OPTICALLY ACTIVE NITROGEN COMPOUNDS, 223 XXVII1.-Optically Active Nitrogen Compounds. d-. and 1-Phenylbenzylmethylethyla-mmonium Salts, By HUMPHREY OWEN JONES. THE first definite proof of the existence of optically active compounds in which the optical activity is due to asymmetric atoms other than carbon atoms was given by Pope and Peachey (Trans., 1899, ’75,1127), when they resolved Wetiekind’s a-phenyl benzylmethylallylammoniurn iodide into dextro- and {avo-rotatory forms in which the activity is caused by the asymmetzy of the quinqrievalent nitrogen atom. Up to the present time, the salts then described, together with those described later by Pope and Harvey (TIans., 1901, 79, S%), derived from the Same sub-tance, are the only active nitrogen compounds known. The a- and P-phenylbenzylmet hylally lamrnonium salts are also unique i n that they are the only ones which have been shown to exist in definite stable isorrieric foi ms.The phenomena observed iq the phenyl trrethylett-iylHllylnmmonium iodides by Wedekind (Bey., 1903, 36, 3791), if they can be considered as due t o isomerism a t all, are of quite a different order, since the differences vanish when the compounds separate from solution and the amorphous product becomes crystalline. It becomes therefore a matter of importance to determine whether the existence of optical activity is in any way connected with the existeoce of isomerides or whether it is dependent on the presence of any special radicles. The resolution of one of these compounds was undertaken by the author i n order to show that the negative result of the attempts to prepare active forms of salcs of the type NR’R’R,”’X was not due t o the absence of ordinary isomerides or to the particular radicles used.A preliminary notice of the partial resolution of phenylbenzyl- methylethylammonium compounds has already been published (Trans,, 1903, 83, 1419). This subject acquired additional interest after the publication of a paper by Wedekind (Zed. physikal. Chem., 1903, 46, 235), in which unsuccessful attempts to resolve two salts, namely, p-tolyl- benzylmethylallylammonium and p-tolylmethylethylallylammonium d-camphorsulphonates, are described. The question of the existence of isomerides of phenylbenzylethyl- methylammonium iodide was first investigated by preparing this compound in the three possible ways and carefully comparing the properties of the products.2 2 4 JONES : OPTICALLY ACTIVE NITROGEN COMPOUNDS.Pherqlbenna: ylmethy Zethylammoniunz Iodide. (1) A mixture of methylethylaniline (27 grams) and benzyl iodide (43 grams) in molecuIar proportion set almost completely to a solid gum in the course of about two hours. The gummy mass, when dis- solved in rectified spirit, crystallised slowly ; the crystals which aeparated were short, almost colourless prisms melting at 139--140°, the melting point being slightly raised by repeatedly crystallising the substance from alcohol. Unfortunately, with this compound, as with some of the other substituted ammonium salts pi-eviously investigated, the melting point is not a constant, but depends on the rate of heat- ing, and was sometimes found to have been lowered a degree or two when the substance was left for some time in a, stoppered tube, although in this case no visible change had taken place.The carefully purified salt, when slowly heated, becomes slightly brown a t 140°, and melts either at, or just below, 144-145'. If, however, the tube con- taining the salt is not introduced until the bath is at about 140°, the salt may not melt below 146-148'. These differences are apparently due to partial decomposition below the melting point. Hence, in the comparison of melting points described below, two tubes containing the substances to be compared were always heated side by side in the same bath. (2) Benzylethylaniline, which was prepared as described by Fried- lander (Bey., 1889, 22, 58s) by the action of ethyl iodide on benzyl- aniline at looo, boiled at 2S6-288' under 750 mm, and at 190' under 20 mm.pressure. A mixture of this base (14.4 grams) with methyl iodide (10 grams) in molecular proportion gradually deposited the ammonium iodide in a crystalline form, yielding about 1.6 per cent. after three days. The cold alcoholic solution, when diluted with ether, slowly deposited prismatic crystals which melted a t 145' when slowly heated and at 147-148* when quickly heated, or a little higher than the compound produced by method (1). A mixture of the two preparations melted at the same temperature as the product of the second method ; more. over, the crystalline form of the t w o products is the same, go that the two are apparently identical.(3) A mixture of benzylmethylaniline ( I 9.6 grams) and ethyl iodide (15-6 grams) in molecular proportion deposited, with extreme sIowness, a crystalline solid and a little gummy matter, not more than about 2 per cent. of the salt being obtained after ahout a fortnight. The melting pointof the cruEe substance was 135-137', and on recrystallisatlon from alcohol the melting point rose t o 143-144', being identical with that of the product from the first method; a mixture of the two Qelted atD- AND L-PHENYLBENZYLMETHYLETHYLAMMONlUM SALTS. 225 the same temperature, so that they are evidently the same. The quantity of material available was insufficient to furnish crystals large enough for crystallographic examination.Hence no definite stable isomerides can be produced in this way. It is possible that the gummy product produced at first by method (1) and also the amorphous phenylmethylethylallylammonium iodide of Wedekind (Bey., 1903, 36, 379 I ) may represent the untransformed, and, in these cases, the unstable addition product, which, when deposited from solution, undergoes transformation into the more stable form, but which is stable under special conditions, as in the case of certain compounds investigated by Kipping, and the a- and P-phenylbenzylmethylallylammonium compounds of Wedekind (com- pare Trans., 1903, 83, 1405). A specimen dried in a vacuum desiccator was employed in the following analysis : 0,1350 gave 0.2695 CO, and 0.0691 H,O. (2-54.4; H=5*60. C,,H,,NI requires C = 54.4 ; H = 5-66 per cent.Resolution of the d- und l-Cal?z,~horszcZpho.Izutes. The camphorsulphonates were made in the usual manner by boiling the silver salt of the acid with the calculated quantity of the ammon- ium iodide prepared by method (l), and a mixture of ethyl acetate and a little alcohol. I-Camphorsulphonic acid was prepared from I-borneol as described by Pope and Harvey (Trans., 1901, 79, 76) and was found to melt a t 193'; its ammonium salt in aqueous solu- tion gave [MI, = 51.5". The I-camphorsulphonate of the I-base was prepared both from the inactive iodide and from crude 2-iodide obtained from the most soluble portions of the d-camphorsulphonate. The camphorsulphonates crystallise readily, are very sparingly soluble in acetone and ethyl acetate, even when hot, but dissolve quite readily in methylene diethyl ether (ethylal) ; a mixture of this with ethyl acetate was therefore used as solvent.The crystallisation was carried out at a comparatively low temperature, as it was found that prolonged heating induced a distinct reti*ogression in the rotatory power of the salt, most probably due to racemisation. Several crystallisations (6-8) were found to be necessary before the rotatory power of the salts became constant ; the resolution therefore proceeds somewhat slowly and is not effected quite so readily as in the case of the phenyl benzylmethylallylammonium salts resolved by Pope and Peachey (Zoc. cit.). The rotatory power of the salts is small compared with that of the last-mentioned compounds, so that on this account a much larger quantity of material must be worked up (about 30-40 grams of the226 JONES : OPTICALLY ACTIVE NITROGEN COMPOUNDS.camphorsulphonates were employed), so that for this reason, and on account of the slight solubility of the salts and the necessity for keeping down the temperature, the process is somewhat tedious. The amount of the rotation observed was in all cases small even with comparatively concentrated solutions, consequently the experimental error is necessarily large, and great care had t o be exercised in order t o secure the highest degree of accuracy. d-PlienylbenxyZmetl~yZethyZaminorzium d-Cnmpho~*suZphonctte. This salt formed lustrous, prismatic crystals melting sharply at 180-181°; its rotatory power in aqueoussolution wasdetermined several times during the process of fractional crystallisation, and [MID was found to approach a constant value of about 71".A specimen mas then dissolved in a mixture of ethyl acetate and ethylal, the solution allowed to evaporate slowly in a desiccator, and the successive fractions examined. The identity of the two results given below shows, within the limits of experimental error, that the compound is pure. First fraction : 1.086 in 25 C.C. gave aD = 1 ~ 3 5 ~ in a 200 mm. tube," Third fraction : 1.063 in 25 C.C. gave uD= 1-32', hence [a],= 15.52' hence 15.54' and [MJD=714'. and [MI, = 70.94O. 0,1355 gave 0.3385 CO, and 0.0935 H,O. C = 68.1 ; H = 7.7. C,,H,,O,NS requires C = 68.2 ; H = 7.6 per cent. 1-Phenylbenxylmet~yleth~larnmoni.um 1-Campliorsulphonacte.The salt has exactly the same properties as the corresponding dd-salt, crystallising in similar lustrous prisms and melting a t 180-181O ; its molecular rotatory power in aqueoas solution gradually became constant at about 71'. A specimen was then dissolved in a mixture of ethyl acetate and ethylal, the solution allowed to evaporate slowly in a desiccator, and the rotatory power of the successive fractions examined. The practical identity of the numbers obtained with one another and with those already given for the dextrorotatory isomeride shows t h a t the salt is homogeneous. First fraction: 0.583 in 25 C.C. gave aD = -0*73O, hence [aID = Second fraction : 1.027 in 25 C.C. gave aD = - 1*27O, hence [ alD = - 15 6' and [ MID = - 71.5". - 15.46' and [MID = - 7 0 - 7 O .* All the following determinations of a, were carried out in a 200 111111. tube.D- AND L-PHENYLBENZYLMETHYLETHYLAMMONIUM SALTS. 227 Third fraction : 1.076 in 25 C.C. gave a,, = - 1-33', hence [ u ] D = Fourth fraction : 1.002 in 35 C.C. gave a, = -1*27', hence [ a ] ~ - 15.45' and [MI, = -70.6'. - 15.8' and [MID = - 7 2 . 3 O . 0.1930 gave 0.4829 CO, and 0-1330 H,O. C26E,,0,NS requires C: = 68.2 ; H = 7.6 per cent. The molecular rotatory power of the d-phenylbenzylmethylethyl- ammonium d-camphorsulphonate is therefore + 71.0' (mean), and t h a t of the corresponding Zl-salt is -71.2' (mean). Since the molecular rotatory powers of the d- and I-camphorsulphonate ions are respectively +51.7' and -51*6O, the molecular rotatory powers of the a?- and 2-phenylbenzylmethglethylarnmonium ions are + 1 9 .3 O and - 19.6' respectively. The value for the d-phenylbenzylmethylethyl- ammonium d-camphorsulphonate agrees fairly well with that already given, namely, + 6 9 O (20~. cit., 1419). The molecular rotatory powers of the d- and I-phenylbenzylmethylallylammonium ions are + 166.4' and - 159' respectively (Pope and Harvey, Zoc. cit.). The remarkably low value obtained for these compounds, which only differ from those last mentioned by the introduction of an ethyl instead of an ally1 radicle, is therefore very surprising. It is dificult to see why such a small difference as this could produce such an enormou8 change in the rotatory powers were it not that it has been repeatedly shown that an ethylene linking exerts a very marked influence in increasing the rotatory power.An attempt was made t o confirm these values by the examination of the bromocamphor- sulphonates, but unfortunately these have not been obtained crys- talline. Even when the d-iodide of the base, obtained from the d-cam- phorsulphonate, was converted into the d-bromocamphorsulphonate, the salt still remained a gum and could not be induced to crystallise. C = 68-2 ; H = 7'66. d-PhenyZbenzylnzethylet~~~Za~n~onizLnz Iodide. The d-iodide of the base was obtained in a crystalline form by add- ing the calculated quantity of a concentrated aqueous solution of potassium iodide to fin aqueous solution of the d-camphorsulphonate. The crystalline salt thus obtained, dried in a vacuum desiccator, was found to melt at practically the same temperature as the inactive iodide, admixture with which did not appreciably change the melting point.The specimen for analysis was crystallised from cold alcohol in the dark, 0.1334 gave 0.2655 CO, and 0.0700 H,O. C = 54.28 : H = 5-82, C,,H,,NI requires C = 54.39 ; H = 5.66 per cent.228 JONES : OPTICALLY ACTIVE NITROGEN COMPOUNDS. The determinations of the rotatory power of the iodide offer con- siderable difficulties. The salt has a very small rotatory power and is very sparingly soluble in alcohol, so that, even with practically saturated solutions, the rotations observed were very small, for example, 0*3-0*4*, and hence the experimental error is very large. Chloroform dissolves a little more of the salt, but the solution shows a peculiar supersaturation phenomenon, the salt first dissolves and then partly crystallises out, in most cases rapidly, but sometimes more slowly, leaving a solution which gives a rotation of about the same magnitude as that of an alcoholic solution. Racemisation also occurs slowly in a chloroform solution, and consequently its use was abandoned.Several determinations were made in alcoholic solution at a concen- tration of about 2 per cent., which represents a nearly saturated solution at the ordinary temperature, The following data represent three of these determinations : 0.5620 in 25 C.C. absolute alcohol gave (XD = 0~37', hence [a], = 8023~ and [MID = 29.0'. 0.5455 in 25 C.C. gave aD = 0*36O, hence [a]D = 895' and [MI, = 29.1'. 0.5560 in 25 C.C.gave aD = 0.38', hence [aID = 8.5" and [MID = 30-1°, It may therefore be concluded that thevalue of [a], for the d-iodide is about 8*3O, although no very great reliance can be placed on these numbers owing to the very large experimental error. I- Phenylbcnxybnaeth yleth ybanznzonium Iodide. This iodide was prepared from the corresponding camphorsulphonate in the manner already described for the biodide. The salt, after re- crystallisation from cold alcohol in the dark, was found to melt, when slowly heated, at the same temperature as the I- and d-iodides, and mixtures of the I- and inactive iodides also meltad a t the same temperature. The following analysis was made on a specimen dried in a vacuum desiccator. 0.1862 gave 0.3701 CO, and 0-0975 H,O. The rotatory power was determined in alcoholic solution in the same 0.547 in 25 C.C.absolute alcohol gave (XD = -0*38', hence [ a ] , , = C =54.2 ; H = 5.81. C1,,H2,NI requires C = 54-39 ; H = 5-66 per cent. way as that of the d-iodide, with the following results : - 6-68' and [MAD = - 30.8'.D- AND L-PHENYLBENZTLMETHYLETKYLAMMONIUM SALTS. 229 0.514 in 35 C.C. absolute alcohol gave all = - 0*34', hence [u]D = - 8.27" and [MID = - 29.2'. 0.526 in 25 C.C. absolute alcohol gave uD = - 0.35', hence [ a ] D = - 8-33" and [MI,, = - 29.4'. The specific rotatory power of the 2-iodide in alcoholic solution is therefore numerically identical, within the limits of experimental error, with that of the corresponding d-salt, namely, about - 8.4'. Automcenaiaation of the Phern ylbenxy Imeth ylet hybammonium Xn Its.The camphorsulphonates retain their rotatory power practically unchanged for weeks in aqueous solution and also in cold alcoholic or ethylal solutions. If, however, the solutions are heated, the aqueous solution becomes turbid, the rotatory power of the other solutions diminishes, and the salt deposited from them has a distinctly low rotatory power. I n one experiment, a specimen of d-phengl benzyl- methylethylammonium d-camphorsulphonate,.having [ MID = 71*4', was recrystallised from a hot mixture of ethyl acetate and acetone, the solution being heated for about 10 minutes in order to bring all the salt into solution; the salt which was deposited on cooling had a much lower rotatory power, namely, [MI,= 56.0'. Boiling in ethyl acetate solution had therefore caused almost complete racemisation of the basic part of the molecule.In the fractional crystallisation of these salts, it is consequently necessary to avoid heating the solutions, and in practice the tempera- ture was usually not rjisecl above about 40'. The solutions of the iodides in alcohol retain their rotatory power in the cold for a long time, especially when left in the dark, On heating, however, the rotatory power diminishes, so that when recrystallising theso com - poands from alcohol the process must be carried out at as low a tetuperature as possible. I n chloroform solution, racemisation takes place slowly in the cold and in absence of light. On one occasion, a fairly strong chloroform solution of the d-iodide(not quite pure), when put into the tube, did not deposit the excess of iodide as all other chloroform solutions did, and was observed to be in the supersaturated state for about a week.The rest of the solution in the flask soon deposited crystals just as the other chloroform solutions had done. 0.859 in 25 C.C. chloroform gave ~,=0*64", hence [a],= 9-31' ; The rotatory power diminished gradually at first a t the rate of about Onlo per day, until it became practically inactive after a little more than a week. These salts therefore behave much in the same way as [M ]D= 33*8'.230 JONES : OPTICALLY ACTIVE NITROGEN COMPOtJNDS. the a-phenyl benzylme thylallglammonium salts investigated by Pope and Harvey (ZOC. cit.). The effect of chloroform in causing racemisation may be due, as suggested by these authors, to a dissociation of the ammonium salt into the tertiary amine and alkyl iodide and subsequent recom- bination to form the quaternary compound.This hypothesis is sup- ported by an experiment made by the authors, which showed that the rate of formation of the salt in chloroform solution is large compared with that in alcohol or ether. Wedekind (Zeit. physikal. Chem., 1900, 6, 23), ondetermining the molecular weight of the iodide in chloro- form solution by the ebullioscopic method, obtained values about one- third of the calculated value indicating dissociation of the salt into its constituents. It is difficult to see how the iodide could dissociate into three sub- stances, the results therefore seem t o indicate that the ebullioscopia method is inapplicable here.Moreover, the slow rate of racemisation indicates dissociation to a slight extent only ; were the dissociation complete, the racemisation ought to be instantaneous. The molecular weights of phenylbenzylmethylallylammonium iodide a n d phenylbenzylmethylethylammonium iodide in chloroform solution were determined by Mr. G. Barger, B.A.., of King’s College, who used his microscopic method. The author is glad t o take the opportunity of expressing his thanks t o Mr. Barger for the care and trouble which he expended on these determinations (compare this vol., p. 286). Phenylberte yZniet?~yZet?~y Zanimoniuna Iodide. A solution containing 23.9 grams per litre was found t o be isotonic with a solution of triphenylmethane containing 0.075 gram-molecule per litre, whence the molecular weight is 319.A solution containing 25.1 grams per litre was found to be isotonic with a solution of azo- benzene containing 0.0725 gram-molecule per litre, which gives a molecular weight 346. The calculated molecular weight is 353, so that at the ordinary temperature the molecular weight is approximately normal. a- PhenyZ6enxyZmeth~ZaZZ~Za~~mmcnium Iodzde. This salt, being much more soluble in chloroform than the fore- going compound, gives better results. A solution containing 59.0 grams per litre was foiind t o be isotonic with a solution of triphenyl- methane containing 0.1 6-0.17 gram-molecule per litre. Hence the molecular weight of the iodide lies between 335 and 381 (mean 358). The calculated value for CI7H,,,NI is 365. So that the molecularD- AND L-PHENYLBENZTLMETHILETHYLAMMONIUM SALTS. 231 weight at the ordinary temperature gives no indication of extensive dissociation. A peculiar phenomenon was observed in t h e case of both salts, which was, however, more marked with the phenylbenzylmethylallyl- ammoniurri iodide than with the other.I n the freshly prepared solution, the molecular weight always appeared slightly too high ; it then gradually diminished, became normal, and went on diminishing until it was somewhat too low. The rate of diminution was much increased by raising the temperature. This behaviour is being investigated, as it may throw light on the cause of the racemisation, and is probably the cause of the anomalous results obtained by the ebullioscopic method.The above results, which show that in both cases there is only a very slight dissociation of the salts in chloroform solution at the ordinary temperature, thus account for the slow rate of race- misa tion. i-P~cnyZ~enxyZmetl~yZethyZammonium Bromide.-This salt was prepared from the iodide by digesting an alcoholic solution with silver bromide. It crystallises from alcohol in prisms closely resembling those of the iodide, which melt sharply a t 155-156'. The rneltiug point of the bromide does not depend on the rate of heating in the same way as the iodide. l-PhenyZ6enzyEmetl~yEethyZammonizLm bromide is not precipitated from the aqueous solution of the camphorsulphonnte by the addition of a concentrated aqueous solution of potassium bromide, and consequently had t o be prepared by the same method as the inactive compound.It crystnllises from alcohol in long prisms melting at 155-156'. A mixture of the I- and i-bromides also melts sharply at the same temperature. A determination of the rotatory power in alcoholic solution was made with the following result : 0.634 in 25 C.C. gave a, = -0668O; hence [.ID = -1304' and The rotatory power is greater than that of tbe corresponding iodide, just as the rotatory power of the active a-phenylbenzylmethylallyl- ammonium bromide is greater than that of the iodide. The rotatory power of the bromide diminishes on recrystallising from alcohol. C = 62.36 ; H- 6.4. [MI, = - 4101~. 0.212 gave 0.4850 CO, and 0.1200 H,O. C,,H,,NBr requires C = 62.7 ; H = 6.53 per cent.232 JONES : OPTICALLY ACTIVE NITROGEN COMPOUNDS.CrystaEZine Form of the d-, 1-, and i - P ? ~ e n y l b e n , y ? / l i , a e t h y l e t ~ ~ l a ~ ~ ~ Iodides. The crystallographic examination was undertaken in order to decide whether the inactive form was a racemic compound or merely an inactive mixture. These salts crystallise from alcoholic solutions in beautiful, lustrous prisms, the faces of which, although appearing quite bright, nevertheless give bad reflections on a goniometer, so that i t was found difficult t o get good measurements. The crystals could only be obtained of very small size, 1-2 mm. i n length, and were therefore troublesome to measure. The dominant form mas always the prism nz(llO), the pinacoid FIG. 1. I u5- ~(100) was sometimes present as a very narrow face, but b(010) was never observed.Two distinct types of crystals mere observed, see Figs. 1 and 2, one sort exhibited the domes e(011) and n(012) with the basal plane c(OOl), often as a mere line, arid o( 11 2) was sometimes present; the other kind showed a well-developed basal plane and no domes, but o(112) was always present, and cc{lOO) was occasionally observed. No general forms were present, 80 that the question of hemi- hedrism could not be definitely settled. From analogy with the a-phenyl- benzylmethylallylammonium salts and the phenylmethylethylallyl- ammonium salts which belong t o the sphenoidal class of the prismatic system, it is probable that this salt also belongs to the same class;D- AND L-PHENYLBENZYLMETHYLEF€iYLAMMONIUM SALTS.233 this is supported by the scanty evidence which was obtained by examining the etched figures produced on the faces m(110) and m”’(lT0) by dilute alcohol. The few definite figures observed were devoid of a plane of symmetry, and those on rn and m”’ were inter- changeable by rotation about the dyad axis. The crystals of the d-, I-, and i-salts are similar in habit and appearance and have identical angles, so that the inactive salt is either an inactive mixture or a pseudoracemic compound. The fact F I G . 2. of the melting points of mixtures of d- or h a l t with the i-salt being the same as that of any one of the salts also supports this conclusion. Crystalline system. Prismatic. Sphenoidal Class (1). a : b :~=0*7456 : 1 : 1.1409. Forms observed : a(100), mfllO>, cfOOl), e(011), 4 0 1 2 ) , and 01112). The following angular measurements were made : ! Number of observations.Angle. I--___ ; -- m. m’”= 110 : i i o ’ 20 rnm’ = i i o : i i o 19 a m =100:110 I 4 c c =001:1)11 1 10 c 7a = 0 0 1 : 0 ~ 2 10 c e’ =011:011 6 n V! =012 : oi.2 4 F n. =011:012 1 8 c ‘)tf =Oil :012 I 7 m n =110:012 1 3 m”n = i i o : o i 2 I 3 c 0 =001:112 1 9 na 0 =110:112 I 10 Limits. - 0 13 0’- 73”55’ 106 0-10850 36 20-- 36 55 48 35 - 48 56 29 1s - 29 56 97 14- 97 46 58 50- 59 50 18 25- 19 20 72 40 - 73 30 106 39 -107 20 20 20 - 21 19 69 0 - 69 30 78 10- 78 56 hi e m . 106 34 1 36 40 48 46 ~ 29 37 1 97 32 1 59 12 19 0 78 28 78 13 / 106 56 20 56 i 69 14 I 1 Calculated. I 106’35‘ ! 36 37 I 97 32 ’ 59 34 19 4 ~ 78 28 78 3 ‘ 106 57 1 6 9 8 20 58234 SUDBOROUGH AND ROBERTS : The author desires to express his thanks t o Mr. A. Hutchinson for kindly placing at his disposal the goniometer with which the above measurements were made. It may therefore be concluded that (1) the ammonium compounds of the type NR’R”RR””X can be resolved into enan tiomorphously related optically active forms, although the resolution may often be difficult owing to the feeble rotatory power of the salts aud the ease with which racemisation takes place, and that (2) the existence of optical activity is in’dependent of the existence of ordiuary isomerides of the salt. Racemisation of the active salt,s takes place very readily on heating or on allowing Folutions in chloroform to remain in the cold ; the cause of the racemisation is probably dissociation into tertiary amine and alkyl iodide, followed rapidly by recombination, as suggested by Pope and Harvey. The expenses of this invest,igation have been met by a grant from the Government Grant Committee of the Royal Society, for whtch the author is glad to make this acknowledgment. UNIVERSITY CHEMICAL LABORATORY, CAMBEIDGE.