This compound was prepared by me by iodising parabromorthonitrophenol, m. p. 88°, in alcoholic solution with iodine and mercuric oxide. After purification by means of its potassium salt, and subsequent decomposition of the latter and crystallisation from alcohol, it was obtained in yellow scales of a somewhat dirty appearance melting at 104°. The calcium salt forms dark-red plates containing 4 mols. H2O. Analysis:

I. 0-2295 gram lost 0·0204 gram at 190° and gave 0.0397 gram CaSO4.

II. 0.3441 gram gave 0·3652 gram Ag(Br,I).

Of the foregoing calcium salts of orthonitrophenol-derivatives, it is to be remarked that they all pass from a red to an orange colour with the loss of their water of crystallisation; this is at variance with the observations of Carnelley and Alexander, who state (Proc., 1888, 64) that in the metallic derivatives of the nitrophenols the colour passes towards the red end of the spectrum as the water of crystallisation diminishes. It is also noteworthy that calcium dichlororthonitrophenol has less tendency to unite with water than the other calcium dihalogen orthonitrophenol-derivatives; for example, the monohydrate (an orange salt) is formed with ease, whilst the red salt containing 4 mols. H2O is only obtained by slowly evaporating the solution over concentrated sulphuric acid.

As regards the colour of the salts of the nitrophenols, it is to be observed that those from the alkali metals, potassium and sodium, &c., of derivatives of orthonitrophenol are red, whilst those of paranitrophenol are yellow; also that those of ortho-orthodinitrophenol and its derivatives are red, and those of orthoparadinitrophenol are yellow. In a paper by Nölting and Pick, published in a recent number of the Berichte (21, 3158-3160), two isomeric dinitroxylenols, from ortho-xylene, are described; one melting at 127°, having the NO-groups in the two ortho-positions with respect to the OH-group, and the other melting at 82°, and having the NO,-groups in ortho-para-positions with respect to the same. Now, according to the authors, the salts of both these compounds are yellow, and thus it would seem that the introduction of two methylgroups in the place of two hydrogen-atoms brings about a con

siderable modification in the colour in the case of ortho-orthodinitrophenol, removing it further from the red end of the spectrum. Ortho-orthodinitrophenol and its halogen substitution-derivatives melt at a lower temperature than orthoparadinitrophenol and its corresponding derivatives; but this state of things seems to be reversed by the introduction of the two methyl-groups into the molecule, and the derivative of ortho-orthodinitrophenol then melts. at the higher temperature.

XII.-On Berberine. (Part I.)

By W. H. PERKIN, Jun., Ph.D.

BERBERINE, the yellow alkaloïd of Berberis vulgaris, was discovered in the year 1826 by Chevallier and Pelletan (Journ. de Chim. Médicale, 2, 314) in Xanthoxylum clava Herculis, and described by them under the name of xanthopicrit.

Subsequent researches have shown that this alkaloid is a constituent of a large number of plants, in some of which it is present in considerable quantities.

It has been found in Cocculus palmatus, or Columbo root (Boedecker, Annalen, 66, 384; 69, 40); in Berberis vulgaris (Buchner, Ann., 24, 228); in Colline polycarpa (Stenhouse, Ann., 95, 108; 105, 360); in the wood of Coscinium fenestratum from Ceylon, and in Xanthorrisa apii folia, a North American plant of the Ranunculacea order (Perrins, Ann., 83, 276); in Hydrastis canadensis (Mahla, Sill. Am. J. [2], 33, 43 (1862)); another North American plant of the Ranunculacea order, which, as it contains about 4 per cent. of crude alkaloïd, and is readily obtainable in England, can be used as a source of the alkaloid (Perrins, Ann., Suppl. 2, 173).

Berberine has also been found in Woodumpar, a yellow wood from Upper Assam, in the root of Coptis tecta or Mahmira-a plant growing in Hindoostan and China, and which is much valued in India on account of its tonic properties. Perrius (loc. cit., p. 174), who, examined this source, obtained as much as 8 per cent. berberine from the root- -a far larger quantity than has up to the present been found in any plant.

The French chemists Chevallier and Pelletan, the discoverers of berberine, although they carefully and accurately described some of its properties, do not appear to have made any analyses of their

product, and it was not till 1835 that any attempts were made to ascertain the composition of this interesting substance. In this year, Buchner and Herberger (Annalen, 24, 228) published the results of an important series of experiments on this alkaloïd, together with a number of analyses of the base itself, and also of its compound with oxide of silver. These experiments led them to the supposition that berberine was represented by the formula C33H3Ń 2012 (C = 6 : 0 = 8).

At a later date, Fleitmann made a careful examination of berberine, and showed that instead of being a weak acid, as Buchner had suggested, it was in reality a strong base, forming with acids stable salts, and in all respects showing the properties of a well-defined alkaloïd.

Fleitmann analysed not only berberine itself but also several of its salts, and the results of his experiments led him to the conclusion that the formula proposed by Buchner was erroneous, and that the composition of this base was CH18NO, (C = 6 : 0 = 8).

In the year 1861, J. Dyson Perrins (Annalen, Suppl. 2, 176) published the most accurate investigation of berberine which had up to that time been made, in the course of which he analysed a large number of its salts, and showed, in a most conclusive manner, that Fleitmann's formula was inaccurate, and that the true formula was CHNO, (C 12: 0 16).

=

Although the analytical results obtained by Perrins agree so closely with this formula, I still thought it necessary, before starting an investigation on the constitution of this interesting alkaloïd, to repeat some of these analyses with specially purified substances, so as to leave no doubt as to the relation of the carbon, hydrogen, nitrogen, and oxygen-atoms in it. The berberine used in these experiments was purified in the following manner :—

Crude berberine hydrochloride (from Merck in Darmstadt) was recrystallised twice from water acidified with hydrochloric acid, the beautiful yellow, silky needles dissolved in a little boiling water, and the solution rendered alkaline by the addition of an excess of carbonate of soda solution.

On cooling, the liquid became filled with brownish, silky crystals. These were filtered from the dark brownish-red mother-liquor, washed with water till the washings were only slightly yellowish, and dried on a porous plate. The beautiful silky mass thus obtained was next dissolved in boiling 80 per cent. alcohol (carefully purified methylated spirits may also be used) from which, on slowly cooling, long, yellowish-brown needles separated. After repeated crystallisation from dilute alcohol, these crystals entirely lose their original brownish colour, and are at length deposited in the form of long, silky needles of an intensely yellow colour. This substance was repeatedly analysed under varying conditions, but in spite of this no

satisfactory results could be obtained. One reason for this is that the substance contains several molecules of water of crystallisation, part of which is given off slowly at ordinary temperatures and part at 100°.

The substance dried at 100° is, however, never free from water of crystallisation, and if attempts are made to drive this off at a higher temperature, slight decomposition invariably takes place sufficient to spoil the analytical results. Another reason for the difficulty experienced in obtaining concordant numbers with berberine prepared in this way, is that the substance almost invariably contains traces of chlorine.

Experience showed that the nitrate, hydriodide, and platinochloride of berberine were far more suitable for analytical examination than the base itself.

Berberine Nitrate, C20H1;NO,,HNO,.-This salt has been described. by both Fleitmann and Perrins. It is readily prepared by adding an excess of dilute nitric acid to a warm, concentrated solution of the base. The nitrate is thus precipitated as a yellow crystalline mass, which is collected, well washed with water, and repeatedly recrystallised from this solvent. In this way beautiful yellow needles were obtained which, after drying first over sulphuric acid in a vacuum and then at 100°, gave the following results on analysis :—

I. 01724 gram substance gave 00722 gram H2O and 0.3808 gram CO2.

II. 0-1638 gram substance gave 0.0679 gram H2O and 0.3621

gram CO2.

III. 0-2056 gram substance gave 12.5 c.c. N. Bar 760 mm. t =

These numbers agree well with those previously found by Fleitmann and Perrins, but not so well with those given by Henry.

This salt is readily soluble in hot water, sparingly in cold, and

almost insoluble in dilute nitric acid.

If the hot solution of berberine nitrate be mixed with a small quantity of ammonia, on cooling a remarkable compound separates in sparingly soluble golden plates. I have not as yet been able to determine the nature of this substance, but from the analyses already carried out it appears to have the formula C2H20N3O9.

I am at present engaged in the investigation of this substance. Berberine Hydriodide, CHNO,,HI.-This very insoluble salt is obtained as a yellow precipitate when an aqueous solution of berberine is acidified with a few drops of aqueous hydrogen iodide-or when potassic iodide is added to a solution of nitrate of berberine. The salt was collected, well washed with water, dried on a porous plate, and recrystallised from a large quantity of boiling 80 per cent. alcohol.

In this way, beautiful satiny needles were obtained which on analysis gave the following numbers:

I. 0·1460 gram substance gave 0.0523 gram H2O and 0.2767 gram CO2.

II. 0-3561 gram substance gave 01795 gram AgI.

Perrins and Henry, who also analysed this salt, obtained numbers agreeing closely with the above.

Berberine Platinochloride, (CH17NO4)2, H2PtCl-This double salt has, on account of its insolubility and suitability for determining the molecular weight of berberine, been repeatedly examined. It is readily prepared by adding platinum chloride in excess to a slightly acid solution of berberine hydrochloride. The yellow precipitate thus thrown down was collected, well washed with water, dried on a porous plate over sulphuric acid, and then at 100°, and analysed with the following results: