MOLECULAR COMPLEXITY OF NITROGEN COMPOUNDS. 2069CCXVII1.-The Molecular Complexity, in the LiquidState, of Terlcalenh Niti-oyen Compounds.By WILLIAM ERNEST STEPHEN TURNER and ERNEST WYNDHAMMERRY.IN a review of the possible causes of association in the amides byDr. A. N. Meldrum and one of us (Trans., 1908, 93, 876), it wassuggested that the association observed might well be due to thepresence of the tervalent nitrogen atom or the oxygen atom of thegroup *CO*NH,, and the observations recorded in this com-munication were, in the main, carried out in order to test thissuggestion.Evidence bearing directly on the problem should be obta.inableby an examination of the molecular complexity of the amines, forin these substances the nitrogen atom can pass with great readinessfrom the ter- to the quinque-valent state.Such data as existed atthe beginning of the work were but meagre, consisting of deter-minations, either in solution or in the liquid state, of the molecularweight of aniline, two or three of its derivatives, and of thetoluidines (Auwers and Pelzer, Zeitsch. physilcal. Chem., 1897,23, 449; Auwers and Dohrn, ibid., 1899, 30, 542; Dutoit andFriderich, Compt. r e d . , 1900, 130, 327). It is well known, how-ever, and our research confirms the fact., that the molecularassociat.ion of an aromatic substance is either considerably less thanin the corresponding aliphatic compound, or is non-existent. Wedecided accordingly, whilst including a number of aromatic sub-stances, to make a study of the aliphatic amines.The research was extended, however, beyond a comparison of theamines and amides, and was made to include a survey of othernitrogen-containing compounds.Of these, some, notably acete,propie, butyrs, and benzo-nitrile, had already been examined byseveral investigators (Ramsay and Shields, Trans., 1893, 63, 1089 ;Dutoit and Friderich, Zoc. c i t . ; Guye and Baud, Arch. Sci. phys.nat., 1901, [iv], 11, 449; Renard and Guye, J. C'him. phys., 1907,5, 97), although their results do not in all cases agree as well asmight be expected. We have included in our work redeterminationswith these four substances, but have not given full results in eachcase.As regards the nitro-compounds, Ramsay and Shields (Zoc. c i t . )have proved, by a comparison of nitroethane and nitrobenzene, thataliphatic nitro-compounds are associated, whilst aromatic compoundsare non-associated.The observations on this class of substances wehave not extended, but the possibility of association connected wit2070 TURNER AND MERRY : THE MOLECULAR COMPLEXlTY, IN THEthe nitroscgroup, it5 revealed by the tendency of a few derivativesof hydrocarbons to form double molecules at low temperatures(Piloty, Ber., 1898, 31, 452; Bamberger and Rising, Ber., 1901,34, 3877), attracted us to examine-three nitrosoamines, of whichdimethylnitrosoarnine is of particular interest on account of its con-siderable conductivity, solvent power, and dielectric constant(Walden, Zeitsch. physikal. Chem., 1903, 46, 103).From the highvalue of the dielectric constant we expected to find association.The determination of the molecular complexities in the liquidstate, rather than in solution, was attended by certain advantages,for our results render it possible to make a comparison, in somecases, with the extent of association in solution; and we have alsobeen able to include a number of substances the solubilities of whichin benzene and similar solvents are very slight, the advantage ofbeing able to include formamide being considerable. As far aspossible, the experiments have been carried out over the same rangeof temperatures.Against these advantages, we have to set off the fact that thecalculation and interpretation of the results by the Ramsay andShields’ method, which has been used in its original form, is opento objection, and our own determinations support those of Dutoitand Friderich and of Guye and his pupils in demonstrating thatthe Ramsay and Shields’ formula can only be applied.within some-what circumscribed limits. We have not been able, owing to lackof data in most cases, to employ any of the alternative methods ofdetermining the molecular complexity, such as have been suggestedby Ramsay (Proc. Roy. Soc., 1894, 56, 175) and Walden (Zeitsch.physikal. C’hem., 1909, 65, 184).EXPERIMENTAL.The mea.surement of surface tension wm carried out inapparatus of the U-tube type, the capillaries used in the construc-tion being previously carefully tested and calibrated at differentpoints.I n form, the apparatus was very similar to that employedby Hewitt and Winmill (Trans., 1907, 91, 441)-reference to theirdiagram will serve to explain our arrangement-but differed fromit in two respects: the capillary tube was backed by enamel, and,more important still, the end of the capillary tube, to the extent ofabout one inch, was bent sharply downwards, and a piece of quilltubing of the same diameter fused on to it, this added tube beingbent sharply upwards so as to be parallel both with the Capillaryand with the wide limb of the U-tube. The object of this elbowof quill tubing was twofold. It served, in the first place, as it traLIQUID STATE, OF TERVALENT NITROGEN COMPOURDH. 2071for particles falling from the rubber connexions, and, in the secondplace, as a reservoir into which liquid could be t,ransferred from thecapillary.Constant temperatures were obtained by the use of large bathsand a carefully regulated flame. A t the higher temperatures weemployed a bath of paraffin wax (m.p. 60°), finding it muchpreferable to concentrated sulphuric acid, and we were able to useit at temperatures up to 2 1 0 O .In most of the experiments, the air was removed from theapparatus by exhaustion to a low pressure, although it has beenshown by Renaud and Guye (Zoc. cit.) that unless the substance iseasily oxidisable, measurements of the surface tension made both inpresence and in absence of air agree to within about4 per cent.A t each temperature, three readings of thecapillary rise were made by means of a readingmicrometer, the differences never amounting to morethan four hundredths of a millimetre. A freshsurface was obtained for each reading by t'ilting thetube and allowing liquid to run from the capillaryinto the elbow of quill tubing. The substances ofhigh melting point could not be treated in this way,and in these cases a fresh surface was obtained bytemporarily forcing the liquid from the capillaryinto the wide tube.Two independent series of deter-minations of the surface tension were made witheach substance, in many cases by employing two tubesand placing them side by side in the same bath.The density determinations were mainly made inspecific gravity bottles of 10 C.C. capacity, calibratedat five or six different temperatures between 20° and90°, so that the volumes at other temperatures couldbe deduced.A t the higher temperatures considerabledifficulty was experienced in using this form ofpyknometer, and the simple form shown in the diagram was devisedand employed, the narrow tubing being capillary of 1-1.5 mm.diameter of quill tubing of about the same bore. The cup wascharged with the solid, and the pyknometer transferred to theheating bath. The liquid level was adjusted by air pressure to themark b , excess above the level a removed by fine capillary tubes, andthe cup wiped out before weighing. All densities are comparedwith that of water at do.With the exception of benzamide and salicylamide, the materialswere obtained by purchase, mainly from Kahlbaum, and all wer2072 TURNER AND MERRY: THE MOLECULAR COMPLEXITY, IN THEaubjected to purification save a pure specimen of phenylhydrazineobtained from Kahlbaum.The liquid amines were treated with solid sodium or potassiumhydroxide, and afterwards distilled, benzyl-, dibenzyl-, and triamyl-amine under diminished pressure, the others under atmosphericpressure.Distillates of constant, or almost constant, boiling pointwere obtained and used throughout. Diphenylamine and tribenzyl-amine , were crystallised from alcohol until of constant meltingpoint,. Wherever possible, the densities obtained were comparedwith those of other investigators. The density of the anilineemployed agreed exactly with that found by Bruhl (Zeitsch.physikal.Chem., 1895, 16, 193) ; and those of propyl-, dipropyl-,and tripropyl-amine were in very close agreement with the numbersgiven by Perkin (Trans., 1889, 55, 693).The nitrosoamines (from Schuchardt) were treated as follows :phenylmethylnitrosoamine with freshly heated sodium sulphate,the two others with recently ignited potassium carbonate, and sub-sequently distilled under diminished pressure.Of the nitriles, acetonitrile, benzonitrile, phenylacetonitrile, andmtoluonitrile were treated with phosphoric oxide and afterwardsdistilled, the acetonitrile under atmospheric, the others underdiminished pressure. Propio-, butyro-, and isobutylaceto-nitrile werekept over sodium sulphate, and then distilled under atmosphericpressure. Lacto-, mandelo- and o-toluenitrile were merely distilledunder diminished pressure.The melting point of the p-toluonitrilewas 28*4O, and the substance was used without further purification.The densities at 20° of acetonitrile and benzonitrile agreed veryclosely with the values found by Bruhl (Zoc. cit.), but those ofpropio- and isobutylaceto-nitrile were a little higher, that of phenyl-acetonitrile a little lower, than Bruhl found.The formamide was well dried over sodium sulphate and after-wards distilled under diminished pressure. Its density at ZOOwas in good agreement with that given by Bruhl. Theother amides, the anjlides and urethanes were purified by methodsalready described (Meldrum and Turner, this vol., p. 1607).A t the conclusion of the surface-tension measurements, the meltingpoints of the salicylamide and phenylacetamide were tested, andfound to agree exactly with those of the original substances.I n addition to the amides on which we have been successful inmaking measurements, we endeavoured to bring into the scope ofthe work a number of diamides and ethyl oxamate, but found themall unstable.Carbamide decomposed at its melting point ; ammoniafrom malonamide could be detected even before melting; whilstethyl oxamate, when gradually heated, gave ammonia at 130OLIQUID STATE, OF TERVALEKT NITROGEN COMPOUNDS. 2073Neither could we find a suitable solvent for these substances. Theirinsolubility in benzene, ether, etc., is well known. Lachman(Zeitsch. physikal. Chern., 1897, 22, 170) found that ethyl oxamateis soluble in methyl oxalate, and, in this solvent, gave molecular-weight values indicative of decided association.We used methyloxalate, ethyl oxalate, diphenylamine, and phenylurethane inattempts to dissolve carbamide, oxamide, malonamide, and succin-amide, but found all these amides either insoluble or only slightlysoluble. We possess indirect evidence, however, showing that thesesubstances should be classed as associated.I n the following tables are recorded the molecular weight of thesubstance, the temperature ( t ) , the capillary rise in cm. (h), theradius of the tube in mm. ( T ) , the density (p), the surface tension(y), the values k1 and k, of the Ramsay and Shields' constantscalcula.ted from the separate experiments, k, the mean values, andthe degree of association (x).I n the case of formamide the mean values of k are derived fromfour sets of experiments.to.10"20304520"30456020"3045607520"30456020"30456075h .3.6463-5063.3683'1643.4623'3343'1522-9623.3423.2383.0792-9192.7603 '4623'3523.1843'0183-3093'3103.1723.0342-892VOL. XCVII.A mine 8 .n-PT-opgZa7ni71c (h1.W. = 59).T* P.7- k,.0.1779 0'7271 23.13 -0.1779 0.7185 21.98 1'830.1779 0 7081 20.81 1-830 1780 0'6894 19'04 1-82Dipyopylamine (M.W. = 101).0.1779 0 7390 22'32 -0 1799 0.7299 21.23 2.420.1780 0-7164 19-72 2-270.1780 0-7i25 18,17 2'37Tripropytanzine (31. W. = 143).0.1850 0.7571 22.96 -0.1851 0.7493 22.03 2-560.1851 0.7373 20'61 2-640.1851 0.7252 19'22 2'630.1851 0.7130 17.87 2.61isoAniylantine (M.W. = 87).0'1851 0'7506 23.59 -0 1851 0.7417 22-57 1-990.1851 0,7277 21.04 2.030.1851 0.7128 19-53 2 00Triamylanziae (KW. = 227).0.1851 0.7859 24-25 -0.1851 0.7790 23'41 3.090.1851 0.71376 22-11 3'180.1851 0.7568 20.85 3'140'1851 0'7461 19-59 3-20k2. -1.851 *841.86-2.142-312.33-2'432 $32 *652.65I1.922 '032.03-3 *123.113 *073-17k.1.841 '8351 -84--2-282'292-35-2.4952.6352'642.63-1.9552-032.0153'1053.1453'1153-185G Ux,-1.241.241.24-0 -900.890-86-0.780 '720.720.72-1-131 *071.08-0'560.550.560'52074to.60"759010520"3045607520"304560i 595"10512013520"30456020"3045607520"3045607520"3045607590TURNER AND MERRY: THE MOLECULAR COMPLEXITY, IN THEh.4'2684-1504'0233'8904,3884.2984 '1 554.0153.8724-3624.2804.1504.0263 8913.7073 6283-5043.3784'5i24 4804.3224'1794.2694'1734.0203.8583.6923.8383.7583.6283.4903.3504-3944.3104'1824.0483.9163'782A mines (continued.).Biphenylantine (M.W. = 169).r. P. Y- P I . k2.0.1777 1.0547 39-23 - -0.1778 1.0435 37.77 2'37 2.660.1778 1'0326 36.23 2.55 2'460.1778 1'0217 34.66 2.63 2.55Benzylamine (M. W. = 10i).0'1850 0'9813 39.07 - -0.1850 0.9727 37'94 2 09 2.070.1850 0.9597 36.17 2.19 2-050 1850 0'9463 34'49 2-12 2'050'1850 0.9338 32.81 2.28 2.17Dibenzyhnine (M.W. = 197).0.1850 1'0276 40'68 - -0.1850 1'0199 39'61 2.90 2.710'1850 1.0083 38'01 2.92 2-780.1880 0-9963 36 40 2.93 2-820-1850 0.9844 34.79 3'01 2.99TribencgZanzine (M.W. = 287).0.1850 0.9912 33'34 - -0-1850 0'9850 32'43 3-42 3-400.1850 0.9741 30.97 3-59 3'460.1851 0.9632 29'54 3.57 3'59Phenylhytbrazzne (M.W. = 108).0.1850 1.0978 45-55 - -0.1850 1.0899 44'31 2.18 2.160-1850 1.0777 42'27 2-47' 2-270.1850 1.0653 40'40 2.23 2.16DimetJbyZnitrosonminc (M. W. = 74).0-1850 0.9965 37'73 1-74 1.750'1850 0.9813 35'80 1.85 1'860'1850 0.9654 33'80 1-94 1'930.1850 0.9491 31.80 1.97 1'920.1850 1.0059 38.97 -Disthylnitrosoamine (M.W.= 102).0.1850 0.9422 32.81 - -0.1850 0'9331 31.82 1.79 1.770.1850 0.9197 30.28 1-90 1'890,1851 Om9061 28'71 1.97 2'000.1851 0.8919 27.13 2.01 2,02Phenylmethylnitrosoctmilac! ( B I . W. = 136).0.1850 1.1275 44.96 - -0.1850 1'1187 43.75 2.38 2-300.1850 1.1055 41'95 2.40 2'430.1850 1.0919 40'11 2-49 2-510-1850 1-0782 38-31 2-45 2'480.1850 1'0644 36-53 2-47 2-55k.2.5152-5052.59--2 -082.122.0852.225-2.8052-852.8753 *OO-3-413.5253.58-2-172 *372.195-1.7451 35.5 z .9351.945-1 *781'8951.9852.015I2 -342.4152 $02-4652-512.-0-770 7 80.74-1-031 -001 -030.97-0.660 '640 '630.59-0 '490'470'46-0.970-850.95-1 '341.221.151.14-1 '301'181-101.08-0-860-820 *780-800.7i?.20"30456020"30456020"30456020"30456020"30456020"30456020"3045607520"3045607530"456075LIQIJID STATE, OF TERVALENT NITROGEN COMPOUNDS.h.4.0593.9893.8793.7653.8923.7803,6083.4313.6373.5483.4103'2684.4074 3154'1764-0104.4884'4044.2764.1364.5154.4504.3464 -2344.1894.1003.9613.8093.6564.0894-0123.8903.7603.6244-1123.9783'8463-701iVitciles.Lactonitrile (M.W. = 71).r. P- Y. k,. k2.0'1850 0.9377 36.38 - -0.1850 0.9788 35-43 1.27 1.240.1850 0'9656 33-99 1-31 1'330-1850 0.9525 32-54 1-35 1.30Butyronitrile (M.W. =69).0.1780 0'7936 26.97 - -0-1780 0.7842 25-88 1.73 1-700.1779 0'7701 24.25 1.77 1.790.1579 0'7556 22.62 1-78 1.77is0 Butylacetonitrile (M. W. = 97).0.1850 0.8035 26.53 - -0-1850 0'7955 25 61 1-80 1-810.1851 0.7827 24-23 1.83 1.850-1851 0.7699 28-84 1.89 1.87Benaonitrile (M. W. = 103).0'1776 1*0051 38.59 - -0.1777 0.9974 37-61 1.93 1.950.1777 0.9831 35.78 2-03 2-080.1778 0.9692 33.89 2-32 2.25Phenylacetonitrile (M. W. = 117).0'1850 1,0157 41.36 - -0.1850 1'0076 40'27 2.09 2.070.1850 0-9939 38-56 2'14 2-130-1850 0'9792 36-75 2-32 2-29Mandelonitrile (M. W. = 133).0.1776 1.1165 43.91 - -0.1776 1.1086 42.98 1.78 1.870.1776 1'0966 41.52 1-87 1.890.1777 1'0844 40'02 1-96 1.93o-Toluoititrile (M. W, = 117).0.1850 0.9955 37'84 - -0.1850 0.9863 36'70 2-21 2-210.1850 0.9i37 35.00 2.25 2.330'1850 0.9596 33.17 2'45 2-270'1850 0-9481 31-46 2-39 2.32m- Toluonitrile (M.W. = 117).0'1778 1'0316 36.79 - -0.1778 1'0235 35'81 1'84 1-840.1778 1.0122 34.34 1-91 1.910-1778 0-9997 32'78 2.03 1'990.1779 0.9872 31-22 2.07 2.05p-Toluonitrile (M. W. = 117).0.1850 0.9785 36.51 - -0-1850 0'9640 34-80 2'22 2.210.1850 0 9512 33.20 2'13 2 2 20.1850 0'9390 31-54 2'29 2.178.1 -2551'321 -325--1.7151.781-775-1 *8051.841'88-1.942-0552-285-2.082.1352.315-1.8251.881'945-2.212-292 -362 *355-3. '841.912.012 -06-2.2152'1752 *232075X .2 '212 '042 03--1.381.301-31-1.271'241 -20-1.141-050'89-1 *030.990.88-1 -251.201'14-0.940.890.850.85-1'241-171.081'04-0 -940-960-086 u 2076to.20"3045607585"95105120aoo9010512080"903 05120130"140150160170160"170180140"15016017060"7590105120'130145160TURNER AND MERRY: THE MOLECULAR COMPLEXITY, IN THEh.5.8125'7705'7045'6345.5294.3004'2264-1474.0203-7963.7183,5993.4784.2724'2264.1524'0763.8863.8043.7563'7023.8463 -6443.5873'5223-7853.7463.7003.6483.8713'7903 *7003.5993.9383.8683-7623'6527..0-17720.17720.17730.17730.17730.18500 18500.18500*18500.17780 17790.17790.17790.18500.18500.18500.18500.18500.18500-18500.18500.1850Amides.Formnmide (M.W. = 45).P * Y- k,.1'1350 57'35 -1.1267 56-51 0.651'1142 65.27 0.641:1015 53.94 0.731.0892 52-36 0.93Acetamide (af.W.=59).0'9904 37.98 1-180'9822 36 96 1'240'9703 35-39 l . 3 1Propionamide (M. W. = 73).0 9517 30.88 1.2909395 29-50 1-350'9272 28 14 1.36Lnctanzule (M.W. = 89).1'1301 43'34 1.061'1181 42-13 1.121'1062 40 91 1'130.9986 33'96 -0'9597 31.77 -1.1381 44'12 -Bewzxamide (M. W. = 121).1.0792 88.06 -1.0717 37'40 1'111.0641 36.73 1-161.0565 36.01 1.291.0489 35-23 1'43k,.-0.610.650.730.941-141.171 -27-1.331-311'37-1.041 *131-15-1 *081-261-251'43PhengZncetmtide (M.W. = 135).0.1850 1,0179 33.66 - -0-1850 1-0105 32.89 1'58 1.510.1850 1'0029 32.05 1.77 1-70Sall'cylanaide (M.W.= 137).0.1850 1'1749 40'35 - -0.1850 1'1663 39'64 1'22 1.180.1850 1.1578 38.87 1'40 1'360.1850 1.1493 38'04 1-55 1-55Anitid es and Urethanes.E'ormanilide (M. W. = 121).0.1850 1'1115 39.04 - -0.1850 1.0971 37.73 1'49 1-480-1850 1'0866 36'48 1'55 1-550*1860 1'0743 35-08 1.75 1-70Acetanilide (M.W.=135).0'1778 1'0261 35-24 - -0.1778 1.0179 34'34 1-86 1.880.1778 1.0055 32.99 1-87 1.890.1779 0'9933 31.65 1.88 1-93k.0 %30.660-740.95--1-161 *2051 -29-1.311 -331.3651.051.1251-14-1.0951.211'271.43-1 -5451.735-1 *201 -381.55-1.4851 *561.725-1 '871 *881.9052. -6'185.764-853'34-2-472.332.11-2'062'011'9412 872 5 92-54-2.702.322.161.81-1-611 -35-2.351-901'59-1 -691 -601'36-1 -211-201.1LIQUID STATE, OF TERVALENT NITROGEN COMPOUNDS.207 7Anilides and Urethanes (continued).Methylacetanilide (M. W. = 149).to. h. T. P. Y. k,. k2. k. X.105" 3.524 0.1850 1.0036 32.09 - - I -115 3'448 0-1851 0'9951 31.15 2-14 2.22 2.18 0.96120 3.406 0-1851 0-9910 30.65 2-22 2.21 2'215 0'94130 3.330 0.1851 0.9528 29.71 2-24 2.24 2.24 0.92145 3.206 0.1851 0.9703 28.24 2 3 3 2'27 2'30 0'88The ranges of temperature for which the values of k are calculated are :-105-115" ; 105-60" 3,80075 3.68190 3'554105 3'42460" 3.31475 3'20590 3.095105 2.98060" 3.68075 3.57690 3.468105 3.384-120" ; 115-130" ; 130-145".Ethylacetanilide (1f.W.= 103).0.1850 0'9938 34.27 -0.1850 0'9798 32-73 2-460.1850 0 9657 31-15 2.580.1851 0.9516 29.58 2.59Ethylurethane (M. W. = 89).0.1851 1'0459 31.47 -0.1851 1.0313 30.01 1'510.1851 1'0162 28-56 1.530.1851 1.0005 27'07 1.58Phrnylzcrethane (M.W. =165).0.1850 1'0792 36-04 -0'1850 1.0677 34'65 2.180'1851 10538 33-18 2.250.1851 1.0399 31'67 2.39-2.462'492.60-1-481.511.56-2-152.192-41-2 '462.5352.595-1'4951-521-57-2-1652'222'40-0.800 *760.74-1.691'651.57-0-970-930.83The following results, which we do not consider it necessary togive in full, have also been obtained. Acetonitrile: 20-30°,x = 1.58, in good agreement with the values of Dutoit and Friderich(Zoc.cit.) and Renard and Guye (Zoc. cit.). Prop'onitrile: 20-30°,x = 1-44, agreeing substantially with the results of Ramsay andShields and Renard and Guye, but not with those of Dutoitand Friderich. Aniline : 20-45', k = 1.695, x = 1-40 ; 45-75',k = 2.005, x = 1.09, in agreement with Dutoit and Friderich.We have also confirmed the abnormal result obtained by Dutoitand Friderich for diphenylamine.In the case of ethylurethane, we have obtained values whichagree well with those of Guye and Baud, but entirely differentresults with phenylurethane. The authors mentioned found,between 63'8' and 108'8', k = l * 3 8 ; and 108*So and 152.8O, L=1*81,values indicative of pronounced association.In benzene solution,phenylurethane is but slightly associated, much less so than ethyl-urethane, which, at the lower range, Guye and Baud did not findassociated as much as phenylurethane.k = 1-47', x = 1.73 ; 3 0 4 5 O , k = 1.53, x = 1-63 ; 45-60', k = 1-56,Ic = 1-63, x = 1-48 ; 30-45', k = 1.63, x = 1.48 ; 45-60', k = 1.66,The general results are discussed in the sections below2078 TURNER AND MERRY: THE MOLECULAR COMPLEXLTY, IN THEn-Propylamine ..................... 1-84Dipropylamine.. ................... 2.29Tripropylamine .................. 2.635GoAmylaniine ..................... 2 03Benzylamine ............ 2.12Dibenzylamine ......... 2-85Tribenzylamine.. ...... 3.58 (120-135"LIQUlD STATE, OF TERVACENT NITROGEN COML'OUNDS.207 9isodmylamine (M.W. = 87'1).Solvent : 15'03 grams.zu (grams). A'. M.W. (0bs.j. 2.Longinescu's method ( J . Chirn. phys., 1903, 1, 296), and from thecryoscopic observations of Freundler (BUZZ. SOC. chirn., 1895, [iii],13, 1055).Other methods of testing the molecular complexity in the liquidstate are, however, not in favour of the idea that dissociation occurs.Walden (Zoc. cit.), using a formula which he had found to begenerally valid for non-associated liquids, showed that it number ofthe substances having high values of the Ramsay and Shields'constant did not differ from the well-recognised normal substances ;whilst Kurbatoff and Eliseeff ( J . Russ. Phys. Chem. SOC., 1909, 41,1422) have pointed out that the esters examined by Homfray andGuye are normal according to the values of Trouton's constantwhich they possess.Evidence of dissociation in the liquid state should be revealed,perhaps to a less extent, in solution, and the dissociation shouldincrease with the concentration. We have, accordingly, determinedthe molecular weights, in benzene solution, of the amyl- and benzyl-ITrianzyla?mhc (M.W.= 227-3).Solvent : 14.83 grams.w (grams'. Aa. M.W. (obs.). x.3l.W. (obs.)M. W. (calc.)'amines. In the tables, the association factor, x,=Benzylnmine (M. W. = 107.1).Solvent : 16.06 grams.0.5562 1.568 110.4 1-030.7489 2'055 113.5 1-061.348 3'485 122.4 1-141.497 3*7W 122'9 1-15Dibcnzylaminc (M.W. =197*1).Solvent : 14.88 grams.0'1395 0-253 185.3 0'940'4995 0,871 192.7 0.981.187 2.021 197.3 1-001-490 2,536 197'4 1.00ITribciazyZaminc (M.W. = 287.2).Solvent : 15.23 grams.0.3406 0.402 278.3 0-9703660 0.926 271.3 0.951-173 1'406 273 8 0 '951'534 1'861 270 7 0.94Our results afford further evidence that the primary amines areassociated, but there is no evidence of any dissociation of triamyl-amine or dibenzylamine. The results with tribenzylamine are low,but in no way commensurate with the apparent dissociation in thefused state.w (grams). AO. M.W. (obs.). z2080 TURNER AND MERRY: THE MOLECULAR COMPLEXITY, IN THEAs a final test, we plotted the values of molecular surface energyagainst the temperature for triamylamine, dibenzyl- and tribenzyl-amine, since Dutoit and Friderich (Zoc.c i t . ) found that thecoefficient of the molecular surface energy of the normal liquidswhich they examined was independent of the temperature. Thefollowing values were used :Triamylamine, 1060.7 ; 1029.8 ; 982.1 ; 935.0 ; 887.0.Dibenzylamine, 1353.0 ; 1324.0 ; 1280'2 ; 1257.7 ; 1190.5.Tribenzylamine, 1460.0 ; 1425.8 ; 1372.0 ; 1318.4.The temperatures are given in the tables (pp. 2073, 2074). Ineach case, the straight line joining the end-points passed, almostperfect'ly, through all the points.We must conclude therefore that the abnormal results underdiscussion are due to the non-validity of the Ramsay and Shields'formula.The Nitrosoamines and the Nitrites.As in the nitrclcompounds, so also in the nitroso-compounds hereexamined, association occurs only in the aliphatic series.The causeof association is to be connected with the nitrosegroup, since thesecondary and tertiary amines are non-associated.The tendency of the nitriles to associate is also only marked inthe aliphatic series. Benzonitrile has a slight tendency toassociation, and the property is exhibited distinctly by m-toluonitrile.The other aromatic nitriles, save mandelonitrile, which is associated,exhibit abnormally high values of k.It will not escape notice that the introduction of a hydroxylgroup into the substance considerably raises the association, as inlactonitrile and mandelonitrile (see also lactamide).The Arnides.The amides in the liquid state are very strongly associated, and,unlike the nitriles, nitro- and nitroso-compounds, this associationextends to the aromatic as well as to the aliphatic compounds.Indeed, the extent of association in benzamide and salicylamide isstriking.From the following table of association factors, it will be seenthat the extent of association is roughly of the same order as thatexisting in the hydroxyl-containing substances-the organic acidsand the alcohbls.The data for the acids, alcohols, water, andphenol are taken from the papers of Ramsay and Shields (Zoc. cit.)and Ramsay and Aston (Trans., 1894, 65, 168). Since data aLIQUID STATE, OF TERVALENT NITROGEN COMPOUNDS. 2081exactly comparable temperatures are not available, the actualtemperatures are quoted :H,O (20-30') 3-81C,H,*OH (16-46") 2-25 C,H,*CO,H (16-46") 1.77CH,*OH (16-46") 3-43 H*CO,H (16-46") 3-61C,H,*OH (16-46") 2-74 CH3'00,H (1 6-46") 3'62C,H,'OH (46-78") 1-43H*CO'WH2 ( 20- 30") 6.18CH3*CO*XH2 ( 85- 95") 2-47CzH5'CO~NHz( 80- 90") 2.08C,H,'CO*NH, (130-140") 2.70The outstanding feature of the results recorded is undoubtedlythe high associative power exhibited by f ormamide.Walden(Zeitsch. phpsikd. Chem., 1906, 54, 180) expressed the belief thatthis substance is strongly associated, but made no measurement ofits complexity save in aqueous solution, in which it possessed thenormal molecular weight. Again (Proc. Faraday Soc., 1910, 156),he states that "formamide appears to reproduce nearly all thevaluable qualities of water." Save certain fused salts and sulphuricacid, formamide is more strongly associated than any other liquidyet examined. Its molecular complexity, however, decreases rapidlywith rise of temperature. Between 20° and 75O, its complexity dropsfrom 6.18 to 3-34, whilst that of water falls only from 3-44 to 2.9.We suggested (Proc., 1910, 26, 128) that the solvent power offormamide for salts ig connected with its high molecular complexity.Acetamide has also been found a solvent for salts (Walker andJohnson, Trans., 1905, 88, 1597; Menschutkin, J .Russ. Phys.Cbem. SOC., 1908, 40, 1415). Formamide and acetamide can also,like water, produce hydrolysis of antimony trichloride (Bruni andManuelli, Zeitsch. Elektrochem., 1905, 11, 554).Paasing to the anilides and urethanes, we note that associationis diminished by substitution of hydrogen in the amidsgroup.Theresult with acetanilide was unexpected. The measurements ofAuwers, of Beckmann, and of Meldrum and Turner, made onsolutions of the anilides, show clearly that acetanilide is distinctlymore associated than formanilide. Quite a different result isobtained on comparing the two substances in the liquid state.The question whether the tervalent nitrogen atom is responsiblefor the association in the amides can, we believe, be regarded asanswered in the negative. Methyl- and ethyl-acetanilide are non-associated, as also are secondary and tertiary amines, in all of whicha tervalent nitrogen atom is still present. It is also obvious, fromthe results with methyl- and ethyl-acetanilide, that the oxygenatom does not bring about association (compare MeIdrum andTurner, 1910, 97, 1616), a conclusion in agreement with what isalready known concerning esters, acid anhydrides, and ether2082 MOLECULAR COMPLE Y l T Y OF NITROGEN COMPOUNDS.Whatever the properties of chemical combination possessed bythe nitrogen or oxygen atom, it appears clear that they cannot beheld to be the cause of molecular association, and, in most cases,perhaps in all, association only occurs when these elements arepresent in distinct electronegative groups.I n the case of the amides, it appears that association is onlypossible when hydrogen is still present in the amide group.Thepower of molecular association disappears only when the hydrogenis eliminated from this group. Formamide and acetamide, also,like the hydroxylic substances methyl and ethyl alcohols, ethyleneglycol, and glycerol, can combine with salts in the same way aswater enters into union as water of crystallisation (Titherley, Trans.,1901, 79, 413 ; Walker and Johnson, Zoc.cit. ; Menschutkin, J . Russ.Phys. C'hem. Soc., L906, 38, 1010; Grun and Bockisch, Ber., 1908,41, 3465 ; Rohler, Zeitsch. Elektrochem., 1910, 16, 419).Such facts as these might be used as evidence in favour of thehydroxylic constitution of the amides. But the arguments againstthis theory are very weighty (Meldrum and Turier, Trans., 1908,93, 890), and we have to remember that not only do water andf ormamide possess like properties, but liquid ammonia, an associatedliquid, closely resembles water, can produce hydrolysis (ammonolysis,Franklin, J .Amer. Cliem. SOC., 1905, 27, 820), and, like water, cancombine with salts.It is difficult to locate the exact cause of the association in theamides. The apparent connexion between association and powerof producing hydrolysis indicates another method by which thecause of association in the amides might conceivably be tested. I f ,in water, for example, the hydroxyl group is responsible both forthe association produced and also for the hydrolysing power ofwater, then we might assume that the grour, in the amides whichproduces hydrolysis is also the cause of association. Bruni andManuelli (Zoc. cit.) have found that when antimony trichloride ishydrolysed by formamide or acetamide, the entering group, whichis equivalent t o one chlorine atom, and therefore to the hydroxylgroup, is R*CO*NH.Evidence of the nature of the action, if any,of the anilides and urethanes is desirable in this connexion.Finally, our results bear out the general connexion between thedegree of association and the dielectric constant of a liquid. Quit8recently, Walden (Zeitsch. physikal. Chem., 1910, 70, 569) haspointed out that all substances with high dielectric constants possesscertain electronegative groups, as OH, NO,, CO, CN, NH,, etc.Such groups we know to be present in those carbon compoundswhich exhibit association, and we should expect to find the dielectricconstant and the degree of association run parallel. I n the paperDECOMPOSITION OF PEKSULPHURIC ACID, ETC. 2083already referred to, Walden has made comparison of the twoprope?ties. We quote the following values of the dielectric constant,in connexion with the fresh data brought forward in this com-munication : Formamide (20°), >84 ; acetamide (83O), 59.2 ; di-methylnitrosoarnine (20°), 53.3 ; lactonitrile (ZOO), 37.7 ; formanilide(liquid), 20.5 ; acetanilide (liquid), 19.5 ; phenylacetonitrile (21*5O),18-2.These numbers, and the more extensive comparisons by Walden,show that i t is generally true that associated substances have highdielectric constants. The converse is by no means true, although,as may be seen, the substances of highest dielectric constant arethose which have the highest association factors.As regards dimethylnitrosoamine, it is quite possible, bearing inmind the abnormal results obtained with secondary and tertiaryamines, that its degree of association is greater than our measure-ments reveal.The connexion between the dielectric constant and the degree ofassociation, although at best approximate, leads us, when taken inconjunction with the fact that the elements nitrogen and oxygenwith unsaturated valencies do not appear of themselves to causemolecular association, to the conclusion that association in liquidsis due to eIectrical rather than, as supposed by Guye and Baud(Compt. rend., 1901, 132, 1555), to chemical forces.The cost of the materiaIs used in this work was in part defrayedby a, grant from the Research Fund Committee of the ChemicalSociety, for which we desire to express our best thanks.CHEMISTRY DEPARTMENT,THE UNIVERSITY, SHEFFIELD