Xug. 20, 1956
~ 0 M h I U " I C A T I O N STO THE EDITOR
TABLE I ISOTOPIC DISTRIBUTION PATTERNS Number of D atoms
Addition t o 1-hexene
42O 0
1 2 3 4 3
6 7
0.6% 4.9 89.5 4.3 0.7
150"
1.9% lL4 d3 6
Exchange with hexane
801"
95.6% 4.2 0.2
19.7 5 .8 1.7 0.8
8
9 10 11 12 13
.
11
.3 .2
350 73 3 17.6 2.7 0.82 ,72 .80 .89
.84 .69
.57 .-I2 .29
.22 .10 . 02
11
417 1
X simple modification of the cell reported for fractionation purposes' adapts it for analytical work with 5-10mg. of substance. In this the inside glass tube nearly displaces all the volume in the sacZ and leaves only space for a thin film of solution. Stirring with nitrogen is now not required and a faster rate of dialysis is obtained by the resulting increase in the membrane area with respect to the inside solution volume. The outside solution can be replaced by fresh solvent at arbitrary intervals and analyzed for determination of the escape rate. Under these conditions the escape rate expressed as decrease in precetitage of the original remaining in the sac with time should be independent of the amount taken provided a single solute is present which behaves ideally. A plot of the logarithm of the percentage against time should give a straight line. This has been found to hold experimentally with many solutes. Thus in Fig. 1, curve 1 gives the escape rate of bacitracin d (solvent is 0.1 S acetic acid), curve 2 that of aspartic acid.
At 15O0, however, the distribution pattern is broader (Table I) and somewhat resembles patterns observed with nickel catalysts. Intermediate 10 temperatures yield intermediate patterns. 9 At O o , ethylene gives ethane-ds, but, a t - i s o , 8 2 ? reaction is incomplete and the product is ethane-dz and undeuterated ethylene. x6 Table I presents distribution patterns for iso3 5 topic exchange between hexane and deuterium. c 4 .Correction for multiple adsorption shows that only M .f 3 a single deuterium atom is exchanged upon adsorpr: tion of hexane. The multiple exchange of low inc tensity a t the higher temperature probably results z z from dehydrogenation to hexenes followed by re\ hydrogenation. Metallic catalysts give quite different patterns characterized by extensive mul. tiple e x ~ h a n g e . ~ ~ ~ 20 40 60 80 100 120 140 Acknowledgment.-This investigation is supMinutes. ported by the Petroleum Research Fund of the F1g 1 *\rnerican Chemical Society. Mr. S. Xeyerson of the Standard Oil Company (Indiana) assisted us For the separation of mixtures greatly improved with mass spectroscopy. selectivity can be obtained by choice of solvent, the particular size of tubing, treatment of the (4) S. 0. Thompson, J . Turkevich and A . P. Irsa, T H I SJ O U R K A I . , 73, 5213 (1951). membrane, etc. As an example curves 1 and 2 are ( 5 ) R. L . Burwell, Jr., and W. S . Briggs, ibid., 74, 5OY6 (1932); escape rates respectively of aspartic acid and baciH. C. Rowlinson, R. L. Burwell, Jr., and R. H. Tuxworth, J . Phrs. tracin A (mol. wt. 1421) through single membranes Chem., 59, 225 (I955), of Visking 15/32 seamless cellulose tubing while 3 (M) C. Kernball, P Y C CR o y . Soc i L o i z d o n ) , 2 2 3 8 , :I61 (1954). and 4 are those found for double membranes (one DEPARTMENT O F CHEMISTRY sac inside the other). NORTHWESTERN UNIVERSITY ROBERTL. BURWELL, JR. EVANSTON, II,I,INOIS A S T H O N Y B.LITTLEWOOD That the improved selectivity holds for a mixture is shown by comparison of Figs. 1 and 2 . RECEIVED J U N E 21, 1956 In Fig. 2 a mixture of 10 mg. each of bacitracin and aspartic acid was studied. Curve 1 is a weight FRACTIONAL DIALYSIS WITH CELLOPHANE curve. Curve 2 is an optical density curve ( 2 5 5 MEMBRANES mp) referring only to bacitracin. Curve 3, that Sir. of aspartic acid, can be calculated from the values Preliminary data' concerned with fractional in 1 and 2. The significance of this general approach for dialysis with cellophane indicated the possibility of unexpected selectivities for the separation of studying homogeneity with respect to size and/or mixtures of dialyzable solutes. Further study has shape (and perhaps charge) under extremely mild confirmed this indication and shown that consider- conditions is self evident. Comparison with simiable improvement in selectivity, adaptability and lar solutes of known size gives a good estimation of interpretability is possible particularly with solutes size for the unknown. -Although different sizes of Visking have shown of larger molccular size which scarcely pass through the membrane. great variation in selectivity and permeability .r
' ~2
(1) I.. C . Craig a n d T. 1'. King,
T H I S JOURPAL,
77, BtiSO il95.j)
( 2 ) \V H Seegars, J L a b Clrii .\fed, 28, 897 ( I O U )
1172
COMMUNICATIONS TO THE EDITOR
Vol. 78
Merritt and Lanterman’s dimethylglyoxirne paper contains an obvious error, since the latter authors stated that both of the above angles equal 75.9’, an acit”QciJ7 x 6 impossible situation since the two oxime groups are related by a center of symmetry; Dunitz and =uF! 5 Robertson then assumed, apparently by inspection of the published projection of the structure on (OOl), that the angle N-0. . .N’ was smaller and equal to 75.9’) and that the angle 0-N. . .O’ was its supplement, and therefore closer to that expected for the covalent bond. Unfortunately, there is yet another error, for the published parameters give instead for these angles 85’ for N-0. . .N’ and 95’ for L 0-N. . .O’, and these are close enough together to a“ 1’ ’ 20 ’ 40 ’ 60 ’ BO ’ I b O I h O 110 cause the argument to lose considerable force. Minutes. Additional information relative to this question Fig. 2. has appeared recently, namely, the results of the different membranes prepared from the same roll crystal structure of formamidoxime.6 For this appear remarkably reproducible. The most per- molecule the situation is somewhat more commeable one thus far studied (No. 20/32 “Dialysis plicated because there are more than just the two t ~ b i n g ” has ) ~ been found to pass insulin, ribonu- tautomeric structures possible. Nevertheless, declease, lysozyme and chymotrypsinogen in 0.1 N tailed considerations show that the hydrogen bondacetic acid a t characteristically different rates ing in this crystal is consistent with only two struc(50y0 escape times in same cell are 1.6,3.5, 3.5 and 5 tures, I : NH2-CH=N-OH, and I1 : -NH-CH = hr., respectively). The membrane permeability N-+OHz. (Each of these has more than one resocan be increased by mechanical stretching so that nance form.) The structure NHz-CH=+NH-Ois in particular eliminated because both atoms ovalbumin readily passes. A full account of this work will be reported soon. which form hydrogen bonds with the oxime nitrogen atom lie so far from the molecular plane that (3) Light and Simpson, Biochem. Biophrs. Acta, 20, 261 (1956), this atom must be the acceptor atom in these two found t h e 20/32 size t o pass insulin. hydrogen bonds. Structure I1 above, may be reLABORATORIES OF THE LYMANC. CRAIG jectede on the ground that the relative electroROCKEFELLER INSTITUTE FOR MEDICAL RESEARCH XErv YORK,N. Y . TE PIAOKING negativities of oxygen and nitrogen will render both of its resonance forms unstable. We are thus left RECEIVED JULY 5 , 1956 with Structure I, and the observed bond lengths indicate that the formamidoxime molecule is best represented as a resonance hybrid, the predominant THE STRUCTURE OF OXIMES forms being NH2-CH=N-OH and +NHz=CHSir : Pitt,’ in a review article, has suggested that N--OH, which contribute about equally, and perhaps a small contribution of the form NHz--CHoximes should be represented as RzC=+NH-ON=+OH. rather than, as ordinarily written, RzC=N-OH. At present, therefore, it appears that the usual However, the weight of more recent evidence, set forth below, favors the classical formulation of oxime structure, R2C=K-OH, is correct, but obRzC=N-OH. The basis for Pitt’s suggestion was viously a direct location of the hydrogen atoms in a preliminary report of the crystal structure deter- an oxime by use of accurate three dimensional mination of syn-$-chlorobenzaldoxime. The hy- X-ray data, or by neutron diffraction, would be drogen bonding in this crystal was said to be: highly desirable. (6) D. Hall a n d F. J. Llewellyn, ibid., 9 , 108 (1956). N. . .O distances of 2.82 A., with the angles: 0-N OF CHEMISTRY . . .O’ = 101.4’ and N-0. . .N’ = 82’. Since the DEPARTMENT CALIFORNIA former is closer to that expected for a covalent UNIVERSITYOF SOUTHERN JERRY DONOHUE bond angle (the hydrogen atom is assumed to lie Los ANGELES7 , CALIFORNIA RECEIVED APRIL9, 1966 on or near the N. . .O’ and 0. . .N’ axes), the structure RzC=+NH-O- is indicated. The final results of this structure determination have not yet been THE STRUCTURES OF SINIGRIN AND SINALBIN; AN ENZYMATIC REARRANGEMENT published. Dunitz and Robertson3 later discussed Pitt’s suggestion, pointing out that the structure Sir : found for a ~ e t o x i m e ,with ~ angles 0-N. . .O’ of The rnyronate ion, isolated as the potassium salt 124’ and N-0. . .N’ of 11l0, was compatible with sinigrin in 1839, is the precursor of the isothioeither structure, but the results in the case of cyanate of black mustard and horseradish and the dimethylgly~xime~ probably supported Pitt’s sug- prototype of mustard oil glucosides. The curgestion. As pointed out by Dunitz and R o b e r t ~ o n , ~rently accepted structure (I, R = H*C=CHCHz), proposed in 1897,’ rested on the enzymatic hydrol(1) G. J. P i t t , Annual Reports ofthe Chemical Society, 47, 457 (1960). ( 2 ) B. Jerslev, Nature, 166, 741 (1950). ysis of sinigrin t o allyl isothiocyanate, D-glucose (3) J. D.Dunitz and J. H. Robertson, Annual Reportsofthe Chemical and bisulfate ion, the cleavage by silver nitrate t o Society, 49, 378 (1952). (4) T.K. Bierlien a n d E. C. Lingafelter, Acta C r y s t . , 4, 450 (1951). glucose and silver sinigrate, C4H504NS~Ag2,a mer3
i
(5) L. L. Merritt and E. 1.anterman. ibid., 5, 811 (1952).
(1) J. Gadamer, Arch. Pharm., 186, 44 (1897).
