84 FLURSCHEIM : THE RELATION BETWEEN THEX.-The Relation between the Strengths of Acidsand Buses, and the Qzcuntitative Distribution ofA@nity in the Molecule. Part 11.By BERNEURD FLURSCHEIM.IN Part I (Trans., 1909, 95, 718)* it has been shown how thehitherto inexplicable influence which many substituents exerciseon the dissociation constants of acids and bases can be approxi-mately foreseen i f the electropolar nature of the substituent istaken into consideration, and also the amount of chemical forcerequired for its linking and the steric effect exercised by it on theelectrolytic equilibrium. Mention was also made of the factthat all known constants of amines are in harmony with thetheory, with the exception of the chloro- and bromo-anilines andpaminophenol. The latter compound and p-anisidine will bedealt with in Part 111; the present paper is concerned with thehalogen-substituted anilines, and, in view of conflicting valuesgiven by different authors, also with the toluidines. Further, agraphic method has been devised by which the superposition of thethree factors can be better illustrated than by the tabular arrange-ment previously given.Lastly, an analysis of the constitution ofsome derivatives of triphenylcarbinol, etc., is intended todemonstrate how these views may be applied with advantage tothe elucidation of some disputed problems.1.-The Dissociation Constants of p-Toluidine, m-Toluidke,pChloroaniline, m-Chloroaniline, p-Bromoaniline, and m-Bromo-aniline.The following values have been hitherto obtained :p-Toluidine .........1'56 x lodg (at 25", kw=1.18 x 1O-l4) (Bredig, Zeitsch.physikal. Chem., 1894, 13, 303); 1 ' 1 3 ~ l O - ~ (at 25",kw=1*18 x (Farmer and Warth, Trans, 1904, 85,1713) ; 4.5 x 10-lo (at 15") (Veley, Trans., 1908,93,2122) ;2.2 x (at 25") (Denison and Steele, Trans., 1906, 89,999, 1386).m-Tolnidine ...... 5.9 x 10-lo (at 25") kw=1'18 x (Bredig, Zoc. cit.) ;2-9 x 10-lo (at 25") kw=1-18 x lO-l4) (Farmer and Wltrth,Zoc. c i t . ) ; 3.9 x 10-lo (at 14") (Veleg, Zoc. cit.).p-Chloroaniline . . . 1 '49 x 1 0-lo (at 25") (Farmer and Warth) ; 1-24 x 10-l1 (atlo") (Veley).* In Part I, page 727, line 11 from below, the passage " the strength of linkingsis more affected . . . ." should read "the strength of linkings is less affected .. . ."This principle has been correctly applied in the tables. Also page 729, line 3," OmO0158 " should read '' 0*00149." Some printers' errors have been correctedin the list of " Errata.STRENGTHS OF ACIDS AND BASES. 85m-Chloroaniline,p-Bromoaniline ,m-Bromoaniline ,.. 6-58 x (at 10") (Veley).,..,.2'07~10-~~ (at 18") (Veley) ; 1*04~10'~~ (at 25") (Farmer9.5 x 10-l' (at 19") (Veley).and Warth).All these constants have been determined by hydrolysis, eitherby electrical conductivity (Bredig), or by an indicator (Veley),distribution (Farmer and Warth), and velocity of migration(Denison and Steele). I n addition, hydrolytic values have beendetermined for some of these bases electrolytically by Walker(Zeitsch.physikal. Chem., 1889, 4, 319), and colorimetrically byLellmann and Gotz (Annalert, 1893, 274, 139), but no constant,has been calculated. According t o Walker, the relative strength,beginning with the weakest, is : m-chloroaniline, pchloroaniline ;p-toluidine, aniline, m-toluidine, and, according to Lellmann andGiitz : m-chloroaniline, m-bromoaniline, pchloroaniline, p-bromo-aniline; aniline, ptoluidine. After due allowance has been madefor the effect of the differing temperatures chosen by Veley, hisseries would be : mchloroaniline, p-chloroaniline, m-bromoaniline,p-bromoaniline ; aniline, m-toluidine, p-toluidine. The series ofFarmer and Warth is : pbromoaniline, p-chloroaniline ; m-tohidine,aniline, p-toluidine ; and that of Bredig : aniline, m-toluidine,ptoluidine.If this mass of contradictory evidence is sifted in the light ofthe present theory, m-toluidine should be stronger than aniline(compare Part I, tables).Similarly, m-bromoaniline should bestronger than m-chloroaniline, for the same reason that mbromo-benzoic acid is weaker than m-chlorobenzoic acid (compare Part I,tables). The relative strength of the met% and para-halogen-substituted compounds is, however, theoretically not quite so simplydeduced. It is well known that many substituents exercise astronger polar influence from the par* than from the me&position, notwithstanding the greater number of intervening atoms,which, in an open chain, reduce the polar effect of a substituent.This can only be due to a direct neutralisation of residual affinitybetween the para-atoms, as has been assumed in the benzeneformula of Claus.The strength of this diagonal bond is, however,variable, for it depends on the amount of affinity which the carbonatoms of the nucleus have to offer t o the substituents. I n para-monosubstituted anilines, the diagonal bond between the sub-stituted carbon atoms is the weaker the more unsaturated thesubstituting atom is, that is, the more affinity it can neutralise.An independent proof for this has been given by means of thedirecting influence on introduction of a substituent ( J . p . Chern.,1905, [ii] 71, 502). Whereas steric considerations would lead oneto expect that the ratio of para- to ortho-compound would b86 FLURSCHEIM : THE RELATION BETWEEN THEgreater when bromobenzene than when chlorobenzene is nitrated,the reverse has been observed.It might be argued that this isdue to the greater polarity of chlorine, which might inhibit theformation of an ortho-compound; but this view is excluded by thefact that other strongly negative groups, such as NO, and CO,which are mainly substituted in the meta-position, always yieldmore ortho- than para-di-derivatives as by-products. The onlyexplanation that remains is therefore to assume that an unsaturatedsubstituent, like chlorine, by making a great demand on theaffinity of the carbon atom t o which it is linked, primarily weakensthe diagonal bond with the par%atom to a greater extent thanthe normal bonds with the ortho-atoms, with the result that thefree affinity on which substitution depends is more increased inthe par& than in the ortho-position; the reverse happens whenan oversaturated atom, such as quinquevalent nitrogen in thenitro-group, is linked t o the nucleus, which it cannot bind asstrongly as the hydrogen atom that it has displaced.Moreover,since chlorine has a, stronger para-substituting power, in similarconditions, than bromine, it must reduce the diagonal exchange ofaffinity t o a greater extent, that is, it must be a little more stronglylinked to the nucleus, a deduction which is in agreement withother facts and views briefly outlined in a recent preliminary note(Proc., 1909, 25, 261), also with the constants of chloro- andbromo-substituted aliphatic acids (Part I, tables).I f it follows, however, as a necessary deduction from observedfacts, that the strength of the diagonal bond is variable, itremained an open question whether the transmission of a polareffect from atom to atom also varies with the amount of affinitybound by their linking.The a 5 i t y va.lues of the chloro- andbromo-anilines are capable of supplying an experimental solution.For none of the meta- or para-compounds is there a steric factorto be considered. The quantitative factor would tend to makep-nitroaniline a stronger base than the meta-isomeride (Part I,tables) ; nevertheless, p-nitxoaniline, where the strong negativeinfluence is transmitted by a strong diagonal bond, is four timesweaker than m-nitroaniline.On the other hand, taking thequantitative factor alone, the para-halogen-substituted anilineswould be weaker than the meta-isomerides. From analogy t o thenitroanilines, the polar factor would have the same effect if thetransfer of polarity were independent of the strength of thediagonal bond. If, however, that transfer showed a variationparallel to that of the bond, it is conceivable that with a weakeningof the diagonal linking, as in p-chloroaniline, a point may bereached where the diagonal influence becomes smaller than thSTRENGTHS OF ACIDS AND BASES. 87difference between the polar effects in the y- and &positions in anopen chain; in that case the polar factor becomes smaller for thepara- than for the met*position, and this difference may againconceivably be so great that it outweighs the quantitative effect inthe opposite direction. The result would be that p-chloro- andbromo-aniline would be stronger bases than their isomerides.Thesame considerations enable one to foresee the relative strength ofp-chloro- and pbromo-aniline in either case. I f the polar influencetransmitted through a bond is independent of the strength of thelatter, chlorine, being a little more strongly linked and also morepolar, would, in the para-position, weaken aniline more than wouldbromine; in the reverse case, the very fact that chlorine is morestrongly linked, would, by weakening the diagonal bond, alsoreduce its polar effect, so that p-chloroaniline might become astronger base than p-bromoaniline.Moreover, if the polar factor out-weighed the quantitative factor as regards the relative strengthof the meta- and para-isomerides, it would have to do the samewith regard to that of the para-compounds among each other. Inother words, p-chloroaniline not only might, but would, be strongerthan p-bromoaniline if pchloroaniline were stronger than m-chloro-aniline; the sequence of values of Lellmann and Gotz, and ofVeley, would therefore be impossible.To decide these important points, it was necessary to devise amore trustworthy method for the determination of the affinityvalues of very weak bases. The vastly differing results quotedabove show that no such method existed. It is well known thatvalues based on electrical determinations of hydrolysis become lessaccurate for very weak bases, whereas the colorimetric methodsdepend greatly on physiological factors, and can claim but a quali-tative usefulness.A good distribution method, however, whichwould yield values in ageement with those obtained by the con-ductivity method for stronger bases, where the latter method istrustworthy, could also be relied on to give accurate results for veryweak bases, where the conductivity method fails. Such a method,simple and suitable also for organic laboratories, would be that dueto Farmer and Warth (Zoc. c i t . ) . The results published by theseauthors for m- and ptoluidine, however, differ considerably fromthose obtained by Bredig by the conductivity method (ZOC.cit.).This has been ascribed to association. Having had some experienceof the difficulties affecting the quantitative determinationof volatile amines, the author thought that the incorrect resultsof Farmer and Warth might be due either to actual experimentalerror in the determination of the amines, or to the too highconcentrations used, which again may have been necessitated b88 FLURSCHEIM : THE RELATION BETWEEN THEthe difficulty of accurately estimating small quantities of theamine. As may be seen in the experimental part, this was foundto be the case, and the introduction of an accurate method ofdetermining the amine gave results in agreement with the deter-minations by conductivity where such were available, and closelyagreeing with each other in all the other cases.The followingvalues were obtained, a t 25O and for k,=1*18: mtoluidine,5.48 x 10-10; p-toluidine, 1-48 x 10-9; m-chloroaniline, 3.45 x 10-11;m-bromoaniline, 3.82 x 10-11; p-chloroaniline, 9.9 x 10-11; p-bromo-aniline, 8.8 x 10-11. The relative values for m- and p-toluidine andthose for mchloro- and m-bromo-aniline agree with the theory.The relative values for the meta- chloro- and bromo-compounds onthe one hand, for the para-compounds on the other, constitute, onthe basis of the present theory, an experimental proof that atom$transmit their polarity to each other in proportion t o the quan-titative strength of the bond by w h k h they are linked. Lastly,the relative values for p-chloro- and bromo-aniline also agree withthe theory.2.--GraphicaZ Illustration of t h e Three Factors zohiclb Determinethe Uissociation Constant.When a radicle is introduced two or three times successively inthe same position, the magnitude of the effect on each introductionis generally not the same.This is in accordance with thetheoretical postulate that all three factors, if they differ from 1,must vary with each substitution. When two atoms A and Bare linked, by partly transferring the polarity of each to theother, they reduce the difference in their specific polarities (compareProc., Zoc. cit.). If a second atom B is then linked to A, thedifference of polarity being now less than it was for the first B,the polar effect of the second B on A ihust be smaller than thatof the first B.Similarly, since the force with which atoms arelinked is the result of an equilibrium ( J . pr. Chem., 1907, [ii],76, 185; Proc., Zoc. cit.), that equilibrium capnot be displaced somuch by the second atom as by the first. The steric factor, onthe other hand, shows the reverse change. The behaviour ofdi-ortho-substituted when compared with mono-ortho-substitutedcompounds, and that of tertiary, secondary, and primary aliphaticacids on esterification leave no room for doubt that the stericeffect is relatively greater for each subsequent introduction of thesame substituent in the same position.All this has been duly considered in the tables given in Part I,but it can, of course, be much more accurately representedgraphically, as exemplified by the accompanying figures.ThSTRENGTHS OF ACIDS AND BASES. 89logarithms of the factors and the total effect are plotted asordinates, and the total number of substituting groups as abscissae ;the logarithms of the factors, when added, give the logarithm ofthe total effect exercised by a substituent. This total is thequotient of the dissociation constant of the substituted by that ofthe unsubstituted compound. The individual factors, althoughthey cannot, of course, be determined with mathematical accuracy,Sztbstitwtion of Hydroxyl in Bcnzoic Acid.Substitution of Chlorine in Acetic Acid. Substitution of NO, in, Beiazoic Acid.can still be approximately arrived at, especially as regards theirrelative values, by conforming to the three postulates just men-tioned, and by comparing a great number of compounds, andgenerally conforming to the principles laid down in Part I.As the curves for ammonia show, the hitherto inexplicable effectof methyl in first raising, then lowering, the constant is readilyaccounted for.For trichloroacetic acid, the older value b90 FL~RSCHEIM : THE RELATION BETWEEN THEOstwald (K=121) has been adopted, since the curves obtainedwith the constant given by Drucker (K<40) left no doubt that thisvalue is too low, a conclusion which is supported by comparisonswith other chlorinated acids. (Owing to the exigencies of space,the different figures are on different scales.)3.-The Constitution and Relative Stability of Some QuaternaryBases, Salts, and Ions.The constitution of numerous coloured organic salts is still amuch discussed point, the question being generally whether the saltsand ions have the same constitution as the corresponding base, orwhether the ionisation of the latter has been accompanied byintramolecular rearrangement.In many cases, at least in solution,differences of constitution have been established ; but in othersonly the inconclusive evidence afforded by the colour of thecompounds has been available. The present theory suggests thefollowing deductions.A linking can be broken by ionisation if the amount of affinityavailable for its formation is relatively small (see Part I), or if therespective atoms are of pronounced and opposite polarity.Now theformer of these conditions is common to all organic ‘‘ halochromic ”bases, both in their normal and pseudo-forms. I n triphenylcarbinol.for instance, the aliphatic carbon atom of the benzenoid form is to agreat extent saturated by the residual affinity of the three benzenenuclei (Thiele) ; little affinity is therefore left for the hydroxyl,which becomes ionisable (Walden). In the quinonoid form, thehydroxyl group would be attached to a carbon atom saturated bythe great residual affinity of the atoms a t the end of a chain ofcontiguous ethylenic linkings (compare pentadiene, Thiele), andtherefore also weakly bound and ionisable. Hence it is evidentthat when such a compound is dissolved in an ionising solvent,the great mobility of the electric charge must cause an equilibriumin which ions of either form are present.is determined mainly by the polarity of the atom linked to thepositive corpuscle in either case, in the sense that the electronprefers the atom most strongly heteropolar or least isopolar toitself.Which of the two isomeric forms separates as a solid fromsuch a solution depends on the product:The constantiE, = Cquinonoid ion/cbenzenoid ion(1). k3 x k,k2If it is greater than 1, the quinonoid form separates, and viceversa. In this expression k, and k, are the dissociation constantssolubility of undissociated benzenoid formsolubility of undissociated quinonoid form- x . . STRENGTHS OF ACIDS AND BASES. 91of the quinonoid and benzenoid forms respectively.These dependon the quantitative polar and steric factors, and their relativemagnitude can therefore be ascertained by means of the presenttheory. It is seen that, whereas in the ion both forms aresimultaneously present, the solid salt or hydroxy-compound may,a priori, correspond with either one of them.When the quinonoid form of two compounds, I and 11, givesa base of alkaline strength (oxonium, azonium, etc.), and thereforepractically non-hydrolysed salts, it can be shown that the relativedegree of hydrolysis of their benzenoid salts, with strong acids andfor equivalent dilutions, very approximately corresponds with theequation :(2). . . . . . Chydrololysed I-- Chydrolysed 11 = &,, x k , ,The tendency toward hydrolysis therefore is also a functionof all the three factors, as the following few examples maydemonstrate.(a) The Polar Factor.-The dimethylamino-group is much moreelectropositive than methoxyl.I n consequence, k, is much greaterfor p-dimethylamino-substituted triphenylcarbinols than for thecorresponding methoxy-derivatives, and the former are hydrolysedto a much less extent than the latter. For the same reason,N-methylquinolinium salts are more stable than benzopyriliumsalts. Since, for polar reasons, k, for a hydroxide or acetate, etc.,is invariably much lower than for a chloride, sulphate, etc., whereasfor the oxonium-, azonium-, etc., form, k, is of the alkaline order forthe hydroxide as well as for all the salts, and since k, is independentof the anion, prcduct (1) leads one to expect that the solidhydroxide, etc., will be benzenoid in many cases where the chloride,etc., are quinonoid.( b ) The Quantitative Factor.-An unsaturated substituent inthe para-position raises the quantitative factor of k,, as shown bythe formula:An independentsuch cases groupslabile and reactiveproof for this is afforded by the fact that inattached to the methane-carbon atom becomeeven where ionisation and a quinonoid changeare excluded.Thus phenylisoamylcarbinol can be distilled at 1 3 2 O(Grignard, Ann. Univ. Lyon, 1901, No. 6, page 1), whereas its p-di-methylamino-derivative gives off water at 1 20° (Sachs and Weigert,Ber., 1907, 40, 4365), on account of the weakened linking of thehydroxyl group.When, accordingly, methoxyl is introduced intoevery para-position in triphenylcarbinol, each successive substitutio92 FLURSCHEIM : THE RELATION BETWEEN THEmeans, for the respective nucleus, a corresponding rise of k,, and,through the polar factor, also of k,. By equation (2), the suc-a3I o 11 i c equilibria.cessive decrease of hydrolysis is therefore approximately geometrical(compare v. Baeyer and Villiger’s ‘‘ Potenzengesetz,” Bey., 1902,35, 3021). Similar considerations apply t o the introduction ofone, two, and three dimethylamino-groups in the para-positions oftriphenylcarbinol (Hantzsch and Osswald, Ber., 1900, 33, 278), orof one and two of these groups in the para-position with respectto nitrogen in an azoxonium or azothionium salt (compareKehrmann, Ber., 1906, 39, 923).( c ) The Steric Factor.-If the ionic equilibrium for fluoresceintrimethyl ether :@ G,H,*CO,MeOMe C,H3<:> C,H,* OM0\/ C6H4-C02MeOMe C,H,<$>C,H, OMe63C6H4-C02Me-- .- OMWC,H,<~>C,H,:OM~ (5I 63is compared with that for the similar compound not containing thecarboxymethyl group, k3 is nearly the same for both, since thesteric and polar influence of the carboxyl group counterbalanceeach other. But k,, depending only on the polar factor, is muchgreater for the fluorescein, which is accordingly less hydrolysed,notwithstanding the presence of the additional negative substituent(Kehrmann, Ber., 1909, 42, 870).It is a disputed point whether these and similar ions (azoxonium,azothionium, etc.) are ortho- or para-quinonoid; it is seen thatthey are both.The relative preponderance of the competingequilibria depends on the respective products k, and k,.If the ionic equilibria for benzopyrilium salts are compared :(A-)CHZCH - CH:QHe3 a3C6H4<o-(!7~ - C6H,<o =CH ;I STRENGTHS OF ACIDS AND BASES. 93Cf3 Cf3i t is seen that for k3 in B both the steric and quantitative factorsare greater than in A ; and the polar factor, and therefore k,, notbeing greatly different for A and B, B is hydrolysed to a lessextent than A (Decker and Fellenberg, Ber., 1907, 40, 3818;Decker and Felser, Ber., 1908, 41, 3002).In the ionican increase ofaccordingly, inequilibrium,the size of the anionthese compounds the+ --- CPh,,raises the steric factor for k,;bromides are better conductorsIc3than the chlorides (Walden).The above ionic equilibrium also facilitates the interpretationof the reactions of these compounds.If they were merely benzenoidin solution, the lability of para-substituting halogen atoms indissociating solvents would be inexplicable ; if a direct migrationof the acidic radicle in the undissociated molecule were assumedto account for it (Gomberg), then triphenylmethyl chloride insulphur dioxide would be transformed into pchlorotriphenyl-methane, since hydrogen is much more mobile than chlorine.According to the theory advanced in this paper, the exchange ofhalogen exclusively occurs in the quinonoid ion, and, the electriccharge being much more mobile than hydrogen, the former onlymigrates to the central atom.It also follows from this that saltsand double salts obtainable from a benzenoid carbinol or halogenidein a non-dissociating solvent must t,hemselves be benzenoid, whatevertheir colour (compare v. Baeyer, Tschitschibabin). The same appliesto the corresponding derivatives of distyryl ketone and similarcompounds, These deductions apply to many other ionic equilibria,for instance : diazonium salts * :H H* The reasons why Cain's formulae have been adopted in this paper are, inaddition to those adduced by Cain (Trans., 1907, 91, 1049, and recent discussionwith Hantzsch), the following :1. The fazt thar; only the normal, but not the iso-derivative, can directly change tothe diazonium form ; this excludes sterecisomerism, and is explained by Cain'sformula94 FLURSCHEIM : THE RELATION BETWEEN THEacridinium salts :e!Ietc.EXPERIMENTALIn order to effect a quantitative estimation of the volatile aminesobtained in the benzene layer by Farmer and Warth’s distributionmethod, it is not feasible to precipitate the amine by hydrogenchloride and then t o collect it.For, apart from partial thermicdissociation on drying, the mechanical losses preclude a sufficientaccuracy, especially when it is remembered that the weight of theamine in the benzene layer, and therefore any experimental error inits determination, is multiplied by 20, after which the weight ofthe amine in the aqueous layer is obtained by subtraction.(a) Determination as Acetates.-By adding a small excess ofacetic anhydride to the benzene layer, and then distilling off the2.The fact that only the isodiazohydroxide, but not the normal one, changes toa nitrosoamine excludes stereoisomerism, and is explained by Cain’s formula.3. The quantitative and polar factors being the same for 8yn- and unli-isomerides,but the steric factor being greater for the syn-, the latter would be the stronger acid.The reverse is, however, the case, notwithstanding the fact that the secondarychange of the undissociated hydroxide, which would shift the electrolytic equilibriumin its favour, is generally more pronounced for the iso- than for the normalcompound.4. An objection against Cain’s formuls has now been removed by the recentdiscovery of azomethane, whereas no aliphatic diazonium compound has yet beenobtained.5.The fact that diazonium salts do not show the characteristic reactions of theethylenic bonds of a quinone is in perfect harmony with Cain’s views ; according touniversal experience, especially as regards benzene substitution, tervalen t nitrogenis more unsaturated than an ethylenic linking, so that bromine, for instance, is firstadded by tervalent nitrogen, whereby the quinonoid configuration disappears, and aperbromide of Chattaway’s formulation is produced. Similarly, hydrogen is firstadded by tervalent nitrogen in the diazonium ion, and the bridge being therebyapened, only a hydrazine, but not a diamine, can result. Oxygen, also, is primarilyadded by the tervalent nitrogen of the diazonium ion, the bridge being broken, anda nitroarnine produced.6.Other formulae that have been put forward seem to be highly improbable.There are steric objections against Euler’s ,formula, whereas Morgan’s seems anunnecessary complication. Cain’s views leave room, of course, for an analogousortho-quinonojd formula ; but since Hantzsch (Ber., 1903, 36, 2069 ; Hantzsch andSmythe, Ber., 1900, 33, 505) has shown that the replacement of o- and p-substituentsin the nucleus of diazo-compounds probably occurs in the isodiazo-form, it need nothave any bearing on the constitution of the normal diazo- and diazoniumcompoundsSTRENGTHS OF ACIDS AND BASES. 95benzene and drying a t looo, the amine may be determined in afairly exact way in some cases, whereas no constant weight isobtained in others.Thus, with p-bromoaniline, the weight of theacetyl derivative was invariably too low.( b ) Determinution as Hydrochlorides.-The benzene was distilledoff in a current of dry hydrogen chloride, and the residue, togetherwith the flask (which had been weighed when dry), heated for anhour in a current of the same gas at looo; before opening theflask, the hydrogen chloride was expelled by dry carbon dioxide.A t first a cork stopper was used, and two tests gave the followingresults :p-Chloroaniline : 0.5774 gave 0.7394 hydrochloride ; calculated,0.7424 ; difference, 0*0030 = 0.4 per cent.p-Chloroaniline : 0-4822 gave 0.61 79 hydrochloride ; calculated,0-6199 ; difference, 0.0020 = 0-32 per cent.Subsequently it was found that, after repeated use, traces ofthe cork dissolved in the benzene under the action of the acid,and a washing flask with ground-in stopper was therefore used.(c) Determination as Picrates.-An experimentally simpler and,in view of reduced thermic dissociation, also better method consistsin heating about 1 gram of picric acid to constant weight, at looo,in a wide-necked 120 C.C.flask. The benzene solution is thenintroduced, the solvent distilled off, the flask heated for aboutforty-five minutes a t looo, and the weight of the amine obtainedby subtraction. The flask is then heated .for another hour, whenthe loss should be less than 0*0010 gram. I f it is more, heatingshould be continued until it is less (per hour), and until the lossbecomes constant (due to thermic dissociation only) ; this constantloss per hour, multiplied by the time of heating, must then beadded to the final weight.m-Bromoaniline : 0.0962 gave 0.0965 after heating for thirtyminutes, and 0.0956 after seventy minutes.p-Anisidine: 0.1166 gave 0.1169 after six hours; in this caseprolonged heating was necessary, whereas nearly always forty-fiveminutes were sufficient, since, probably through traces of oxidation,the mass became coloured and tenaciously retained some benzene.As the anisidine picrate showed no thermic dissociation whatever,the weight became finally constant and was correct, notwith-standing the prolonged heating.Since the picrate method gave results as accurate as could bedesired, the results by that method are alone given in detail, thoseby the hydrochloride method being shortly mentioned for com-parison.For p-anisidine, however (see Part 111), the latter methodwas fonnd to be preferable, there being no oxidation in that case96 FEURSCHEIM : THE RELATION BETWEEN THEDetermination of A finity Values.-In each case specially purepreparations were procured. The values were calculated by theformulze given by Farmer and Warth (Zoc. cit.). The results bythe picrate method do not vary by more than 5 per cent., a limitwhich has been declared by Ostwald to be admissible even for thedirect conductivity method (Zeitsch. physikal. Chem., 1889, 3 1,73).They differ by less than 10 per cent. from Bredig's con-ductivity values, which, in one case, differ by 18 per cent. fromeach other.The letters again signify : F = distribution coefficient ; c1 =initialconcentration of acid ; c2 = total initial concentration of amine;x =hydrolysis ; v = dilution ; s =substance used (in grams) ; T = sub-stance obtained from benzene layer. Where salt is added unders, the hydrochloride of the amine was used; in the other cases, thefree amine and the amount of N-hydrochloric acid correspondingwith c1 were employed.m-To1uidine.-A specimen wasacetylation, nitration, hydrolysis,and reduction :S. r.0.1 623 0.09070-0763 0'0428prepared from p-toluidine byelimination of the amino-group,F. :i::6 Average, 32.S. r . c,. C1. 9. X. k.0.9299 0.0357 0.00648 0-00648 154'3 0.056 5'48 x( salt)Specimen from Kahlbaum, which was redistilled :2'1794 0.0561 0.0152 0'0152 65-8 0-037 5'46 x(salt)1.8753 0.0518 0'0131 0-0131 76.3 0.0397 5-49 x(salt) Average, 5-48 x 10-lo.By Hydrochloride Method:~ = 1 * 5 9 ; ~ = 6 7 ; ~ = 0 ' 0 3 5 3 ; k= 6.10 x lO-]".p-Toluidine (Kahlbaum) :S. r. 11:0'0952 0.0528 31 '08. I'. C> C1' 9. 2, k.2-6247 0.0376 0.0183 0.0183 54.6 0'0207 1'48 x(salt )m-Chloroaniline (Schuchardt) :S. r. F.0.2370 :::: Average, 55.2. 0'33160'3169 0-22793. r. 4. c1. V. z. k.0.7793 0'2423 0-00613 0.00613 163 0'21 3-45 x locJx0'7984 0-2465 0.00626 0-00626 160 0-208 3'45 x lo-"Average, 3-46 x lo-"STRENGTHS OF ACIDS AND BASES. 97By Hydrochloride Xethod:9=1'8143 ; s570.4 ; x=0*142 ; k=3*56 xm-Bromoaniline (Schuchardt) :5. T. 2.162.3 155.5 Average, 158.9 0'1916 0,14930.3156 0-2449S. T. Czs C1' 9. X. k.0.2988 0.1673 0'00174 0-00175 575 0.345 3.73 x0.3397 0,1571 0'00163 0-00163 613.8 0-347 3'92 x 10-l'(anlt) Average, 3.88 x 10-l1.By Hydrochloride Method:~=0'3180 ; v=540*5 ; ~=0'351; k=3*36 x 10-l'.p-Chloroaniline (Schuchardt) :8. r . F. ii:: Average, 64 0.2177 0.14780.2030 0'1379S. T. c2. C1. V. X. k.0.8173 0.1222 0.00498 0.00498 201 0'143 9.9 x lo-"By Hydrochloride Method:(salt)S= 1.9702 ; V = 64.73 ; x = 0.084 ; k= 9.9 =lo-".p-Bromoaniline (Kahlbaum) :S. r. F.0.1929 0.1447 113.8 115.8 Average, 114.8.0.1908 0*1428S. r. C,. C1' V. 2. k.0.3443 0'1141 0.00165 0.00165 606 0.248 8.8 x 10-l1.(salt)By Hydrochloride Method :s=2.0975; ~ = 8 2 * 0 ; ~=0'0975 ; k=9*2 x lo-".It will be noticed that by the picrate method, dilutions varyingfrom 54.6 to 613.5 could be employed, whereas Farmer and Warth'sdilutions were much smaller, varying from 10.0 to 64.1.The author is again indebted to Dr. Senter for kind criticism,FLEET, HANTS.VOL. XCVII