36 T.-C. TSAO, K. BAILEY AND G. S. ADAIR I95I

REFERENCES

Adair, G. S. (1925). Proc. roy. Soc. A, 108, 627.Adair, G. S. (1928). Proc. roy. Soc. A, 120, 573.Adair, G. S. (1949). In Haemoglobin, ed. F. J. W. Roughtonand J. C. Kendrew. London: Butterworth's ScientificPublications.

Adair, G. S. & Adair, M. E. (1938). C.R. lab. Carl8berg, 22, 8.Adair, G. S., Bailey, K. & Tsao, T.-C. (1949). Biochem. J.

45, v.Adair, Cx. S. & Robinson, M. E. (1930a). Biochem. J. 24,

993.Adair, G. S. & Robinson, M. E. (1930b). Biochem. J. 24,

1864.Adair, G. S. & Robinson, M. E. (1931). J. Phy8iol. 72, 2P.Astbury, W. T. (1941). J. Soc. Chem. Ind., Lond., 60, 491.Astbury, W. T., Reed, R. & Spark, L. C. (1948). Biochem. J.

48, 282.Bailey, K. (1937). Biochem. J. 31, 1406.Bailey, K. (1948). Biochem. J. 43, 271.Bailey, K. (1951). Biochem. J. 49, 23.Bailey, K., Gutfreund, H. & Ogston, A. G. (1948). Biochem.

J. 43, 279.Boeder, P. (1932). Z. Phy8. 75, 259.Brohult, S. (1940). Nova Acta Soc. Sci. up8al. 12, no. 4.Brohult, S. (1947). J. phy8. colUoid. Chem. 51, 206.Burk, N. F. & Greenberg, D. M. (1930). J. biol. Chem. 87,

197.Chibnall, A. C. (1942). Proc. roy. Soc. B, 131, 136.Chibnall, A. C., Rees, M. W. & Williams, E. F. (1943).

Biochem. J. 37, 354.Edsall, J. T. & Mehl, J. W. (1940). J. biol. Chem. 133,

409.Goodeve, C. F. (1939). Tran8. Faraday Soc. 35, 342.Kroepelin, H. (1929). KoUoidzechr. 47, 294.

Mehl, J. W., Oncley, J. L. & Simha, R. (1940). Science, 92,132.

Middlebrook, W. R. (1949). 1st Int. Congr. Biochem. Abstr.,p. 141.

Neale, S. M. (1937). Chem. & Indudtr., p. 140.Neuberger, A. (1950). J. Sci. Food Agric. no. 3, 80.Oncley, J. L. (1941). Ann. N.Y. Acad. Sci. 41, 121.Ostwald, Wo. (1933). Kolloidzwchr. 68, 61.Philippoff, W. (1935). Kolloidzwchr. 71, 1.Philippoff, W. (1936). Kolloidzwchr. 75, 142.Philippoff, W. (1938). KoUoidzwchr. 83, 163.Portzehl, H., Schramm, G. & Weber, H. H. (1950). Z.

Naturfor8ch. 5b, 61.Robinson, J. R. (1939). Proc. roy. Soc. A, 170, 519.Roche, J., Roche, A., Adair, G. S. & Adair, M. E. (1932).

Biochem. J. 26, 1811.Sanger, F. (1945). Biochem. J. 39, 507.Scott Blair, G. W. & Schofield, R. K. (1931). Phil. Mag. 11,

890.Scott Blair, G. W. & Schofield, R. K. (1934). Phil. Mag. 17,

225.Signer, R. & Sadron, C. (1936). Helv. chim. Acta, 19, 1324.Simha, R. (1940). J. phys. Chem. 44, 25.Steinhardt, J. (1938). J. biol. Chem. 123, 543.Tsao, T.-C. (1950). Unpublished observations.Tsuda, S. (1928). Kolloidzsch. 45, 325.Tsvetkov, V. N. & Frisman, E. (1945). Acta Physicochimica,

U.R.S.S., 20, 61.Walker, J. (1949). J. chem. Soc. p. 1996.Weber, G. (1950). Unpublished observations.Weber, H. H. (1950). Proc. roy. Soc. B, 137, 50.Wu, H. (1931). Chin. J. Phy8iol. 5, 321.Wu, H. & Yang, E. F. (1932). Chin. J. Physiol. 6,51.

Studies on the Analysis of Vitamins D1. THE REACTION OF VITAMINS D AND RELATED SUBSTANCES

WITH IODINE TRICHLORIDE

BY J. GREENWalton Oak8 Experimental Station, Vitamin8 Ltd., Tadworth, Surrey

(Received 18 September 1950)

The problem of the chemical analysis of vitamins Din natural oils and synthetic irradiation products ofergosterol and 7-dehydrocholesterol has receivedconsiderable attention over the past 15 years.Relatively little success, however, has been achievedin attempts to find a substitute for bioassays fordetermining vitamin D potencies, although someprogress has been made with the analysis of highpotency synthetic concentrates. Ewing, Kingsley,Brown & Emmett (1943) have given a very fullsummary of the literature dealing with the chemical;analysis ofvitamin D. So far, themost useful reagenthas undoubtedly been the improved antimony

trichloride reagent of Nield, Russell & Zimmerli(1940). Since the paper by Ewing et al. (1943), themain advances have been due to Sobel, Mayer &Kramer (1945) with their discovery of the glyceroldichlorohydrin reaction, and De Witt & Sullivan(1946), who used a chromatographic separation onmagnesia and diatomaceous earth and a modifiedantimony trichloride-acetyl chloride reagent inethylene dichloride for determining vitamin D inseveral products. Ewing, Powell, Brown & Emmett(1948) developed two methods *for determiningvitamin D in irradiated ergosterol products, one bydirect estimation with the improved antimony tri-

INTERACTION OF VITAMINS D AND IC13chloride reagent of Nield etal. (1940) and the other,after a chromatographic separation on 'Superfiltrol',by direct measurement of extinction at 265 mnu.The four chiefproblems in the accurate analysis of

vitamin D in natural products are (a) the eliminationof large quantities of interfering sterols, particularlycholesterol, (b) the elimination of vitamin A, (c) theneed to measure minute amounts of vitamin D,because of its great biological potency, and (d) todevise a sufficiently reproducible and specific re-agent. No reagent so far described is very satis-factory. Spectrophotometric measurements at265 mu. are of doubtful value, unless irrelevantabsorption is low; the various antimony trichloridereagents must be meticulously prepared to give evenmoderately reliable results, and no hitherto de-scribed procedure is based on a stoicheiometricreaction.The present work describes a new reagent which

undergoes a stoicheiometric reaction with thevitamins D. Part 1 shows how the reaction may beused for quantitative titrations and is the basis (withother work to be described later) of methods ofchemical analysis for vitamin D in natural andsynthetic products. These will be described insucceeding parts.

In the course of investigations into the action ofinorganic halogen compounds on vitamin D, it wasfound that a dilute solution of iodine trichloride inchloroform, when added to a solution of calciferol inthe same solvent, gave an immediate reddish violetcoloration, which disappeared on addition of excessofthe reagent. The reaction occurred in the cold, andappeared to take place instantaneously. When thechloroform was replace j by a more polar solventsuch as acetone, no co-our appeared. It seemed,from these observations, that the reaction was asimple addition of chlorine to the unsaturatedlinkages ofthe calciferol molecule, with the liberationof free iodine.The difficulty of estimating free iodine in the

presence of iodine trichloride by chemical methodsprevented an unequivocal test of this hypothesis,but a study of the reaction in the visual spectro-photometer showed that the red-violet solution gavethe typical absorption spectrum of iodine, with amaximum at 518 mP.The reaction was obviously different from all

previously described reactions of iodine mono-chloride or trichloride with unsaturated molecules,in all of which there is iodine addition rather thanliberation.

EXPERIMENTAL

Material8 and apparatwu(1)Calciferol,m.p.115-116o. SuppliedbyGlaxoLaboratoriesLtd., 'Puriss. in Nitrogen' grade. (2) VitaminD8. Suppliedby Glaxo Laboratories Ltd., and also by E. I. Dupont de

Nemours, U.S.A. (3) Lumisterol, m.p. 117-118°. Obtained,by alkaline hydrolysis, from a sample of lumisteryl dinitro-benzoate, supplied by the National Institute for Medical Re-search, London. Recrystallized from acetone. (4) Tachy-steryl dinitrotoluate, m.p. 1510. Supplied by G. Merck,Darmstadt. (5) Suprasterol II, m.p. 110-111°. Obtained, byalkaline hydrolysis, from a sample of the dinitrotoluatesupplied by G. Merck, Darmstadt. (6) Suprasterol I. Iso-lated as the allophanate from the irradiation of calciferol inether by a mercury arc lamp. (7) Sitosteryl acetate. (8) Pyro-calciferyl acetate, m.p. 79-80°. (9) isoPyrocalciferyl acte,m.p. 1100. (10) Ergo8terol, m.p. 158-1600. Supplied byGlaxo Laboratories Ltd. Recrystallized from methanol.(11) 7-Dehydrocholeaterol, m.p. 1470. Obtained pure, byfractional crystallization from ether of a 60% concentrate,supplied by Glaxo Laboratories Ltd. (12) ,8-Carotene. Byfractional crystallization of a pure commercial grade,supplied by British Crop Driers Ltd. (13) Vitamin A con-centrate. Molecular distillate 1,015,000 i.u./g. supplied byBritish Drug Houses Ltd. (14) Iodine trichloride. Re-chlorinated until pure. (15) Carbon tetrachloride. B.P. grade,washed with H%504, six times with water, dried over MgSO4,and redistilled in glass apparatus.

Instruments used for colorimetric measurement were theSpekker absorptiometer, the Hilger-Nutting visual spectro-photometer, and the Hilger medium quartz ultravioletspectrophotometer.

Preliminary qualitative 8tudie8

Several classes of substance, containing differentunsaturated systems, were tested to see if they gavethe IC13 reaction under the same conditions as calci-ferol (i.e. simple mixing in CHI13 solution).

Sterol8. The reaction was given by calciferol,vitamin D3, ergosterol, 7-dehydrocholesterol, lumi-steryl and tachysteryl dinitrotoluates, suprasterols Iand II, pyrocalciferyl acetate, and isopyrocalciferylacetate.The reaction was not given by cholesterol or

sitosteryl acetate, although the former very slowlyliberated iodine on shaking with a solution of IC13.

Vitamin A gave a positive reaction.f-Carotene. The reaction of IC13 with a dilute

solution of fl-carotene was found to follow an un-usual course. When the reagent solution is addedslowly to the carotene solution, a transient greenish-blue colour occurs at the point of mixing and dis-appears in less than a tenth of a second. As thereagent is added, no pink coloration is observed, butthe golden solution of the fl-carotene becomes pro-gressively paler. Finally, with the addition of onedrop of reagent, a threshold is reached and the wholesolution very suddenly turns a deep pink, as the freeiodine is liberated.The transient greenish-blue colour is, in all proba-

bility, related to an initial reaction similar to thechromogenic reactions with carotenoids given byother inorganic halides, such as SbCl3; this reactionis then followed by a second reaction involvingaddition of the whole ICl3molecule to the conjugated

.37Vol. 49

system, and this in turn is followed by a third re-action involving liberation of iodine. The second-stage reaction product, formed by the addition ofthewhole IC13 molecule, appears to be highly unstablenear the iodine-liberation threshold, since evensudden shaking of the solution at this point serves toliberate the iodine.

Vegetable and animal oils. Almost all natural oilsand fats tested reacted to some degree to liberateiodine from IC13. In addition, most oils absorbedvery large amounts of the reagent. The only ex-ception found was castor oil, which absorbed largeamounts of the reagent, but liberated no iodine.

Aromatic hydr6carbons, and heterocyclic bases. Verypure benzene, in dilute CC14 solution, liberatediodine only very slowly from IC13. Ordinary re-agent benzene reacted positively, as did toluene andxylene. When solid IC13 was thrown into benzene ortoluene, iodine was liberated with almost explosiveviolence.

Pyridine and quinoline, in solution, immediatelyabsorbed the reagent, with the separation of yellowprecipitates.

Benzene derivatives. In compounds such asaromatic phenols, ketones, etc., no reaction takesplace.

Mono-ethenoid compounds. Hydrocarbons such ascyclohexene absorb the reagent without iodineliberation. Styrene, which contains a side-chaindouble bond in conjugation with the benzene ring,reacts similarly. Non-hydrocarbon molecules, suchas maleic anhydride, do not react.

Di-ethenoid compounds. Isoprene (two conju-gated double bonds) absorbed IC13 without libera-tion of iodine. Methyl linoleate (two conjugateddouble bonds) reacted similarly.

Tri-ethenoid compound. Methyl linolenate (threeunconjugated double bonds) absorbed the reagentwith no iodine liberation. Methyl elaeostearate(three conjugated double bonds) liberated iodineimmediately from IC13, without absorbing an excessof reagent.

Preliminary quantitative studiesFor quantitative work CC14, purified as described,

was found to be the best solvent. IC13 in this solventis far more stable than in CHC13. CC14 is also thesafest solvent for preparing stable dilute solutions ofprovitamins and vitamins D, as Paterson & Harvey(1944) have already demonstrated for ergosterol.

First studies of the reactions were made by esti-mating liberated iodine on the Spekker absorptio-meter, using O.B. 2 filters and cells of 10 ml.capacity and 1 cm. optical depth.

(a) Stability of colour. Solutions of calciferol inCC14 (15.5 mg./50 ml.) and of C10, in CC14 (46-8 mg./50 ml.) were prepared. Equal volumes (4 ml.) ofeach solution were mixed in the test cell and the

EEN I95I

reading taken on the absorptiometer, matchingagainst pure CC14. The cells were covered withground-glass slides to prevent evaporation, andreadings were taken at intervals. Table 1 shows thatthe reading is reasonably stable for at least I min.,after which a drift occurs.

Table 1. Stability of iodine colour, obtained by inter-action of calciferol and IC13, measured on absorp-tiometer

(For details see text.)

Time(min.)00.250500-75105*0

10.015-020*025-030*0

Reading(E)

0*1210-1210-1220*1220-1220-1300-1340-1390-1440-1440-150

(b) Shape of titration curve. 0014 solution (1 ml.),containing 2 mg. of vitamin D3, was titrated in thetest cell with a solution of IC13 in CC14 (0-052 g./100 ml.). The procedure was to add to the 1 ml. ofvitamin D3 solution a measured volume, from aburette, of the reagent, make up to 10 ml. with pureCC14, and immediately read on the absorptiometer.

Table 2 shows the typical titration results ob-tained.

Table 2. Vitamin D3 (2 mg.) titrated with IC13. Totalvolume made up to 10 ml. with CC14

(For details see text.)

ICl, solution added(ml.)1-02-03-03.54-04-55*05-56-06-57-07-5

Reading(E)00550-1140-1800*2070-2350-2630-2890*2920-2810*2930-2710-202

Stepwise addition of reagent produces a peakreading which is quickly followed by a second peakvery nearly identical in maximum reading, andfinally there is a rapid fall-off in readings as excessreagent re-absorbs the iodine liberated. Such two-peak curves are nearly always observed when thetitration is carried out using small stepwise additionsof reagents.

38 J. GR:

INTERACTION OF VITAMINS D AND IC133The slope of the upward portion of the curve is

independent of the original amount of vitaminDtitrated.

(c) Attempts to rernove excess reagent. Attemptswere made, with only moderate success, to removeexcess IC13 quantitatively from the reaction mixturewithout removing the already liberated iodine.Small quantities of a substance such as castor oil orcyclohexene were added to the titration solution, butalthough these substances did not absorb free iodinefrom simple CC14 solution, in the presence of IC13both the latter reagent and iodine were alwaysabsorbed together. On addition of sufficient IC13to a solution of cyclohexene or castor oil and iodine,complete absorption may be obtained, leaving acolourless solution. If vitamin D3 is titrated, withthe addition of a small amount of cyclohexene beforeIC13 addition, both substances compete for the re-agent and the maximum quantity of iodine liberatedis considerably lower than that liberated by vitaminD3 alone. If the stepwise titration is carried out,adding cyclohexene after the IC13, liberated iodine isonly absorbed beyond the end point and, in this case,the maximum quantity determined is only slightlylower than the theoretical amount. Further experi-ments on these lines might lead to a simpler quanti-tative titration method, but it was not considerednecessary to investigate the matter in detail, sincethe end point is relatively easy to determine.

It is possible that pure vitamin D3 could beelectrometrically titrated against IC13 in a suitablesolvent, since, for example, IC13 is conducting innitrobenzene, whereas iodine is not. Electrometrictitration in CC14 should not be impossible, usingperhaps high frequency (electrodeless) titrationmethods.

Quantitative 8pectrophotometrwC 8tudie8Examination of the mixture of IC1d and calciferol

in CC14 solutions with the Hilger-Nutting visualspectrophotometer showed, in the presence ofa slightexcess of calciferol, the normal iodine absorptionspectrum, with a maximum at 518 m,p.: there wereno other bands to be seen in the visible region. Theultraviolet absorption of the same solution showeda strongband at 265 mix. and a faint band at 310 m,.The absorption spectrum of the chlorinated calci-ferol is unknown and may account for either or bothof the latter two bands. No peak was observed at330 m,.: this indicates the absence of free chlorinefrom the reaction mixture. Pure iodine in CC14 wasfound to have a molar extinction coefficient of 1116,which figure was used throughout the followingcalculations. Earlier determinations by Brode(1926) gave considerably lower values: the dis-crepancy could not be accounted for.

(a) Vitamin D3. The vitamin D3 molecule containsa conjugated tri-ethenoid system. If the IC13 re-

action is quantitative and stoicheiometric, theequation followed will be

'D'+ 2IC13 - 'D3' C1 + I2

For experimental test of the equation, the IC13solution used contained 91 1 mg./250 ml., and thevitamin D3 solution 40-2 mg./50 ml. The vitamin D3solution (1.5 ml.), mixed with 4 ml. of the reagent,theoretically gives an IC13/D3 molar ratio of 2.

Table 3. Determination of 1C13:12:D3 molar ratiO8

Vitamin D3 C1d, added(mg.) (mg.)1-206 1-4561-206 1-602

Total vol.(ml.)5.55.9

E318 MI.1-211-25

I2 liberated(mg.)0-7660-849

The vitamin D3 solution (1.5 ml.) was placed in atest cell (2 cm. optical depth) and 4 ml. reagentadded. Liberated iodinewas estimatedbymeasuringthe optical density of the solution at 518 m,. Theexperiment was repeated using a 10% excess ofreagent. Table 3 shows the results obtained. Themaximum amount of I2 was formed when 10%excess C1I3 was added: the molar ratio, 12/vitamin D3was. then 0-849 x 384=1-06

1-206 x 254

The purity of the IC13 was checked by measuringits extinction at 455 m,u. at an optical depth of 4 cm.The solution was found to contain only 88% of pureIC13. Using this correction, at the maximum pointthe molar ratio IC13/vitamin D3 is therefore,

1-602 x 384 x 881-206 x 254x 100 1

Although this figure is slightly low, it is in goodagreement with the value expected from the equa-tion above. The slight difference between the figurefound in this experiment and the theoretical value ofthe molar ratio is due to the presence of some mono-chloride in the reagent.

It was clear, from these results, that the reactionwas stoicheiometric and could be used for titration,using the maximum iodine liberated as a means ofmeasuring the end point. Fig. 1 shows the fulltitration curve obtained by titrating 3 ml. of thevitamin D3 solution with ICdl. The technique usedwas to place the vitamin D3 solution in a small conicalflask, add a measured amount of reagent and mix byswirling. The absorption at 518 mSt. was then rapidlymeasured in the test cell. Titration was continueduntil the peak had been passed. Simple calculationgave the iodine liberated at each point and-this wasplotted against volume of reagent added. Thecharacteristic double peak is clearly shown. Beyondthe end point, the absorptipn falls off slowly for aconsiderable time, since IC13 itself contributes aboutone-third of its maximum absorption at 518 m,u.

VoI. 49 39

0*J. GREENThis makes it relatively easy to determine the preciseend point, although, of course, immediately after thepeak in Fig. 1, the measured 'iodine liberated' ismade up from an 'iodine + IC1' component.

(b) Iodine trichtoride-iodine back reaction. Excessreagent combines with liberated iodine to form IC1:if this reaction were stoicheiometric, equimolecularproportions of '2 and IC13 would be necessary to formthree molecules of IC1. The complete vitamin D3-IC13reaction to monochloride formation would then in-volve 50% excess reagent over the amount necessaryfor quantitative iodine liberation.

2-0

1~5 _ /1

E;0.

0 5-

X 1X0 /

0

_ I * I I I6 12 18 24

12 liberated (mg.)

Fig. 1. Full titration curve produced by titrating 2-412 mg.of vitamin D. with 1C13. Measurements carried out onvisual spectrophotometer.

The vitamin D. solution (1.5 ml.) was mixedrapidly with 6-6 ml. of the IC13 solution and themixture examined spectroscopically at 2 mm.optical depth. Only the iodine band at 518 mju. wasobserved, and indeed, visually the solution stillappeared to contain much free iodine. Experimentshowed that several molecular proportions excessreagent were necessary to remove all the iodine, andthe solution then showed the typical band at 460 mu.

due to IC1 and an additional band at 330 m p. (due todissociation of excess IC10). The 1C13-12 back reactionis thus not stoicheiometric: iodine is re-absorbedslowly, but eventually completely.

(c) Calciferol. The calciferol molecule contains aconjugated tri-ethenoid system in the sterol nucleusand one double bond in the sterol side chain. Inmost samples of calciferol examined, only the con-jugated system could be titrated with IC13, andanalytical results obtained were almost identicalwith those described for vitamin D3. However,anomalous titration of all four double bonds has beenfound to occur with one sample of calciferol. Studiesof the calciferol reaction are referred to in moredetail later.

(d) Vitamin A. The reaction of vitamin A wasstudied, using the molecularly distilled concentratehaving 1.015 x 106 i.u./g. A solution' (0.0400 g./25 ml. CCIl) was titrated with the same IC13 solution,in 1 ml. quantities. Since the theoretical potency ofpure vitamin A is 3-a x 106 i.u./g., the solution con-tains 0-488 mg. ofvitamin A/ml. Table 4 presents thetitration data.

Table 4. Titration of vitamin A with IC13 on theVi8ual 8pectrophotometer

(1 ml. vitamin A solution taken.)

ICd3 added(ml.)30404-5505-56-06-570

Total vol.(ml.)405-05.56-06-57-07.58-0

E518m,.1-21-21-261-151-061-060-940-84

I2 liberated(mg.)

0-7890-7870-8480-8060-768

Calculation gives the 12/vitamin A molar ratio atthe end point as 1-71. The reaction,

3'A'+ 10IC135I2,requires a ratio of 1-66. In view of the impurity ofthe vitamin A source, the experimental resultsuggests thatthe stoicheiometric reaction is probablyobeyed in the case of vitamin A.

Table 5. Titration of #-carotene with IC13

IC13 added Total vol.(mnl.) (ml.') E518 miA.0-5 5-0 3-31-0 5-5 1-651-5 6-0 0-852-0 6-5 0-802-5 7-0 0-803 0 7-5 0-763.5 8-0 0-704 0 8-5 0-685-0 9.5 0.59

(4.5 ml. carotene solution taken.)I2 liberated

(mg.) RemarksNil Colour, golden. General absorption below 580 mu.Nil General absorption below 495 m,u.Nil0-594 Colour, pink. No general absorption0-6400-6530-6400-6610-652

AA I951j

INTERACTION OF VITAMINS D AND IC13(e) f-Carotene. The reaction of ,8-carotene was

studied by titrating 4-5 ml. of a solution containing0-0880 g./100 ml. CCI . The extraordinary reactionof ,B-carotene with ICl, has already been discussed,and is shown clearly by the spectrophotometricstudy, as presented in Table 5.A study of the table shows'that almost all the

theoretical quantity of iodine is liberated at onepoint, shortly before the end point is reached:further addition of reagent then liberates normalquantities of iodine up to the end point, which isprolonged.

Calculation gives the 12/carotene molar ratio inthis experiment as 3-48. Within the limits of experi-mental error, therefore, the reaction is followedaccording to the relationship

3'carotene'+ 22 IC13 -* 11 I2.

Quantitative determinations, using theSpekker micro-absorptiometer

In the reaction between vitamins D and IC10, themaximum absorption reading at 518 m,u. is constantover a fairly prolonged end point. The estimation ofvitamins D by titration depends on an accuratedetermination of the maximum, which can, undersuitable conditions, be readily carried out with aSpekker absorptiometer. Determinations of ab-sorption must, because ofinherent difficulties causedby reagent instability, slight drift of readings andvolatilization, be made within about a minute.The importance of determining the smallest

possible amounts of vitamin D has necessitated theuse of the micro-absorptiometer for titration purposes, with the use of 0-5 ml. optical cells. However,for larger quantities of material, the Sensitive Modelabsorptiometer, using 5 ml. cells, may be used.

It has been found convenient to take all readingson the instrument within the range 0-075-0-250 onthe density drum. Although this is a more restrictedand lower range than generally used with thisinstrument, its use facilitates the determination ofminimal amounts of material, gives satisfactorystraight line standard curves, and precision ofreadings is usually within limits of± 3%.

(a) Preparation of reagent. The preparation of astable fully chlorinated reagent is essential. Asusually obtained, IC13 is a dry, bright-orange solid,the'precise colour of which depends on its degree ofpurity. In air, it quickly loses chlorine to yield thedark-brown, liquid monochloride: it also loseschlorine in all solvents, at varying rates depending onthe solvent. Up to 15% decomposition of the solidreagent may occur before this state is visuallynoticeable by the appearance offlecks ofbrown liquidmonochloride.

In practice the procedure adopted for obtaining asuitable reagent has been as follows. The solid

(5-10 g.), ground beforehand in a mortar, is placedin a clear glass, wide-mouthed bottle, fitted with aground glass stopper. The contents of the bottle areperiodically rechlorinated by passing a stream ofdrychlorine for about 10 min. or until the original bright-orange colour is regained.'The bottle is kept in thedark, but, since it is of clear glass, early changes ofcolour in the reagent can be noticed.

(b) Titration method. The vitamin D solution ismade up in CC1I. For use on the micro-absorptio-meter, the concentration should be 100-250 p&g./ml.The solution is filled into a 5 ml. burette, and a 25 ml.accurately graduated burette is filled with pure CC14 .The reagent solution should contain about 60-

80 mg. IC13/100 ml.; the exact strength is notcritical. Dilute solutions of this kind are unstableand must be prepared immediately before titration,in either of the following ways. (i) Between 250 and500 mg. is weighed, as rapidly as possible, directlyinto a few ml. of CC14 in a stoppered weighing bottleor small flask and dissolved in 50 ml. of CC14, thesolid being always kept covered with solvent: somesmall insoluble yellow residue usually remains.A portion is diluted to working concentration. Thefinal solution will give a reading of0-150-0-200 on theabsorptiometer when matched against pure CClI in1 cm. optical depth, using O.B.2 filters. (ii) To10-25 ml. of CC14 in a clean dry beaker, a fewcrystals of the solid reagent are rapidly added and,after stirring gently with a glass rod, the supernatantis decanted into a second beaker when the desiredconcentration above the undissolved solid is reached.This may be estimated by a quick reading on theabsorptiometer, or more conveniently, after a littlepractice, visually. Decantation may be followed byfurther dilution, if necessary, but the concentrationshould not be increased by adding further solidreagent. The main essential of either method of pre-paration is speed. Method (ii) is more convenient,takes about 30 sec. to complete, and is recom-mended for accurate titration.The prepared solution is transferred to an

accurately graduated 5 or 10 ml. burette, whichshould be filled only once with the reagent: if morereagent is needed, it must be prepared afresh bymethod (ii) or by sub-diluting the strong solution inmethod (i). All the burettes used must have fine-borejets and well fitting taps: the latter should belubricated with a grease resistant to organic solvents:an excellent preparation is the starch-glycerol pastedescribed by Herrington & Starr (1942).The titration is carried out in the 0-5 ml. cells of

the micro-absorptiometer, which have an opticaldepth of 1 cm. Filters may be O.B.2 or spectrumgreen 604 (using, with the latter, heat-resistingfilters of Calorex or Chance's ON. 19). Only one ofthe micro-mounts of the instrument is used, as theremay be a slight variation in the dimensions of

Vol. 49 41

J. GREENaperture between the two mounts provided, andeach cell is placed in turn in the same mount. Theblank cell is filled with pure solvent. To the test cell isadded first 0-05-0-20 ml. of the vitamin D solution(25-50 jg. vitamin D), then a measured volume ofthe reagent, and finally pure CC14 from the thirdburette to make the volume up to 0-5 ml. A smallground-glass cover slip is placed over the cell, whichis inverted twice to mix the contents and thenplaced in the micro-mount for measurement ofcolour. The cell is then drained and the process re-peated with increasing additions of IC1I to a constantvolume of test solution, until the end point has beenpassed, as indicated by a fall in readings, and,visually, by re-absorption of liberated iodine.

It may be necessary to repeat several readingsaround the end point, using minimal increments ofreagent in order to obtain, as accurately as possible,the true maximum reading. All readings should-bemade rapidly, preferably without zeroing the gal-vanometer each time. The cells must be quite clean,and the optical surfaces should not be touchedduring the titration procedure. It has been foundadvisable, after making up the contents of thetitration cell and closing it with the glass cover slip,to wipe the optical surfaces quickly with a piece ofsilk fabric before inserting the cell into the instru-ment.

(c) Effect of filter. The spectrum green 604 filtersgive readings about 30% higher than the O.B. 2filters, leading to a corresponding increase insensitivity. However, owing to the narrower trans-mission range of the former filters, the end point ismuch sharper and rather more difficult to determine:using the 604 filters, therefore, the reagent, near theend point, should be added in the smallest possibleincrements.

Titration results and standard curves

Vitamin D3. Table 6 shows typical titrationresults at two levels of vitamin D3, using spectrum

Table 6. Titration of vitamin D3 onmicro-absorptiometer

(Total volume 0-5 ml.)Weight of D3

(KAg.)52-0

35.3

IC13 added(ml.)0-080-100 120-140-160-180-100-120-130-140-160-18

Reading(E)

0-1820-2000-2080-2150-218 (peak)0-2000-1260-1320-148 (peak)0-1390-1340-126

green 604 filters, and Fig. 2 shows standard curvesobtained by plotting maximum titration readingsagainst micrograms of vitamin D3, using both typesof filter. The curves are linear over the range used.

20 40Weight of vitamin D3 (Lg )

Fig. 2. Standard curves forvitaminD3. Titration carried outwith ICI3 on the Spekker micro-absorptiometer, using twotypes of filter. Total volume, 0-5 ml.; optical depth, 1 cm.*_, spectrum green 604; A-A, O.B. 2.

Calciferol. As already stated, most samples ofcalciferol tested (at least twelve in number, withdifferent origins) only three double bonds weretitrated with ICI3. With one sample, however, whichwas checked many times by titration on the micro-absorptiometer, a four double-bond titration wasalways obtained. Fig. 3 shows the two standard

Weight of calciferol (p-g.)Fig. 3. Standard curves for calciferol, titrated on micro-

absorptiometer; total volume, 0- ml.; optical depth 1 cm.;O.B.2 filters.

curves constructed by plotting maximum readingsagainst ,ug. of calciferol, using O.B. 2 filters. Curve Bcorresponds to the three-bond titration and is, ofcourse, almost identical with the vitamin D3 curve.

42 I95 I

INTERACTION OF VITAMINS D AND IC13Table 7. Titration of 8terots on the micro-ab8orptiometer

(0.5 ml. cells. Spectrum Green 604 ifiters.)

SubstanceWeigh

0Ergosterol 4Lumisterol 4Lumisteryl dinitrobenzoatePyrocalciferyl acetatei8oPyrocalciferyl acetate 4Suprasterol II 5Tachysteryl dinitrotoluate 9Tachysterol 6

A number of experiments were carried out todetermine the reason for the existence of two calci-ferol curves. First, chemical purification of thesamples by fractional crystallization of the freesterols and their dinitrobenzoates failed to disturbthe titration results. Exhaustive drying at 0 01 mm.was similarly without effect. The possibility ofperoxide catalytic reaction ofthe fourth double bondwas considered, but titration of normal samples inthe presence of benzoyl peroxide or anti-oxidantssuch as ethyl gallate, quinol, etc. produced no changein the reactions. At present, no explanation can beoffered.

Other 8terols. The titration ofother sterols and theirpure esters was carried out as described for thevitamins D. Calculation of the number of doublebonds titrated was made by computation from thestandard vitamin D3 curve. Table 7 presents thetitration data and calculations for ergosterol,lumisterol, pyrocalciferyl acetate, iwopyrocalciferylacetate, tachysterol, suprasterol II and 7-dehydro-cholesterol.

DISCUSSION

Iodine trichloride in carbon tetrachloride solutionundergoes two types of reaction with certain classesof unsaturated substances. The first type involvesimmediate and quantitative addition of the iodinetrichloride molecule and is apparently limited tohydrocarbons and fatty acids which contain one or

more isolated double bonds or two conjugateddouble bonds. The second-type reaction, which hasbeen found to occur with substances containingsystems of three or more conjugated double bondsand, in addition, a group of sterols related to thevitamins D, involves chlorination and liberation ofiodine. Iodine monochloride, in general, reactssimilarly: but, with this reagent, the second-typereaction is not so specific: cholesterol, for example,which only contains one unsaturated linkageliberates iodine from iodine monochloride. The twotypes ofreaction are, for several reasons, remarkable,and the iodine liberation reaction is of a uniquecharacter. Although there is a woalth of literaturedealing with the action of iodine chlorides on un-

t titrated Maximum titration4Lg..) reading40-0 0-17047*0 0-12667-0 0*13658-0 0*20418-0 0-18354-0 0.15195 0 0*14557-0 0-122

No. of doublebonds titrated

3*051*972-102-853*102-051X731-35

saturated substances, all the reactions hithertodescribed have involved iodination. The presentreactions also differ from those previously describedby taking. place under the mildest possible condi-tions, in cold carbon tetrachloride solution. Finally,although occurring in pure non-polar solvent, thereactions are instantaneous and stoicheiometric.Temperature -is without effect either on the type,

the velocity, or the quantitative character of thereaction, which proceeds identically at room tem-perature or at -23° (the melting point of carbontetrachloride). There is no peroxide effect involvedin either type of reaction: their occurrence is inde-pendent of the presence of substances such as

benzoyl peroxide or powerful antioxidants, such as

quinol, pyrogallol, or ethyl gallate.The type II reaction (iodine liberation) gives rise

to speculation as to the mechanism involved. Thereaction R + 2IC13 -+ RC16 + I2,

where R is a molecule containing three conjugateddouble bonds, is one of semi-addition to R, or sub-stitution of the IC13 molecule.The absence of reaction between IC13 and chole-

sterol, in contrast to the reaction of ICI with thelatter substance, indicates that the chlorination doesnot necessarily proceed through a first step ofdissoci-ation of the trichloride to the monochloride and freechlorine (although this dissociation, of course,

always occurs). It is also improbable that, incarbon tetrachloride solution, sufficient ionizationoccurs to enter into the reaction mechanism.

It is believed that the most important indication.of the reaction mechanism may be given by theobserved reaction with P-carotene. The transientgreenish-blue colour formed at the point of contactwith the reagent is almost certainly due to an initialmesomeric change in the ,-carotene molecule underthe influence of an iodine trichloride molecule or a

Cl- or [C1J]- ion: this is akin to the type of meso-

meric change (with resultant strong colour forma-tion) produced by the action ofantimony trichlorideon carotenoid structures and vitamin D. Thetransient colour reaction of carotene with iodinetrichloride is at once followed by addition of the

43

J. GREENreagent, iodine being liberated only just before theend point. The conjugated chain is therefore almostcompletely saturated with iodine and chlorine atomsat first (no free chlorine is detectable spectro-scopically at any point of the reaction before the endpoint), and the halogenated compound formed is,in the case of P-carotene, relatively stable beforethe iodine-liberation threshold is reached. Iodineliberation then takes place perhaps by chlorine sub-stitution by excess reagent or by intramoleculararrangement to a more stable structure, with loss ofiodine. It is possible that some sequence of eventssuch as this, but very much quicker, occurs in thereaction with vitamins D and other reactive sub-stances which liberate iodine, in these cases thefirst-formed iodine-chlorine complex being extremelyunstable. This hypothesis of a two-stage reacstionmechanism may be extended to cover the type Ireaction. Thus, in the reaction with cyclohexene, ifthe reacted solution is allowed to stand for 24 hr.,iodine is slowly liberated: this occurs even in theabsence of excess reagent and it is therefore difficultto see how the second-stage reaction, in the caseof cyclohexene, occurs without intermolecular re-arrangement, since the products of the reaction arepresumably dichlorohexahydrobenzene and iodo-chlorohexahydrobenzene.

It is tentatively suggested, however, that the twotypes of ICI, reaction are basically of the samecharacter, the apparent difference being due to theextremely varying stabilities of the first-formediodine-chlorine addition compounds. The prefer-ential iodine-liberation reaction which takes placewith systems of more than three conjugated doublebonds and the series of sterols derived photo-chemically from ergosterol indicates that a stericfactor determines these stabilities: accommodationofthe large iodine atom in their molecular structuresis probably too difficult for certain types of unsatur-ated compounds.These hypotheses reconcile the observed reaction

with theoretical requirements concerning the knownpolarizability of iodine chloride.The quantitative titration of sterols on the micro-

absorptiometer (Table 7) gives some interestingresults. In all cases, with the exception of tachy-sterol, an integral number of bonds is titrated(within experimental error). Although, in the caseof calciferol only three bonds usually react and theside-chain double bond is preferentially unreactive,in ergosterol, its provitamin, which contains oneunconjugated andtwo conjugated double bonds, it isfound that all three bonds react, and no anomalousbehaviourhasbeendiscovered. The side-chain doublebond, which is preferentially non-reactive in calci-ferol, is reactive in ergosterol. Even more curiously,lumisterol, which is an ergosterol diastereomer,reacts to the extent of only two double bonds, as do

the acetates of pyrocalciferol and i8opyrocalciferol(free sterols not tested). The suprasterols have not sofar been assigned a definite molecular structure-:according to Muller (1935), suprasterol I containsthree double bonds, but these are not conjugated.Suprasterol II also appears to have three uncon-jugated double bonds. Although the suprasterolsare isomers of ergosterol, they do not contain thesterol skeleton. They react to the extent of twodouble bonds with iodine trichloride, and, therefore,if the tentative suggestions as to their structure arecorrect, are the only pure substances so far testedwhich liberate iodine from iodine trichloride andwhich do not possess a conjugated system.

7-Dehydrocholesterol contains two conjugateddouble bonds and is the only substance with less thana total of three double bonds which liberates iodinefrom iodine trichloride.The titration of the purest obtainable specimen of

tachysterol dinitrotoluate gave a figure of only1-73 bonds titrated. Tachysterol itself, isolated fromthe dinitrotoluate by alkaline hydrolysis, was foundto be veryunstable, and themaximum bondtitrationvalue obtained was 1-35. Tachysterol contains, likethe vitamins D, a system of three conjugated doublebonds, and would be expected to show a bondtitration figure of about 3. The discrepancy appearsto be too large to be entirely accounted for by theknown instability of tachysterol, especially in viewof the very low result obtained by titration of thedinitrotoluate.The titration of many samples of calciferol and

vitamin D3 on the micro-absorptiometer has a highdegree of reproducibility, and the overall accuracy iswithin limits of ± 5 %. This is considered very satis-factory in view of the errors involved in reading arestricted range of the instrument drum, inherentslight drift of readings, troubles caused by volatilityin the cell, the measurement of the small quantitiesof solutions, and inter-sample variation. Thetitration of a single solution of vitamin D is repro-ducible within. limits of + 3%. Standard curvesshould be checked on at least two independentsamples of vitamin D3 and calciferol.The limiting amounts of vitamin D that can be

estimated by direct titration are governed solely bythe difficulties of estimating the small amounts ofiodine liberated by the reaction.

SUMMARY

1. Iodine trichloride, in carbon tetrachloride,reacts quantitatively with the vitamins D, theirradiation sterols, vitamin A, ,-carotene, and someother substances, to liberate iodine. The chemistryand stoicheiometry of the reaction have beenexamined and discussed, and a theory ofthe reactionmechanism proposed.

AA I95I

Vol. 49 INTERACTION OF VITAMINS D AND IC13 452. A method for the quantitative estimation of

the vitamins D has been developed. The determina-tion has an overall accuracy of ± 5 %.

The author wishes to thank Prof. R. A. Morton, F.R.S., forhis very generous help with the spectrophotometric studies.He is also greatly indebted to the donors of the samples ofsome of the rather rare sterols used.

REFERENCES

Brode, W. R. (1926). J. Amer. chem. Soc. 48, 1877.De Witt, J. B. & Sullivan, M. X. (1946). Indu8tr. EnyngChem. (Anal. ed.), 18, 117.

Ewing, D. T., Kingsley, G. V., Brown, R. A. & Emmett, A. D.(1943). Indu8tr. Engng Chem. (Anal. ed.), 15, 301.

Ewing, D. T., Powell, M. J., Brown, R. A. & Emmett, A. D.(1948). Anal. Chem. 20, 317.

Herrington, B. L. & Starr, M. P. (1942). Indu8tr. EngngChem. (Anal. ed.), 14, 62.

Muller, M. (1935). Hoppe-Seyl. Z. 233, 223.Nield, C. N., Russell, W. C. & Zimmerli, A. (1940). J. biol.

Chem. 136, 73.Paterson, R. B. & Harvey, E. H. (1944). Industr. EngngChem. (Anal. ed.), 16, 495.

Sobel, A. E., Mayer, A. N. & Kramer, B. (1945). Industr.Engng Chem. (Anal. ed.), 17, 160.

Studies on the Analysis of Vitamins D2. THE ANALYTICAL PURIFICATION OF VITAMIN D BY DIFFERENTIAL SOLUBILITY,

PRECIPITATION REACTIONS, AND CHROMATOGRAPHY

BY J. GREENWalton Oak8 Experimental Station, Vitamin8 Ltd., Tadworth, Surrey

(Received 18 September 1950)

Attempts to analyse synthetic irradiation productsand fish-liver oils for calciferol and vitamin D3 arefraught with unusual difficulties. These arise, in themain, from the very high biological potency ofvitamin D, the wide range ofpotency in materials tobe assayed and the low specificity ofknown chemicalreactions. The difficulties have not been generallyappreciated, and the literature shows that, with fewexceptions, published methods are empirical. Therehave thus been few attempts to deal fundamentallywith the analytical separation of the vitamins Dfrom known interfering substances.In fish-liver oils, the chief interfering substance is

vitamin A, the removal ofwhich is the main problemto be solved. Vitamin A may be present, weight forweight, in at least a hundred times the concentrationof vitamin D, which it masks in all known reactions.Milas, Heggie & Reynolds (1941) attempted toremove vitamin A and carotenoids by preliminarytreatment of concentrates with maleic anhydride indioxan. Ewing, Kingsley, Brown & Emmett (1943)developed chromatographic methods for the elimi-nation of vitamin A and claimed good results.Muller (1947) also described a complicated chro-matographic procedure.The analysis ofvitamins D in irradiation products

appears at first sight to be easier and advances havebeen made, particularly by Ewing, Powell, Brown &Emmett (1948). It is often assumed that most of thesterols in an irradiation product of ergosterol or

7-dehydrocholesterol do not interfere with theanalysis of vitamin D by the best known colori-metric methods. Experience has shown, however,that without the best attainable purification ofvitamin D by removal of sterols, such methods donot give results agreeing with biological assays,when applied to widely varying products. This is sofor the method of Ewing et al. (1948).In the present work, two reactions have been used

to study the analytical behaviour of calciferol andvitamin D.. Since the two substances can differ inbehaviour, they have been studied independently,and conclusions drawn from study of one have notbeen, without test, accepted for the other. The tworeactions are the antimony trichloride-acetyl chloridereaction of Nield, Russell & Zimmerli (1940) and theiodine trichloride reaction described by Green (1951).The antimony trichloride reaction is the more

specific for vitamins D, but is interfered with, tovarying degrees, by the presence of cholesterol, 7-dehydrocholesterol, tachysterol (and possibly pro-tachysterol), vitamin A and carotenoids. A dis-advantage in its use is the lack of reproducibilityof the reagent, which makes it difficult to obtainprecise results even with fairly pure vitarnin D3.The iodine trichloride reagent is less specific, butgives a high degree of precision, and is additionallyuseful in analytical studies ofthe sterols produced byirradiation which do not react with the antimonytrichloride reagent.