remained undecided whether every difference in the groups attached to the nitrogen atom was sufficient to cause activity. Contrary to expectation, I have been unable to obtain optically active forms of the asymmetric salts of the above-mentioned paratoluidine series. The failure was probably due to unfavourable solubility conditions, and also to a tendency to autoracemisation. The latter phenomenon was observed by Pope and by myself in the case of the active a-phenylmethylallylbenzylammonium iodide.

This phenomenon was formerly explained by assuming the dissociation of the salt into benzyl iodide and allylmethylaniline, according to the equation :—

(C ̧н ̧) (C ̧H ̧) (CH ̧)N (C,H1)I ✈ C ̧H ̧ (C ̧H ̧) (CH ̧)N + C ̧H, I.

That is to say, by the easy passage, under certain conditions, of pentavalent into bivalent nitrogen, and the consequent destruction of space asymmetry, and, therefore, also of optical activity. As a matter of fact, the iodide is largely dissociated in boiling chloroform solution. Whether this is also the case in chloroform solution at ordinary temperature could not then be decided. H. O. Jones, however, by using Barger's microscopic method for determining molecular weights, was able to show that such salts possess normal molecular weights in chloroform solution at ordinary temperature. The explanation of the mechanism of autoracemisation in the case of the active asymmetric ammonium salts becomes, therefore, as difficult as that of the spontaneous racemisation of the active esters of brom-fatty acids, unless it is assumed that the degree of dissociation at any instant is at ordinary temperatures so small as to escape measurement by the methods employed. This small amount of dissociated material would on recombination form the racemic salt. In the next instant another small quantity of the active salt would be dissociated and racemised, and so on until the whole mass was racemised. It is evident that, in spite of this process, an approximately normal molecular weight might be found. The velocity of the change that is to say, the amount of active salt dissociated in the unit of time-depends upon the strength and duration of the light falling on the solution, and also on the temperature. I am at present engaged in the measurement of the velocity of autoracemisation under various conditions, in the hope of elucidating this problem.

The important question whether all asymmetric ammonium salts, independent of difference in groups, could be obtained in active forms has at last been answered. Jones, after failing to obtain ammonium salts of the type N.a.a.b.c.X., and also cyclic salts in active forms, succeeded in resolving the phenylethylmethylbenzylammonium base at almost the same time. I prepared the iodide and the dextrocamphorsulphonate of the same base, and succeeded in resolving the latter by a single recrystallisation from methyl formate ([M] of the dextro-phenylbenzylethylmethyl d-camphorsulphonate = +69°). By employing the same useful solvent I have also recently been able to resolve the homologous propylphenylbenzylmethylammoniumhydroxide. The d-camphorsulphonate of the dextro base forms transparent rhombohedra, which attain a diameter of one or more centimetres. The highest rotatory power hitherto observed is [M] D= D= +62°. The iodide prepared from the camphorsulphonate was active. Activity did not result when acetone or acetic ether was employed. I am at present working on the resolution of the homologous isobutyl base.

1

The problem of resolving salts containing two asymmetric nitrogen atoms appeared particularly interesting. For this purpose I have converted ethylene dikairolinium iodide 2 into the di-dextro-camphorsulphonate.

The single fractions showed the same specific rotatory power and a molecular rotation of 103° to 104°. The di-d-brom-camphorsulphonate crystallises remarkably well, and seems to offer a better chance of resolution. The acetic ester of kairolinium d-camphorsulphonate ([a]P=11·7°) has not been obtained in active forms either by Jones or myself.

The problem of examining the so-called inactive isomers of asymmetric nitrogen offers great difficulties. Anyone who has examined so great a series of asymmetric systems for isomers without success as I have, will understand me when I call the single case of isomerism in the series of benzylallylphenylmethylammonium salts (which I discovered several years ago)' remarkable,' and to a certain extent a 'puzzle.' Kipping and Aschan have both taken exception to this expression. I think, however, that everyone will agree with me that the extraordinary rarity of such isomers-predicted by most theories-is most striking, especially as in the observed case the isomers possess about equal stability. Since then I have found a new case in the series of asymmetric ortho-tolyl-ammonium salts, but on account of experimental difficulties I have not been able to establish it with absolute certainty. Since among the homologues of the asymmetric aniline salt, with this exception, no isomers have been found, I have commenced experimenting with the asymmetric phenetidine and anisidine bases. Paraphenetylbenzylallylmethylammonium iodide (and the corresponding d-camphorsulphonate) are beautifully crystallised salts when made by the combination of allyl or benzyl iodide with the corresponding tertiary bases. The third method-addition of methyl iodide to benzylallyl-p-phenetidine-leads to an amorphous salt. Experiments are already in progress to prove if this is really a case of isomerism.

O. Aschan has recently discovered a new case of isomerism in the series of the diacid ammonium salts which differ in solubility. At the time of his publication I was working on the ethylene bases of the tetrahydroisoquinoline series. I therefore tried whether the above-named bases, which in contradistinction to the piperidine derivatives are quite unsymmetrical, possess analogous powers of reaction. The reaction between trimethylene bromide and ethylene di-tetrahydroisoquinolide takes place almost quantitatively at 100° C. The equation probably is:

According to this, the salt, which is rather easily soluble in water, would contain a seven-atom heterocyclic system. The prospects of successfully preparing this system by the second method are not good, since, according to my experience,

I intend to try to resolve the fatty asymmetric base of Le Bel, N(CH ̧)(C2H ̧) C2H2)(¿.C ̧H ̧)OH by the new method.

ethylene bromide seldom reacts with tertiary bases in the normal manner. kairolin itself with ethylene bromide gives almost exclusively the salt:

Iso

It has not yet been found possible to bring the Br in the -CH,Br group into reaction with a second molecule of isokairolin.

I do not consider that our present knowledge is sufficient for successful theoretical speculation on the configuration and isomerism relations of the pentavalent nitrogen atom. This, however, is certain, viz.: in active ammonium salts the tetra-atomic radical N. a. b. c. d., the centre of activity, must possess tetrahedric grouping. This radical occurs in solution as the free active cation. The fifth valency, which is not always satisfied, cannot-apart from other groundspossess equal value. I picture it in a rectilinear lengthening of a tetraheder axis, as the following sketch shows:—

This configuration is that proposed long ago by van't Hoff, and since then revived by Aschan. The latter has also deduced the inactive isomers (prepared by Le Bel, Kipping, Aschan, and Wedekind) from this scheme. In my opinion, the change of places of the different radicals, or, still more, the lack of change of places of the different radicals (appearance of isomerism), is so little understandable that even with the help of this model we cannot as yet form any clear ideas of the intramolecular reactions among the ammonium salts. I incline more and more to the opinion that, in the case of nitrogen which shows such different behaviour, one must discard the idea of a fixed valency. The theories which Werner has developed for carbon may be useful here. They allow, too, of an explanation of autoracemisation without the assumption of dissociation. Let us consider affinity as a force acting uniformly from the nitrogen atom, considered as a sphere, towards the surface. Then we can imagine the five radicals fixed as five valency positions' on the surface of the sphere. Four of these could take up a tetrahedral grouping, if in this position the greatest exchange of affinity took place. Now, the intra-molecular movements of the radicals in substituted ammonium salts are particularly evident. This is proved by the tendency towards change of place. These movements appear, then, as pendulum-like oscillations about the valency position, and are increased by rise of temperature, by the action of sunlight, and by other unknown causes, until, finally, a change of place occurs, just as Werner assumes in the case of compounds containing asymmetric carbon. (Explanation of autoracemisation of brom-succinic acid.)

I do not consider it, however, impossible that the five radicals would cause a different grouping of the valency positions, when the force acting between them is of a different nature. Such groupings are perhaps the pyramid formula, or the double tetrahedron of Willgerodt. The latter might occur when two negative radicals are combined with a tertiary amine. The attractions and repulsions

between the radicals would now be quite different from those in the normal ammonium type. The configuration set forth by the double tetrahedron is the labile one, and a tendency therefore exists for it to go over into the stable ammonium form (tetrahedron or pyramid), one of the negative radicals being replaced by hydrogen or alkyl.

I think that such pictures are useful in helping us to understand the easy change of position of the radicals in quaternary ammonium salts. Further, the complex behaviour of pentavalent nitrogen becomes thereby more easily explicable.

7. On the Products obtained by the Action of Tertiary Bases on some Acid Chlorides. By Professor E. Wedekind.

The author has previously shown ('Annalen,' 318, 99; 323, 257) that, in spite of the violent reaction which occurs between the chlorides of powerful acids and strong tertiary bases, no quaternary salt of the type

RNCO.

is produced, but that the hydrochloride of the tertiary amine,

is obtained in quantitative yield. This fact is in so far of importance in connection with the stereochemistry of nitrogen as it indicates the reluctance of the trivalent nitrogen atom to take up two acidic or negative groups; the trivalent nitrogen atom exhibits a kind of striving to assume the most stable condition, that, namely, of which ammonium chloride is the type.

The question at once arises, in connection with the above reaction, as to what becomes of the residue of the acid chloride molecule; the solution of this problem presented extraordinary experimental difficulties, but its study has led to the discovery of several interesting facts, which may here be briefly mentioned. The author has already shown that by the action of acetyl chloride on triethylamine, pyridine, &c., dehydracetic acid is produced (Annalen,' 323, 247); in this reaction four molecules of each component take part, in accordance with the equation 4CH3.COCI + 4N(C2H ̧), = 4N(C2H ̧) ̧HCl +¤ ̧Н ̧O1(4C2H2O);

but when the acetic chloride is replaced by propionyl, phenylacetic or hydrocinnamic chloride, products are obtained which have only three times the molecular weight of the hypothetical substance which must be supposed to be first formed by the splitting off of hydrogen chloride from the one molecule of the acid chloride. Thus the condensation product from propionic chloride, CH,.CH2.CO.Cl, has the empirical formula C,H,,O, or 3(CHO), and that from phenylacetic chloride, CH.CH.CO.Cl, corresponds to the formula CHO, or 3(CHO).

12

The latter substance proved to be fairly easily obtainable, and the first supposition that it might be a phloroglucinol derivative-the symmetrical triphenylphloroglucinol--was found untenable, because the material can under no conditions be reduced to triphenylbenzene, and is relatively very stable even in alkaline solutions. The new compound shows simultaneously the behaviour of a lactone, a monoketone, and a primary alcohol; thus it gives with soda a monosodioderivative, with hydroxylamine a monoxime, with acetic chloride a monoacetyl compound, and with benzoyl chloride a monobenzoyl derivative. On treating it with ammonia under pressure it yields a very stable pyridine derivative, and this reaction shows it to be a simple homologue of pyronone (the previously known pyronone compounds contain, like dehydracetic acid, a carboxyl group in the side chain).

On heating with alkalies it yields analogous products of hydrolysis to the symmetrical trialkylphloroglucinols.

The mechanism of the reaction by which the substance is formed (compare J. N. Collie, Trans. Chem. Soc.,' 77,971) would seem to be that three molecules of the acid chloride first condense with evolution of two molecules of hydrogen chloride, yielding the chloride of an ay-diketonic acid in accordance with the following scheme:

C.H..CH,.COC1 + CH..CH,.CO.C1 + CH..CH,.COC
CH.CH.CO.CH.CO.CH.CO.CI
CH, CH

This hypothetical chloride then undergoes total or partial conversion into an enolic form, from which, by subsequent loss of hydrogen chloride, in which the hydroxyl group is involved, the closed pyronone ring is formed:

Since this pyronone derivative differs from those previously studied, in that it gives an oxime as well as a monobenzoyl derivative, it must be regarded as tautomeric in that it can assume the above ketonic form as well as the following enolic or hydroxylactonic constitution:

The latter formula naturally gives rise to the acidic derivatives. The remarkable power which the carbonyl group in the closed ring possesses of reacting with hydroxylamine must be attributed to the multiplication of unsaturated groups in the molecule-namely, to the presence of two phenyl groups and two double bonds

in the closed chain.

As was to be expected from the known behaviour of pyrone or pyronone compounds, the action of ammonia gives benzyldiphenyldihydroxypyridine:

CH.C

он

CH.CH.C-N=C.OH

Lastly, it may be mentioned that the author has made a remarkable observation upon the interaction of isobutyric chloride with tertiary amines; in this reaction an extremely volatile substance crystallising in colourless needles and having an odour of menthol and camphor is obtained. This product is formed by the condensation of only two molecules of the acid chloride, and is not a pyronone derivative but a diketone; it is in all probability a tetramethylene derivative, from which the author hopes to prepare the parent hydrocarbon.

8. Sur les Manganates et les Permanganates. Par Dr. A. ETARD.

Le permanganate de potassium est parfaitement connu, mais le manganate_vert est, je crois, considéré comme une masse fondue très riche en potasse. Cependant, Mitscherlich a décrit et mesuré des cristaux de MnO4K. Il ne semble pas qu'on les ait préparés depuis. Mon but n'est pas de signaler ce sel, connu autrefois, mais de décrire ses relations avec le permanganate. Quand du permanganate cristallisé