and deoxybenzoin, were described, and it was suggested that they were the analogues of the substances obtained by Schiff from ethyl acetoacetate and considered by him to be derived from the ketonic, enolic, and ketoenolic modifications of that ester. For the sake of convenience, these additive products are termed in this paper a-, B-, 7, that obtained from the pure ketone being the a-, that obtained by the action of piperidine the B-, and that by the action of sodium ethoxide the y-modification. The fact that they are slightly basic has made it possible to effect the isolation of unstable hydrochlorides which are different in the case of the a- and ß-, but identical in the case of the B- and y-modifications. It was previously pointed out that these isomeric derivatives resemble one another very closely in appearance and properties, and that the differences in melting point alone differentiated them from each other, consequently the isolation of different hydrochlorides is important. The behaviour of these hydrochlorides with absolute alcohol is interesting; the a-salt dissociates, yielding a mixture of a- and y-base, whereas the ẞ- or y-salt is transformed almost completely into the a-base. The additive products from dibenzyl ketone have been more thoroughly examined than those from deoxybenzoin, and it appears that the a-form is the most stable, the others being converted into it by the action of heat. It was further noticed that whereas piperidine readily converts the a-, and less readily the the B-form, phenylhydrazine, which appears to act in a similar way, converts the a- into the ß-, but not the y- into the B-modification. It should be remarked, however, that in all these cases of transformation, the only criterion of the change is the alteration in melting point, and although so far as the work has gone at present these changes appear to be similar in the analogous substances employed, yet further investigations, which are in progress, are necessary before the question can be considered definitely settled and theoretical deductions drawn.

Benzalaniline and Dibenzyl Ketone.

(i) a-Dibenzyl Ketone Benzalaniline. This substance, prepared as previously described, and melting at 164-165°, showed a slight rise in the melting point on repeated recrystallisation from boiling ethyl or methyl alcohol, in which it is only slightly soluble, the maximum rise being from 165° to 168°. If heated in a sealed tube at 150°, complete decomposition into aniline, benzaldehyde, and dibenzyl ketone takes place. When melted, allowed to cool, and then recrystallised from benzene, it showed an unchanged melting point.

a-Hydrochloride, C2H2ON,HCI.-If the a-additive product is dissolved in excess of toluene and dry hydrochloric acid gas is passed in,

the salt crystallises out in quantitative amount. following result was obtained:

HCl found 8.37, calculated 8:53 per cent.

On analysis, the

It is a white, crystalline powder melting at 155.5°, and dissociating with water, giving the a-base melting at 164.5°. If treated with absolute alcohol, dissociation takes place in the cold, but if the solution is heated, the base which has separated out in the form of a white, bulky, crystalline mass gradually passes into solution, complete decomposition having taken place. The base recovered after the treatment with cold alcohol has a much higher melting point, namely, 174°, and on recrystallisation from benzene was found to melt fairly constantly at 172-173.5°. The following combustion showed that it had the same composition as the a-modification melting at 164-165°.

Found C-85.90; H=6·68.

C28H25ON requires C=85·93; H=6.39 per cent.

When this substance is treated with piperidine, the melting point rises from 172-174° to 183-183.5°. The product may be recrystallised from chloroform and light petroleum, and shows an unchanged melting point, but after a few recrystallisations from pure benzene the melting point falls to 178°. It gives a hydrochloride, when treated as previously described, which melts at 158.5°. When this is treated with absolute alcohol, the resulting base melts indefinitely between 174° and 176°. Its behaviour is very similar to that of a mixture of equal amounts of the a- and y-modifications, as is shown in the following table:

(ii) B-Dibenzyl Ketone Benzalaniline.-This modification is obtained by the action of piperidine on the a-form as previously described, and

melts at 174-175°, but if kept for several minutes at a temperature of 163-164°, it slowly melts; also if melted in a tube and then recrystallised from boiling benzene, it melts at 163-164°. The change from the a- to the ẞ-form may also be brought about by means of phenylhydrazine either alone or dissolved in benzene; the resulting substance melts at 174°, and since no change takes place on treating this with piperidine, this modification and that obtained by using piperidine are presumably alike.

B-Hydrochloride. This salt is prepared in a similar way to the a-compound. A determination of the hydrogen chloride gave the following result:

HCl found 8.17; calculated 8.53 per cent.

It is a white, crystalline powder melting at 162°, and is dissociated by water, giving the B-base melting at 174°. On treatment with absolute alcohol, dissociation also takes place; the resulting base melts at 171-172°, but on several recrystallisations this drops to 163-164°, and the product presumably is the a-modification, since on treatment with piperidine the melting point rises to 174°. If the hydrochloride is kept for a few minutes at a temperature of 155°, or the temperature at which the a-salt melts, it also slowly melts.

It

(iii) y-Dibenzyl Ketone Benzalaniline was obtained from either the a- or B-modification by the action of a trace of sodium ethoxide. melted at 181-182°, but when kept at the melting point for several minutes and then recrystallised from benzene, it shows a melting point of 164°, being changed to the a-modification. Piperidine with some difficulty, but not phenylhydrazine, gives the B-form melting at 174°.

==

y-Hydrochloride.-The hydrochloride (HCl = 8:46 per cent.), obtained in a similar manner to the others, appears to be the same as that from the B-modification. This is shown in the following table:

(i) a-Deoxybenzoin Benzalaniline.-As previously noticed, in the formation of the a-modification melting at 154° small quantities of a substance possessing a much higher melting point are formed, but sufficient quantities of this for a satisfactory analysis have not been obtained. If, instead of keeping molecular quantities of the two constituents at a temperature of 60-70° until the additive product separates out, a solvent such as light petroleum, or ethyl or amyl alcohol be employed, insoluble substances of very high melting point are produced; these have not yet been investigated.

a-Hydrochloride. -The additive products do not appear to be so basic as those previously described, and seem capable of forming salts containing varying amounts of hydrogen chloride. If benzene is used as solvent and gaseous hydrogen chloride is passed in, small, colourless needles melting at 104° separate out. These may be recrystallised from benzene and light petroleum, and show an unchanged melting point. Two analyses gave the following results:

HCl found 6.68 and 6.70 per cent.

C27H23ON,HCl requires HCl = 8.82 per cent.
4CHON,3HCI
HCI = 6.77

If this salt is dissolved in chloroform and the solution saturated with hydrogen chloride, on the addition of light petroleum small quantities of large, well-formed, colourless rhombohedra melting at 124° slowly separate. Even these only contain HCl=7-10 per cent.

and rapidly dissociate on exposure to air. The first salt is much the easier to obtain. When seen in bulk, it is a white, crystalline powder which dissociates with water, giving the a-base melting at 154-155°. In an analogous manner to the corresponding a-salt of dibenzyl ketone benzalaniline, it also dissociates with alcohol, the melting point of the resulting base being 158.5-159°, whereas the a-modification melts at 154-155°.

(ii) B-Deoxybenzoin Benzalaniline.-This modification, prepared as previously described, and melting at 164-165°, is transformed into the a-form by the action of heat in a corresponding manner to the dibenzyl ketone derivative. The hydrochloride, obtained in a similar manner to the a-salt, melts at 122.5°, and is dissociated with water, regenerating the B-base melting at 163-164°. A determination of the amount of hydrogen chloride present gave the following result:

HCl found 6.68 per cent.

4C27H23ON,3HCl requires HCl = 6.77 per cent.

(iii) y-Deoxybenzoin Benzalaniline.-This modification, melting between 173° and 174°, like the corresponding dibenzyl ketone derivative, gives a hydrochloride which appears to be identical with that obtained from the B-form. It melts at 123°, whereas the B-salt melts at 122.5°, and on treatment with water gives the ẞ-base melting at 164.5°.

Further experiments are being carried out with these derivatives of deoxybenzoin, and with dibenzyl ketone, using benzal-p-toluidine instead of benzalaniline; it has been noticed that benzal-o-toluidine, under precisely similar conditions, does not form additive products.

CXII. Condensation of Methyl Acetonedicarboxylate. Constitution of Orcinoltricarboxylic Esters.

By F. W. DOOTSON, M.A.

MANY years ago, it was shown by Cornelius and Pechmann that ethyl acetonedicarboxylate, when heated in the presence of sodium, condenses to triethyl orcinoltricarboxylate (Ber., 1886, 19, 1446). More recently, the subject has been investigated by Pechmann and Wolman (Ber., 1898, 31, 2014) and by Jerdan (Trans., 1897, 71, 1106; 1899, 73, 808), and a variety of methods have been described for the preparation of the triethyl ester. The most successful of these methods depends upon the so-called catalytic action of sodium ethoxide, and by