during 5 hours. The product, when cold, being very viscous, was mixed with a small proportion of ethyl acetate, and this precipitated a white solid which, when filtered and drained on porous earthenware, weighed 175 grams. Extraction with 500 c.c. of boiling ethyl acetate removed a substance which crystallised from the cooling liquid in snowwhite needles, melting and evolving gas at 241°; more of this compound, which contained 7.8 per cent. of nitrogen, was obtained on extracting the residue with a further quantity of ethyl acetate, but the specific rotatory power being [a], -30.5°, it was evidently a different substance from the one obtained in the first experiment. The residue, after being extracted repeatedly with boiling ethyl acetate until no more soluble matter could be removed, was hydrolysed with alcoholic potash, yielding a base which melts at 184°, and gives [a] - 43·7° in a 4 per cent. solution in absolute alcohol, and [a] -27.0° in a benzene solution of the same concentration. We regard this substance as pure neobornylamine, but are unable to identify either the insoluble compound which yields it on hydrolysis, or the substance melting at 241°, because the base obtained on hydrolysing the latter derivative melts at 176°, and gives [a]D in absolute alcohol, appearing therefore to be a mixture of bornylamine and neobornylamine.

It will be noticed that the specific rotatory power of neobornylamine, obtained by the method indicated, is 12° higher than that of the specimen prepared previously, and the hydrochloride of the purified base has [a]-44-2° instead of [a] - 39′0°. It will not be necessary, however, to revise the description which has been already given of the crystalline acyl derivatives of the base. The acetyl compound, for instance, obtained from the specimen having the higher rotatory power, crystallises from petroleum in long, slender needles melting at 144° and having [a] 19·1° in a 2 per cent. solution; moreover, this derivative, and the benzoyl compound, when hydrolysed, yield the base which gives [a] -43.7°.

Having several grams of this material at our disposal, we converted a portion into the benzylidene derivative, which was obtained as a pale yellow, viscous oil boiling at 180° under 25 mm. pressure. The substance does not crystallise when cooled, or after a long interval, and its oily nature accounts for the fact that the specimen of benzyl. idenebornylamine obtained by Griepenkerl was liquid (Annalen, 1892, 269, 353), whilst the purified base is a crystalline solid (Forster, Trans., 1899, 75, 1152). Benzylideneneo bornylamine does not yield a crystalline methiodide when heated in a sealed tube with methyl iodide, and consequently it has not been possible to obtain methylneobornylamine from it.

It cannot be claimed that the process here indicated for isolating

neobornylamine is very successful, because it can be applied only to mixtures having a specific rotatory power not less than [a] - 10°, and therefore already rich in the lavorotatory base. An attempt to deal with a mixture of approximately equal quantities of the two bases led to the isolation merely of two crystalline substances, which differ from those already described. The procedure was substantially the same as before, the product of the action of ethyl oxalate on the base being exhausted twelve times with 1000 c.c. of boiling ethyl acetate. The solid deposited by the first extract differed quite perceptibly from the eleven succeeding fractions, which approximated to one another in appearance, melting point, and specific rotatory power. It was therefore dealt with separately, being dissolved in boiling absolute alcohol, which deposited compact, transparent crystals with a small proportion of long, silky needles; the latter, being specifically light, were easily separated by mechanical means, and the remainder was then recrystallised until the specific rotatory power of two successive fractions was the same. The compound thus obtained appeared a perfectly definite substance, and analysis pointed to the empirical formula of dibornyloxamide :

0-1868 gave 0.5010 CO2 and 0·1717 H2O. C=73·14; H=10·21. 0.2624 18.4 c.c. of nitrogen at 19° and 760 mm. N = 8.23. C22H6O2N2 requires C=73·33; H=10·00; N=7·77 per cent.

In spite of this result, however, the appearance and properties of the substance were totally distinct from those of dibornyloxamide. It crystallised from petroleum in well-formed, transparent prisms melting at 100°, instead of lustrous leaflets melting at 140°; alcohol deposited it in large transparent crystals, which become opaque on exposure to air, instead of long, silky needles. Moreover, the specific rotatory power was higher, being [a] -36-4° in a 4 per cent. solution in absolute alcohol. Yet, although these crystals had every characteristic of a definite substance, the base obtained on hydrolysis melted at 164°, and gave [a] +30.7°.

The substance associated with the foregoing compound was isolated by recrystallising together all fractions subsequent to the first, until the melting point underwent no change. In this way, long, silky needles were obtained, melting and evolving gas at 251°. No conclusion regarding the composition of this compound could be drawn from analysis. Several concordant determinations of carbon gave nearly 4 per cent. less than the amount required by the formula for dibornyloxamide; the quantities of hydrogen and nitrogen, however, were in agreement with those required by that expression. In spite of this, the base obtained on hydrolysis was practically pure bornylamine.

Although several hundred grams of the mixture were manipulated

in one experiment, no derivative of neobornylamine could be isolated. The substance remaining undissolved after twelve successive extractions with 1000 c.c. of boiling ethyl acetate was not a definite compound, as the base resulting on hydrolysis gave [a] +23°. We had therefore convinced ourselves that it is hopeless to deal by this method with mixtures of bornylamine and neobornylamine, unless the latter base preponderates considerably, and further experiments in this direction were therefore abandoned.

ROYAL COLLEGE OF SCIENCE, LONDON.
SOUTH KENSINGTON, S. W.

CVI.-Aminoamidines of the Naphthalene Series. (Third Communication on Anhydro-bases.)

By RAPHAEL MELDOLA, F.R.S., and LEWIS EYNON, A.I.C. THE azo-derivatives of ethenyltriaminonaphthalene were incidentally referred to in a former communication by one of the authors and Percy P. Phillips (Trans., 1899, 75, 1016). In continuing the investigation of these compounds we had occasion to prepare some ethenyltriaminonaphthalene by the process of Markfeldt (Ber., 1898, 31, 1174) for comparison with the original anhydro-base described in 1887 (Meldola and Streatfeild, Trans., 51, 691). It was found to our surprise that the phenylazo-derivatives obtained from the bases by combination with phenyldiazonium salts were isomeric and not identical, and a further examination of the bases has revealed the fact that the isomerism of the azo-derivatives is due to the isomerism of the ethenyltriaminonaphthalenes which are not identical as assumed by Markfeldt in the paper referred to. This isomerism of the anhydro-bases is certainly unexpected, and is the more remarkable since they are both prepared from the same dinitroacetnaphthalide, the only difference being the nature of the reducing agent, Markfeldt having substituted iron and hydrochloric acid for tin and this acid, as originally employed in 1887. That the isomerism is entirely due to the nature of the reducing agent has been proved by repeated experiments with the same specimen of dinitroacetnaphthalide.

In the present communication, we give a detailed account of the results of our investigation of Markfeldt's base and some of its salts and azo-derivatives, so as to bring into prominence the differences between these and their isomerides obtained from the original ethenyltriaminonaphthalene of 1887. With respect to the nature of the

isomerism, we have come to the conclusion that it is most probably due to structural difference and not to space configuration (stereoisomerism), since the characters of the two sets of compounds are profoundly different. As was indicated in 1887, two constitutional formulæ are possible :

Which of these represents our original base and which Markfeldt's isomeride must for the present remain undecided. The main point established by the present research is that tin and iron behave differently as reducing agents towards 2 : 4-dinitro-1-acetnaphthalide. The constitution of the isomeric bases will form the subject of future investigation, and it is proposed to extend the study of the difference in the behaviour of the two metals as reducing agents to other nitro-compounds, since evidence has already been obtained that this peculiarity is not manifested only in the naphthalene series.

Meldola and Streatfeild's anhydro-base is so soluble in water that it has not yet been isolated (Trans., 1887, 51, 698). Markfeldt's base is much less soluble, and separates out as an oily resin on adding sodium hydroxide solution to a strong aqueous solution of the hydrochloride. The resinous mass becomes crystalline on standing, and the free base can be crystallised from hot water, in which it readily dissolves. The pure compound is probably white, but all our preparations consisted of flat, slate-coloured needles, the colour being no doubt the result of atmospheric oxidation. The melting point of the base, allowed to dry in air at the ordinary temperature, is 84-85°. On analysis:

0.1799 gave 0.4072 CO, and 0·1016 H2O. C=61·73; H=6·28. 0.1178 17.95c.c. moist nitrogen at 12.2° and 765 mm. N=18·17. 0.2408, air-dried, lost 0·0354 H2O in water-oven. H2O=14·70.

C12H11N,2H2O requires C = 61.80; H=6·44; N = 18.03;
H2O=15.45 per cent.

The water of crystallisation is not quite driven off at the temperature of the water-oven. The anhydrous base begins to soften and

shrink at 128° and fuses at 135°. Analyses of a specimen, dried as above, gave the following results:

0.1638 gave 0.4386 CO2 and 0·0896 H2O. C=73·03; H=6·08. 0.1112 19.9c.c. moist nitrogen at 12.2° and 773-7 mm. N = 21.58. C12HIN, requires C=73·10; H=5·58; N=21·32 per cent.

11 3

This aminoamidine is more distinctly basic in character than its isomeride. It gives a precipitate with silver nitrate in ammoniacal solution, but it is not sufficiently acid to dissolve in aqueous sodium hydroxide. The silver salt is gelatinous, and could not be obtained pure enough for analysis.

Salts of Markfeldt's Base.

Hydrochlorides.-The salt prepared by Markfeldt's process, namely, by crystallisation in presence of excess of hydrochloric acid, is a dihydrochloride of the formula C12H11N3,2HCl, H2O.

It consists of pinkish, flat, silky needles, the colour being doubtless the result of atmospheric oxidation. The specimen used for analysis was dried over lime in a desiccator, and had lost some hydrogen chloride under these conditions:

0.2435 gave 0.2322 AgCl. Cl=23.59.

0.1385

17.05 c.c. moist nitrogen at 11.8°and 748·1 mm. N=14.38. C12H11N3,2HCl, H2O requires Cl = 24·65; N=14·58 per cent.

11

The crystals become opaque on heating in the water-oven, and at 150-160° (in an air-oven) the water and 1 mol. of hydrogen chloride are driven off, leaving a monohydrochloride :

0.210, dried over lime in a desiccator and heated at 150-160° until constant in weight, lost 0.039618.86 per cent. The loss corresponding to C12H11N3,2HCl, H2O – (HCl + H2O) = 18.92 per cent.

2

The monohydrochloride can also be prepared from the dihydrochloride by the partial removal of hydrogen chloride by means of dilute cold aqueous ammonia. The first specimen prepared in this way was analysed under the impression that it was the free base, and the results showed that only half the acid is removed under these conditions:

0.2926 gave 0.6187 CO2 and 0·1516 H2O. C=57·67; H=5·76. 0.0886 12.3 c.c. moist nitrogen at 11° and 766 2 mm. N=16·67. 0.5503 0-3260 AgCl. Cl = 14.66.

The above specimen had been crystallised from water from which it separates as small, opaque, white needles melting at about 282° and