1978 CHICK AND WILSMORE : THE POLYMERISATION OF KETEN.CCX.-The PoZymerisatiorz o f Keten. cycloButan- 1. : 3-dione (" Acetylketen ' I ) .By FRANCES CHICK and NORMAN THOMAS MORTIMER WILSMORE.IN a previous paper (Trans., 1908, 93, 946) we stated that whenketen, either in the liquid or the gaseous state, is left to itself atthe ordinary temperature, a brown substance is formed, the chiefconstituent of which is a colourless liquid of a very pungent odour,boiling at 126--127O/760 mm. (corr.), and which, after havingbeen frozen, melts at - 7 O to -6O; and we showed by combustionana.lyses, by determination of the vapour density according to theHofmann method, and by measurement of the freezing point ofits solution in benzene, that this liquid is a polymeride of keten,formed according to the equation 2CH2:C0 = C4H402 We alsoshowed that the substance undergoes a series of reactions whichindicate a close relation with acetoacetic acid.Thus, with waterin the cold, acetoacetic acid itself is produced:on boiling with alkalis, acetates are formed:C4H402 + Ba(OH), = Ba(C,H30,)2 ;with aniline, a very vigorous reaction takes place, resulting in theformation of acetoacetanilide :C4Hd0, + C6H5*NH2 = CH3*CO*CH2*CO*NH*C6H5 ;and with phenylhydrazine, a phenylhydrazone-phenylhydrazide ofacetoacetic acid is obtained :C4H40, + 2C6H5*N,H3 =C4H402 + HzO = CH3*C0CH2*@O2H ;CH,*C(:N*NHPh)-CH,*CO*NH*NHPh + R,O.The hydrazide character of the latter substance was shown bythe formation of a hydrochloride and of a platinichloride.Additionof pyridine to the polymeride caused a violent reaction, with theformation of brown resins; but when the reaction was allowed totake place in benzene solution, dehydracetic acid w i ~ s the chiefproduct.All these facts led us to conclude that the polymeride stilCHICK AND WILSMORE : THE POLYMERISATION OF KETEN. 1979contained a keten group, and we therefore assumed it to be acet-yl-keten, having the formula CH,*CO*CH:CO. A t the same time, weconsidered the possibility of its being cyclobutan-1 : 3-dione :as this would agree better with our observation that it did notappear to react with alcohol. Nevertheless, the great reactivityand very unsaturated nature of the substance led us to prefer theketen formula, and a measurement of the refractive index alsoseemed to support this choice:M.R.CH,*CO*CH:OO .................. 29.489CH,/ \CH, ..................... 18,782CH~C(oH))CIH .................. 20.664Found ................................. 20 -075co\co/\C(OH)/Here, however, we had to suppose that the *CH:CO group hada depressing action on the refractive index, an assumption which,on account of lack of experimental evidence, we made with allreserve.After the publication of the previous paper, the work was for atime interrupted, and it was not until the session just closed thati t could be seriously resumed. I n the meantime a paper byStaudinger and Bereza has appeared (Bey., 1909, 42, 4908), inwhich the polymerisation of the disubstituted keten, ethyl ethyl-ketencarboxylate, CO:CEt*CO,Et, is described, the productaccording to these authors being undoubtedly a cyclobutanederivative, namely, ethyl 1 : 3-diethylcyclobutan-2 : 4-dione-1 : 3-di-car b oxylat e,CO,Et*CEt<Eg>C Et-CO,Et,and on this and other grounds they conclude that the substancediscovered by us must be A1-cycZobuten-l-ol-3-one,The arguments of Staudinger and Bereza did not, however, seemto us to be necessarily convincing.Thus, ethyl 1 : 3-diethylcyclo-butan-2 : 4-dione-l : 3-dicarboxylate reacts with two molecular pro-portions of aniline to form two molecules of ethyl ethylmalonanilate :C0,Et*CEt<CO>CEt*C02E:t co + ZC,H,*NH, =2C02Et-CH Et*CO-N H*C,H,.Also, on heating, the diethyl ester dissociates with regeneration6 0 1980 CHICK AND WILSMORE : THE POLYMERISATION OF KETEN,of ethyl ethylketencarboxylate. Our substance, on the other hand,reacts with only one molecular proportion of aniline, and it showsno tendency to dissociate on heating-quite the contrary.Similarremarks apply also to the reactions of truxilic acid referred to byStaudinger and Bereza in this ccnnexion. Again, the comparativestability of the polymeride of keten, both in the pure state andin alcoholic solution, which Staudinger and Bereza cite as beinginconsistent with the keten formula, was not conclusive, moreespecially as our previous observations on these points were incom-plete. We have since found that alcohol reacts with the polymerideif a trace of mineral acid be present, forming acetoacetic ester:C4H402 + C,H,*OH = CH3*CO*CH,*C0,-C2H, ;and that, when the pure substance is kept for some weeks in sealedtubes at the ordinary temperature, dehydracetic acid is formed,apparently by direct polymerisation.Moreover, Staudinger andBereza show that disubstituted ketens become much less reactivewhen one of the substituents is an acidic group, and, conseguently,it was not unreasonable to assume, as we did, that the substitutionof acetyl for a hydrogen atom in keten should have a similar effect.So far therefore there was not sufficient evidence available toallow of a definite conclusion being drawn between the open chainand the cyclic formula for the first polymeride of keten. Most ofthe reactions described above involved a transference of a hydrogenatom from the reagent t o the polymeride, and there was nothingto show into which part of the molecule of the latter this atomhad entered.It occurred t o us that a study of the action of bromine on thepolymeride might throw some light on its constitution.We foundthat bromine combined directly with the substance, forming amonobromoacetoacetyl bromide, which, on treatment with alcohol,gave the corresponding ethyl monobromoacetoacetate. If thepolymeride had the acetylketen structure, addition of bromineshould give the a-bromoacetoacetyl bromide :CH,*COCH:CO + Br, = CIH,*CO*C'HBr*COBr,whereas from cyclobutan-1 : 3-dione only the y-derivative couldresult :Since we have actually obtained the y- and not the a-derivativeunder conditions which rendered a wandering of the bromine atomfrom the a- to the y-position highly improbable, it follows that theevidence in favour of the cyclic structure for the first polymerideof keten is now conclusive.It may be noted, however, in passing,that the substance may still be regarded as an internal anhydridCHICK AND WILSMORE : THE POLYMERISATION OF KETEN. 1981of acetoacetic acid. The true acetylketen has therefore yet to bediscovered, but we venture to prophesy that, when prepared, it willresemble cyclobutan-1 : 3-dione in many of its reactions.Incidentally, by treating the above-mentioned y-bromoacebacetyl bromide with aniline, we have obtained the same bromoaceto-acetanilide as was prepared by Knorr (Annulen, 1886, 236, 79) bythe bromination of acetoacetanilide, and, by digesting this withconcentrated sulphuric acid as described by him, we also obtainedhis bronio-2-hydroxy-4-methylquinoline (Zoc.cit., p. 91). Knorr alsoprepared a substance, which he believed to be identical with thelatter compound, by the action of bromine water on Z-hydroxy-4-methylquinoline itself. As under these conditions the brominewould almost certainly occupy the 3-position, Knorr assumed thathis bromoacetoacetanilide was the a-derivative, the condensationwith sulphuric acid taking place according to the equation:Me60 Me,/'We now find, however, that the compound formed on brominating2-hydroxy-4-methylquinoline is not identical with, but is an isomerideof, the bromo-2-hydroxy-4-methylquinoline produced by the actionof sulphuric acid on bromoacetoacetanilide ; and, since our synthesisshows that the latter substance undoubtedly has the bromine inthe y-position, it follows that the compound obtained from it bythe action of sulphuric acid is o-bromo-2-hydroxy-4-methylquinoline :CH,BrCH The semi-enclic formula, CO<cG>C*OI-I, proposed by Stsudingerand Bereza agrees better than the diketonic foimula, CO<Ez2>CO,with the molecular volume, refractive index, and dispersion of thesubstance, as we shall show later; but we can find no chemicalevidence for it.cycZoButan-1: 3-dione does not appear to reactwith sodium, with acetyl chloride, or with phenylcarbimide. I nfact, a mixture of it with phenylcarbimide may be separated intoits constituents by fractiona.1 dist,illation.It may be mentioned,however, in this connexion that the substance gives a red colourwith ferric chloride in alcobolic solution. The semi-enolic formula,1982 CHICK AND WILSMORE : THE POLYMERISATION OF KETEN.is also inconsistent with the reaction with phenylhydrazine, thissubstance reacting normally with B-diketones, which can exist in thesemi-enolic form to form a pyrazole ring. The reaction with phenyl-hydrazine is supported by that with semicarbazide, two molecularproportions of which react with cyclobutan-1: 3-dione to form asemicarbazone-semicarbazide of acetoacetic acid :CH2<CO>CH, co + BCO*N,H, =H,O + CH,*C(:N*NH*CO*NH,)*CH2*CO*NH*NH*CO*NH2.Dry ammonia combines with cyclobutan-1 : 3-dione at low tem-peratures to form acetoacetamide, previously prepared by Claisenand Meyer (Ber., 1902, 35, 583) by the prolonged action of diluteaqueous ammonia on acetoacetic ester :CH,<~~>CH, UO + N H ~ = CH,*CO*CH,*CO=NH~.By the further action of dry ammonia, acetoacetamide is convertedinto a yellow oil.This oil could not be purified, but it appearedto consist mainly of the hitherto non-isolated amide of P-amino-crotonic acid, CH,*C(NH,):CH=CO*NH,, for, on heating it in acurrent of dry hydrogen, a crystalline substance was produced,melting a t 197O, and having the empirical formula C,H,,O,N,.From analogy to the condensation of acetoacetamide described byClaisen and Meyer, this crystalline substance is probably 4-amino-2 : 4 - dimethyl - A2 - tetrahydro-6-pyridone-3-carboxyZamide, formedthus :NH, CH, NH, CH,\/C\/CFrom its reactions the stability of the cyclobutan-1 : 3-dione ringis thus seen t o be very small, even when compared with that ofcyclobutane prepared by Willstatter and Bruce (Ber., 1907, 40,3979).I n fact, the substance behaves as if the ring were notcompletely closed, that is to say, as if its constitution were:QH2* GO-CO*CH,-Since, however, there is no precedent for such a formula as this,we are not prepared seriously to propose it.The action of Grignard’s reagent (in this cam magnesium methyliodide) gave for the most part only by-products which could not bCHICK AND WILSMORE THE POLYMERISATION OF KETEN.1983identified, but there was some indication of the formation ofdiacetone alcohol, which was probably produced somewhat asfollows :Reduction of cy clobutan-1 : 3-dione with hydrogen in the presenceof platinum black gave an unlooked-for result, n-butyraldehydebeing formed, The reaction may possibly take place in the followingstages :or C,H,O, + 6H = CH,-CH,*C~*CHO + H,O.As the intermediate compounds have not been isolated, it is notpossible to say exactly at what stage the ring is broken. Theelimination of water would be facilitated by the dehydrating actionof unchanged cydobutan-1 : 3-dione.The further polymerisation of cyclobutan-1 : 3-dione to de-hydracetic acid may perhaps be represented by the followingscheme, although many others are possible, especially if cydobutan-1 : 3-dione be given the open-chain structure suggested above:(JH.. . . . . . . . -. -. . . .CO<~,H;>C:C<~~>CH, + H,O = C O < ~ ~ ~ \ CJ33 c(oH):c<~~>cH,= CH,. CO~CH,~ cq0- CH*CO C O > ~ ~ z .Collie’s formula for dehydracetic acid (Trans., 1891, 59, 179)has here been chosen. The condensation would be much moredifficult to follow if dehydracetic acid had the formula assigned toit by Feist (Annalen, 1890, 257, 253).Much work has been done with a view to discovering the con-stitution of the yellow substance which is formed from cyclobutan-1: 3-dione in the presence of quinoline; but, on account of thesmall quantity available, not much success has been obt~ained.I n our former paper we stated that “acetylketen” combinedwith sodium ethoxide in dry alcohol to form sodium ethylaceto-acetate, which, on boiling with hydrochloric acid, gave carbondioxide and methyl propyl ketone.Since the polymeride of ketenis in reality qczobutan-1: 3-dione, the compound formed withsodium ethoxide must have been sodium butyrylacetate, which1984 CHICK AND WIISMORE : THE POLYMERISATION OF KETEN.however, would also give methyl propyl ketone on boiling withhydrochloric acid :CH2<E>CH, + Na0*C2H5 = CH,*CH,*CH,*CO*CH,*CO,NaCH,~CH,*CH,*CO*CH,*CO,Na + HCI =NaCl + CO, + CH,*CH,*CH,*CO*CH,.We propose to study this type of reaction more fully.cgcZoButan-1: 3-dione does not react with hydrocyanic acid, evenwhen the two substances are heated on the water-bath in sealedtubes, and it does not react with liquid cyanogen at the ordinarytemperature.EXPERIMENTAL.Preparatbn of cycloButan-1: 3-dione.We have not succeeded greatly in improving the yield of cyclo-butan-1 : 3-dione, about 5 grams from 150 grams of acetic anhydridebeing the best hitherto obtainable.It is not necessary to frac-tionate the crude keten, beyond allowing the temperature to riseto about - 80° to - 70°, before distilling it into the pressure tubes.There seems to be little risk of explosion if the tubes are carefullymade, and they may be handled with safety after polymerisationhas taken pIace, as the internal pressure will then have diminished.They should, however, be re-cooled to about -8OO before beingopened, or loss of substance may be caused by the rush of gas.Diluting the keten with ether to diminish the violence of the poly-merisation did not noticeably improve the yield; but, on the otherhand, surrounding the tube containing the keten with a bath ofalcohol cooled to -30°, or lower, and allowing the whole slowlyto attain the temperature of the room, materially reduced theformation of brown resins.To ensure complete polymerisation, thetubes should be kept for st day or two before being opened. Thedistillation of the cyclobutan-1 : 3-dione is conveniently carried outunder about 30 mm. pressure, when the bulk of the substance passesover as the temperature of the bath rises from 40° to 60°. Thetemperature should, however, finally be raised to about looo.Theliquid then remaining in the distilling flask consists of dehydraceticacid and brown resins. The distilling flask should be heated by awater- or oil-bath, and not by a naked flame, or the distillatemay be contaminated with decomposition products from the residue.The whole apparatus should, of course, be carefully dried and filledwith dried air before commencing the distillation; and only driedair should be allowed to enter the apparatus during and at theclose of the operation. These precautions are especially necessarywhen it is intended to study the physical properties of the subCHICK AND WILSMORE : THE POLYMERISATION OF ICETEN. 19S5stance. The preparation of cyclobutan-1 : 3-dione may be recom-mended as a useful exercise for students in laboratories where asupply of liquid air is available.Physical Properties of cycloButan-1: 3-dione.We have confirmed the boiling point of cyclobutan-1: 3-dioiicunder 760 mm.pressure previously given. Under 100 mm. pressurethe substance boils at 69-71O (corr.). The freezing point of a,carefully prepared sample was found t o be -7.9O t30 -7.5O.The density of two samples of the substance was determined atvarious temperatures. As a mean of four weighings (0.7432, 0.7433,0.7432, 0.7433), the pyknometer contained 0.74325 gram of waterweighed in air at lao. The weights of cyclobutan-1: 3-dione andthe densities calculated from these were as follows:t.9 -5"10.514.114'317 718.721'622.725 '326.729.3Weight of substance in air.I.11.e-0.82380.82260.81960.81930.81 650.81570.61300'81230-80980.80880'806446t. r. 11. v 1'10531.10081'09691.09211,08771.08301.10121.09581.09121.0864The temperatures were measured by means of a thermometerwhich had been carefully compared with a standard thermometerfrom the International Bureau of Weights and Measures, and theyhave been corrected to the hydrogen scale. The densities in eachseries lie very nearly on it straight line, but there is a difference offrom 2 to 4 parts in 10,000 between the two series. The densityat 2 3 O , given in our previous paper, is in line with those in thefirst series. Taking the mean of the two series by graphic inter-polation, the density of cyclobutan-1 : 3-dione between loo and 30°is given by the equation :*& = 1.3 000 - 0.001 18(t - 15').Taking the molecular weight its 84.03, the molecular volume at15O is 76.39.Assuming Traube's values for the atomic volumes at15O (Traube, Grulzdriss d e r physikalischerc Chemie, 1904, p. 120;Smiles, Chemical Constitution and Physical Properties, 1910, p.125) : C = 9.9 ; H = 3.1 ; Oco = 5.5; OOH = 2.3 ; ring formation = - 8.11986 CHICK AND WILSMORE : THE POLYMERISATION OF KETEN.and molecular co-volume = 25.9, we obtain the following calculatedvalues :CH,*Cd.CHCO ........................ 88.9Found.. .................................. 76 -4This property is therefore decisively in favour of a cyclic formula.The enolic formulz appear to agree better than the diketonicformula, but it is quite possible that the contraction due to theformation of the cyclobutane ring is more than 8.1 units, thisnumber having been obtained mainly from a study of benzenoidsubstances.*The refractive index of cyclobutan-1: 3-dione has been re-determined, using portions of the two samples just mentioned, withthe result that the value for the D line given in our former paperwould seem to have been too low. This may have been due to theprecautions to exclude moisture having been insufficient. Thedensities were interpolated from the values given above :First Sample.Line. t . N. 46t. M. R.C 18.7" 1'43490 1.0957 20'007D, 19'2 1'43?69 1.0951 20,131G' 18.7 1.45179 1.0957 20'680Se5cond Sampie.C 18'6" 1 -43547 1.0958 20-02918-3 1.43877 1.0961 20'15718.7 1.45236 1.0957 20.702D, G'Both sets of numbers give as the value for the molecular'' dispersion," 0.673.For Comparison, the values for the molecularrefractivities and dispersion, calculated from the atomic refrac-tivities given in Landolt and Bornstein's '' Tabellem " for thevarious isomeric formulR, will be found in the table herewith :CH;CO'CH:COFound.. ..........C. Dl G' - C'.18-53 18.78 0.4719-54 19.72 0-6420-56 20'66 0.8020.36 20'49 0.7020.02 20'14 0.6CHICK AND WILSMORE : THE POLYMERISATION OF KETEN. 1987From these figures it would appear that of the possible ringformulze the semi-enolic modification is the most probable. Thevery unsaturated nature of the substance, however, makes itlegitimate to amume the existence of an exaltation of the refractiveindex and dispersion which would raise the values calculated forthe diketonic formula by an amount rather Iess than that due to adouble bond.The study of the absorption spectrum has been kindly repeatedfor us by Professor E.C. C . Baly, using various thicknesses of a0.1 molar solution in dry ether. The result was to confirm theobservation of Mr. H. E. Watson given in our former paper.There was no selective absorption, and only it slight generalabsorption in the ultravioletAntalysis of cycloButan-1: 3-dione.- -As a further check on thecomposition, it third analysis of the substance was made:0-1298 gave 0.2704 CO, and 0.0599 H,O.C4H,0, requires C=57.1; H=4.8 per cent.Action of Alcohol on cycloButan-1: S-dione.-As stated earlierin the paper, cyclobutan-1 : 3-dione combines with ethyl alcohol inthe presence of a trace of mineral acid.To about 0.5 gram of thecompound dissolved in ethyl alcohol, a small quantity of a dilutealcoholic solution of sulphuric acid (ethylsulphuric acid) was added.After two days the odour of cyclobutan-1 : 3-dione had disappeared.The free acid was neutralised by shaking with moist calciumcarbonate, and the mixture was dried with calcium chloride, filtered,and distilled-towards the end under diminished pressure. The lastfraction had the odour of acetoacetic ester, gave the characteristicviolet colour with aqueous ferric chloride, and boiled a t 1 8 2 O /760 mm.(corr.), pure acetoacetic ester boiling at 1819C=56.8; H=5.1.,4 ction of Bromine on cycIoButnn-1: 3-dione : y-BrornoacetoacetylBromide.In order to study the action of bromine on cyclobutan-1 : 3-dione,these substances were diluted with carbon tetrachloride, which hadbeen dried over calcium chloride and distilled. The bromine waspurified by washing with aqueous potassium bromide and thenwith water, and distilling from phosphoric oxide. In a preliminaryexperiment a, dilute solution of cyelobutan-1 : 3-dione was titratedwith a bromine solution, using the colour of the bromine asindicator. The strength of the bromine solution was found inthe usual way by shaking with aqueous potassium iodide andtitrating with thiosulphate.The titrstion of the cyclobutan-1 : 3-dione was not very sharp, as towards the end the reaction becamevery sIow, but it WAS sufficient to show that the substances reacte1988 CHICK AND WILSMORE : THE POLYMERISATION OF KETEN.in molecular proportions. 0.384 Gram of cyctobutan-I : 3-dionerequired 0.707 gram of bromine, whilst according to the equationC,H,O, + Br, = C4H4O2Br2, 0.732 gram should have been used. Tocarry out the bromination on a larger scale, cyclobutan-1 : 3-dione,in weighed quantities of from 2 to 5 grams, was dissolved in about50 C.C. of carbon tetrachloride contained in a 100 C.C. measuringflask. An equivalent amount of bromine, dissolved in 20 to 30 C.C.of carbon tetrachloride, was then run in very slowly from a burette,the flask meanwhile being kept immersed in cold water and wellshaken.The reaction appeared to be over when a little more than95 per cent. of the bromine had been added. The solution of theacid bromide fumed strongly on contact with moist air, and reactedvigorously with water, giving hydrobromic acid. To prepare theester, alcohol, which had been dehydrated by means of metalliccalcium, was added to the solution of the acid bromide. Thesubstances reacted at once with evolution of heat, and the hydrogenbromide which was formed, being much less soluble in carbontetrachloride than in the ester, '' salted out " the latter, whichformed a light brown layer on top of the tetrachloride. We havefound that the same thing occurs if carbon disulphide is used assolvent in place of the tetrachloride. This phenomenon must, havebeen observed in brominating ethyl acetoacetate in the ordinaryway, but we have not seen it mentioned.On shaking the mixturewith water, the ester at once dissolved in the carbon tetrachloride.The washing with water was continued until all free acid had beenremoved. The solution was then dried over calcium chloride, thecarbon tetrachloride and any alcohol remaining were evaporated,and the residue was distilled under 10 mm. pressure, the bulk ofthe ester passing over between l l O o and 115O. This is about 1 5 Olower than the boiling point of the y-bromoacetoacetic ester under10 mm. pressure given by Epprecht (AnnuZen, 1894, 278, 77);but, as this author does not state that the mercury in his manometerhad been properly boiled out, the pressure in his apparatus wasprobably rather higher. The presence of bromine in the eiter wasproved by warming with dilute sodium hydroxide, sodium bromidebeing formed.With copper sulphate and sodium acetate, a brightgreen copper salt was produced. The y-position of the. bromine inthe ester was also shown by the reaction with thiocarbamide, asdescribed by Epprecht (Zoc. c i t . ) , ethyl 2-aminothiazolyl-4-acetate,melting at 93-94O, being obtained according to the equation:CH,Br*CO-CH,*CO,Et + CS(NH,), CHICK AND WILSMORE : THE POLYMERISATION OF KETEN. 1989Ethyl a-bromoacetoacetate would have given with thiocarba.mide,ethyl 2-amino-4-methylthiazole-5-carboxylate, melting at 175O :It is well known, however, that in certain circumstances thea-bromo-ester may be converted into the y-bromo-ester by the actionof hydrogen bromide, and it occurred to us that this might possiblyhave taken place when the alcohol was added to the acid bromide.The “salting out” of the ester by the hydrogen bromide wouldcertainly facilitate such an interchange.Accordingly, the synthesisof the ester was repeated in a slightly different way. The solutionof the acid bromide was run very slowly into excess of alcohol, inwhich finely ground, dry sodium acetate was suspended, the mixturebeing kept well stirred during the process. The bromination ofthe cyclobutan-1 : 3-dione and the reaction with alcohol were bothcarried out a t Oo. This time no “salting o u t ” of the ester tookplace.The solution was washed with water, dried, and concentratedM before, but the ester was distilled under 5 mm. pressure, andpassed over below looo, leaving only a very small residue in thedistilling flask. The yield of ester was about 75 per cent. of thatcalculated from the weight of cyclobutan-1 : 3-dione taken. Theester thus obtained was quite colourless. The thia.zole derivativewas also very nearly colourless, but it melted as before at 9 3 O(corr.), even after repeated recrystallisation from ether.To prepare y-bromoacetoacetanilide, a little less than onemolecular proportion of aniline dissolved in carbon tetrachloridewas added slowly to a solution of y-bromoacetoacetyl bromide inthe sa.me solvent.The anilide separated at once in the form ofan almost white precipitate, which, after washing with water anddrying, melted at about 134O. On recrystallising from alcohol,colourless crystals were obtained, which melted and decomposed a t138O (corr.), as given by Knorr (Annalen, 1886, 236, 79) for thebromoacetoacetanilide which he obta’ined by brominating aceto-acetanilide. The identification was confirmed by taking a mixedmelting point with a sample of bromoacetoacetanilide prepared byKnorr’s method. When more than one molecular proportion ofaniline was added to the acid bromide, the second bromine atom wasalso attacked, and tarry substances were formed, from which a pureproduct could not be isolated. On dissolving the bromoaceto-acetanilide in cold concentrated sulphuric acid, keeping the solutionovernight, and then pouring into water, bromo-2-hydroxy-4-methyl-quinoline wits precipitated, which, after recrystallising from alcoholand drying, formed the curious felt-like mass described by Knorr,which melted and decomposed at 260° (corr.).Knorr (Zoc. cit.1990 CHICK AND WIISMORE : THE POLYMERISATION OF KETEN.p. 92) gives the melting point as circa 258O.” For comparison, asample of bromo-2-hydroxy-4-methylquinoline was prepared by theother method described by Knorr (Zoc. cit., p. 91). 2-Hydroxy-4-methylquinoline, obtained by digesting acetoacetanilide in thecold with concentrated sulphuric acid, was suspended in warmwater, and to the mixture bromine water was added until a per-manent yellow colour was produced, slightly more than one molecularproportion of bromine being required.The resulting bromo-2-hydroxy-4-methylquinoline was washed with water and r ecrystallised repeatedly from alcohol. I n appearance it was indis-tinguishable, even under the microscope, from the bromohydroxy-methylquinoline prepared from bromoacetoacetanilide, but itmelted, apparently without decomposing, at 273-275O (corr.),although it began to soften at 269-270°. A mixture of the twopreparations melted a t 240O.The y-bromoacetoacetyl bromide wm evidently an unstable sub-stance, as its solutions began to turn brown after a, few hours.Nevertheless an attempt was made to isolate it in a pure state. Asolution of it in carbon tetrachloride was prepared as before,starting with about 5 grams of cyclobutan-1 : 3-dione.The carbontetrachloride was removed by distilling under diminished pressure(100 to 30 mm.), and the liquid remaining in the flask was distilledunder 5 mm. pressure in a current of dry hydrogen. Between 1 and2 grams of an oily liquid passed over between 105O and l l O o , buthydrogen bromide was evolved at the same time, and a largeamount of a dark-coloured residue remained in the flask. Theliquid appeared colourless while on the end of the condenser tube,but was ligEt brown when observed in the test-tubes placed tocollect it. The best sample was sealed up as quickly as possible.It gradually darkened on keeping, and after a few days was nearlyopaque. The other samples reacted vigorously with water, withalcohol, and with aniline, and fumed strongly in moist air.Theresidue in the distilling flask appeared to consist largely of freecarbon, A small proportion, however, was soluble in chloroform,and the solution, when placed on the skin and washed with water,gave rise to a brilliant violet stain, which was very stable.Action of Sernicarbazide on cycloButan-1: 3-dione.To prepare the semicarbazide derivative, cyclobutan-1 : 3-dionewas treated with an aqueous solution of semicarbazide hydrochloridecontaining an excess of sodium acetate. A crystalline precipitatewas formed on keeping, which separated from warm water in smallCHICK AND WiLSMORE : THE POLYMERISATION OF KETEN. 1991rhombic ( 1) crystals, melting and decomposing slightly a t217-218O.The substance was insoluble in non-aqueous solvents,but was very soluble in dilute hydrochloric acid, and from thissolution platinic chloride precipitated a sparingly soluble platini-chloride. Owing to the small solubility of the base in cold water,and to the fact that it was decomposed by boiling water, themolecular weight could not be determined. Analysis, however,showed that it was the semicarbaaone-semicarbazide of acetoaceticacid:0.1210 gave 0.1448 CO, and 0.0626 H,O. C = 32.6 ; H = 5.6.0.1135 ,, 0'1388 CO, ,, 0.0570 HiO. C=33*4; H=5*6.0.1106 N=39.0.C,H,,O,N, requires C = 33.3 ; H = 5.6 ; N = 38.9 per cent.,, 37.1 C.C. N2 (dry) at 1 7 O and 751 mm.Action of Ammonia on cycloButan-1; %&one.To follow the reaction with ammonia, cyclobutan-1: 3-dione inquantities of about 3 grams was dissolved in dry ether -and placedin a U-tube, which was immersed in a freezing mixture of ice andsalt, kept at -loo to -15O.A slow current of ammonia gas,produced by warming aqueous ammonia, and dried by passing firstthrough a 50 per cent, solution of potassium hydroxide and thenover freshly ignited lime, was then passed through the solution inthe U-#tube. Almost immediately a solid substance commenced toseparate out; but after about twenty minutes oily drops of a yellowcolour began to make their appearance. The reaction was thenstopped, and the solid substance was quickly scraped on to a porousplate, which had previously been cooled to Oo.If the temperaturewas allowed to rise, or if the action of the ammonia was prolonged,the solid was completely converted into the yellow oil, from whichit could not again be recovered. The solid substance, after dryingon the porous plate, was usually pure (m. p. 5 4 O , corr.). It wasinsoluble in ether, but very soluble in water, alcohol, or glacialacetic acid. It could, however, be recrystallised most convenientlyfrom a mixture of acetone and light petroleum. The crystals werecolourless. The aqueous solution gave the violet colour characteristicof acetoacetic ester. On warming with aqueous alkalis, ammoniawas evolved, and if this solution was boiled with hydrochloric acid,carbon dioxide and acetone were given off, the latter being identifiedby the iodoform reaction, the alkaline mercuric chloride test, andthe formation with benzaldehyde of distyryl ketone, melting at 112O(corr.).Heating the substance directly with hydrochloric acidcaused decomposition with formation of brown compounds. O1992 CHICK AND WILSMORE : THE POLYMERISATION OF KETEN.adding ammoniacal copper sulphate to the aqueous solution of thesubstance and ncutralising the ammouia, a grccn copper salt wasformed :0.1071 gave 0.1871 CO, and 0.0667 H,O.0.0765, in 19-59 glacial acetic acid, gave A t = - 0'157O.C,H70,N requires C=47.5 ; H = 6.9 per cent.C =47*6 ; I3 = 6.9.M.W. = 97.0.0798, ,, 19.59 ,, 9 , ,, ,, A t = -0.158O. M.W. = 101.M.W. = 101.The substance is therefore identical with the acetoacetamidedescribed by Claisen and Meyer (Ber., 1902, 35, 583), for which,however, they give a melting point of 50°.They did not determinethe molecular weight.The yellow oil formed by the further action of ammonia onacetoacetamide appears to bear some resemblance to the substanceobtained by Duisberg (dnnalen, 1882, 213, 174) by the action ofammonia on acetoacetic ester, from which, however, this author wasunable t o isolate a compound having a definite composition. It isbest prepared by the prolonged action of ammonia on cyclobutan-1 : 3-dione a t -loo to -15O. I f the reaction is started at theordinary temperature, much heat is evolved, and coloured decom-position products are formed. The U-tube may, however, beremoved from the freezing mixture as soon as all the solid aceto-acetamide has disappeared.The oil thus obtained always containedexcess of ammonia, which could not be completely removed, evenby passing a current of dry hydrogen through it for many hours,or by leaving it over sulphuric acid in a vacuum desiccator forseveral months, although the pressure was sometimes reduced to2 to 3 mm. As it was not soluble in any solvents which did notmix with water, it could not be purified by washing. It wasreadily soluble in water or alcohol, less readily so in acetone, andinsoluble in ether, benzene, light petroleum, chloroform, or carbontetrachloride, On exposure to moist air, a crystalline substancewas formed, which, after drying on a porous plate, melted at 147O.A quantity of this sufficient for an analysis could not, however, beobtained.On passing a current of dry hydrogen through the yellowoil a t l l O o , ammonia and water were given off. A t the end of anhour the substance had become very viscous, and portion takenout on a glass rod solidified on cooling. The remainder was thenwashed out of the tube with warm acetone, and, on cooling thesolution, a colonrless, crystalline compound separated out, whichmelted at 1 9 7 O (corr.). Recrystallisation from acetone did notalter the melting point. The compound was readily soluble inwater, alcohol, or glacial acetic acid, sparingly so in cold acetone,and insoluble in ether, benzene, or light petroleum. On warmingwith aqueous alkalis, ammonia was evolved. For a quantitativCHICK AND WJLSMORE : THE POLYMERISATION OF KETEN.1993estimation of the ammonia, the substance was distilled with2N-potassium hydroxide solution, the ammonia being caught instandard acid :0.1133 gave 0-2176 CO, and 0.0743 H,O.0.1063 ,, 21.5 C.C. N, (moist) at 19O and 752.8 mm.0.0765, in 17-75 glacial acetic acid, gave At = - 0.092O.0-0989, ,, 17.75 ,, 7, ,, ,, At = -0.128'. M.W. 4 7 0 .C,H,,O,N, requires C = 52.5 ; H = 7.1 ; N = 22.9 per cent.M.W. = 183.C =52.4; H=7.3.N=22*9.M.W. = 183.0.2369 gave 0-0233 NH,, that is, 183 grams C,H,,O,N, gave18.0 grams NH,.As stat,ed on p. 1982, we consider the substance to be 4-amino-2 : 4-d~methyZ-A2-tetrahydro-6-pyridone-3-carboxy~ade, 183 gramsof which should give 17 grams of ammonia on distilling with dilutealkalis.On acidifying the alkaline solution from which theammonia had been distilled with nitric acid, the free 4-amino-2 : 4-dimethyl-A~-tetrahydr~6-pyridone-3-carboxylic acid was not pre-cipitated, but the silver and barium salts appeared to be insolublein water. We propose to prepare larger quantities of the amidefrom acetoacetamide obtained by Claisen and Meyer's method (Zoc.cit .) .Action of Grignard's Reagent on cycloButan-1 : 3-dione.To study the action of Grignard's reagent (magnesium methyliodide), from 3 to 4 grams of cyclobutan-1: 3-dione were dissolvedin about 50 C.C. of dry ether contained in a large test-tube, andrather more than two molecular proportions of the reagent, dis-solved in about 100 C.C. of ether, were placed in a second tube ofabout 200 C.C.capacity. The reagent was kept at -78O by meansof a bath of solid carbon dioxide and alcohol, and the solution ofcyclobutan-1 : 3-dione was cooled to about -50°, a lower tem-perature than this causing the solute to crystallise out. Thereagent was kept vigorously stirred by means of a current of airdried over phosphoric oxide, and the solution of cyclobutan-1 : 3-dione was run in very slowly in a fine stream by means of awash-bottle arrangement, dry air under pressure being used to drivethe liquid over. Even at the low temperature of -78O, a brightyellow precipitate was at once formed. On allowing the tem-perature to rise, the precipitate turned brownish-red at about-40°, and became almost black above Oo. In this condition onlyresinous substances appeared to be formed when the mixture wasrun into ice-water.It was found preferable to add the etherealmixture slowly, as soon as it had reached -40° to -30°, to a2N-sulphuric acid solution, containing a slight excess of acid, whichVOL. XCVII. B 1994 CHICK AND WILSMORE : THE POLYMERISATION OF KETEN.was kept partly frozen during the mixing by a freezing mixture.A t the end of the reaction, the ethereal solution was separated,and the aqueous solution was shaken with fresh ether, which wasadded t o the first. The ethereal solution was dried over calciumchloride and distilled, the pressure being reduced to 50-60 mm.after the ether had been removed. A small amount of liquidpassed over a t 80-looo, and a few drops of a bright yellow liquid,having a pleasant aromatic odour, at 180-200°. The last portionof this yellow liquid solidified in the condenser tube on cooling toyellow needles, which, after drying on a porous plate, melted a t145-146O.On adding water to the yellow liquid, more of theyellow solid was precipitated, leaving a nearly colourless solution.This was extracted with ether, the ethereal solution dried withcalcium chloride, and evaporated, when a small quantity of a highboiling liquid remained. This gave an orange-coloured solutionwith concentrated sulphuric acid, and, on pouring this into water,a small quantity of an oil was formed, which had an odourresembling that of mesityl oxide. The fraction of the lower boilingpoint also had an odour resembling that of mesityl oxide, and itgave a bright red colour with concentrated sulphuric acid.Aconsiderable amount of residue remained in the distilling flask,and resinous substances were also formed on neutralising andevaporating the aqueous solution.Owing to the poor yield, it was not possible to obtain a liquidhaving a definite boiling point. It is, however, probable from theforegoing that the reaction proceeds, a t any rate to a small extent,in the direction indicated on p. 1983. It is evident that some of theintermediate compounds, possibly including the compound formedwhen only one :CO group has been attacked, are very unstable,readily condensing to form resinous substances.Reduction of cycloButan-l : 3-diolze.This reduction promised to be of interest, but the choice ofmeans was limited.On account of the chemical nature of thesubstance, the more usual reducing agents were inadmissible, and,owing to the tendency to polymerise on heating, Sabatier andSenderens’ method was not promising. Accordingly, the methodof reduction by means of hydrogen and platinum black was selected.The hydrogen was prepared from zinc and dilute sulphuric acid,which were free from arsenic. The gas was passed through a tubepacked with copper gauze and kept at a red heat to absorb tracesof oxygen, which would otherwise have been reduced to water incontact with the platinum black, and it, was dried by means ofsulphuric acid contained in a spiral wash-bottle.The platinuCHICK AND WILSMORE : THE POLYMERISATION OF HETEN. 1995black was prepared by Loew's method (Ber., 1890, 23, 289). Itwas washed until free from chlorides, and was dried in a vacuumover sulphuric acid. The reduction vessel consisted of a large test-tube fitted with a rubber stopper, t'hrough which passed inlet andoutlet tubes for the hydrogen and a tap funnel. The inlet tubereached to the bottom of the reduction vessel. About 2 grams ofplatinum black were placed in the reduction tube, and the air inthe latter was replaced by hydrogen. The platinum black was thenagain dried by warming the tube to about 60°, and alternatelyexhausting with a Fleuss pump and filling with hydrogen. About4 grams of cyclobutan-1 : 3-dione, dissolved in dry ether, were thenintroduced through the tap funnel, and a slow current of hydrogenwas turned on, the lower end of the reaction vessel being placed ina water-bath to prevent cooling during the evaporation of theether.After about two hours, most of the ether had been drivenoff, and a spiral condenser fitted with taps, such as was used forcollecting keten, was attached to the outlet tube, and was surroundedby a bath of solid carbon dioxide and alcohol, in order to catchany volatile compounds which might otherwise be carried away bythe hydrogen. The water-bath was now warmed to about 6 5 O ,and was ma.intained at that temperature by means of an electricheater. After two days the odour of cyclobutan-1: 3-dione couldno longer be detected a t the outlet? tube of the reduction vessel ondetaching the Condenser, but a strong odour of butyric acid becamenoticeable when the connecting tubes were opened to the air.Thereduction vessel was then completely immersed in the water-bath,and the temperature was raised to about 9 5 O . After a furthertwelve hours, all the liquid had passed over into the condenser,which was found to contain a white, crystalline solid. A smallquantity of a brown residue remained with the platinum. I norder to ascertain whether any compounds had been formed whichwere gaseous at the ordinary temperature, the condenser wasattached to a manometer and to the Fleuss pump and exhaustedto about 5 mm. pressure. The tap leading to the pump was thenclosed, and the temperature was allowed to rise.The solid melteda t about -20°, but the pressure did not) rise above 50 mm., evenwhen the apparatus had attaine.d the temperature of the room.The condenser was then filled with dry air, after which it had tobe set aside for a few days. On again opening the taps, it wasnoticed that a, reduction of pressure had taken place, indicatingabsorption of oxygen. On transferring the contents of the con-denser to a distilling flask and distilling, the major portion passedover at 70--80°, and on reducing the pressure to 20 mm., a furtherportion passed over at about 80°. The first fraction had an odour6 ~ 1996 CHICK AND WILSMORE : THE POLYMERTSATTON OF KETEN.resembling that of acetaldehyde, but, on exposure to air, a strongodour of butyric acid was’ developed.This fraction also gave pro-nounced aldehyde reactions. The fraction of higher boiling pointhad a strong odour of butyric acid. It was readily soluble inwater, was acid to litmus, and liberated carbon dioxide fromsodium carbonate with formation of a very soluble crystallisablesalt. It boiled at 163--164O/757 mm. (corr.), and, after solidifyingin a freezing mixture, i t melted at. -7O to -6O. The substancewas therefore n-butyric acid, formed by the oxidation in air ofthe aldehyde. n-Butyraldehyde was accordingly the final productof the reduction of cyclobutan-1 : 3-dione by the method employed.The intermediate stages of the reaction are suggested on p. 1983.The water formed would, of course, combine with excess of cyclo-butan-1 : 3-dione to form acetoacetic acid, which, at the temperatureof the experiment, would decompose into carbon dioxide and acetone.It was anticipated, from analogy to the reduction of acetoaceticester with sodium amalgam, that aldol or ay-dihydroxybutane mighthave been formed, but no trace of these could be detected.Condensation of cycloButnn-1: S-dione.acetylketen ” slowly turnedbrown on keeping at the ordinary temperature, and that brownresins were formed when it was distilled under the atmosphericpressure. We have since found that when cyctobutan-1 : 3-dioneis heated in sealed tubes on a water-bath to 80-90°, it is convertedchiefly into dehydracetic acid, the reaction being complete aftertwo to three hours.A certain amount of brown resins and ofcarbon dioxide are, however, formed at the same time. Also,in some samples, which had been sealed up in glass tubes and keptat the ordinary temperature, crystals of dehydracetic acid madetheir appearance after several weeks ; but here again brown resinsand carbon dioxide were also produced.The yield of dehydraceticacid was, however, the greater the lower the temperature at whichcondensation took place. Condensation did not. appear t o beaccelerated by the action of light, for a sample of cyclobutan-1 : 3-dione, which had been sealed up in a bulb of “ uviol ” glass andexposed to bright sunlight’, did not turn brown more rapidly thanother samples kept in the dark at the same temperature. On theother hand, cyclobutan-1: 3-dione is the more stable the lower thetemperature at which it has been distilled. From the formationof brown resins and of carbon dioxide along with the dehydraceticacid, it is evident that the condensation follows more than onecourse.On a.ccount of the manifold reactions of the multiple ketenI n our former paper we stated thaCHICK AND WILSMOHE : THE POLYMERISATION OF KETEN. 1997groups, for the study of which dehydracetic acid has been thechief starting point, notably in the hands of Collie and his pupils(compare Trans., 1907, 91, 1806), the proof of the formation ofthis compound from cyclobutan-1 : 3-dione, and consequently fromketen itself, was of special importance. After recrystallisation, thedehydracetic acid melted at 109O (corr.), and the melting point wasnot altered by mixing with a sample of pure dehydracetic acid fromother sources.Under the microscope the two samples appearedto be identical when they had been crystallised in the same way.The dehydracetic acid from cyclobutan-1: 3-dione also gave anorange colour with ferric chloride. Finally, it was converted bythe usual method into diacetylacetone, which was identified by theyellow barium salt and the characteristic violet colour with ferricchloride.Condensation of cycloButawl : 3-dione in Presence of Quinoline.The condensation of cyclobutan-1 : 3-dione in the presence ofquinoline has also been further investigated. Quantities of about5 grams of cyclobutan-1 : 3-dione were mixed with an approximatelyequal amount of pure quinoline in a wide test-tube, to which wasfitted a rubber stopper and a mercury valve to allow of the escapeof gas and at the same time to exclude air.After two or threedays yellow crystals appeared in the tube, and the supernatantliquid became dark brown, while carbon dioxide was slowly givenoff. After ten days to a fortnight the reaction appeared to becomplete, and the contents of the tube, now become very viscous,were scraped out into a mortar and ground up with acetone, inwhich the yellow crystals were not soluble. The yellow substancewas then collected and washed thoroughly with acetone, after whichit was dissolved in hot glacial acetic acid and reprecipitated by theaddition of water. It was usually pure after one such re-crystallisation, melting at 2 4 4 O (corr.).(In our former paper wegave 231O as the melting point, but the substance then availablewas not quite pure.) The yield was not good, being only about20 per cent. of the weight of cyclobutan-1: 3-dione taken, and thiswas much reduced if impure quinoline had been used. The acetonefiltrate from the yellow compound was dark reddish-brown whenviewed by transmitted light, and had a marked green fluorescenceby reflected light. It contained only resinous substances, fromwhich a pure compound could not be isolated.It was difficult to obtain satisfactorily consistent analyticalnumbers, probably owing to the formation of met.hane on heating :0.0848 gave 0.2258 CO, and 0.0367 H,O. C =72.6; H =4-8.0'1100 ,> 0'2931 CO, ,, 0.0468 HiO. C=72.7; H-4.71998 CHICK AND WILSMORE : THE POLYMERISATION OF KETEN.0.0683, in 18.94 glacial acetic acid, gave A t = - 0'066O.M.W. = 213.0.0693, ,,* 17.63 ,, ,, ,, ,, A t = - 0'068O. M.W. = 225.0.0531, ,, 17.85 ,, ,, ,, ,, E=0.032. M.W.=235.CI3Hl0O3 requires C = 72.9 ; H = 4.7 per cent. M.W. = 214.The substance was readily soluble in hot glacial acetic acid, butmuch less so in the cold acid, crystallising on cooling in brightyellow needles or plates. The solutions were bright yellow, with agreen fluorescence. It was slightly soluble in benzene, almostinsoluble in alcohol or acetone, and insoluble in water, chloroform,light petroleum, or ether. It was soluble in dilute alcoholic potash,forming a red solution from which it could be reprecipitated in asomewhat impure form by hydrochloric acid.A suspension of thesubstance in water was neutral t o litmus, but addition of ferricchloride gave a red colour, which was not discharged by dilutehydrochloric acid. It was insoluble in aqueous alkalis, and it didnot react with benzoyl chloride, so that hydroxyl groups wouldappear to be absent. It dissolved on warming with 75 per cent.sulphuric acid, forming a deep yellow solution with a greenfluorescence, but the original substance could not be recovered byneutralising the acid. Some reaction had therefore taken place,probably the removal of a side chain.As the yellow compound had a strong resemblance to certainnaphthalene derivatives discovered by Collie (Trans., 1893, 63, 329 ;1896, 69, 293), an attempt was made to prepare the parent hydro-carbon.Owing to the smal! quantity of material available,distillation with zinc dust did not offer much prospect of success,and electrolytic reduction was therefore tried. The solution of thesubstance in sulphuric acid was diluted until the strength of theacid was about 25 per cent., and was placed in a porous pot,together with a cathode of platinum gauze, which could be rapidlyrotated. To prevent formation of oxidising substances, an amal-gamated zinc plate was used for the anode, a slow stream of 25 percent. sulphuric acid being made to flow through the anode vessel toremove the bulk of the zinc which dissolved. The current densitywas about 0.1 ampere per sq.cm. After electrolysis had proceededfor some hours, the cathode solution wi~s shaken with benzene, and,on evaporation of the latter, a minute quantity of a colourlesssubstance, melting at 96-97O, wm obtained. The yield could notbe improved. Substitution of a zinc cathode for the platinum gaveonly oily substances, which could not be caused to crystallise, andwhich were too small in quantity for effective distillation.A suspension of the substance in carbon tetrachloride decoloriseda solution of bromine in the same solvent. A crystalline substanceappeared to be formed at first, but it dissolved on further additioCHICK AND WILSMORE : THE POLYMERISATION OF KETEN. 1999of bromine. On shaking the mixture with water, hydrobromicacid could be detected in the latter.I n the quantitative experi-ments weighed quantities of the substance were shaken with amoderate excess of bromine solution, and, when the substance hadcompletely dissolved, the excess of bromine was determined bymeans of potassium iodide and thiosulphate in the usual way. Thefree acid was then found by titration with standard alkali. Thebromine solution contained 0.0275 gram equivalent per litre andthe alkali 0.0384 equivalent:0.0707 gram required 25-64 C.C. bromine solution and 15.99 C.C.alkali.Or 214 grams reacted with 2.13 equivalents of bromine, forming1-86 equivalents of acid.0.0757 gram required 26.37 C.C. bromine and 16.68 C.C. alkali.Or 214 grams reacted with 2.05 equivalents of bromine, forming1-81 equivalenh of acid.The formation of nearly two equivalents of acid from twoequivalents of bromine would seem to indica.te the formation of anacid bromide, as in the reaction between bromine and keten(Trans., 1907, 91, 1941), or in that between bromine and cyclo-butan-1 : 3-dione described above, this acid bromide on addition ofwater giving one equivalent of hydrobromic acid and one of anotheracid.The yellow compound was attacked by dilute permanganate inthe cold, but only acetic acid could be detected among the productsof oxidation.It also reacted with phenylhydrazine in glacialacetic acid, forming a compound containing nitrogen, which wasinsoluble in benzene but very soluble in alcohol, from which it couldbe crystallised in colourless needles or plates, melting sharply at238O. A quantity sufficient for analysis could, however, not beobtained.On treatment with an alkaline solution of hydroxylamine, anoxime was formed. The yellow compound was dissolved in asolution containing 10 grams of potassium hydroxide t o 20 C.C. ofwater with 120 C.C. of alcohol. Excess of hydroxyIamine hydro-chloride dissolved in a little water was added, and the mixture waswarmed for five minutes on the water-bath. On heutralising withhydrochloric acid, a yellow substance slowly separated, precipitationbeing complete in about two hours. It was washed with water andrecrystallised from hot water. It was also soluble in alcohol. Itmelted and decomposed a t 200° (corr.). The results of analysisseem to show that we did not succeed in getting this substance ina completely pure state; in fact, it could not be dissolved withoutshowing signs of decomposition 2000 APPLEOEY: THE VISCOSITY OF SALT SOLUTIONS.0.1072 gave 0.2336 CO, and 0.0576 H,O. C= 59.4 ; H = 6.0.0.1030 ,, 0.2342 CO, ,, 0.0517 H20. C=61*5; H=5*6.0.1084 ,, 0.2436 CO, ,, 0.0555 HiO. C=61.3; H=5.7.0.1126I. C13Hl,03N, requires C = 63.9 ; H = 4.9 ; N = 11.5 per cent.Formula I results from the reaction of hydroxylamine with two,, 9-98 C.C. N, (moist3 at 18-5O and 761.6 mm. N= 10-2.11. C13H1404N2 ,, C=59*5; H=5*3; N=10.7 ,,:CO groups according to the equation:C13HIoO3 + 2NH,*OH =C13H12O,N, + 2H2O.Formula I1 represents the reaction of two molecules of hydroxyl-amine, one reacting normally with a :GO group, and the otheradding on to a double bond:C13H1003 + 2NH,*OH = C13H1404N2 + H20.The aqueous solution of the hydroxylamine compound was acidIt was titrated with standard alkali containing 0.0384 to litmus.equivalent per litre :0-0777 gram required 6-23 C.C. alkali.That is, 244 grams C13H,,03N2 require 0.75 equivalent of alkali,and 262 grams C13H1404N, require 0.81 equivalent of alkali.It will be seen fram the foregoing that so far we have found noclue to the const'itution of the yellow condensation product of cyclo-butan-1 : 3-dione, and the yield in the preparation of the substanceand of all its derivatives is so poor that very much larger quantitiesof the raw material (that is, of cyclobutan-1: 3-dione) than havebeen hitherto available would be required before this portion ofthe research could be resumed with any chance of success.DNIVERSITY COLLEGE,UNIVERSITY OF LONDON