case the yield of lophine was 46 per cent. of the theory; in the second 93 per cent. I have repeated Radziszewski's experiment, and had no difficulty in obtaining lophine, but cannot confirm his statement as to the almost theoretical yield. On the contrary, the yield was in my hands but small.

With regard to the above oxygenated compound which I hoped to obtain from benzil, benzaldehyde, and ammonia, it is in the highest degree probable that this compound has been known for a very long time, without, however, its true nature being recognised. (The existence of this compound has, as I shall endeavour to show presently, a distinct bearing upon the question of the constitution of lophine.) Zinin (Annalen, 34, 190), by adding aqueous ammonia to a warm alcoholic solution of benzil, obtained a compound to which he assigned the formula C2H20N2O2, and to which the name azobenzil was afterwards given. Halving this formula, we arrive at C2H5NO, the formula of the compound sought for. Zinin draws attention to the simultaneous production of ethylic benzoate in the reaction, and remarks that the formation of azobenzil is accounted for by this fact, without, however, explaining more precisely in what way this is the case. The formation from benzil and ammonia of a compound of Zinin's formula corresponds to a reduction, and Zinin probably only meant that this reduction was accounted for by the simultaneous oxidation of a portion of the benzil to benzoic acid.

Zinin's reaction may, I think, be interpreted in the following manner:-In the first place, a portion of the benzil is decomposed in presence of alcohol and ammonia, with formation of ethylic benzoate and benzaldehyde :

C.H.-CO

C&HS-CO

+ C,H,O = CH.—COOC.H + CH–CHO.

The benzaldehyde then reacts with a second molecule of benzil and one molecule of ammonia, yielding azobenzil:

C,H,-CO
C.H,-CO

| + CH–CHO + NH=

CH-C-O
C&H,-C-N
Azobenzil.

C-CH, +20H2,

the reaction taking place according to equation I of the phenanthraquinone series.

In order as far as possible to test the correctness of this supposition, and to ascertain the nature of this compound, a quantity of it was prepared by Zinin's method. Aqueous ammonia was added to a warm alcoholic solution of benzil till a precipitate was produced; this

was then left in contact with the liquid at a temperature of about 70° for ten hours. Instead, however, of crystallising the substance from alcohol, it was found advantageous to extract the white crystalline powder, which formed the product of the reaction, with boiling light petroleum, in which the azobenzil readily dissolved, but the other substances present were practically insoluble. The solution on cooling deposits the pure compound in groups of fine colourless prisms. By crystallisation from boiling alcohol, it was obtained in the very lustrous long thin needles described by Zinin. which is not given by Zinin, was found at 115°. above the range of the mercurial thermometer. A small quantity was boiled for some time in a test-tube without suffering the slightest decomposition-a behaviour which scarcely points to a compound containing 42 atoms of carbon in its molecule. Concentrated hydrochloric acid converts it into a gummy hydrochloride. Heated under pressure with the acid to 250°, it yields benzoic acid, ammonium chloride, and a resinous mass.

Analysis confirmed Ziniu's results:

The fusing point,

The substance boils

I

Substance.

0.1157

CO2. 0.3607

OH.. 0.0547

II. 0.1572 gram burnt with copper oxide in a vacuum gave 6'6 c.c. moist nitrogen at 22° and under 754 mm. pressure.

The vapour-density of the substance was determined by Victor Meyer's air-displacement method, heating in a lead-bath, with the following result:

0-0881 gram displaced 7.3 c.c. air measured moist at 19', and under 757.5 mm. pressure.

Calculated for CH15NO. 10.28

Found.

10.23

The substance, therefore, possesses the formula and molecular weight here assigned to it, and the above may be regarded as the most probable account of the mechanism of the reaction in which it is formed.

If, on the other hand, in the reaction of benzil with ammonia, the benzaldehyde which is formed by the decomposition of 1 mol. of benzil were to react, together with two molecules of ammonia, upon a second molecule of benzil, lophine would be formed, the reaction taking place according to equation II of the phenanthraquinone series. In fact, Radziszewski has shown that lophine is obtained in small quantity by the action of ammonia upon benzil.

What I wish to point out is that in these cases-just as in the corresponding reactions between the methyl ether of salicylaldehyde, phenanthraquinone, and ammonia-we have the two reactions I and II taking place simultaneously with the formation of two compoundsone containing 1 atom, the other 2 atoms of nitrogen.

Now there is only one at all probable mode of formulating the compounds containing 1 atom of nitrogen, consistently with their formation from 1 mol. of double ketone, 1 of aldehyde, and 1 of ammonia. We must assume in them the existence of the complex of atoms—

-C-O
-C-

This is also in accordance with their decompositions. Thus benzenylamidophenanthrole and azobenzil, when heated with concentrated hydrochloric acid, are split up into benzoic acid and ammonia, whilst the phenanthrene and stilbene portions of the molecule are resinised under the conditions of the experiment.

We have therefore to assume that during the formation of the compounds containing 1 atom of nitrogen, an intramolecular re-arrangement occurs the two carbon-atoms of the double ketone-group, -CO-CO-, become united by double bonds. This corresponds with what occurs when a quinone of the ortho-series-also a double ketone is converted by the action of reducing agents into a quinol.*

With phenanthraquinone, three cases of this intramolecular re-arrangement, occurring under the influence of reducing (or hydrogenating) agents, or of substances equivalent in their action to reducing agents, are known: the conversion of phenanthraquinone into phenanthraquinol by the direct addition of two atoms of hydrogen; the conversion of phenanthraquinone into the monethylic ether of phenanthraquinol, by the successive action of zinc-ethyl and water, the reduction in this case consisting in the indirect addition of the equivalent of two atoms of hydrogen in the shape of one atom of hydrogen and one ethyl-group; and lastly, the action of aldehydes, together with ammonia, upon phenanthraquinone. The formation of the double compound of phenanthraquinone with hydrogen sodium sulphite might perhaps be added to this list. Since in all such reactions the above-mentioned re-arrangement occurs in the carbon linkings of the closed lateral chain, no reactions in which phenomena of reduction are involved can be employed in determining the constitution of phenanthraquinone. Practically, all the arguments in favour of Graebe's formula for phenanthraquinone have been drawn from some such source. I hope

In the present case the reducing agent is an aldehyde, and, when the aldehyde has done its work, we have no longer an aldehyde-residue, but an acid-residue in the molecule of the new compound. The occurrence of this re-arrangement has been proved for three double ketones-phenanthraquinone, chrysoquinone, and benzil.

As regards the compounds containing 2 atoms of nitrogen in the molecule, it seems to me that the simplest way of formulating these is to assume in them the existence of the complex of atoms

This is what I have done in the work on phenanthraquinone and in the paper on lophine published in conjunction with Mr. Robinson. I take for granted an intramolecular re-arrangement, such as occurs in the formation of the oxygenated compound.

Radziszewski, on the other hand, assumes in the latter class of compounds the existence of the complex

Here the assumption of an intramolecular re-arrrangement is dispensed with, and this is so far a point in favour of Radziszewski's formula. This mode of formulating these compounds did not escape me; but I rejected it for the reasons above given. It seemed to me a more probable assumption that in two reactions of the same classboth condensations of double ketones with aldehydes and ammonia, both of the class of condensations in the ortho-series (employing the term "ortho" in an extended sense)-occurring simultaneously in the same operation, an intra-molecular re-arrangement which must occur in the one and which is conditioned by the reducing action of the aldehyde, should also occur in the other, the same condition being again present.

Up to this point I have described the grounds of analogy which led me to prefer the anhydro-base formula for lophine to that of Radziszewski. It now remains to regard the two formulæ from the point of view of the reactions of lophine.

Radziszewski finds a confirmation of his formula in the fact that lophine, when fused with potassium hydrate, yields benzyl alcohol and benzoic acid; and he ascribes the formation of these substances to the action of the alkali upon benzaldehyde furnished by the decomposition

shortly to be able to lay before the Society an account of some reactions which are not open to the above objection, and which appear to me to decide in favour of Fittig's formula.

of the lophine. I cannot find that this reaction decides either way. A compound of the anhydro-base formula would split up under the influence of the alkali into benzoic acid, ammonia, and the hypothetical compound

C1H2-C(OH)

C&H2-C(OH)

a compound which corresponds with 2 mols. of benzaldehyde, and would be decomposed by the alkali with formation of the benzoic acid and benzyl alcohol obtained by Radziszewski.

By careful oxidation, lophine yields benzamide and dibenzamide, according to the equation: C2H16N2 + OH2 + O2 = C6H¿.CO.NH. + NH(CH¿.CO)2 (Fischer and Troschke). If we adopt the anhydrobase formula, this reaction is readily accounted for. It is only necessary to assume that, as is usual in the oxidation of unsaturated compounds, the separation of the parts of the molecule occurs at the points where the atoms are connected by double bonds:

It is difficult to see in what way Radziszewski's formula can account for this reaction.

One point in which the anhydro-base formula appeared to satisfy all requirements was the way in which it accounted for the formation of compounds containing alcohol-radicles; for example: Kühn's ammoniam-compound, diethyl-lophinium iodide, C21 H15(C2H2)2N2I (cf. this Journal, Trans., 1882, 329), a compound corresponding with Hübner's diethylanhydrobenzdiamidobenzene iodide, CH,(CHS)2N2I. In Radziszewski's lophine formula there is no replaceable hydrogenatom attached to nitrogen, so that the formation of this compound of Kühn's cannot be accounted for. Radziszewski perceives this difficulty, but I think that he underrates it. He says that the fact that lophine, although containing no hydrogen directly attached to nitrogen, yields compounds with alcohol-radicles "cannot surprise any one who has studied Hofmann's beautiful researches on the exhaustive action of methyl iodide upon conine and piperidine." The cases are, however, scarcely comparable. Dimethylconine and dimethylpiperidine were obtained by the destructive distillation of the corresponding ammonium-hydroxides-a process very different from that