EXCHANGE 
 
innivcreit? of Chicago 
 
 
 The Molecular Rearrangement of Symmetrical 
 Bis-Triphenylmethylhydrazine 
 
 A DISSERTATION 
 
 SUBMITTED TO THE FACULTY OF THE OGDEN GRADUATE 
 
 SCHOOL OF SCIENCE IN CANDIDACY FOR THE 
 
 DEGREE OF DOCTOR OF PHILOSOPHY 
 
 DEPARTMENT OF CHEMISTRY 
 
 BY 
 
 RALPH LYMAN BROWN 
 
 Private Edition Distributed by 
 
 THE UNIVERSITY OF CHICAGO LIBRARIES 
 
 CHICAGO, ILLINOIS 
 
 Reprinted from the Journal of the American Chemical Society, Vol. 44, No. 6, June, 1922. 
 
Gbe inniversits of Chicago 
 
 The Molecular Rearrangement of Symmetrical 
 Bis-Triphenylmethylhydrazine 
 
 A DISSERTATION 
 
 SUBMITTED TO THE FACULTY OF THE OGDEN GRADUATE 
 
 SCHOOL OF SCIENCE IN CANDIDACY FOR THE 
 
 DEGREE OF DOCTOR OF PHILOSOPHY 
 
 DEPARTMENT OF CHEMISTRY 
 
 BY 
 
 RALPH LYMAN BROWN 
 
 Private Edition Distributed by 
 
 THE UNIVERSITY OF CHICAGO LIBRARIES 
 
 CHICAGO, ILLINOIS 
 
 Reprinted from the Journal of the American Chemical Society, Vol. 44, No. 6, June, 1922. 
 
EXCHANGE 
 
THE MOLECULAR REARRANGEMENT OF SYMMETRICAL 
 BIS-TRIPHENYLMETHYLHYDRAZINE 1 
 
 Organic hydrazine derivatives resemble derivatives of hydroxyl- 
 amine in all fundamental characteristics, except that, with a single 
 exception, they have not been found to undergo molecular rearrange- 
 ments of the same type as the Beckmann rearrangement of oximes 
 or the Lessen rearrangement of hydroxamic acids. Stieglitz and Sen- 
 ior, 2 who called attention to this difference, made a series of a.t- 
 tempts to effect such rearrangements of hydrazine compounds, but all 
 their efforts were unsuccessful except when they used sym.-bis-triphenyl- 
 methylhydrazine, (CeHs^C.NH.NH.CCCGl^a. When they heated this 
 compound to 250 to 300 with anhydrous zinc chloride and hydrolyzed 
 the reaction product, aniline was obtained, which proved conclusively 
 that one of the phenyl groups of the triphenylmethyl radicals migrates 
 under these conditions from carbon to nitrogen, which is the typical shift 
 of the Beckmann-Lossen rearrangements, and of the analogous rearrange- 
 ments in the triphenylme thane series discovered by the one of us and his 
 collaborators. 3 The only other product identified in the preliminary 
 investigation was triphenylmethane, a product of a parallel thermal de- 
 composition of the hydrazine. 
 
 The present paper reports the results of a continuation of the investi- 
 gation, which was carried out to throw light on the mechanism of the re- 
 arrangement. Unexpected results obtained toward the end of the in- 
 vestigation have made it impossible, at the present moment, to give a com- 
 plete theory of the reaction, but as the one of us (Brown) was called 
 to service in the U. S. Army, we wish to present now the lines of thought 
 
 1 This paper was published by Ralph L. Brown and Julius Stieglitz in the /. Am. 
 Chem. Soc., 44, 1270-92 (1922). 
 
 2 Proc. Nat. Acad. Sci., 1, 207 (1915); and J. Am. Chem. Soc., 38, 2727 (1916). 
 
 3 See the literature references in J. Am. Chem. Soc., 36, 272 (1914) and 38, 2051 
 (1916). 
 
 O O i O *> 
 
followed as working hypotheses and the relation of the results obtained, 
 to these views. 
 
 By analogy to other rearrangements of this type, two possible courses 
 of the rearrangement reaction are suggested at the outset. 4 In the first 
 place, according to the theory of the one of us, 5 we might have as the primary 
 action offering least resistance a decomposition of the hydrazine compound 
 into triphenyl-methylamine and triphenyl-methylimide, a univalent ni- 
 trogen compound which would undergo the rearrangement proper 
 
 I. (C 6 H 6 ) 8 C.NH.NH.C(C 6 H 6 )3 + (C 6 H 6 ) 3 CNH 2 + (C 6 H 5 ) 3 C.N (1) 
 and 
 
 (C 6 H 5 ) 8 C.N > (C 6 H 5 ) 2 C:NC 6 H6 (2) 
 
 By hydrolysis of the phenylimido-benzophenone thus formed, aniline 
 and benzophenone would be produced. 
 
 (C 6 H 6 )2C: NC 6 H 5 + HOH + HC1 > (C 6 H 6 ) 2 CO + CeHsNHsCl (3) 
 
 The alternative course of the reaction would be a rearrangement akin 
 to the rearrangement of triphenylmethyl-peroxide, as investigated by 
 Wieland. 6 
 (C 6 H 6 )3C.O.O.C(C6H 6 ) 8 >2 (C 6 H 6 ) 3 C.O > 2 (C 6 H 6 ) 2 C.OC 6 H5 > (C 6 H 6 ) 2 COC 6 H 6 
 
 (C fl H 6 ) 2 COC 6 H 6 (4) 
 
 #*s-triphenylmethylhydrazine might rearrange in an analogous way as 
 follows. 
 
 II. (C 6 H 6 ) 3 CNH.NHC(C 6 H 5 )3 > 2(C 6 H 5 ) 3 C.NH > 2(C 6 H 6 ) 2 C.NHC 6 H6 
 
 (C 6 H 6 ) 2 C.NHC 6 H 5 
 
 (C 6 H 8 )2C.NHC 6 H 6 (5) 
 
 Hydrolysis of the resulting product, if it occurred at all, would lead to 
 the formation of aniline and benzopinacone. 
 
 [(C 8 H 6 ) 2 C(NHC 6 H 6 )] 2 + 2H 2 + 2HC1 > [ (C 6 H 6 ) 2 C(OH) ], + 2C 6 H 5 NH 3 C1 (6) 
 
 In the experimental part of this report, it will be shown that, besides 
 aniline, which was again identified after hydrolysis of the reaction prod- 
 uct, benzophenone was isolated and identified, a result which decidedly 
 favors the mechanism of the reaction expressed under I. The third prod- 
 uct of the reaction required by Scheme I is triphenyl-methylamine. 
 This was not found, but it was suspected that any triphenyl-methylamine 
 formed would be decomposed by the zinc chloride at the temperature of 
 the reaction. For that reason, the effect of zinc chloride on the amine was 
 studied. As anticipated, when triphenyl-methylamine reacts with zinc 
 
 4 The interpretation from the point of view of the electron conception of valence 
 will be given further on in this paper. 
 
 6 See the literature references given by Stieglitz and Leech, /. Am. Chem. Soc., 36, 
 272 (1914) and by Stieglitz and Stagner, ibid., 38, 2051 (1916). 
 
 8 Wieland, Ber., 44, 2550 (1911). 
 
chloride, the chief products of decomposition are ammonia, phenyl-di- 
 phenylenemethane and triphenylmethane. For instance, we may have 
 
 H 
 
 I /C 6 H 4 
 3 (C 6 H 6 ) 3 C.NH 2 > 3NH, + 2 C 6 H 5 C< | + (C 6 H 6 ) 3 CH (7) 
 
 This action is analogous to the decomposition of triphenylmethyl 
 chloride and bromide by heat and of the corresponding carbinol by phos- 
 phorus pentoxide. 
 
 Returning now to the examination of the products of the rearrange- 
 ment of fo'5-triphenylmethylhydrazine, we have found without difficulty 
 that ammonia and phenyl-diphenylenemethane, besides triphenylmethane, 
 are indeed formed in the reaction. Furthermore, quantitative determina- 
 tions, which will be discussed at greater length presently, showed that the 
 amount of ammonia, i. e., triphenyl-methylamine as the primary prod- 
 uct formed is of the order demanded by the equations given under I. 
 
 Thus far, all the experimental evidence, therefore, favors the conception 
 that the rearrangement of fo's-triphenylmethylhydrazine by zinc chloride 
 follows the course through a univalent nitrogen derivative which seems 
 to be the same as that commonly followed as the path of least resistance 7 
 in the rearrangement of oximes, hydroxamic acids, chloro- and bromo- 
 amides of the acids, acyl azides, triarylmethyl-azides, -hydroxylamines 
 and -chloro-amines, as grouped together by the theory of the one of us. 
 Exhaustive examination of the reaction products brought to light further 
 facts, some of which have raised novel and interesting questions. Before 
 these are presented, it seems wisest to complete the discussion of the re- 
 arrangement proper by referring as briefly as possible to the interpretation 
 of the rearrangement from the point of view of the electron conception 
 of valence. 8 The discussion from this fundamental point of view, will 
 greatly facilitate the consideration of the further experimental results 
 to be presented below. 
 
 For fo's-triphenylmethylhydrazine we have the electronic structure 9 
 (C 6 H 5 +) J =C+ -N+- N- +C=(C 6 H 5 +) J (8) 
 
 +H H + 
 
 The atom with an unstable positive charge, as postulated by the one of 
 us as occurring in all of the rearranging compounds of this character, 
 in the present instance is one of the two otherwise apparently symmetrical 
 nitrogen atoms: for the sake of convenience we have designated the un- 
 stable nitrogen atom with the positive charge in question by putting it 
 in heavy type as N + . In the formation of a rearranging univalent ni- 
 trogen derivative, either of the following courses may be followed: 
 
 7 Stieglitz, Proc. Nat. Acad. Sci., Ref. 2. Stieglitz and Stagner, Ref. 5. 
 
 8 Stieglitz and Leech, Ref. 5. Jones, Am. Chem. J., 50, 440 (1913). 
 
 9 Only those charges are indicated which bear on the discussion. 
 
6 
 
 (1) the true ammonia nitrogen atom ~Nl deprives the nitrogen atom IN+ 
 of its hydrogen atom, forming triphenylmethyl-amine (C 6 H 5 )3C.NH 2 
 and triphenylmethyl-imide (C 6 H 6 )3C . Ni which by a shift of the electron 
 would form (C 6 H 5 )3C.N; (2) more probably the disturbance originates 
 in the unstable nitrogen atom IN+ which captures 2 electrons from the 
 neighboring nitrogen atom. With the accompanying migration of the hy- 
 drogen atom lost by ~NI as a result of the loss of its electrons, this gives 
 
 (C 6 H 5 + ) 3 ^C + -NI( + H) 2 and N~+C = (C.H.+), (9) 
 
 As a basis of choice between these two possible courses we have no experi- 
 mental information except that the second course would be exactly analo- 
 gous to the changes occurring in the closely related rearrangements of 
 halogen amides, oximes, etc. A final decision between these two paths 
 would be, we believe, of great importance, since therein most likely will 
 be found the key to the interesting fact that this symmetrical hydrazine 
 rearranges, whereas all the unsymmetrical hydrazines examined by Senior 
 and the one of us are not rearranged. 10 
 
 In either case, we have then, further, the normal course for univalent 
 nitrogen rearrangements 11 
 
 (C 6 H 6 '+), = C+-N > (C 6 H 6 + ) 2 ZCj:rN- > (C 6 H 6 + )2lCt:N-(C 6 H 6 + ) (10) 
 
 ^C 6 H f 
 
 C 6 H 6 + 
 
 Two electrons migrate from the methyl carbon from the valence indi- 
 cated by the sign in heavy type to the univalent nitrogen atom, the released 
 positive phenyl group migrating to the free negative charge on the nitrogen. 
 
 Returning now to the further observations made in the course of the 
 experimental work, we would mention first that the formation of tri- 
 phenylmethane and nitrogen 
 
 (C 6 H5)3C.NH.NH.C(C6H 5 ) 3 > N 2 + 2(C 6 H 6 ) 3 CH (11) 
 
 observed by Senior and the one of us, was conclusively shown to be a 
 thermal decomposition. The electronic interpretation is given in the pre- 
 liminary paper 12 and shows that the decomposition is the result of an in- 
 tramolecular oxidation-reduction in which the electrons migrate from 
 the nitrogen to the methyl carbon atom, whereas the molecular rearrange- 
 ment just discussed, which is also the result of an intramolecular oxida- 
 tion-reduction reaction, involves a migration of electrons from the methyl 
 carbon to the nitrogen, this course being the direct reverse of that followed 
 in the thermal decomposition. The decomposition by heat and the re- 
 arrangement are therefore the results of two parallel and competing oxi- 
 
 10 The interpretation is given by Stieglitz and Senior, Ref. 2 [/. Am. Chem. Soc.\ p. 
 2729. Since this was written, Messrs. E. C. Gilbert and J. F bmith in this Laboratory 
 have effected the rearrangement of a number of hydrazine derivatives, including benzo- 
 phenone-hydrazone. (May, 1922). J. S. 
 
 11 See Stieglitz and Leech, and Stieglitz and Stagner, Ref. 5. 
 Ref. 2 (J. Am. Chem. Soc., p. 2731). 
 
dation-reduction reactions of an intramolecular character. One of the 
 main experimental difficulties encountered was the problem of reducing 
 the loss of material by the thermal decomposition to a minimum. At 
 best a yield corresponding to the rearrangement of somewhat more than 
 a third of the substance was finally obtained. 
 
 The next observations made were not anticipated and have raised 
 new questions of interest. Among the reaction products phenol was ob- 
 tained in appreciable quantities. It was at first thought that its presence 
 might be due to some decomposition of the aniline group hi phenylimido- 
 benzophenone, the zinc chloride, when it is present hi large excess, possibly 
 reversing the well-known change of phenol into aniline by zinc-ammonium 
 chloride. 13 Experiments made with phenylimido-benzophenone and 
 zinc chloride proved that this is not the source of the phenol. Nothing 
 was found in the literature indicating "the formation of phenol from tri- 
 phenyl-methylamine or its carbinol, from benzopinacone (Equation 6) 
 or its decomposition product, benzohydrol. Furthermore, it was then 
 found that when fo's-triphenylmethylhydrazine is heated with zinc chloride 
 in the absence of air, for instance in an atmosphere of carbon dioxide, no 
 phenol is formed, but the formation of diphenyl is strongly suggested by 
 its odor. The carrying out of the action in the presence of carbon dioxide 
 in place of air led to the further unexpected observation that under these 
 conditions there is no rearrangement to an aniline derivative; at most, 
 small quantities of aniline were obtained in the working up of the prod- 
 ucts, which might be due to the presence of occluded air in the reagents 
 used. These results open up a series of interesting questions, which 
 can only be suggested here as the basis on which further investigation is 
 being planned. 
 
 It is evident, in the first place, that the production of phenol is due 
 to the oxidation of a phenyl radical at some stage of the action at the ex- 
 pense ultimately of the oxygen of the ah-. Two interesting possibilities 
 suggest themselves. Phenol may be formed by the capture of oxygen 
 by some of the phenyl radicals before they have reached their destination, 
 the nitrogen atoms (see the middle phase of the action represented in 
 Equation 10). Whether the phenyl radicals, originally positive, first 
 capture some of the migrating electrons which lead to the rearrangement, 
 and escape because they have become electrically neutral is a matter of 
 fascinating speculation which can only be suggested here. In support 
 of this conception we have the fact that the formation of diphenyl is indi- 
 cated, as yet only by its odor, especially when oxygen is excluded. As 
 far as we can discover, the formation of compounds of the type of phenol 
 and diphenyl has never been observed before in the numberless rearrange- 
 ments in which an aryl or an alkyl group migrates from carbon to nitrogen, 
 18 Merz and Weith, Ber., 13, 1299 (1880). 
 
8 
 
 but none of these has been carried out, as far as we can find, at so high a 
 temperature as 300. 
 
 In the second place, Wieland 14 found that when azo-triphenylmethane 
 is formed by oxidation of fo's-triphenylmethylhydrazine, it decomposes 
 at once, even at 0, into triphenylmethyl : this compound is oxidized by 
 air to its peroxide, which in turn is rearranged by heat into the diphenyl 
 ether of benzopinacone (Equation 4). This phenol derivative may be the 
 source of the phenol obtained in our reaction. Wieland found, however, 
 that the hydrazine is not oxidized to the azo compound by air, but re- 
 quires more powerful oxidizing agents. It is possible, nevertheless, that 
 at 300 a temperature never attained in Wieland's experiments, oxidation 
 by air does occur. Furthermore, part of the azo compound might undergo 
 the rearrangement leading to the formation of phenylimido-benzophe- 
 none, as follows. 
 
 III. (C 6 H S ) 3 C.N.N.C(C6H 5 )3 + O * H 2 + (C 6 H S ) 3 C.N:N.C(C 6 H S ) 3 (12) 
 
 H H 
 and 
 
 (C 6 H 6 ) 3 C.N:N.C(C6H 5 )3 > 2(C 6 H 5 ) 3 C.N > 2(C 6 H 5 ) 2 C: NC C H 5 (13) 
 
 Electronically, we would have 
 
 +H H + +H 2 (14) 
 
 Two electrons leave the nitrogen at the point indicated by the negative 
 sign in heavy type, with the formation of water and azo-triphenylmethane. 
 The dissociation of the latter into two molecules of triphenyl-methylimide 
 (C 6 H 5 ) 3 C.N, needs no further explanation and the rearrangement of the 
 imide would follow the course indicated in Equation 2 above. 
 
 No rearrangement of this kind was ever observed by Wieland in his 
 attempts to prepare azo-triphenylmethane, and while the reaction which 
 he did observe (decomposition into nitrogen and triphenylmethyl) was 
 carried out in the cold (usually at 0), a higher temperature would pre- 
 sumably simply accelerate this decomposition. Nevertheless, some of 
 Wieland's work will be repeated as soon as circumstances permit, and 
 efforts especially made to isolate azo-triphenylmethane in order that it 
 may be subjected to sudden heating to 300. 
 
 This interpretation still fails to account for the formation of the de- 
 composition products of triphenyl-methylamine. To account for these 
 we would have to assume that at 300 still another decomposition of the 
 hydrazine occurs, never observed in its study, but analogous to the well- 
 known decomposition of hydrazobenzene into aniline and azobenzene. 
 
 2(C6H 5 ) 3 C.NH.NH.C(C6H 5 )3 -> 2(C 6 H 6 ) 3 C.NH 2 + (C6H 5 ) 3 CN:NC(C 6 H 6 ) 3 (15) 
 14 Wieland, Ber., 42, 3020 (1909). 
 
9 
 
 In that event, the azo compound must be formed and rearrangement 
 should occur even in the absence of air, which does not seem to be the case. 
 If we assume, therefore, for the present simply as a working hypothesis 
 that the course of the action is that given in Scheme I and that the forma- 
 tion of phenol is actually due to the oxidation of phenyl groups escaping 
 rearrangement, then for every molecule of triphenyl-methylamine formed 
 there should be produced either a molecule of phenylimido-benzophenone 
 (Equations 1 and 2) or a molecule of phenol. The quantity of triphenyl- 
 methylamine produced is found by the quantitative determination of its 
 decomposition product, ammonia, and the quantity of phenylimido- 
 benzophenone produced is found from a volumetric assay of the aniline 
 formed by its hydrolysis. In addition, the quantity of phenol obtained 
 was also determined volumetrically. Now, according to what has been 
 developed above, the interpretation of the rearrangement outlined under 
 I demands that the sum of the aniline and the phenol formed should be 
 roughly of the order 15 of the amount of ammonia (triphenyl-methylamine) 
 obtained. This result is well borne out by the experimental values as 
 collected in Table II. 
 
 Experimental Part 16 
 
 I. Preparation of syw.-fo's-Triphenylmethymydrazine, (C 6 H 5 ) 3 CNH.- 
 NHC(C 6 H 5 )3- For the purpose of obtaining better yields, the compound 
 was prepared by the treatment of triphenylmethyl bromide rather than 
 of the corresponding chloride, 17 with hydrazine hydrate. 
 
 The triphenylmethyl bromide was prepared according to the method of Allen and 
 Kolliker 18 and recrystallized from benzene. To the hydrazine hydrate (8 g.), covered 
 with absolute ether (100 cc.), the triphenylmethyl bromide (27.4 g., 95% pure) was 
 added in small portions throughout the course of about an hour, the mixture being 
 constantly shaken. The reaction began immediately on the introduction of the bromide 
 and proceeded at ordinary temperatures with evolution of heat and continuous precipi- 
 tation of the s;ym.-fo'5-triphenylmethylhydrazine and hydrazine hydrobromide. The 
 action was practically complete a short time after the last portion of triphenylmethyl 
 bromide was added, but the reaction mixture was allowed to stand for a day to insure 
 completion. The precipitate was collected on a filter and washed, in turn, with ether, 
 
 15 Every molecule of triphenyl-methylamine formed gives a corresponding mole- 
 cule of ammonia, our experiments having shown that the ammonia formation is practi- 
 cally quantitative when triphenyl-methylamine is heated with zinc chloride. But 
 some of the phenyl groups may well escape combination with oxygen or with nitrogen. 
 Thus, the formation of diphenyl was strongly indicated in the last experiment under- 
 taken. The sum of the aniline and phenol formed should, therefore, be rather somewhat 
 smaller than greater, than the amount of ammonia produced. 
 
 16 I wish to use this opportunity to express my appreciation of the skilful and 
 painstaking manner in which my young collaborator, Mr. Brown, has handled the 
 rather complex experimental material which this problem has presented. All the ex- 
 perimental work was carried out by Mr. Brown. J. S. 
 
 17 Ref. 14, pp. 3021, 3025. Ref. 19, p. 2727. 
 
 18 Allen and Kolliker, Ann., 227, 110 (1885). 
 
10 
 
 alcohol, water, very dilute sodium hydroxide solution, water, alcohol and ether. The 
 yield was 12.9 g. of crude material melting at 213. Further purification by the method 
 of Wieland gave 11.2 g. of the compound melting at 216. A second recrystallization 
 raised the melting point to 219-220. Since decomposition takes place at the melting 
 point, the melting-point tube containing the material was introduced into the temper- 
 ature bath when it was 5 below the melting point of the compound. 
 
 II. A Study of the Decomposition of s;yw.-fo's-Triphenylmethyl- 
 hydrazine at its Melting Point 
 
 The fact that the only evidences of a rearrangement had been obtained 
 when the compound had been heated with zinc chloride at a temperature 
 higher than that of its melting point, made a study of the melting-point 
 decomposition imperative. At its melting point in air, fo's-triphenyl- 
 methylhydrazine showed a marked evolution of gas and gave as a residue 
 a dirty-white solid melting without purification at 85. This residual 
 solid mixed with triphenylmethane 19 melting at 92, gave a melting point 
 of 87-90. A sample of 1.5 g. of the fo's-triphenylmethylhydrazine was 
 then heated at 220-230 in an atmosphere of dry carbon dioxide until 
 all evolution of gas ceased and the gas had been swept into a nitrometer 
 filled with 50% potassium hydroxide solution. The residual material, 
 while molten, was an amber colored liquid and in the solid state was a 
 yellowish white substance. 
 
 The gas was examined first. It showed no loss of volume when passed 
 through a palladium black tube, such as is used for determinations of hy- 
 drogen by absorption, nor when passed into potassium hydroxide, ammoni- 
 acal cuprous chloride and phosphorus pipets. This eliminates the possi- 
 bility of hydrogen, unabsorbed carbon dioxide, carbon monoxide and 
 oxygen being present. The gas was then mixed with several volumes 
 of pure electrolytic hydrogen, and run into a eudiometer over mercury 
 with a little cone, sulfuric acid on its surface. Continuous sparking across 
 platinum electrodes within the tube produced a continuous reduction of 
 volume and the deposition of a white solid on the acid-wet portion of the 
 inner walls of the tube. In one experiment 0.7610 g. of fo's-triphenyl- 
 hydrazine gave 33.38 cc. of gas (0, 760 mm.). The theoretical value for 
 the evolution of all the nitrogen as N2 is 33 . 02 cc. A portion of the gas 
 (20 cc.) was mixed with pure hydrogen (67 cc.) and sparked until the acid 
 threatened to touch the electrodes within the eudiometer. The residual 
 volume was 10.2 cc. or 3.2 cc. over the 7 cc. excess of hydrogen added. 
 The quantity 3 . 2 cc. is that portion of the original stoichiometric mixture 
 of 80 cc. (20 cc. of N2 + 60 cc. of H 2 ) which had not reacted to form am- 
 monia. This amount (3.2 cc.) is 4% of the original 80 cc. Now, the 
 acid solution washed from the eudiometer and analyzed gave 0.0250 
 g. of ammonia. Twenty cc. of pure nitrogen, measured under the same 
 
 19 Stieglitz and Senior, Ref. 2, p. 2732. 
 
11 
 
 conditions as prevailed for the 20 cc. of gas which was mixed with hydrogen 
 and sparked, would give 0.0261 g. of ammonia. The ammonia found 
 (0.0250 g.) is thus 4.2% less than the theory demands, (0.0261 g.). 
 Since the gas unconverted to ammonia is 4% of the original amount taken 
 and the ammonia found is 4.2% less than theory demands for this same 
 original amount of nitrogen and no other gases were found, the con- 
 clusion is drawn that the gas from the thermal decomposition of the 
 hydrazine was all nitrogen. 
 
 The solid product of the decomposition of foVtriphenylmethylhydrazine 
 appeared in part as a snow-white sublimate and in part as a residual solid 
 in the bottom of the test-tube. Samples of sublimate were needle-like 
 crystals and melted at 92. When mixed with pure triphenylmethane 19 
 melting at 92 they melted at 92. The identity of the sublimate is thus 
 established. The remainder of the residual solid was crystallized from ben- 
 zene and fused to free the crystals from the molecule of benzene present 
 as benzene of crystallization. The product then melted at 92 and mixed 
 with triphenylmethane melting at 92 gave a melting point of 9192. 
 A sample of 7610 g. of fos-triphenylmethylhydrazine decomposed under 
 the conditions previously stated, *'. e., in an atmosphere of carbon dioxide 
 by heat in a bath at 220 to 230, gave 0. 6968 g. of residue in the test-tube 
 and 0.0191 g. of sublimate which total 0.7159 g. The reaction equa- 
 tion calls for 0.7197 g. of triphenylmethane. Repeated tests for aniline 
 and aniline derivatives in these residual solids were without exception 
 negative. 
 
 These results indicate that fo's-triphenylmethylhydrazine at its melting 
 point decomposes practically quantitatively into nitrogen and triphenyl- 
 methane. 
 
 HI. A Series of Attempts to Secure Rearrangement of sym.-bis-Tri- 
 phenylmethylhydrazine below its Melting Point or in Solution 
 
 In view of this decomposition at the melting point, an effort was made 
 to secure rearrangement in solution or at temperatures below that at which 
 the compound melted. Although the results were negative, the brief 
 description of several may possibly save future repetition. 
 
 One-half g. of fo's-triphenylmethylhydrazine was triturated with 8 g. of anhydrous 
 zinc chloride and allowed to stand protected from moisture for 2 months at ordinary 
 temperature. The mixture darkened during this time, but no trace of aniline could be 
 detected when the mass was worked up and tested according to the method of Stieglitz 
 and Senior. 19 Analogous samples which had stood for shorter lengths of time gave the 
 same negative results. 
 
 One-half g. of fos-triphenylmethylhydrazine and 8 cc. of dry benzene were heated 
 in a sealed tube at 250 for 30 minutes. The contents of the tube after cooling were 
 found to be under marked pressure. The dark colored solution showed no traces of 
 aniline derivatives but contained nearly pure triphenylmethane. The pressure in the 
 
12 
 
 tube was then due to nitrogen, the same thermal decomposition having taken place 
 as occurs when the solid compound is heated. 
 
 A sample of the hydrazine was heated with finely ground dry zinc chloride for 8 
 hours at 100*. During this time the yellow mixture went through a continuous color 
 change and caked considerably. Although no gas evolution was visible, the residual 
 hydrocarbons melted well below 100 indicating formation of triphenylme thane. No 
 aniline could be found when the test previously referred to was applied. 
 
 The fo's-triphenylmethylhydrazine with ground zinc chloride was heated at 170- 
 180 for 3 hours. Slight evolution of gas could be detected and the mass softened. As 
 before, no aniline could be detected. Triphenylmethane was identified as a product. 
 This indicates that the thermal decomposition of the hydrazine can be secured at temper- 
 atures below its melting point by continued heating. 
 
 A mixture of fo's-triphenylmethylhydrazine and zinc chloride was placed in dry 
 benzene and boiled for 7 hours under a reflux condenser at the boiling point of benzene. 
 The condenser had been thoroughly dried and was protected by a calcium chloride tube. 
 Part of the yellow color of the mixture was taken up by the benzene. No aniline could 
 be detected in the resulting solution or products. Triphenylcarbinol was identified as a 
 product. It melted at 158-159. Mixed with pure carbinol (m. p. 159-160) it 
 melted at 158-159. 
 
 Sw-triphenylmethylhydrazine (0.065 g.) was mixed with phosphorus pentoxide and 
 heated in a test-tube immersed in an acid bath at 105-110 until no further change in 
 appearance was noticeable (about 2 hours). No aniline or aniline derivatives could be 
 detected in the residue. A portion of the original compound was recovered unchanged. 
 
 A sample (0.062 g.) of the hydrazine was put into water (50 cc.) along with a 
 trace of specially prepared platinum black, 20 and held at 50 for 30 hours. The mix- 
 ture was acidified, heated, made alkaline and distilled with steam. No aniline could 
 be detected in the distillate. The fos-triphenylmethylhydrazine was recovered un- 
 changed. 
 
 A sample (0.09 g.) of the hydrazine, approximately 0.04 g. of the platinum black and 
 water (20 cc.) were heated in a sealed tube at 170 for 1 hour. Tested as above, no 
 aniline could be detected in the contents of the tube. The original compound was re- 
 covered unchanged. 
 
 Some of the hydrazine (0.5 g.) was heated with anhydrous aluminum chloride 
 (5 g.) at 180 for 10 minutes. The hydrolyzed products gave no test for aniline. The 
 products were not investigated further. 
 
 IV. A Series of Experiments to Determine Conditions for a Maximum 
 Rearrangement by Zinc Chloride 
 
 Realizing that a continuance of the experiments to effect the rear- 
 rangement at temperatures below that of the melting point of the com- 
 pound might result only in a vain search for a catalyzer, we decided to 
 return to the original method of Stieglitz and Senior and to secure a greater 
 degree of rearrangement, by a rigorous study of all the conditions involved, 
 than had been obtained in this earlier work. 
 
 Accordingly, experiments were, carried out under varying conditions 
 and the degree of rearrangement was determined in each case by titra- 
 tion of the aniline recovered with 0. 1 N sodium bromate solution. 21 In 
 
 20 Loew, Ber., 23, 289 (1891). Also see Tanatar, Z. physik. chem., 40, 475 (1902) 
 and 41, 37 (1902). 
 
 21 Curme. J. Am. Chem. Soc., 35. 1162 (1913). 
 
13 
 
 .articular the temperature and the time of heating and especially, 
 the shape of the vessel in which the reaction was earned 
 out 'and the time required to heat the mixtures up to the re- 
 action temperature were varied. The results are given in the table 
 that follows. The significance of the first 8 columns is obvious. 
 In the last column is given the result of heating phenylimido- 
 benzophenone under the corresponding conditions with zinc chloride^ 
 The first ! experiments (Nos. 12 and 17) were carried out m hard 
 glass tubes, the next 4 in an Erlenmeyer flask of 60 cc. capacity 
 on the floor of which the mixture was spread in a thin layer and in the last 
 4 experiments, for the securing of still more rapid heating, a lOOcc. Erlen- 
 meyer flask was used. A metol bath was employed and the temperature 
 of the bath taken with the aid of a thermocouple. Since phenyhmido 
 benzophenone, (C 6 H 5 ) 2 C = NC 6 H 5 , is probably the actual product formed 
 by the rearrangement and the aniline is obtained by subsequent hydrolysis, 
 parallel experiments under the same conditions were made with this pr< 
 uct in order to determine whether it is destroyed by zinc chloride under the 
 conditions of the experiment. 
 
 TABLE I 
 DEGREE OF REARRANGEMENT BY ZINC CHLORIDE 
 
 Aniline recovered 
 Aniline 
 
 Expt. 
 
 Sample 
 G. 
 
 ZnCli 
 G. 
 
 Temp. 
 
 Time 
 Min. 
 
 Found 
 G. 
 
 Calc. 
 G. 
 
 (CH*)j = C = N C 
 
 12 
 17 
 
 0.50 
 0.50 
 
 5 
 5 
 
 300 
 330 
 
 10 
 
 5 
 
 0.005 
 0.015 
 
 0.09 
 0.09 
 
 6 78.5 
 17 90.4 
 
 19 
 
 0.61 
 
 6 
 
 330 
 
 3 
 
 0.024 
 
 0.11 
 
 22 .vl'- : 
 
 21 
 
 0.59 
 
 6 + 
 
 360 
 
 3 
 
 0.032 
 
 0.11 
 
 30 
 
 22 
 23 
 
 0.51 
 0.52 
 
 6+ 
 
 7 
 
 400 
 
 390+ 
 
 3 
 
 2 
 
 0.030 
 0.033 
 
 0.09 
 0.09 
 
 33 
 37 91.3 
 
 04 
 
 0.55 
 
 7 
 
 390+ 
 
 4 
 
 0.031 
 
 0.10 
 
 31 
 
 
 ^r 
 
 25 
 28 
 29 
 
 0.51 
 0.78 
 0.57 
 
 7 
 
 7 
 7 
 
 390 + 
 450 
 450 
 
 2 
 2 
 
 0.028 
 0.038 
 0.029 
 
 0.09 
 0.14 
 0.10 
 
 31 
 27 
 29 
 
 approxi- 
 mately 
 90 
 
 30 
 
 0.57 
 
 7 
 
 450 
 
 ! X /6 
 
 0.035 
 
 0.10 
 
 35 
 
 
 32 
 
 0.52 
 
 7 
 
 450 
 
 ! 1 /6 
 
 0.031 
 
 0.09 
 
 34 
 
 These values are calculated on the basis of the assumption (Scheme I, p. 4 
 of the introduction) that each molecule of hydrazine yields one molecule of aniline i 
 not two molecules (as demanded by Scheme II). 
 
 The method used in Expts. 30 and 32 was chosen as the best under the 
 working conditions employed. Expt. 23 gave a better yield, but 60cc, 
 flasks of the variety which after careful annealing would stand being 
 plunged repeatedly into a metal bath at the temperatures used were 
 then immediately available. 
 
 The method employed in Expts. 30 and 32 was essentially this, 
 dry ^5-triphenylmethylhydrazine was weighed into the specially 
 
14 
 
 flask, the anhydrous zinc chloride finely ground in a hot dry mortar 
 added to it and the two were thoroughly mixed. If the fo's-triphenyl- 
 methylhydrazine is pure and both substances are thoroughly dry, little, if 
 any yellow color appears when the two are mixed. The flask was fitted 
 with an empty 30cm. calcium chloride tube and that closed with a short 
 filled calcium chloride tube. The flask whose contents, protected from 
 the moisture of the atmosphere, had thus been thoroughly mixed and 
 shaken down into a thin layer over the bottom of the flask, was plunged 
 into the molten metal bath at 450. After 70 seconds it was removed 
 and allowed to cool. The method for working up the product is essentially 
 that described below (p. 18) for the quantitative examination. 
 
 Examination and Identification of the Products of Rearrangement 
 
 Aniline. The identity of this compound was established by Senior 19 
 by the analysis of its chloroplatinate. Besides the observation of the char- 
 acteristic purple color developed on the addition of calcium hypochlorite 
 solution to the first drops of the steam distillate, the tribromo-aniline ob- 
 tained from the titration of several of the acidified steam distillates 
 containing aniline, was washed with water, and recrystallized from alcohol. 
 This product melted at 118119. When mixed with tribromo-aniline, 
 melting at 119^120, it melted at 118-120. 
 
 Phenol. The quantitative preparation of phenol as tribromo-phenol 
 is described below (p. 19). The white precipitates from several titra- 
 tions 22 were collected on a filter, washed with water and recrystallized 
 from dilute alcohol. The fine, long, white needle crystals melted at 90. 
 Mixed with tribromo-phenol, melting at 94, the product melted at 90-92. 
 The identity of phenol as a product of the reaction is thus established. 
 
 Benzophenone. The quantitative separation of benzophenone in 
 the form of its oxime is described below (p. 19). Our qualitative 
 work by similar methods yielded a considerable quantity of almost pure 
 white crystals of benzophenone-oxime which, when dry, melted at 137- 
 138. Mixed with benzophenone-oxime melting at 141, the substance 
 melted at 138139. This established the presence of benzophenone 
 among the hydrolyzed rearrangement products. 
 
 Diphenylene-phenylmethane and Triphenylme thane. The hydro- 
 carbons obtained (see below, p. 20) melted at 82-83. The melting 
 point of this material 19 was not raised by further recrystallization from 
 alcohol or benzene. However, in concentrated ether solution by a careful 
 regulation of the rate of the spontaneous evaporation of the solvent, 
 two distinct types of crystals were formed. One consisted of groups of 
 fine needles growing from a common center; the other was made up 
 of a relatively small number of large bar-like crystals. Both were colorless 
 
 22 See Koppeschaar, Z. anal. Chem., 25, 162 (1886). 
 
15 
 
 and the two were mechanically separated with ease. The first melted 
 at 143-144 without purification. Recrystallization from alcohol gave 
 it a constant melting point of 145-146. Diphenylene-phenylmethane, 
 prepared according to Kliegl 23 and freed from triphenylmethane, formed 
 at the same time, and by repeated crystallization from alcohol, melted at 
 145-146. An equal part mixture of the two compounds of this melting 
 point melted at 145-146. This established the identity of the diphenyl- 
 ene-phenylmethane produced in our reaction. 
 
 The larger crystals after being pressed between two filter papers melted 
 at 92 without further purification. Mixed with triphenylmethane melting 
 at 92, they melted at 92 and hence were triphenylmethane. 
 
 Triphenylmethane and Gum Resin. From the resins left by steam 
 distillation, only triphenylmethane (m. p. 91, identified as usual) was 
 isolated. 
 
 V. The Decomposition of Triphenyl-methylamine at 450 in the 
 Presence of Zinc Chloride and the Establishment of the Existence of 
 Triphenyl-methylamine as an Intermediate Product in the Rearrange- 
 ment of syw.-fos-Triphenylmethylhydrazine 
 
 Despite the thorough search made for triphenyl-methylamine and its 
 hydrolytic product, triphenylcarbinol, among the rearrangement products, 
 neither could be found. The properties of triphenyl-methylamine are 
 such that quantities of it along with some triphenylcarbinol should have 
 been found without difficulty, principally in the gummy residue of the 
 steam distillation made in the determination of the aniline, if it were 
 present in the melt resulting from the heating of fo's-triphenylmethyl- 
 hydrazine with zinc chloride to effect the rearrangement. Since none was 
 found and its formation in the rearrangement was suggested by the theory 
 under investigation, its decomposition by zinc chloride after its formation 
 was indicated. Hence, it became essential to know the effect of heating 
 triphenyl-methylamine with zinc chloride under conditions identical with 
 those used to produce rearrangement of fo's-triphenylmethylhydrazine. 
 Accordingly, triphenyl-methylamine was prepared from triphenylmethyl 
 bromide and dry ammonia. 24 To insure its being free from the hydro- 
 bromide, it was extracted in a Soxhlet apparatus with low-boiling ligroin. 25 
 It melted at 103 after crystallization from the ligroin. The following 
 experiments were then carried out. 
 
 Triphenyl-methylamine (1.5 g.) was mixed with finely grotmda nhydrous zinc chlo- 
 ride (8 g.) and heated at 450 for 70 seconds, exactly as described in the previous section 
 for the rearrangement of the foVtriphenylmethylhydrazine. The melt was treated with 
 
 23 Kliegl, Ber., 38, 287 (1905). 
 
 24 Elbs, ibid., 16, 1276 (1883). 
 
 26 This method of purification of triphenyl-methylamine is due to Mr. H. C. Trimble 
 of this Laboratory, whose dissertation is, as yet, unpublished. 
 
16 
 
 ordinary ether and the resulting dark red solution was filtered from a dirty-white in- 
 soluble solid portion (A). 
 
 The solution was shaken with cone, sodium hydroxide solution, the ether portion 
 separated, dried with anhydrous potassium carbonate and solid sodium hydroxide, and 
 evaporated to dryness. The residue was dried in vacua over solid potassium hydroxide. 
 It was taken up in absolute ether and, because the solution was not quite complete, 
 the ether extract was filtered. Dry hydrogen chloride, passed into the solution, gave a 
 flocculent precipitate (B). The solution after filtration was partially decolorized with 
 charcoal and evaporated to dryness. The residual gum was freed from ether in vacuo 
 and dissolved in a little benzene. Crystallization took place rapidly. A portion of the 
 crystals pressed between filter paper, fused on the steam-bath and allowed to solidify, 
 melted at 90-91. A mixture of this with triphenyl methane in equivalent quantities 
 melted at 91-92. 
 
 The insoluble material 26 (A) when covered with a little cone, sodium hydroxide 
 solution and warmed gave copious quantities of ammonia. It evidently consisted of a 
 zinc ammonium chloride. 
 
 The precipitate (B) was a clay colored, slimy material, which caked on the filter 
 paper as it dried. It was very slightly soluble in alcohol and insoluble in water, in ether 
 and in ligroin. It dissolved in chloroform and benzene, but crystallized from neither. 
 On evaporation of these solvents both spontaneously and by heat no crystals could be 
 obtained and the residue was a resin. When precipitated from either of these solvents 
 by the addition of low-boiling ligroin, the substance was a clay colored, amorphous mate- 
 rial. Examination under the microscope showed no crystalline material. It gave 
 no test for chlorine or the chloride ion. Repeated treatment with cone, ammonium 
 hydroxide produced no change in its properties. It possessed no melting point but 
 darkened and shrank into a hard resin above 220. Even at 290 no fusion had taken 
 place. This cannot be the hydrochloride of unchanged triphenyl-methylamine. 
 
 In a further experiment the melts from two more samples of the fusion of triphenyl- 
 methylamine with zinc chloride were washed into separate flasks, treated with alkali and 
 distilled for ammonia in the usual way. 
 
 Analyses. Subs., (1) 0.5805, (2) 0.5402. Calc. for triphenyl-methylamine: 
 NH 3 , 0.0381, 27 0.0355. Found: 0.0366, 0.0345. 
 
 The melt from a third sample was dissolved in ether and dil. hydrochloric acid. 
 Each of the two solutions, after being well shaken and separated, was washed thoroughly 
 with the other solvent. The portions of like solvents were combined. The aqueous 
 portion containing the zinc chloride and the ammonia was rendered strongly alkaline 
 and distilled for ammonia. 
 
 Analysis. Subs., (3) 0.5916. Calc. : NH 3 , 0.0388. Found: 0.0384. 
 This result, in view of the separation of the ether and aqueous portions before dis- 
 tillation and the fact that no triphenylcarbinol was obtained from the distillation residue 
 in (1) and (2), eliminates the possibility of any hydrolysis of unchanged triphenyl- 
 methylamine in the first two determinations. Hence, we conclude that triphenyl- 
 methylamine, when heated under the conditions previously stated, gives up its nitrogen 
 as ammonia practically quantitatively. 
 
 In the distillate of Determinations 1 and 2 were small quantities of solid hydro- 
 carbons which, on crystallization from the ether with which they were extracted, melted 
 at 143-144. Recrystallized from alcohol and mixed with diphenylene-phenylmethane 
 
 26 The exact nature of this zinc chloride ammonia complex was not investigated. 
 
 27 This value was calculated on the basis of one molecule of NH 8 for one molecule 
 of triphenyl-methylamine. 
 
17 
 
 melting at 145, it gave a melting point of 142-143 . Diphenylene-phenylmethane is 
 thus identified as one of the products of decomposition. 
 
 As in the other experiment the residue from the ammonia Distillations 1 and 2 and 
 the ether solution of 3 yielded triphenylmethane and gum or resin. 
 
 A sample of triphenyl-methylamine heated alone at 450, as before, gave a strong 
 odor of ammonia and a residue of triphenylmethane, diphenylene-phenylmethane and 
 resin. This decomposition is entirely analogous to the thermal decomposition of tri- 
 phenylmethyl bromide and triphenylmethyl chloride in which, respectively, hydrogen 
 bromide and hydrogen chloride together with triphenylmethane, diphenylene-phenyl- 
 methane and resin are obtained ; 28 and the decomposition with zinc chloride present is not 
 unlike the decomposition of triphenylcarbinol when heated with solid phosphoric acid, 21 
 except that, in the latter case, water is removed instead of ammonia, the quantity of 
 diphenylene-phenylmethane formed is much greater and the quantity of resin is very 
 small. 
 
 In a further experiment a sample of triphenyl-methylamine (0.294 g.) with zinc 
 chloride (8 g.) was heated as before at 450 for 70 seconds. The melt was dissolved 
 in ether and hydrochloric acid, and the ether portion was separated and shaken with 
 20 cc. of 2 N sodium hydroxide solution. The alkali extract was rendered acid and 
 distilled with steam. The first portion (1 cc.) was divided into three parts and tested 
 for phenol as follows: (1) by the addition of ferric chloride; (2) by the addition of 
 Millon's reagent and then nitric acid; 29 (3) by boiling with dil. nitric acid which will re- 
 veal traces of phenol by the odor of 0-nitrol-phenol in the hot solution. All tests failed 
 to detect any phenol. A parallel experiment with phenylimido-benzophenone (0.220 g.) 
 yielded negative results in all three tests. 
 
 All the decomposition products of triphenyl-methylamine had been recovered from 
 the melt of tas-triphenylmethylhydrazine and identified as such except ammonia. 
 .Bw-triphenylmethylhydrazine (0.901 g.) and finely ground dry zinc chloride (10 to 11 g.) 
 were therefore heated under the conditions and hi the manner previously described for 
 this compound. The melt was treated with ether and a small quantity of water. The 
 aqueous portion when made strongly alkaline with concentrated sodium hydroxide and 
 warmed, gave a strong odor of ammonia. The gaseous ammonia turned moistened red 
 litmus paper blue and gave a heavy brown precipitate with Nessler's reagent. 
 
 Another sample of fo's-triphenylmethylhydrazine (0.5 g.), without being mixed 
 with zinc chloride, heated at 450 gave no detectable ammonia, neither was any ani- 
 line to be detected in the hydrolyzed products; while a sample (0.2295 g.) of phenyl- 
 imido-benzophenone 30 heated in the same way gave 83% of the theoretical amount 
 of aniline. The ether solution of the hydrocarbons from this heating of fos-triphenyl- 
 methylhydrazine, after decolorization with a little charcoal, yielded only triphenyl- 
 methane on complete evaporation. This melted at 90-91. Apparently fos-triphenyl- 
 methylhydrazine does not rearrange when it is heated in the absence of zinc chloride, 
 and ammonia is not produced when the rearrangement does not take place. 
 
 28 Elbs, Ber., 17, 701 (1884). Hemilian, ibid., 7, 1208 (1874); ibid., 11, 837 (1878). 
 E. and O. Fischer ibid. t 11, 613 (1878); Ann., 194, 257 (1878). Nef., ibid., 309, 167 
 (1899). Schwarz, Ber., 14, 1522 (1881). 
 
 ^Almen, Jahresber., 1878, 1079. 
 
 30 This product was prepared by the reaction of Reddelieu, Ber., 42, 4760 (1909), 
 but purified as follows. At the end of the reaction, the melt was shaken with strong 
 alkali, separated and distilled in vacua to remove the excess of aniline. The well-cooled 
 residue was freed from the remainder of the aniline on a cold porous plate and twice 
 recrystallized from dil. alcohol. The product formed beautiful yellow flaky crystals, 
 analyzing on hydrolysis for the calculated amount of aniline. 
 
18 
 
 There can be little question but that triphenyl-methylamine was a 
 primary product of the rearrangement and subsequently underwent the 
 characteristic decomposition of that compound as outlined in the pre- 
 ceding paragraphs. This is established by the following facts holding 
 for the conditions of the rearrangement of fo's-triphenylmethylhydrazine : 
 (1) triphenyl-methylamine gave practically a quantitative yield of am- 
 monia; (2) benzophenone-phenylimide does not yield ammonia; (3) 
 triphenyl-methylamine gave no hydrocarbon products not found among 
 those of foVtriphenylmethylhydrazine ; (4) fo's-triphenylmethylhydrazine 
 did not give ammonia when it did not undergo rearrangement and did 
 give it when it did rearrange; (5) the molecular nitrogen and the tri- 
 phenylmethane from the portion that underwent thermal decomposition 
 and the phenol present could not give ammonia. Further evidence bear- 
 ing out this conclusion is to be found in the quantitative study of the re- 
 arrangement products. 
 
 VI. The Quantitative Study of the Rearrangement Products of sym.- 
 fo's-Triphenylmethylhydrazine 
 
 For a quantitative study of this rearrangement, approximately 15 g. 
 of fo's-triphenylmethylhydrazine was worked up in small portions. For 
 the best yield of aniline under the conditions employed portions of about 
 half a gram should have been used. Individual yields in this series (Table 
 II) were sacrificed somewhat for total quantity of material worked up. 
 Expts. 39 to 47 represent single samples, while in Expts. 47 to 51 two 
 samples were heated separately and then combined for analysis in each 
 experiment. The ratio of the hydrazine to zinc chloride was about 1 :10. 
 The preparation and heating of samples were as given in detail in the pre- 
 vious series of rearrangements. 
 
 After cooling the melt, it was treated with ordinary ether and a little dil. hydro- 
 chloric acid which took most of it into solution. The mixture was heated for several 
 minutes under a reflux condenser to hydrolyze the phenylimido-benzophenone; 31 it was 
 shaken thoroughly and, finally, the two solutions were separated carefully. The 
 aqueous portion was extracted thrice with ether. Each of these ether extracts was washed 
 with dil. acid. The original ether solution was washed twice with dil. acid. The acid 
 portions were combined and likewise the ether portions and the combined portions 
 filtered. This gave three products, (a) the acid portion, (b) the ether portion, and (c) 
 the insoluble portion. 
 
 The acid portion (a) containing the ammonia, zinc chloride and the aniline formed 
 by the hydrolyzed phenylimido-benzophenone was made strongly alkaline in a closed 
 apparatus and distilled with steam into 20 cc. of 6 N sulfuric acid. The distillate (400 
 cc.) was extracted with alcohol-free ether to remove any hydrocarbons, 32 freed from any 
 
 31 The completeness of this hydrolysis was tested in this way: 0.2297 g. of phenyl- 
 imido-benzophenone was treated as indicated in the text and the aqueous portion ti- 
 trated for aniline. Calc. for aniline: 0.0831 g. Found: 0.0828 g. 
 
 32 Despite the complete hydrolysis of the phenylimido-benzophenone and the thor- 
 
19 
 
 dissolved ether and titrated for aniline with standard bromate solution. A slight excess 
 of sodium thiosulfate was added and the tribromo-aniline removed by filtration. The 
 filtrate was then made alkaline and the ammonia distilled in the usual manner. Cochi- 
 neal was used as the indicator 33 
 
 The ethereal portion (b), containing the phenol, all the hydrocarbons and benzo- 
 phenone, was shaken with sodium hydroxide solution (15 to 20 cc. of 2 N) and two like 
 portions of water. The combined aqueous portions containing all the phenol 34 were rid 
 of dissolved ether, made acid with hydrochloric acid and distilled with steam into dil. 
 sodium hydroxide (5 cc.). The distillate (300 cc.) was rendered strongly acid and ti- 
 trated with standard bromate solution for phenol. 
 
 The ethereal solution of hydrocarbons and benzophenone from the whole series or 
 rearrangements was distilled nearly to dryness and distilled with steam until the dis- 
 tillate had entirely ceased to be milky in appearance. Solid hydrocarbons were still 
 coming over but the benzophenone was practically all distilled. 35 
 
 The recovery of benzophenone from the hydrocarbon mixture, extracted from the 
 steam distillate, was carried out essentially as described earlier in this paper. The 
 mother-liquor, after the removal of the benzophenone-oxime, was extracted and this 
 extract, together with the mixture of hydrocarbons previously removed, was treated 
 a second time for benzophenone. A very small second crop of the oxime was obtained. 
 
 The mixture of triphenylmethane and diphenylene-phenylmethane precipitated 
 from the alkaline solution of benzophenone-oxime, after being washed free from alkali, 
 melted at 80-84. This was dissolved in a little ether and by a careful control of 
 the concentration, two crops (1.10 g.) of tripheny hue thane, m. p. 90, were removed. 
 The diphenylene-phenylmethane (0.05 g.) was then crystallized, and recrystallized from 
 alcohol. It melted at 145-146. 
 
 The residue from the distillation with steam was extracted with ether and the ex- 
 tract treated with animal charcoal, filtered, dried with calcium chloride, distilled to 
 dryness and freed from ether in vacua. Benzene was added to this residual gum and 
 
 ough extraction of the acid portion (a) with ether, there was some neutral product 
 distilled at this point. This proved to be benzophenone melting at 45. It 
 was added to the ethereal portion (b). This suggests the probability of formation 
 of a complex oxonium salt of benzophenone. The quantity of this material va- 
 ried greatly. The controlling factors were not investigated but will be determined 
 in this Laboratory. See Maass and Mclntosh, /. Am. Chem. Soc. 33, 71 (1911). 
 
 33 The presence of the ammonium salt in no way affected the aniline titration. 
 (See Kingscott and Knight, "Quantitative Organic Analyses," Longmans, Green and 
 Co., 1914, p. 257; Sutton, "Volumetric Analyses," Blakiston's Son and Co., 1911, p. 
 387). The accuracy of the ammonia determination from the filtrate was suf- 
 cient for the purpose for which it was used. Some blank experiments gave these 
 results: 0.0345 g. of aniline and 0.1164 g. of ammonium sulfate (NH 3 content, 
 0.0300 g.) gave 0.0344 g. of aniline and 0.0303 g. of NH 3 ; 0.0345 g. of aniline 
 and 0.1164 g. of ammonium sulfate gave 0.0344 g. of aniline and 0.0301 g. of NH 3 . 
 
 34 The completeness of the separation, and the absence of condensation, of the 
 aniline and the phenol on being heated with a concentrated solution of zinc chloride 
 and hydrogen chloride were tested. A typical experiment gave these results: 0.0345 
 g. of aniline, 0.0305 g. of phenol and 0.0887 g. of ammonium sulfate (NH S content 
 0.0229 g.) gave 0.0346 g. of aniline, 0.0304 g. of phenol and 0.0232 g. of ammonia. 
 
 35 Benzophenone (0.70 g.) was distilled with steam, the condenser terminating in 
 an adapter which dipped beneath the surface of the distillate as in the aniline, ammonia 
 and phenol distillations. From the first 400 cc. of the distillate 0.672 g. of benzo- 
 phenone was recovered by extraction. 
 
20 
 
 from it two crops of triphenylmethane crystals, totaling 3.3 g., were obtained. The 
 gum after removal of the crystals was again subjected to distillation with steam until 
 800 cc. of distillate had accumulated. This distillate on extraction yielded 0.2 g. of 
 hydrocarbons, which were very largely diphenylene-phenylmethane (m. p. 144). 
 The small amount of contaminating material proved to be triphenylmethane. The 
 residue from distillation with steam (a) was again extracted, dried in vacua and treated 
 with benzene in an effort to obtain further crystallization. The crystals formed were 
 extremely small in amount and were not recovered. 
 
 The insoluble residue (c) left after treatment of the melt with ether and aqueous 
 acid, was subjected, in the original filter papers, to extraction in a modified Soxhlet 
 apparatus with chloroform as the solvent. This gave a carbonized residue (/3) and a 
 chloroform solution of a resin-like material (7) in no way distinguishable from the 
 precipitate (B) described in connection with the decomposition of triphenyl-methyl- 
 amine, except that it was darker in color. This will be discussed later. 
 
 A portion (1.12 g.) of the benzophenone-oxime recovered was dissolved in alcohol, 
 dil. hydrochloric acid added and the solution boiled for several hours; it was then diluted 
 and extracted with ether. The dried extract yielded an oil which crystallized on being 
 seeded with a crystal of benzophenone. This weighed 1.02 g. and melted at 45. 
 Recrystallized from ligroin it melted at 47.5 to 48. 
 
 The identification of each of the compounds obtained was carried through again as 
 follows. 
 
 MELTING POINTS 
 
 Compound 
 
 Known 
 Found compound Mixed sample 
 
 Ammonia 
 
 Tribromo-aniline 118-119 ' 119-120 118-120 
 
 Tribromophenol 91-92 92-93 92-93 
 
 Benzophenone 
 
 as oxime 139 .5 141 140-141 
 
 as benzophenone 47.5-48 48 47 
 
 Triphenylmethane 
 
 (from the steam distillate)... 90 92 90-91 
 
 (from the steam distillation 
 
 residue) 90-91 92 90-91 
 
 Diphenylene-phenylmethane 145-146 145-146 145-146 
 
 The gum or resin (a) residue from the distillation with steam and removal of tri- 
 phenylmethane was apparently identical with that obtained in the decomposition of 
 triphenyl-methylamine and as such was considered as a final product. 
 
 The residue (/3) from the chloroform extraction of insoluble portion (c) from the 
 original melt was in no way distinguishable from carbon in its behavior in a flame and its 
 total insolubility in benzene, a characteristic of free carbon. 
 
 As previously stated the residue (7) after evaporation of the chloroform was iden- 
 tical in every way but depth of color with the material obtained from the decomposition 
 of triphenyl-methylamine. It showed the same solubilities and when precipitated from 
 chloroform with low-boiling ligroin and dried, it gave the caked smear. Repeated pre- 
 cipitation failed to purify or change it. When heated in a melting-point tube, it dark- 
 ened and shrank from 225-230 up (heated to 290) but never melted so as to spread on 
 or stick to the tube. After the heating, the material was shaken out of the tube and 
 formed a hard, dark resin. It was probably a highly condensed or polymerized mass. 
 A sample of the precipitated material finely ground was treated in aqueous suspension 
 
21 
 
 with concentrated ammonium hydroxide. It underwent no change in appearance or the 
 properties evidenced under heat and solvent. Treatment with other alkalies and acids 
 did not affect it. These properties are not unlike those exhibited by the very highly 
 polymerized products of the indene and cumarone series and other partly unsaturated 
 bodies. 
 
 The quantitative determinations of the rearrangement products of this 
 series are given in the table that follows. In this table, Part 1 gives the 
 results for aniline, phenol and ammonia: Col. 1 gives the number of the 
 experiment, Col. 2 the weight of samples used (given in values rounded to 
 centigrams but calculated to milligrams), Col. 3 the aniline found, Col. 4 
 the yield of aniline expressed in percentages of the maximum calculated 
 yield. In Col. 5 the weight of phenol is given and in Col. 6 the equivalent 
 percentage of aniline, calculated on the assumption that each molecule 
 of phenol represents a molecule of aniline lost in the rearrangement (see 
 the introduction). In Col. 7 we have the sum of the aniline actually found 
 and of the aniline equivalent of the phenol found. In Col. 8 the amount 
 of ammonia found is given and in Col. 9 the percentage of the calculated 
 yield found, figured on the basis of the assumption that each molecule 
 of hydrazine gives by way of triphenyl-methylamine a molecule of am- 
 monia. In Part 2, the results for benzophenone are summarized, the 
 calculations being based on the 23.94% of rearranged substance shown 
 by the aniline of the total of Col. 4 of the first part of the table. In Part 
 3 the yields of other products are given in grams. 
 
 TABLE II 
 QUANTITATIVE DETERMINATION OF THE REARRANGEMENT PRODUCTS 
 
 1. Aniline, Ammonia and Phenol 
 1 234 56789 
 
 No. 
 
 Sample 
 
 Aniline found 
 %of 
 
 Phenol 
 found 
 
 An. eq. 
 
 Sum of 
 aniline 
 and 
 an. eq. 
 
 Ammonia 
 
 found 
 
 
 G. 
 
 G. 
 
 Calc. 
 
 G. 
 
 % 
 
 % 
 
 G. 
 
 % 
 
 39 
 
 0.77 
 
 0.040 
 
 28.8 
 
 0.035 
 
 25.1 
 
 53.9 
 
 0.016 
 
 64.0 
 
 40 
 
 0.77 
 
 0.043 
 
 31.0 
 
 0.030 
 
 21.6 
 
 52.6 
 
 0.014 
 
 56.0 
 
 41 
 
 0.65 
 
 0.026 
 
 22.2 
 
 0.032 
 
 27.4 
 
 49.6 
 
 0.013 
 
 61.9 
 
 42 
 
 0..76 
 
 0.028 
 
 20.4 
 
 0.038 
 
 27.8 
 
 48.2 
 
 0.016 
 
 64.0 
 
 43 
 
 1*02 
 
 0.046 
 
 25.0 
 
 0.037 
 
 20.5 
 
 45.5 
 
 0.021 
 
 61.8 
 
 44 
 
 1.35 
 
 0.053 
 
 21.8 
 
 0.067 
 
 27.3 
 
 49.1 
 
 0.031 
 
 69.0 
 
 45 
 
 1.54 
 
 0.038 
 
 13.7 
 
 0.048 
 
 17.3 
 
 31.0 
 
 0.026 
 
 51.0 
 
 46 
 
 0.52 
 
 0.027 
 
 28.7 
 
 0.029 
 
 30.9 
 
 59.7 
 
 0.0115 
 
 67.6 
 
 47 
 
 1.61 
 
 0.071 
 
 24.5 
 
 0.067 
 
 22.7 
 
 47.2 
 
 0.0321 
 
 60.4 
 
 48 
 
 1.46 
 
 0.069 
 
 26.2 
 
 0.074 
 
 28.2 
 
 54.4 
 
 0.033 
 
 68.7 
 
 49 
 
 1.61 
 
 0.072 
 
 24.8 
 
 0.074 
 
 25.2 
 
 50.0 
 
 0.031 
 
 58.5 
 
 50 
 
 1.50 
 
 0.068 
 
 25.2 
 
 0.075 
 
 27.4 
 
 52.6 
 
 0.033 
 
 67.3 
 
 51 
 
 1.55 
 
 0.071 
 
 25.5 
 
 0.080 
 
 28.3 
 
 53.8 
 
 0.033 
 
 64.7 
 
 Total 
 
 15.11 
 
 0.652 
 
 23.94 
 
 6.686 
 
 24.94 
 
 48.9 
 
 0.3105 
 
 62.3 
 
22 
 2. Benzophenone 
 
 Sample Benzophenone-oxime Benzophenone 
 
 taken Found Calc. Found Calc. 
 
 15.12 1.22 36 1.38 1.12 1.27 
 
 3. Other Products 
 
 From steam From steam 
 
 distillate distillate residue Total 
 
 G. G. G. 
 
 Triphenylmethane 1 . 10 3 .30 4 .40 
 
 Diphenylene-phenylmethane .25 . . .25 
 
 Gum resin-steam dist. residue. . . 6 .35 6 .35 
 
 Carbonized residue (0) . . .30 
 
 Resinous material (7) . . . . .35 
 
 VII. Rearrangements of syw.-fo's-Triphenylmethylhydrazine in Other 
 
 Atmospheres than Air 37 
 
 With a view to the study of the rearrangement and the coincidental 
 thermal decomposition by accounting for the whole of the nitrogen of the 
 molecule in the products found, attempts were made to secure the re- 
 arrangement of the hydrazine in an atmosphere of carbon dioxide. It 
 was expected that possibly a relation could be found between the per- 
 centage of the fo's-triphenylmethylhydrazine that underwent the primary 
 decomposition into nitrogen and triphenylmethane and the percentage 
 of the fo's-triphenylmethylhydrazine which underwent the rearrangement, 
 and that by the measurement of the nitrogen evolved and the aniline 
 produced at various temperatures, the extent of the thermal decompo- 
 sition and of the rearrangement would be given. It was proposed to begin 
 at temperatures below 300, this being the only temperature at which, 
 at that time, the arrangement had been effected. 38 Accordingly, the 
 following experiments were carried out. 
 
 A sample (0.514 g.) of Ws-triphenylmethylhydrazine was weighed into a hard glass 
 tube, 5 g. of finely ground anhydrous zinc chloride added and the two intimately mixed. 
 The tube was connected with a source of dry carbon dioxide and by an outlet to a nitrom- 
 eter. When the apparatus had been freed from air, the tube was immersed in an oil- 
 bath at 275 and retained there until nitrogen had ceased to be evolved. The nitrogen 
 collected amounted to 3.7 cc. at 23 and 738.1 mm., uncorrected. The melt was dis- 
 solved out with ether and water, washed well with cone, sodium hydroxide solution, 
 
 36 One g. of benzophenone and 2.3 g. of triphenylmethane yielded . 95 g. of benzo- 
 phenone-oxime and 2.27 g. of triphenylmethane when the mixture was treated as out- 
 lined in the text of this paper. Calc., 1.08 g. of the oxime. Hence 1.22 + 0.13 = 
 1 .35 g. of the oxime calculated (1 .38) in the text is accounted for. 
 
 ? 7 This work was done before the decomposition of triphenyl-methylamine was in- 
 vestigated and the presence of ammonia among there arrangement-decomposition prod- 
 ucts of fo's-triphenylmethylhydrazine was discovered and its quantity taken as the 
 measure of the minimum portion of the fo's-triphenylmethylhydrazine which originally 
 split into triphenyl-methylamine and the univalent nitrogen derivative, which re- 
 arranges to give phenylimido-benzophenone and phenol. 
 
 38 By ourselves and by Stieglitz and Senior, Ref. 19, p. 2731. 
 
23 
 
 hydrolyzed and distilled with steam after the addition of sodium hydroxide. No aniline 
 could be detected in any portion of the distillate with calcium hypochlorite solution. 
 The solid hydrocarbons in the distillate softened from 70 up and melted at 78-84. 
 Beyond this they were not investigated at this time. 
 
 A sample (0.664 g.) of fo's-triphenylmethylhydrazine was treated as above except 
 that a temperature of 300 was used. The evolution of the nitrogen seemed to be 
 complete in about 3 minutes and it had all been swept over at the end of 10 minutes. 
 The gas collected amounted to 6.8 cc., at 24 and 752.5 mm., unconnected. As before, 
 all efforts to detect aniline among the hydrolyzed products failed. The hydrocar- 
 bons extracted from the distillate with steam melted at 78-81 and, mixed with an 
 equal quantity of triphenylmethane melting at 92. melted at 85-88. They were 
 not further investigated. 
 
 After the rearrangement in air had been developed to give a yield of 35% of the 
 calculated amount, the conditions of temperature, time, and container and methods of 
 manipulation were applied to secure rearrangement in carbon dioxide. Four experi- 
 ments were made. A mixture of 0.518 g. of fa's-triphenylmethylhydrazine and 7 g. of 
 zinc chloride prepared as described earlier in this paper was heated in an atmosphere of 
 carbon dioxide, prepared by the heating of pure sodium hydrogen carbonate, and dried 
 by being passed through 2 sulfuric acid wash-bottles. After the flask had cooled it was 
 immersed in an ice-bath and the nitrogen swept into the nitrometer. The melt was 
 lighter in color than those produced in air, being brown, and had no odor of phenol but 
 one resembling that of diphenyl. It was dissolved in ether and water; the resulting 
 solutions were shaken with cone, sodium hydroxide solution and the ethereal solution sep- 
 arated. To this ethereal extract, aqueous and ethereal solutions of hydrogen chloride 
 were added, and the ether was distilled. The residue was warmed for 20 to 30 min- 
 utes on the water-bath, made alkaline and distilled with steam. No trace of aniline was 
 found in the distillate. The hypochlorite test as applied easily detected 0.01 mg. of 
 aniline in the same volume of solution. 
 
 The above experiment was repeated with 0.556 g. of fos-triphenylmethylhydrazine. 
 Fifty cm. of tubing filled with glass beads covered with phosphorus pentoxide had been 
 added to the drying system. The first few drops of the distillate gave a very faint 
 coloration with the hypochlorite solution. The test was considered exceedingly doubtfuf. 
 
 The experiment was again repeated with 0.616 g. of fos-triphenylmethylhydrazine, 
 the carbon dioxide being generated from pure magnesite. The first portion (1 cc.) of 
 distillate in this case gave an appreciable color with hypochlorite solution. The color 
 was not quite the characteristic purple but may have been affected by the opalescence of 
 the distillate due to the hydrocarbons present. 
 
 One more experiment was made with 0.501 g. of the fas-triphenylmethylhydrazine. 
 As in the last experiment, the first portion (1 cc.) of the distillate with steam gave a 
 coloration with the hypochlorite solution. 
 
 In each of these experiments the first 400 cc. of the acid distillate with steam was 
 extracted, freed from dissolved ether and titrated with standard bromate solution. 
 Some bromate solution was used up and varied in amounts equivalent to 2 to 5 mg. of 
 aniline. In no case, however, was there a precipitate of tribromo-aniline, while 3 mg. of 
 aniline gives a marked precipitate of tribromo-aniline. Further, after extraction of the 
 distillate in the fifth experiment, it was diluted to exactly 500 cc. and divided into two 
 250 cc. portions. One portion was rid of dissolved ether and then titrated with standard 
 bromate solution. It required 0.71 cc. of the solution. The second portion (250 cc.) 
 was made alkaline, extracted, freed from ether, acidified and titrated. This portion re- 
 quired 0.72 cc. of the bromate solution. The extract of the second portion, on being 
 evaporated in a very slow stream of dry air, yielded no residue. The distillate of Expt. 
 4 was extracted 10 times with a total of 500 cc. of ether free from alcohol, and required 
 
24 
 
 approximately the same quantity of bromate solution as the other distillates. After 
 the standard solutions used in these titrations had been checked, and 0.0070 g. of aniline 
 out of the 0.0101 g. theoretically possible had been recovered from 0.0279 g. of phenyl- 
 imido-benzophenone on treatment identical with that applied to the fo's-triphenylmethyl- 
 hydrazine, it was concluded that if any phenylimido-benzophenone was formed by re- 
 arrangement in any of these experiments, it was exceedingly minute in amount. Traces 
 of aniline, if proved, could be explained by the presence of the 0.4 to 0.5 cc. of air found 
 to be occluded by the zinc chloride and freed when heated as described above. 
 
 The hydrocarbon extracted from the distillate with steam on evaporation of the 
 ether proved to be triphenylmethane melting at 88-90, while the steam distillation 
 residue in benzene solution yielded triphenylmethane crystals which after being washed 
 with ligroin melted at 88-89. 
 
 The nitrogen evolved in these last four experiments varied from 50 to 75% of the 
 total nitrogen in the sample taken. 
 
 Although a priori reasons make it unlikely that phenol was formed in these ex- 
 periments (see the introduction), the alkaline sodium zincate solutions from the fourth, 
 fifth and sixth experiments were saturated with carbon dioxide and extracted with ether. 
 The extracts were allowed to evaporate spontaneously. The residues showed no test 
 for phenol on the addition of ferric chloride. 
 
 These results, in the light of those from the rearrangements in air, 
 raise the rather serious question as to whether the oxygen of the air is 
 productive of the rearrangement or whether the carbon dioxide exerts 
 an inhibitive influence on it. Qualitative and quantitative determinations 
 of the ammonia in the products of the attempted rearrangements in carbon 
 dioxide will show whether the carbon dioxide prevents completely the 
 primary splitting into triphenyl-methylamine and the univalent nitrogen 
 derivative which rearranges to give benzophenone-phenylimide, or only 
 prevents this latter rearrangement after the splitting has occurred. If 
 the splitting does occur, it is much less in carbon dioxide than in air, 
 in view of the amounts of nitrogen and of the hydrocarbon products 
 found. It is considered doubtful that the splitting occurs, but only ex- 
 perimental study which includes data oh the amounts of ammonia formed 
 can determine that. With data available on the rearrangement in an 
 atmosphere of nitrogen or in vacuo a quantitative study of the ammonia, 
 aniline and phenol as well as the other products from rearrangements 
 effected in mixtures of oxygen and nitrogen, in which the percentage of 
 oxygen is made to vary from zero to that of the air, should shed valuable 
 light on the true part oxygen plays in the rearrangement. 
 
 Two arrangements, one in air and one in oxygen, carried out simul- 
 taneously in the same metal bath and with a lOOcc. Erlenmeyer flask 
 gave these results. 
 
 Time Aniline Phenol Aniline 
 
 Sample ZnClz Temp, heated found found calc. 
 
 G. G. C. Sec. G. G. G. 
 
 Air 0.5192 7 450 70 0.0307 0.0279 0.0936 
 
 Oxygen 0.5701 7 450 70 0.0307 0.0318 0.1028 
 
 Because of difficulties in securing exactly duplicate physical conditions 
 
25 
 
 conclusions cannot safely be drawn on less than a whole series of such 
 measurements, together with the quantitative determinations of the am- 
 monia produced in each arrangement. 
 
 Summary 
 Our conclusions may be summarized as follows. 
 
 1. The formation of an aniline derivative by the rearrangement of 
 syw.-fos-triphenylmethylhydrazine, as observed by Senior and the one of 
 us, was fully confirmed. 
 
 2. Benzophenone was isolated as a product of the hydrolysis of the 
 presumable primary rearrangement product phenylimido-benzophenone. 
 
 3. The decomposition products of triphenyl-methylamine, namely, 
 ammonia and diphenylene-phenylmethane and triphenylmethane, which 
 are formed when triphenyl-methylamine is heated with zinc chloride, 
 were also found in quantity in the reaction products and the conclusion 
 drawn that triphenyl-methylamine is a primary product of the action 
 studied and is then decomposed in the manner indicated. 
 
 4. Triphenylmethane and nitrogen were proved to be products of the 
 thermal decomposition of the hydrazine, the result of an intramolecular 
 oxidation-reduction in a direction reverse to that which causes the re- 
 arrangement. 
 
 5. Phenol was obtained in quantity when the rearrangement reaction 
 was carried out in air but none was found when the action was carried out 
 in an atmosphere of carbon dioxide. The hypothesis has been tentatively 
 formulated that the formation of phenol is due to the capture of oxygen 
 by escaping migrating phenyl radicals. 
 
 6. In support of this view the sum of the aniline and the phenol ob- 
 tained was found to be roughly equal to the total amount of ammonia 
 formed. 
 
 7. The observation was made that there is apparently no rearrange- 
 ment in the absence of air, a result which demands further investigation 
 of the possibility that the rearrangement is the result of an oxidation re- 
 action in which the oxygen of the air takes part. 
 
 8. The interpretation of the reactions from the point of view of the 
 theory of electronic valences is given and accords well with the explanation 
 of other rearrangements of this general type. 
 
 The author takes this opportunity to express his very great apprecia- 
 tion of the kindly interest and valuable suggestions of Professor Julius 
 Stieglitz, under whose direction this research was carried out. It is also 
 his desire to make grateful acknowledgment of the material aid enjoyed 
 during his encumbency of the Swift Fellowship. 
 
UNIVERSITY OF CALIFORNIA LIBRARY