, and antimony; they might, moreover, be confounded with potassa and soda, by which they are scarcely surpassed in alkaline power. Only the deportment of the hydrated oxides, under the influence of heat, distinguishes the derivatives of nitrogen from the corresponding terms of the phosphorus, arsenic, and antimony series.

"If we regard, on the other hand, the compounds belonging to the ammonia type, we observe that the electro-positive character of the substances gradually rises in intensity from the nitrogen to the antimony compounds. Thus trimethylamine and triethylamine are not capable of uniting with oxygen, chlorine, bromine, and iodine; a power which the corresponding terms of the phosphorus, arsenic, and antimony series possess in a high degree. Triethylamine unites with the acids producing compounds of the formula

EN,HCI, EN,HSO,, EN,HNO.

"The corresponding compounds in the arsenic and antimony series do not exist; at all events, chemists have not yet succeeded in preparing them. Triethylamine and triethylstibine only combine directly with oxygen, chlorine, sulphur, &c., producing saline bodies which have the composition respectively

EASO, E ASC12, E ̧ASS2.

ESbO, E3SbCl, ESbS.

"In the phosphorus series, lastly, two classes are represented. Triethylphosphine not only forms compounds analogous to the salts of triethylamine, but also the terms corresponding to the binoxides of triethylamine and triethylstibine. We have, in the first place, the terms

E,P,HC, EP,HSO, E,P,HNO

and, in the second place, compounds of the formulæ

EPO, EPC, EPS.

"The phosphorus compounds, accordingly, hold a position intermediate between the nitrogen compounds, on the one hand, and the arsenic and antimony series on the other. It cannot, however, be denied that the phosphorus compounds stand closer to the arsenic and antimony series than to the nitrogen group.

"Ammonia is a powerful alkali; phosphoretted hydrogen only unites with hydrobromic and hydriodic acids, whilst in arsenietted and antimonetted hydrogen the power of combining with acids has altogether disappeared. In these hydrogen compounds the gradation of properties is, indeed, much more marked than in their trimethylated and triethylated derivatives. On comparing the terminal points of the series, ammonia and antimonetted hydrogen, we cannot fail to be struck by the dissimilarity of properties which at the first glance appears to limit the analogy of the two compounds to a mere parallelism of composition. In the methylated and ethylated derivatives of these compounds, the intensity of the chemical tendencies in general is so much raised that the gradation is no longer perceptible to the same extent."

Hofmann devoted much attention to the study of phosphine derivatives during the three years following the publication of the

paper by Cahours and himself, presenting no fewer than 12 communications to the Royal Society, which were finally embodied in three memoirs communicated to the Society in June, 1860, entitled "Contributions to the History of the Phosphorus Bases (Phil. Trans., 1860, 408-533). It is noteworthy that these contain accounts of a large number of crystallographic determinations by Quintino Sella, whom Hofmann terms "My friend, the celebrated crystallographer of Turin," and whose life he afterwards pourtrayed in a lengthy essay, perhaps the most remarkable and masterly of the numerous biographical notices of which he was the author; his inimitable skill in such matters is especially obvious in this case, as the subject of his memoir was not a chemist; yet he produced an account full of interest to chemists, which must rank as an enduring memorial of his remarkable versatility and literary power.

The first of the memoirs under consideration deals with the specific properties of triethylphosphine, which Hofmann speaks of as "a body in whose chemical relations the leading questions of the day are not unfrequently mirrored with surprising distinctness."

Having recounted his experience of the best method of preparing the phosphine, its oxide is fully considered. This compound exhibits in general but a slight tendency to unite with other substances; nevertheless, we are told, it affords a crystalline compound with iodide of zinc, for example, [(C2H2);PO]2ZnI2, which, strange to say, is formed in presence of a large excess of hydriodic and even of hydrochloric acid. When the anhydrous oxide is added to a concentrated solution of platinic chloride in absolute alcohol, a crystalline platinum compound is precipitated which Hofmann represents by the remarkable formula [(C2H2);PO]3, (C2Hs),PCl2, PtCl, but to which Letts and Collie, who have prepared a similar compound from tribenzylphosphine oxide, assign the formula 4(Et,PO),2HCl, PtCl (Roy. Soc. Edin. Trans., 30, 207). A peculiar crystalline oxychloride, (C2H3)3PO,(C2H5),PCl2, is said to be formed on subjecting the oxide to the action of gaseous hydrogen chloride.

The observations of most interest in the paper relate to the interaction of sulphur compounds and triethylphosphine. Although sulphur unites with triethylphosphine with the greatest readiness, and sulphide of nitrogen is instantly deprived of its sulphur by its action, sulphuretted hydrogen is without action, even in presence of air, the attraction of the phosphine for oxygen preventing the oxidation of the hydride. Mercaptan alone is also without action, but in presence of oxygen converts the phosphine into the sulphide,

Hofmann uses the new notation for the first time in these memoirs. The preliminary communications which had been published in the Proceedings were in the old notation.