THORPE MOLECULAR WEIGHTS OF SUBSTANCES ETC. 141 XV.-On the Belation between the illoleeular Weights of Xubstances and their Xpec$c Gravities wheiz in the Liquid State. By T. E. THORPE Ph.D. F.R.S. ON attempting to trace the development of our knowledge concern-ing the connection which exists between the weights of unit volumes of liquid substances and their relative molecular weights, it will be found that practically nothing had been gained prior to the publication of Hermann Kopp's well known memoirs. In-deed but little was practicable until the necessity of comparing the liquids when under strictly analogous conditions had been recognised as a fundamental condition of the comparison. By dividing the specific gravities of liquids taken at the temperatures at which their vapour-tensions are equal to the standard atmospheric pressurethat is at their ordinary boiling points-into their niole-cular weights certain comparable values are obtained which are known as specific volumes.If the specific gravities are referred to water at 4" C. these values represent the number of cubic centimetres occupied by the relative molecular weights of the liquids expressed in grams at their respective boiling points under the standard pressure. Thus the specific volume of water is 18.8 which may be taken to * The apparatus can be obtained from Messrs. Cetti Brook Sbreet Holborn. VOL. XXXVII. 142 THORPE ON THE RELATION BETWEEN mean that 18 grams of water a t 100" occupy 18.8 C.C. The numbers so obtained were first shown by Kopp to exhibit certain definite rela-tions which may be briefly indicated as follows :-1.In many instances diferences in speciJic volzcme are proportional to diferences in corresponding chemical formzdtx. Thus a difference of CH in a homologous series corresponds to a difference of about 23 in the specific volume or (CHr)a = 22%. On comparing the specific volumes of similarly constituted halo'id compounds it is seen that the substitution of 1% atoms of bromine for an equal number of chlorine atoms increases the specific volume by 5n. 2. Isomeric and metameric liquids have as a rule the same specijic uolwne. Exceptions to be referred to hereafter are exhibited by certain oxygen and sulphur compounds. 3. The substitution of an atom of carbon fw two of hydrogen makes no alteration in the spem'c volwne of members of certain gvoups of organic liquids.It would seem t o follow from Kopp's observations that the specific volume of a liquid formed by the union of two other liquids is equal to the sum of the specific volumes of its components. It may also be inferred that the members of the same family of elements possess identical specific volumes thus the common value of phosphorus, arsenic and antimony would appear to be about 27 ; that of silicon, titanium and tin would seem to be about 35. On the basis of these conclusions Kopp was able to calculate certain fundamental values for the specific volumes of the elements in combi-nation. These values are as a rule constant for the particular element : thus carbon has invariably the value 11 hydrogen that of 5.5.Excep-tions are observed in the case of the chemical analogues oxygen and sulphur. Each of these bodies has two values depending it would seem on its mode of combination or on its relation to the remaining atoms in the molecule. For example acetone and allyl alcohol have each the empirical formula C3H60 but the specific volume of acetone is 78.2 whilst that of allyl alcohol is 73.8. The constitution of acetone may be expressed by the formula-CH3 GEL I II CO ; that of allylalcohol by CH I I CH3 CH,OH. In tthe case of acetone the affinities of the oxygen are wholly satisfied by the carbon that is we have reason to think that the oxygen atom is more intimately associated with one of the carbon atoms than it is with any one of those of the other element whereas in allyl alcohol a moiety of the combining value would seem to be satisfied by carbon an THE MOLECULAR WEXGI-ITS OF SUBSTANCES ETC.143. the remainder by hydrogen. It appears then that when oxygen is united to an element by both its affinities its specific volume is 12.2 ; when it is attached by only one combining unit its specific volume is 7.8. I have already pointed out (PTOC. Roy. Xoc. 160 1875) how these differences in the values for the specific volumes of oxygen and sulphur may be employed to throw light upon the constitution of such bodies as phosphoryl trichloride and thiophosphoryl trichloride. If the phos-phorus in these substances be regarded as trivalent they must possess The corresponding values for sulphur are 28.6 and 22.6.the constitution-c1 I P-0-CI 1 c1 If on the other hand the phosphorus written-Cl I I c1 1 P-s-G1 I c1. is pentavalent they must be c1 I Cl-P=o Cl-P=s I C1 I c1. It is obvious that the knowledge of the specific volumes of PC13, POC13 and PSCls will serve to indicate the mode of combination of the oxygen and sulphur and hence to determine the atomic value of the phosphorus in these compounds provided that the specific volume of the phosphorus is invariable or is independent of its atomic value. This is merely one of the many instances that might be adduced to show that a knowledge of the specific volume of a body is often calculated t o furnish valuable information concerning its constitution. Ac-cording to Kopp in the amines it is 2.3; in cyanogen and certain nitro-compounds it is about 17.No rational explanation of this difference has yet been given. In fact some observations recently pub-lished by Dr. Ramsay (Chew. Xoc. J. No. cci 473) seem to give widely different values for the specific volume of nitrogen in different amines and appear also to show that this element has a distinct value in the pyridine Series of bases. On the supposition that the atomic value of an element is variable Buff has suggested (hm. Chem. Pharm., Suppl. 4 129) that its specific volume is a function of that value, but the experimental evidence which he has adduced in support of this proposition is not very definite. So far as we know however, there is no cZpriori reason why oxygen sulphur and nitrogen should alone possess variable specific volumes.The specific volume of nitrogen also appears to be variable. L 144 THORPE ON THE RELATION BETTVEEN Of late years a fresh interest has accrued to the whole question, on account of the intradependence which has been shown to exist between the atomic weight of an element and its chemical and physical characteristics specific volume included. Our information on this latter point is a t present extremely imperfect but so far as it goes it would seem to indicate that the specific volumes like the other physical and chemical properties of the elementary bodies are periodic functions of their atomic weights. Among the many problems suggested by a review of our present knowledge of the subject the following seem to me to be specially worthy of solution :-I.Is it definitely established that an element in combination has as a rule an invariable specific volume ? May not the volume be modified by the number of the atoms of that particular element in the molecule ? Is it altogether independent of the general complexity of the molecule or may not the specific volume of the molecule be a function of its weight ? 11. Do the various members of a given family of elements possess identical specific voliimes or may not the volume be a function of the atomic weight ? 111. Would a re-examination of the cases of so-called variable atomic value serve t o show that the specific volume of an element is a function of that value as Buff supposes ? IT. The hypotheses of Mendelejeff and Meyer indicate the need for additionad and more exact determinations of the values for the specific volumes of the elementary bodies.V. Lastly it is desirable to multiply examples of the aid afforded by a knowledge of the specific volume of a compound in eluci-dating its constitution. With a view of solving these and certain other points of inquiry which will be referred to hereafter I drew up a scheme of work which involved the determination of the specific volume of the following 32 liquids :-Bromine Br,. Iodine monochloride IC1. Ethylene bromide C 2H4Br2. Ethylene chloriodide. . CzHJC1. Ethylene chloride CZHdC12. Ethidene chloride CzH4C12. Acetyl chloride C,H,O C1. Trichloracetyl chloride CzOCl, THE MOLECULAR WEIGHTS OF SUBSTABCES ETC.145 Chloral CZClsOH. Pentachlorethane CZHC15. Methene chloride Chloroform . . . . . . . . . . . . . . . . . . Chloropicrin Carbon tetyachloride C hloracet oni tril e. . . . . . . . . . . . . . Bromof orm . . . . . . . . . . . . . . . . . . Trichlorobromomethane . . . . . . . . Ethyl cyanide. . . . . . . . . . . . . . . . Propionitrile . . . . . . . . . . . . . . . . CHZCI,. CHCI,. C(NO)ZC13. cc1,. CCl3CN. CHBr,. C B r C13. C,H5.CN. CZHS. CN. E pichlorhydrin . . . . . . . . . . . . . . C3H,OCl. Ally1 alcohol. . C~HGO. Acetone. . C~HGO. Heptane C7Hl6. Ethylamyl . . . . . . . . . . . . . . . . . . C7H16. Octane C8HP33. Di- isobutyl . . . . . . . . . . . . . . . . . . C8H18. Aniline CGHYN. Picoline.. CGHTN. Mono- or triethylamine . . NHrC2H5 or N(C2H5)3. Nitrogen tetroxide . . . . . . . . . . . . N,O4. Silicon tetrachloride . . . . . . . . . . SiC1,. Titanium tetrachloride . . . . . . . . TiCl,. Tin tetrachloride . . . . . . . . . . . . SnCli. Phosphorus trichloride . . . . . . . . Phosphorus tribromide . . . . . . . . Thiophosphoryl chloride. . . . . . . . Phosphoryl bromochloride . . . . . . Phosphoryl chloride . . . . . . . . . . Vanadyl trichloride Ethoxyphosphorus chloride . . . . Triethylphosphine Phosphenyl chloride POCl3. PSCl,. POBrCL. VOCI,. PCl,. C2H50. p ( C2H5)3. P ClCGH5. Arsenic trifluoride ASP,. Arsenic trichloride hcI3 146 THORPE ON THE RELATION BETmEEN Antimony trichloride . . . . . . . . . . SbCI,.Thionyl dichloride s 0 Cl,. Snlphothionyl chloride . . . . . . . . s s CI2. Snlphurylhydroxyl chloride . . . . S0,OHCl. Sulphuryl dichloride SOZC1,. Disulphuryl chloride . . . . . . . . . . S,O,Cl,. Chromyl dichloride. . Cr02C1,. Carbon clisulphide cs,. This list might without doubt have been greatly extended, and in several important directions. One essential consideration in the choice of particular liquids for investigation was the possibility of obtaining them in a state of purity. I was anxious too to exclude as far as possible all liquids boiling above 200° on account of the difficulty of accurately determining their rates of expansion at high temperatures. The results of the observations made with these substances will afford material for the determination of the specific volumes of the following 17 elementary bodies.A toinic weight. Hydrogen 1.0 Fluorine 19.0 Chlorine . . . . . . 35.37 Bromine . . . . . . 79.75 Iodine 126.53 Oxygen 15.96 Sulphur. . . . . . . . 31 -98 Chromium . . . . 52.4 Atoinic weight. Nitrogen . . . . . . . . 14.01 Phosphorus . . . . 30.96 Vanadium 51.2 Arsenic . . . . . . . . 74.9 Antimony . . . . . . 120.0 Carbon . . . . . . . . 11.97 Silicon . . . . . . . . 28.0 Titanium . . . . . . . . 48.0 Tin 117.8 1. METHODS OF OBSERVATION. PrepnmtioiL of the Liquicls. illodes of ascertaiiziiig their Pwity.-In the preparation of the liquids employed in this research I have when-ever possible preferred to make use of methods which would directly yield the wished-for substance immixed with any other product or if that were not practicable I have sought to arrange that the bye-products should be such as could be removed with certainty.All processes of fractional distillation are comparatively valueless for the preparation of liquids of a high degree of purity. Hence in the preparation of such a liquid as thiophosphoryl chloride I have chosen to make use of the reaction-PZS + 3PC1 = 5PSCI3 THE MOLECULAR WEIGHTS OF SUBSTANCES ETC. 147 rather than employ Baudrimont’s method based on the action of phosphorus pentachloride upon antimony tersulphide-3PC1 + SbzS = 2SbClS + SPSCI,, which would involve a separation by distillation. Whenever the mode of preparation was not a safficient guarantee of the purity of the liquid as of course happened not unfrequently, I have sought to establish this either by analysis or by the determi-nation of its vapour-density.In the case of fhe inorganic liquids, which were mainly chlorides or bromides readily decomposed by water this was usually effected by estimations of the amount of the halogen. For the greater number of the organic compounds the method of Tapour-density was employed. Indeed in a number of instances the ordinary methods of organic analysis would be of very little avail in determining the purity of the compounds. Thus an organic combustion as ordinarily made gives practically no clue to tlhe purity of such a substance as heptane. This body may be mixed with 20 per cent. of its next higher or lower homologue without affect-ing the analytical results beyond the errors incidental to the method.I n cases of this character the determination of the molecular weight from the vapour-density affords not only the most accurate but also the readiest mode of ascertaining the degree of purity. As the greater number of the organic compounds given in the fore-going list were of comparatively low boiling point the Gay-Lussac-Hofmann method was generally applicable. The form of the apparatus employed by me differs slightly from that in general use as it admits of all the precision which the process is capable of yielding it may be desirable t o describe it in full. It is represented in Figs. 1-4 about gath the actual size. The trough seen in plan in Fig. 2 and in section at Figs. 3 and 4, is made of bay-wood.It is 200 mm. long and 70 mm. broad and holds about 200 C.C. of mercury. The cavity or well is 50 mm. in diameter and about 12 mm. deep measured from the ledge. This trough can be completely filled with mercury or emptied by raising or lowering the bottle b Fig. 1 which is connected with the trough by a stout caoutchouc tube. By “ kinking” the tube as the mercury flows out the metal may be kept at any desired level in the trough. The trough is rounded at the bottom in order t o economise the mercury, and from its shape allows all the metal to flow out into the bottle when this is lowered with the exception of that contained in the well below the ledge. Through the bottom of the well pass two nickel-plated brass tubes 8 mm. in diameter which are connected with the boiler c, by means of screws one of them d (Fig.2) stands 11 mm. and the other e 8 mm. above the mercury in the well e passes down nearly t 148 THORPE O S THE RELATION BETJ’EEN .PZy. I , Y \ f J l-0-*-4 +j -;E-.- i c I I __ __ TKE MOLECULAR WXIGHTS OF SUBSTANCES ETC. 149 the bottom of the boiler ; d (Fig. a) ends immediately below the upper surface of the boiler. The boiler itself holds about 1,000 c.c. and is less than quarter filled with the liquid (water aniline &c.) employed to vaporise the substance under investigation. The vapour-density tube is 950 mm. long and 16 mm. in internal diameter it is not graduated, but has a thin mark etched round it at about 35 mm. from the upper end. It is surrounded by the wider tube gg 36 mm.in diameter and 960 mm. long this is surmounted by a brass cap and is connected with the condenser ff by a flanged screw. The steam or aniline-vapour passes from the wide tube into the condenser the lower end of which terminat,es above the funnel k so that the condensed liquid flows back into the boiler. The water required for the condenser flows in at p and passes out through s. Att?ached to the condenser, which is made sufficiently rigid to serve as a support for the tubes, is a brass bar t in a slot in which works a brass scale divided into millimetres and at the lower end of which is a steel pointer. This scale with the attached pointer can be moved up and down by means of the rack and pinion seen at e so as to make the pointer just touch the level of the mercury in the well.At n is an arrangement to assist in determining accurately the level of the mercury in the vapour-density tube. It consists of a semicircular strip of brass working vertically along the outer tube. By tightening a screw behind o and working the rack and pinion at o the lower edge of the strip can be brought into exact coincidence with the level of the mercury in the inner tube and its height from the level of that in the trough that is, from the end of the steel pointer read off on the graduated brass scale. The capacity of the vapour-density tube above the etched mark is determined once and for all by weighing with mercury in the ordinary way in the tube actually used it was 64.155 C.C. at 0". The lower part of the tube is then calibrated by pouring in successive equal volumes of mercury as in Bunsen's method of calibrating an eudio-meter and the position of the mercury after each addition is read off on the graduated scale.Knowing the capacity of the little measuring tube employed to fill in the mercury we obtain the value in cubic centimetres corresponding to each millimetre of the vapour-density tube below the etched mark exactly as in an eudiometer. The results are incorporated in a table which gives directly the volume in the tube at every millimetre below the fixed mark. To make a vapour-density determination by means of this apparatus, the brass tubes d and e (Fig. 2) are connected together a t the ends in the trough by a piece of caoutchouc tube and the trough is nearly filled with mercury by raising the bottle b and " kinking " the tube.The A-apour-density tube is next carefully filled with mercury and the weighed quantity of the liquid experimented upon is introduced in th 150 THORPE ON THE RELATION BETWEEN ordinary way in a small stoppered tube or bottle. The vapour-density tube is then brought into the well the bottle b lowered so as toempty the trough t o the level of the ledge the caoutchouc tube connecting d and e (Fig. 2) removed a wedge of cork slipped over the vapour-density tube and the outer glass tube placed in position and put into connection with the condenser. It will be seen from Fig. 2 that the two brass tubes d and e are now between the glass tubes. The lamp is then placed beneath the boiler when the vapour of the liquid passes up by d ; that which condenses between the glass tubes flows back by e whilst the excess passes out at the top and condensing in the tube surrounded by cold water flows back through the funnel 7; connected with e into the boiler.When the height of the flame beneath the boiler is properly arranged the apparatus will work for hours if necessary with perfect regularity and without requiring any attention. As the mercury flows out from the vaponr-density tube it falls over t,he ledge of the well and runs into the bottle. As soon as the volume of the vapour is constant the pointer is brought into coincidence with the level of the mercury in the well and the position of the fixed mark and then the level of the mercury in the tube are determined on the brass millimetre scale.The latter reading gives at once the height of the mercury column in the vapour-density tube and the difference between the two read-ings gives ihe number of millimetres below the fixed mark and hence the volume of the vapour as shown in the calibration table. The height of the barometer is then determined or it may be obtained with sufficient accuracy by a.preliminary reading of the height of the mercurial column in the vapour-density tube before the introduction of the weighed quantity of liquid from this we can calculate the temperature of the steam if water be employed in the boiler with approximate accuracy by the simple expression-t .= 100" + 0.0375 ( h - 760), in which 73 is the barometric height at the time of observation; or obtain it at once from the accompanying Table :- If any other liquid be used it is best t o tie a thermometer to th THE MOLECULAR WEIGHTS OF SUBSTANCES ETC.151 vapour-density tube before the larger tube is placed over it and fo read off the temperature at the moment that the volume of the en-closed vapour is determined. The volume of the vapour is corrected for the error of the meniscus ; this in a tube of the dimensions given was found to be 0.41 C.C. ; and for the expansion of the glass taken as *000025 for 1" C. and the height of the mercurial column is reduced to the staudard temperature by Xendelejeff's extremely convenient formula-The method of calculation calls for very little explanation. Vt = 1 + .00018t + *00000002tS.The height of the barometer is at the same time reduced t o 0" by the ordinary tables. It is unnecessary to make any correction for the tension of the mercury-vapour at about 100" at temperatures much above this it inust of course be taken into account. At 132" (boiling point of amyl alcohol) it is 2.1 mm. ; at 160" (boiling point of oil of turpen-tine) it is 5 9 nim. ; at 183" (boiling point of aniline) it is 11.8 mm. ; at 200" it is 19.9 mm. It will be seen that the vapour required to heat the liquid is intro-duced from below and that the vapour-density tube has no etched scale upon it ; these two circumstances greatly diminish the risk of cracking the tube. Moreover the whole mass of the mercury within the tube is uniformly heated to a known temperature and therefore its reduction to 0" can be made with accuracy.The quantity of the metal which is needed is comparatively small not exceeding 300 C.C. : and the amount of the liquid in the boiler should not be more than 150 C.C. Numerous examples are given in this communication showing hlie degree of accuracy which the method is capable of affording and I may refer to my paper entitled " A Contribution to the Theory of Fractional Distillation " (Clzesn. Xoc. J. August 1879) for an illustra-tion of its value in determining the proportion of the constituents of a mixed liquid which can he rolatilised without action on mercury. In order to ascertain the specific volumes of liquids we require to know their specific gravities under comparable conditions of tempera-ture and in accordance with Kopp's suggestion the temperature of the boiling point of the particular liquid under standard pressure is usually adopted.How far liquids are strictly comparable under these conditions will be discussed hereafter. The direct determination of the specific gravity of the liquid at its boiling point cannot be readily made with the highest degree of accuracy. Dr. Ramsay has recently described a very simple method which admits of a fair approximation to exactitude and by means of this apparatus he has determined the specific volumes of a considerable number of substances (C'hem. Xoc 152 THORPE ON THE RELATION BETWEEN J. Zoc. cit.). 7 have preferred to adopt the principle of the method already employed by Kopp that is (1) to determine the specific gravity of the liquid at some convenient temperature ; (2) to ascer-tain its boiling point with the utmost exactitude ; and (3) to determine with greai care its rate of expansion say between 0" and the boiling point.By operating in this manner and by making use of methods of well established value my work would serve to put on record a number of accurately determined physical data of great importance, and I should be adding to the material required for a discussion of the laws regulating the thermal expansion of liquids even if the main object of my investigation were not attained. 1 proceed therefore to describe the apparatus and methods of observation necessitated by the three distinct series of operations. The Thermometers.-Two sets of thermometers were employed the first set made by Casella consisted of three instruments varying from -9" to 160" ; each instrument WPS graduated into tenths of a degree, and each scale division of 0.1" had a length of about 1 mm.the second set was made by Geissler of Bonn; it also comprised three thermometers graduated like the others and extending from - 14" to 170" 1" on these instruments was about 6 mm. in length. Both sets were compared previous t o use with Kew standards made by the late Mr. Welsh and the divergences were tabulated and applied as correc-tions in the subsequent observations. I am indebted to Sir William Thomson in whose possession the standards are for the opportunity of making the comparison. At a later period of the research the Casella thermometers were carefully calibrated by Bessel's method as modified by von CEttingen and were employed in the comparison of the air and mercurial thermometers undertaken by me in conjunction with my colleague Professor Rucker.In the course of the investigation a large number of determinations of the fixed points of all the instruments have been made and as usually happens these were found to rise gradually in each case, although not to the same extent in all. The position of the lower fixed point was determined after heating whenever practicable in snow ; at other times in finely powdered well-washed ice. The upper point was determined by the aid of the apparatus devised by Regnault for this purpose the temperature of the steam being cal-culated from his tables from the atmospheric pressure at the time of observation.The readings in all cases were made by means of a tele-scope provided with cross-hairs and micrometer screw. The degree of permanent displacement may be seen from the following examples of observations made on the Casella instruments the results are in all cases means of a large number of readings T€lE MOLECULAR WEIGHTS OF SUBSTANCES ETC. 153 Therm. A. The extent of the displacement is evidently dependent on or at any rate is greatly influenced by the degree of heating or in other words by the amount of molecular disturbance to which the glass envelope is subjected ; this seems t o confirm the opinion of Legrand and W. H. Miller that it is not solely due as generally supposed t o the influence of atmospheric pressure on the glass.The rise in the fixed points is graphically represented in Fig. 5, the abscisscae represent the times in months at which the several observations were taken and the ordinates the extent of displacement in hundredths of a degree. The curve of B strikingly resembles one deduced from Despretz’s observations of the change in zero-point of a particular instrument in his possession the comparatively unsymmetrical character of the others is probably owing in great part to the irregular expansion and contraction which as Matthiessen has shown is frequently exhibited by glass exposed to alterations of temperature.* * The bulbs of these thermometers were sealed to the stems. Verdet sap, Dans les thermom&tres dont le reservoir a BtB soud6 ii la tige on observe un deplacement plus consid6rable que clans les thermomhtres dont les reservoir a Qt6 souf36.154 THORPE ON THE RELATIOS BETWEEN Pig. ~.-C?URVES SHOWING RISE OF FIXED POINTS IN TEERMOMETERS. Unless otherwise stated all temperatures are expressed in air-thermometer degrees which are distinguished in order to prevent Confusion by the symbol *.* The conversion from the ordinary scale was made by the aid of ;I table compiled from the observations of Recknagel (Pogg. Ann. 123) for temperatures below lo@" and from those of Regnault made with ;t thermometer of ordinary glass for temperatures above that point,. The following abstract of Dhe table is given in order to allow of the reconversion to the ordinary scale if desired :-Mercurial thermometer.THE MOLECULAR TEIGHTH OF SUBSTASCES ETC. 155 was still further reduced by repeated17 heating and cooling the bulbs before de- f G 100.27 termining their capacities. The number of times was not actually counted but it exceeded several hundreds in every case. It is known that the effect of these alter-ations of temperature is to bring the glass 0 2Al:ii I; into a condition of molecular equilibrium analogous to that which it is suppose 156 THORPE ON THE RELATION BETWEEN The glass tube A was fitted with caoutchouc corks previously well soaked in melted paraffin through the lower one was passed the dilatometer through the upper me were inserted two tubes f and y, the former of which was sealed a t the upper point ; g was drawn out at each end into long capillary tubes the upper length being bent twice at right angles as shown in the figure and sealed at the point e ; the other end was made sufficiently long and fine to pass down the stem of the dilatometer into the bulb.The tubes being in place and the outer surfaces of the corks again covered with melted paraffin the side-tube was connected in the ordinary manner with the Sprengel pump and the whole arrangement exhausted as completely as possible, the flow of the mercury being continued for a couple of hours after the height of that in the fall-tube was practically identical with that of the barometer. A small beaker containing the pure and warm mercury was then brought under 9 and raised until the greater por-tion of the tube was submerged the point e was then broken off when the mercury immediately rushed over into the bulb of the dilatometer, and gradually filled the whole instrument the pump being meanwhile continuously worked.As soon as the mercury flowed from the end of the dilatometer the point o f f was broken in order to admit the air and the action of the pump was stopped. The apparatus was then disc connected and the capillary tube of g carefully withdrawn from the dilatometer. This mode of filling such instruments with mercury is readily executed and is far preferable to the ordinary method of ex-pelling the air by boiling the metal. The dilatometer filled with mer-cury was again weighed and then placed in a bath of water of known temperature and the height of the mercurial column determined in the ordinary way by the telescope.In order t o calibrate the stem, successive lengths of the mercurial column were removed by suction through a 6ne capillary tube inserted into the stem and ending in the small bulb apparatus (l?ig. ‘7) previously weighed. Fig. 7. The increase in the weight of the apparatus gave the amount of mercury so withdrawn; its length in the stem was read off on the graduated scale the fractions of the millimetre being determined by the aid of the micrometer screw of the telescope. Experience showed that it was better to empty the bulb after each weighing thanto retain the mer-cury in the apparatus since the volatilisation of t,he metal caused a THE MOLECULAR WEIGHTS OF SUBSTANCES ETC. 157 appreciable loss of weight in the course of a few hours.During the entire process the dilatometeie was surrounded by water of nearly con-stant temperature flowing directly from the main. The following readings made in the conrse of the calibration of dilatometer N may serve as an example of the character of the variations observed:-Scale reading. It will be noticed that the diameter of the tube becomes gradually less from division 242 to 127 after which it slowly increases. This is a fair sample of what was usually observed although the tubes were carefully selected in the first instance a number of them being tested preliminai-ily by the ordinary method of propnlsion of a short column of mercury I was unable to obtain them in the lengths required of uniformly increasing or diminishing bore ; and a subsequent examina-tion made with the greatest care by the method of propulsion of a number of tnbes of still finer bore has convinced me that it is but very rarely that the variation in the diameter of the capillary tubes of greater length than 20 centimetres can be properly expressed by formulae of the kind employed by Kopp.After the greater portion of the mercury in the stem had been withdrawn the dilatometer was again weighed ; the average of the mean values of the mercury contained in a division wa9 then taken, and that number representing the weight of the unit-volume was divided into the weight of the mercury contained in the bulb together with that in the stem up to the zero-point. The ratios of the mean weights of the mercury contained in one division in various portions of the stem to the weight of the unit-volume were then determined, the intermediate values being obtained by interpolation and from these a calibration table showing the number of unit-volumes cor-responding to each division on the scale was constructed.In the course of the investigation I have in this way made and cali-brated nine dilatometers of varying capacities respectively designated as A B C D E M N 0 and P. The bulbs of A B and C were blown directly from the tube; those of D and E were sealed on to the stem; M and N were fitted with glass stopcocks; 0 and P were of vor,. XXXVII. 15s THORPE ON THE HELATION BET\I%EN considerable capacity and their bulbs were cylindrical and com-paratively narrow. The weight of the unit-volumes in mercury varied from 0-01054 in the case of 0 to 0.01556 in that of E.Their several capacities in unit-volumes were :-A . . . . . . . . . . 1600.19 M . . . . . . . . 338303 B . . . . . . . . 2851.60 N . . . . . . . . . . 3383.84 C . . . . . . . . . . 2964.70 D . . . . . . . . 3283.42 0 . . . . . . . . . . 6182.72 E . . 2868.25 P . . 6174.41 In order to determine the expansion of the glass the dilatometers partially filled with mercury after +he final weighings in the calibra-tion were placed in melting ice or in water of known temperature and the position of the mercury i n the stem accurately determined ; the instruments were then heated by steam from boiling water in the appa-ratus employed to determine the upper fixed point of the thermometers, the temperature of the steam being calculated from the barometric pressure and the position of the mercury column again read.The mean results for I" between 0" and 100" were :-A . . . . . . . . . .0000251 If . -0000244 B . . . . . . . . . . *0000254 N . *0000254 C . . *0000247 D .- -0000213 0 *00002,52 E . . *0000230 9 *0000248 The comparatively small exp.ansions of D andE are possibly connected with the fact that the bulbs of these instrnments were sealed on to the stems and not blown directly out of the tube indeed in the case of D two such sealings were made near the bulb as the stem was acci-dentally broken in the course of the calibrations. The separate results in the case of D were :-(l) *0000214 ; (2) -0000212 ; (3) .0000213.DeterwLination of the Boiling Points of the Liquids.-The liquids were invariably redistilled immediately preceding the determination of their specific gravities and rates of expansion and the boiling points were ascertained in the final distillation. Special care was taken to avoid superheating ; and whenever the nature of the liquid permitted a spiral of platinum wire was placed in the flask to prevent succussive boiling. As the thermometers employed were of unusual length it was not always practicable to immerse the mercurial column entirely in the vapour; hence the correction for the cooled portion of the thread became a t times of considerable importance. As Holtzmann has already pointed out the well-known expression THE MOLECULAR WEIGHTS O F SUBSTANCES ETC.159 in which t = the observed temperature on the thermometer; t' the mean temperature of the cooled column as determined by an attached thermometer ; w the length of the cooled column expressed in degrees ; and 6 = the apparent expansion of mercury in glass for lo vix., *000154 ; over corrects the results especially a t high temperatures a t 200' the error amounts to 0.5". For temperatures at about the boiling point of water the error is not very considerable although it can be perceived by careful reading. The following observations were made in the apparatus employed fo determine the upper fixed points of the thermometers. It will he seen that in all cases where the value of 6 is taken as 400154 the results are over corrected. Landolt and Wiillner (Ann.Chern. Pharm. Suppl. 8 1867) and more recently Mills (Chem. News 31 234) have discussed the various modes of correcting for the cooled column and the former observers have proposed to substitute for Kopp's simple expression a far more complicated formula based on observations made with ther-mometers of Geissler's pattern to which it is more directly applicable than to those which were most frequently employed by me. Perhaps the simplest mode of approximating to the actual temperature is to follow Holtzmann's example and to modify the value of 6 in the formula. The number adopted by Holtzmann is .000135 but as Landolt and Wullner have shown this number gives uniformly low results when w is greater than 60 and t - t' exceeds '75. Mills found that 6 was about *00013 and increased *00001 for every additional 25".In fact 6 = a + an. Although it would doubtless have been more rigorously accurate to have determined the values of a and for the several instruments I find from a large number of observations that M 160 THORPE ON THE RELATIOK BETWEEN the mean value *000143 gives sufficiently accurate results for all tem-peratures up to 200" and for all values of n which are likely to occnr in practice. The numbers in the final column of the preceding table have been calculated by the use of this constant it will be seen that they agree very satisfactorily with the actual numbers contained in the fourth column. As the table of corrections calculated by this constant differs slightly from that generally used it may be desirable to reproduce it here :-160 Table for the Correction of Thwmomater Readings.-THE MOLECULAR WEIGHTS OF SUBSTANCES ETC. 161 The barometer employed was a standard instrument on Fortin's l'attern and had been verified a t Kew. All barometric readings are corrected and reduced t o 0". I n order to make them as nearly com-parable as possible the boiling points corrected for the error of the cooled column are reduced to the standard pressure of 760 mm. Ins means of the expression -0375 ( h - $60) in which h is the height of the barometer a t the time of observation.* This expression, of course assumes t8he validity of Dalton's law but the error intro-duced is negligible. I n two or three cases I have calculated the cor-rected boiling points from vapour-tension observations when these have been a t hand but the difference between the results thus obtained and those given by the above formula has never exceeded 0.05" ; hence we may assume that for the ordinary barometric variatiuns the formula is generally applicable.Determinatiotz of Specz3c Gravity.-The bottles employed had a capacity of from 4 t o 20 c.c. and were fitted with ground glass stoppers. As already pointed out by Kopp other things being equal, there is a probable gain in accuracy in the use of bottles of compara-tively small capacity ; the disadvantage that smaller differences in the relative weights are obtained as compared with the larger flasks being more than counterbalanced by khe rapidity and certainty with which the liquid acquires a constant temperature.I have preferred t o use stoppered bottles to the Sprengel apparatus 011 account of bhe special character of many of the liquids. All the weighings were made by the method of vibrations and are reduced to a vacuum. The weights employed were an excellent set made by Staudinger of Giessen and had been verified with great cure. The specific gravities were usually taken a t the temperature of melting ice and are compared with water at 4". I n cases where reductions were necessary they were made by the formula-V' so = SiT, in which So = the specific gravity a t 0". St = 7 , the higher temperature t. V' = the volume of the liquid at t obtained by the interpo-V = the volume of the water a t t (vol. a t 4" = l) from latiou formule. Rossetti's tables.DetenniiLation of the Thermal Bxpaizsiort of the Lipids.-As many of * The Yorkshire College is in lat. 53" 48' and 186 feet above the sea-level. No attempt has been made to reduce the observations to a standard atmosphere. The liright of the barometer corresponding to the standard atmosphere in Lee& is 739.427 mm. at 0" 162 THORPE ON THE RELATION BETWEEN Fig. 8 THE NOLECULAR WEIGHTS OF SUBSTAKCES ETC. 1 ti3 the liquids experimented upon were strongly fuming corrosive sub-stances readily decomposed by atmospheric moisture it was scarcely possible to introduce them into the dilatometers in the ordinary way without the risk of slight alteration. The arrangement adopted in the case of these bodies is seen in Fig. 8 ; it was found so convenient in practice that it was employed generally even when working with liquids for which it was not originally designed.The liquid was dis-tilled with the proper precautions to exclude moisture directly from the flask in which its boiling point had been determined into the botjtle A which was fitted with a caoutchouc cork containing a short length of glass tube through which was inserted the fine capillary tube s passing down the stem and into the bulb of the dilatometer. Before connecting the caoutchouc tube e the vessel T was filled with mercury and the ghss stopcock I being closed M was lowered as in-dicated in the figure and the cocks a and I slowly opened ; the mer-cury in the upper vessel ran down into 31 and air dried by passing over pumice moistened with oil of vitriol contained in the drying tube, flowed into T.As soon as the mercury ceased to run the cocks a and I were turned and M was raised by a couple of turns of the wheel W up to the pulley. The cock I was opened when the pressure exerted by the mercury drove the liquid from A over into the dilatometer. The flow was so slow that it was easy to regulate the height in the stem to any wished-for position by gradually lowering the brass frame carrying the dilatometer as the liquid passed over. J u s t before the required quantity had been admitted I was closed and a opened ; the flow of the liquid still continued but with great slowness the long leg of the capillary tube acting as a syphon. M was tshen broirght back to its original position and a was closed and the dilatometer lowered until the end of the capillary tube was a t the point in the stern a t which it wais desired that the liquid should stand ; the cock I was again opened and the liquid in the capillary tube quickly drawn back into A when a was opened to prevent the passage of the air through the tube into the liquid.It was found to be quite unnecessary to close the ends of the dilatometers during the process of filling as the extent of surface exposed by the liquid to the air was too small to cause any sensible decomposition. By a simple and obvious modification by which the entire appratus could be filled with dry carbon dioxide or hydrogen i,t would be possible to make use of this arrangement in the case of liquids like zinc- and cadmium-ethyl which ignite in contact with oxygen.The same apparatus was used i'o transfer the liquid to the specific gravity bottles ; these were usually filled immediately after the dilato-Fig-9 164 THORPE ON THE RELATIOX’ BETWEEX meters. Fig. 9 shows the mode in which the bottles were supported on the sliding frame. The dilatometers thus charged were closed with glass rods with qround ends and plunged into oil heated to within a few degrees of the boiling point of the liquid. Occasionally and more particularly in the case of the readily decomposable inorganic chlorides it few minute bubbles of gas formed in the liquid during the preliminary heating : these were easily driven up the sbem by gently tapping the instrument. The dilatometers mere then momentmilg unclosed to liberate the air 3 I I x I.- -.-I I THE JfOLECULhR WEIGHTS OF SUBSTANCES ETC.165 compressed in the stem above the liquid and the glass rods again put into position and firmly fastened down on the ground end of the dilato-meter by strong caoutchouc tubing and copper wire. Of course in the case of the dilatometers provided with st!opcocks the manipulation was even simpler. After standing from 16 to 20 hours a t the ordinary temperature the dilatornetem were placed in melting ice and in about an hour the position of the liquid a t 0" was determined by the aid of the telescope. At the same time the zero point of the thermometer was ascertained. The dilatometers and thermometer were then placed in a brass frame and immersed in water contained in a large copper bath (Fig.lo) fitted with plate glass sides and standing on an iron table the height of which could be adjusted by a screw. The bath held about 22 litres of water which could be kept in con-stant motion by stirrers worked by a small hydrnulic engine. Direct experiments proved that the temperature of the various portions of the water never differed more than by -05". The little engine (for which I am indebted to my friend Mr. Henry Davey C.E.) wasunder perfect uoiitrol and did its work admirably i t entirely dispensed with the necessity for an assistant. The water of the bath was gradually heated by steam blown in from a boiler and the current could if necessary be so regulated that the temperature of the bath could be maintained at any desired point between 30" and 60" to within 0.05" for practically any length of time.I n order to facilitate the subsequent calculation of the empirical formula expressing the raise of expansion the readings were taken a t twelve approximately equidistant temperatures between 0" and the boiling point of the liquid. Up to about 25" the temperature of the water could be maintained absolutely constant but at points higher than this it was found better to take a series of readings with a slowly ascending and descending temperature than to spend time in attempt-ing to obtain i t perfectly nniform. As soon as the temperature of the water of the bath approached the desired point the current of steam was regulated so as to cause a very slow arid gradual increase in tern-peratnre and a series of readings was taken on the two instruments.The current of steam was still further reduced or altogether stopped as the case required and as the temperature slowly fell a second series of readings was taken equal in number to the first and the mean of the double series was considered to represent the uncorrected tempe-rature and volume of the liquid. The readings were made by the telescope furnished with cross hairs and a micrometer screw placed at a distance of about four feet from the bath 44 divisions on the screw corresponded to one division on the diiatometer and the readings could be made with certainty to the &+h of a unit-volume. Th 166 THORPE ON THE RELATION BETWEEN following readings taken indiscriminately from the note book serve to show the general character of the observations :-Temperature: As a rule the liqpids were not heated beyond 65" in the water-bath, partly on account of the difficulty of maintaining the large bulk of water of an approximakely constant temperature,.but more especially from the length of time and large quantity of steam needed to heat the water above this point. The succeeding observations were therefore made in the apparatus represented in Fig. 11. AB is a cylindrical' bahh made Qf sheet iron and holding about 25 litres ; in it is placed a second bath,. CD of 10 litres capacity resting on three short stout rods fitting into tubes fastened to the bottom of the outer-bath both bdhs were filled with cotton-seed oil. The lid e j , when in position is fixed to the bath by screws iii ; it carries a wide brass tube I open at the bottom and pierced with holes; this serves to protect the thermometer t which indicates the temperature in the outer bath ; the three brass tubes 0 p YL act a s guides to the rods of the stirrers.I n order to keep the oil in the upper bath El' as cool as possible and more especially to avoid any rapid changes in its temperature the lid ef,. is made double ; as the various tubes passing through it are soldered to both sides,. the intermediate space is practi-cally an air chamber. EF is a rectangular frame fitted with plate-glass sides; it is fitted to a flange which can be screwed down on to the top of the lid. It was filled with almost colourless rape oil which could be withdrawn when necessary by the tap s communicating with the tube k.The dilatometers x and y carrying tightly-fitting brass tubes stuffed with sninll leather discs and tapped with screws were screwed on to the bottom plates of the lid the thermometer T being passed through the lid from above and held in position by a precisely similar arrangement; the thermometer G fastened to T served to indicate the temperature of the bath EF and hence that of the cooled columns in the thermometer and dilatometers. The oil in all the baths was kept in constant motion by stirrers connected with the cross bar W, and worked by the small engine. I n order to mike an observation the oil in the outer-bath was heate THE NOLECULAR WEIGHTS OF SUBSTAXCES ETC. 167 Fig. 11 16s; THORPE 03' THE RELATION BETWEEN a few degrees higher than the desired point, the exact number of de-grees depending upon the temperature needed in the internal bath.The flame o l the lamp below the bath was then either removed or lowered as the case demanded and as the thermometer T become nearly sta-tionary from the extreme slowness with which the heat was communi-cated from the outer to the inner bath a series of readings was taken in the followiiig order :-(1) of the thermometer T ; ( 2 ) of the dila-tometer; (3) of the thermometer G ; (4) of the dilatometer; (5) of the thermometer T ; (6) of the thermometer G ; the readings being continued in this order with a gradually ascending and descending temperature as in the water-bath observations. The folldwing example of a set of readings will more clearly illustrate their character :-T.89-50 -51 -52 -5.5 -53 -51 *48 -42 Mean 89.50 -Dilatoineter scale reading. 328-10 -34 -59 -63 -70 -66 -63 -52 328.52 As was anticipated and as is evident both from this and from the former example the thermometer is far more sensitive to changes of temperature than the dilahmeter ; but by the method of reading with :t slowly ascending and descending thermometer any error which might have been due to this cause is practically eliminated. It is un-ILecessary to read G to nearer than 0*2" but as the thermometer actually nsed was graduated to tenths the temperature was always observed to the nearest scale division. Reduction of the Observatiom.-In his observations Kopp also em-ployed a double oil-loath but the capacity of the baths (glass beakers) was very much smaller than in the apparatus above described.More-over the sterns of the dilatometess and thermometers were for the greater part of their length simply surrounded by air of which the temperature a t about the middle part of the exposed stems as deter-mined by an attached thermometer was assumed to be that of the nican temperature of the columns. The true volume was calculated by successive approximations by means of the expression THE VOLECULAK. WEIGHTS OF SUBSTASCES ETC. 169 in which V is the observed volume in the dilatometer a t T E the. length of the exposed and cooled column measured in scale divisions, from the surface of the oil in the inner bath to the level of the liquid in the stem t the mean temperature of this column and Vt the volume of the liquid in the dilatometer a t t.The correction thus obtained was added to V and the result V inserted in the above expression, and the calculation repeated until no sensible alteration was obtained. Direct experiments have shown that this method of correcting the observed volumes tends to give too high results from the circumstance t,hat the liquid in the exposed portions of the stem has a uniformly higher temperature than the mean temperature of the air. A dilate-meter was filled with water and the level of tthe liquid carefully deter-mined at various temperatures between 18" and 25" in the water-bath, and it was then heated by steam in the apparatus employed to deter-mine the upper fixed points of the thermometers.The dilatometer was placed so as to expose varying lengths of the stem containing watey to the air and the temperature of the middle point of the cooled column was determined by an attached thermometer exact'ly in the mannei-described by Kopp and with all the precautions adopted by him to prevent overheating and to avoid sudden changes in the temperature t. The results were as follo.cvs :-V,r. t. E. Vt. Calculated vol. - 3301.2 0.0 0.0 0.0 3298-3 23.2 83.3 31 77.8 3301.5 3296.1 22.7 144.9 31'77.5 3301.6 3293.0 22% 220.8 31 $7.4 3301-7 3290% 22.3 286.4 3177-2 3301% 3288.5 21-2 338.2 3176.6 3301.8 It will be noticed that the calculated volumes are in all cases in excess of that directly observed viz. 3301.2.By reducing the value of E by 10 + .015 E in each case results may be obtained which agree almost exactly with the true volume. The calculated values are respectively-3301.2 3301.3 3301:l 3301-2 3301.2 The mean of these values is identical with the true value. From a large number of experiments made with several liquids :It various temperatures and with different lengths of cooled columns I find that in my apparatus E may be taken as the whole length of the column in the upper bath expressed in scale divisions 1 i O THORPE ON THE RELATION BETWEEN The corrected volumes have been calcnlated by the formula-in which V, Vt and E have the same significance as in Kopp's formula. The rate of expansion from 0" to the boiling point of all the liquids experimented upon by me may be represented with sufficient accuracy by a single expression of the form-V = A + Bt + Ct2 + Dd3, the values of the constants being found by substituting successively for V and t the twelve corresponding values observed at equidistant points between 0" and the boiling point of the Jiquid and adding together the resulting expressiom in consecutive groups of three so as to form four equations of oondition.by solving which the four un-known quantities were determined. Each coefficienf was then divided by A giving an expression of the form-V = 1 + B't + C'P + D't3. Correcting for the expansion of the glass 6 of the particular dilato-meter employed this becomes-V = 1 + (B' + 3)t + (C' + B'3)t2 + (D' + C'6)t8, which will be hereafter referred to under the form-Y = 1 + at + bta -+ eta.The labour of reducing t,he observations and more especially of calculating the empirical formuh for so large a number of substances, has been very considerable and it is but just that I should acknow-ledge my indebtedness to my wife for her assistance in this matter: aided by the arithmometer of Thomas (de Co1mar)-an instrument of the greatest service in calculations of this kind-she undertook the greater portion of the very tedious work of computation. I would also here express my obligations to my colleague Professor Rucker for his aid in the mathematical portion of the research. Before beginning the proper work of the investigation I made a series of observations on water mainly with the view of obtaining information concerning the degree of accuracy of my method of experi-ment.The very careful discussions which have been made from time to time particularly by Miller Matthiessen and Rossetti of the numer-ous observations already published made by methods widely divergent in principle go to prove that the rate of expansion of this liquid is known with almost absolute certainty. Moreover I imagined that these preliminary observations would incidentally serve to show ho THE MOLECULAR WXIGHTS OF SUBSTANCES ETC. 1 7 1 far the objection which has been urged against the dilatometrical method by Matthiessen that it tends to give uniformly low kates of expansion is well founded. Two dilatometers 13 and C were charged with recently boiled distilled water in the manner already described.When cold they were put into melting ice for an hour and the levels read with tbe following results :-B . . . . . . C . . . . . . 3051.05 3172.06 The bulbs were then placed in hat water ; the liquid in B expanded through about 100 divisions that in C through about 107 divisions ; when nearly cold the instrumento were again placed in melting ice and the levels again read-B . - . . . . 3051.08 C a . . . . 3172.10 After standing Eor 70 hours the 'levels were-B . . . . . . 3051.08 C . - . . 3172.13 These observahns show that no error due to the liquid adhering to the walls of the tube is to be feared and that the method of closing the dilatome%ers effectually preveats all evaporatiion of the liquid. The results of the observations of the expansion are shown in the following table and are compared with the mean results of the obser-vations of Kopp Pierre Despretz Hagen Matthiessen Kremers, Relative volumes expansion of glass.Dilatometer. correeted for I B. 172 THORPE ON THE RELATION BETWEEN Weidner and Rossetti in column X and in column Y with the means of the fairly-accordant results of Matthiessen and Hagen made by the method of displacement. My own resiilts agree rather better with those of Hagen than with those of Matthiessen which are generally higher than those of pre-vious investigators. The differences however are far too inconsider-able to throw any doubt on the validity of the dilatometrical method of determining the thermal expansions of liquids. I proceed now t,o give the details of the mode of preparation and piirification of the several liquids and the results of the observations made to determine their boiling pnints specific gravities and rates of expansion by heat.I have a t the same time collected together so far as I could all previously existing information on these points and I have in many cases reduced the observations in order to make them more strictly comparable with my own.* II.-sPECIFIC GRAVITIES BOILING POINTS AND RATES OF THERBIBL EXPANSION OF THE VARIOUS LIQUIDS. Bronaine Br,. About a kilogram of commercial bromine was carefully dehydrated by agitation with pure concentrated sulphuric acid and distilled the fraction boiling a t about 60° which amounted to nearly two-thirds of the whole being collected separately.A portion of the distillate was treated with milk of lime and ammonia and the resulting calcium bromide was tested for iodine according to the method employed by Stas but not a trace was found. The remainder of the bromine was placed over powdered potassium bromide and after several months' digestion it was again distilled and the distillate agitated with phos-phorus pentoxide to remove the last traces of water. The purified bromine boiled constantly a t 59.47" under a pressure of '765.2 mm. (n = 0 t = 0). Reduced and corrected boiling point, The specific gravity was found to be 3.15787 a t 9-10A compared with water at the same temperature ; a t OA compared with water a t 4A it is 3.18828. According to Pierre bromine boils at 63" under a pressure of 760.32 mm.and has a specific gravity of 3*18718 at O", 59*2iA. * Some of the observations here given hare been published in preliminary form in a couple of papers contributed to the Proceedings of the Royal Society No. 167, 1876. The details of final results are now however gircn for the first time THE MOLECULAR WEIGHTS OF SUBSTANCES ETC. 1 7 3 compared with water at 4OX (,4nm. de Chirn. et de Phys. [3] 20). Bolas and Groves found 59-6" under a barometric pressure of 751 mm. The following observations of the expansion were made wit!i Dilatometer C :-The rate of expansion of bromine has already been determined by Pierre who has given the following expression to represent the results of his observations :-V = 1 + 0.001 038 186 255s + 0.000 001 711 380 853t2 + OeOOO 000 005 4 4 118t3.This formula gives numbers agreeing fairly well with those afforded by the expression deduced from my observations as the following comparison shows :-15'. 30". 4 5 O . 60". Pierre . . . . . . 1.01608 1.03273 1.05068 146963 Thorpe . . . . 1.01624 1.03326 1.05108 1.06960 Pierre has undoubtedly the credit of having first determined the specific gravity and thermal expansion of approximately pure bro-mine ; but it seems probable in spite of t'he care which was evidently taken in its preparation that the sample employed by him in his ob-servations was not perfectly free from water. This indeed is indi-cated by the order of the divergence between the results of our obser-vations ; and the supposition would seem to be confirmed by the high solidifying point viz.- 7.5 to -8" which he noticed. Baumhauer finds the true freezing point of bromine to be -245" (Rer. 1871 927). Pierre attempted to dehydrate the bromine by digestion with calcium chloride. According to Stas (Nouvekle's Recherches szcr les lois kc., Aronstein's Transl. 179) calcium bromide (which as a desiccating agent is not inferior t o t'he chloride) is incapable of removing the last trace of water from bromine. Iodine iMonochZoride ICI. This remarkable substance has nearly the same molecular weight as bromine and when liquid bears considerable resemblance to that element. It may be prepared by the direct union of iodine and chlorine but is more conveniently obtained by heating an intimat THE MOLECULAR WEIGHTS OF SURSTAKCES ETC.1'75 mixture of iodine and potassium chlorate and distilling the product from powdered potassium chlorate. 1 2 + 3KCIO3 = KC104 + KT03 + KC1 + 0 2 + IC1. A large quantity of the compound prepared in this way boiled con-stantly after repeated distillation between 99.7 and 100*7" n = 40, t = 25". Corrected and reduced boiling point, 101*3*. Iodine monochloride has been variously described as a reddish- brown oily liquid (Gay-Lussac) and a ' hyacinth-red crystalline solid (Schut-zenberger) . Both statements are correct. After distillation espe-cially if kept in sealed tubes the monochloride will remain liquid for many weeks even in a freezing mixture. On the addition of a minute trace of the terchloride solidification a,t once ensues. A quantity of the liquid monochloride placed in an open tube solidifies after a few days owing t o the conversion of a portion of the monochloride into the terchloride and free iodine 3IC1 = ICl + I,.Hannay (Chern. Xoc. J. 1873 815) and more recently Bornemann (AwnnZen 1877, 183) have already made a number of interesting observations on the causes which induce the solidification of this substance. I find the monochloride melts at 24.2'; Hannay found 24.7"; Trapp and also Bornemann 25". Analysis showed that the substance obtained by the above reaction was pure iodine monochloride. Bulbs containing weighed portions of the liquid were broken under dilute sulphurous acid solution and the hydrochloric and hydriodic acids were precipitated by silver nitrate, and the mixed silver salts digested with nitric acid.The results were as follows :-Bar. 7'44.3 mm. I. Amount of IC1 taken 1.2482 grams I. Silver salts obtained 77 7 , From the formula W=- I + * g a + C 1 + k y , I c1 in which W is the weight of mixed silver salts and x and y the amount of iodine and chlorine respectively contained in them we find the per-centage composition of the monochloride t o be :-I. 11. Calculated. Chlorine 21.86 21.98 21-85 Iodine . . . . . . . . . . 78-14 78.02 78.15 A determination of the specific gravity of this compound gave 3.12988 at 17*95" compared with water a t 4" ; at OA its specific gravity is 3.18223 also compared with water at 4A. The following observations on the rate of expansion were made with Dilatometer C in the water-bath :-The observations may be accurately represented by the formula 2997.609 + 2671 46t f 0.002 430 9te + 0.000 008 183t3, by means of which the numbers in trhe third column are calculated.Dividing through by the first term and correcting for the expansion of the glass (.00@0247) we obtain the following formula as expressing the true expansion of iodine monochloride between OA and its boiling paint :-V = 1 + 0.000 9115 896t + -000 000 832 96t2 + -000 000 002 75Ot3. By means of this formula the following table has been calculated :-Etherce Dibromids C2HaBr2. This compound was prepared by passing a stream of well-washed ethene through pure bromine under water treating the resulting oil with dilute soda solution and distilling after dehydration with phos THE MOLECULAR WEIGHTS OF SUBSTANCES ETC.177 phorus pentoxide. The entire amount of liquid boiled hetweon 130.1 and 130.8" ; on redistillation the greater portion came over between 130.6and 130*7" n = 55 t = 11.3". Bar. 766.6 mm. Corrected and reduced boiling point 131.45". The recorded observations on the freezing point of ethene dibro-inide are very discordant ; they vary from as low as -12 or -15" to as high as + 13.1". I found that my preparation solidified at + 9*2", which is almost identical with Regnault's number 9.53". Its specific gravity was found to be 2.19011 at 10*89" compared with water at 4". On the assumption that the liquid contracts regu-larly below its ordinary solidifying point its specific gravity at 0" is 2.21324 compared with water at 4".Other observers have found for the 'boiling and freezing points and specific gravity of this compound :-Boiling poi&. Melting point. Specific gravity. Regnault . . . . 131.6 at 760 mm. 9.53' -Haagen . . . . 131-6 , ? I Pierre . . . . . . 132.6 7 56.9 13.1 P16292 at 20.79" Cahours . . . . 130 ? 0 Reboul . . . . - - I Observations on the rate of expansion made with Dilihmeter C In the watei:-bath.:-gave the following results :-to. 178 THORPE ON THE RELATION BETWEEN These observations may be represented with sufficient accuracy by the expression-2946.754 + 2.735 02t + 0.001 946 3t2 + 0.000 011 584f3, Dividing t'hrough by the first term and correcting for the expansion as the numbers in the last columns of the two tables show. of the glass (*0000247) the above formula becomes :-v = 1 + a-ooo 952 845t + 0.000 ooo 683 45.5~ + 0-ooo ooo 003 947t3, by means of which we obtain the following table ; it shows the true volumes of ethene dibromide at every 5A between OA and 135* :-AC.Pierre who has already determined the expansion of this liquid, found that a single expression of the form V = 1 + at + bt2 + ct3, would not express the results of his observations with sufficient accu-racy. He also selected the volume at 20-09" the lowest temperature at which he made an observation as the unit of volume on the ground that at lower temperatures particularly in the neighbourhood of its freezing point the dilatation of the Iiquid might be irregular a sup-position which is not confirmed by my observations.He accordingly calculated two expressions of the form v = 1 + a0 + 60% + c03, in which 8 = t-20.09". From 20.09 to 100.16" the expansion was expressed by the formula V = 1 + 0.000 952 696 190 + 0.000 001 316 506 858S2 + 0.000 000 001 062 687e3 THE MOLECULAR WEIGHTS OF SUBSTANCES ETC. 1'7'3 and from 100.16 to 132.6" by the formula v = 1 + 0.001 016 765 9888 + 0.000 ooo 102 231 770eZ + 0.000 000 008 788 007e3. These formulze give results differing but little from those afforded by the expression deduced from my observations as will be evident from the following comparison :-0 0 0 0 0 20.09. 50.09. 80.09. 11OQ9. 130.09. Pierre . . 100000 102980 106213 109709 112238 Thorpe 100000 103003 PO6230 109750 112284 Ethene ChZoriodicZe C,H,ICl. This substance was first obtained by Dr.Maxwell Simpson who prepared it by suspending finely powdered iodine in about twice its weight of water and passing chlorine into the liquid (whtich was kept cool and constantly agitated) until the iodine had nearly disappeared. After st'anding for a short tirue the yellowish-brown liquid was de-canted and treated with a stream of ethene washed by passing through soda solution until both the liquid and tbe oil which separated out were decolorised. The chloriodide thus obtained boils nearly constantly after washing and drying ; when first prepared it is colourless but on exposure to light it gradually becomes red owing to the separation of free iodine. In order if possible t>o avoid the use of water in its preparation I attempted to obtain the chloriodide by the direct addition of pure iodine monochloride to ethene matters being so a,rranged that a lnrge excess of the hydrocarbon was present during the reaction.The result was altogether different from what I anticipated. When the two sub-stances are brought together nearly the whole of the iodine is set free, and Dutch liquid is formed only a very small quantity of the chlor-iodide being produced. I varied the conditions of the experiment in several ways but with no very different result. After having spent a considerable amount of time and material in the various trials I found that Geuther had made the same observation (Jahrb. 15 421). The reaction of ethene upon iodine monochloride appears in reality to occur in two phases the chloriodide seems to be first formed but under the influence of a second molecule of the iodine chloride it is decomposed with the formation of Dutch liquid and free iodine.Thus :-(1.) CZH4 + lC1 = C,H,ICl. (2.) CZHaICI + ICl = C2H,C12 + I,. The explanation of the process employed by Dr. MaxwelI Simpson would appear to be as follows :-On treating the powdered iodine sus 180 THORPE ON THE RELATION BETWEEN pended in water with chlorine iodine monochloride is formed but in the outset this substance is q.uickly resolved into hydrochloric and iodic acids and free iodine-5.IC1 + 3H20 = 5HC1 + HI03 + 21,. I n time this phase of the reaction ceases owing partly to the decom-position of the .water and partly to the formation of hydrochloric acid a solution of which dissolves *the iodiiie chloride without altesa-tion.The observations of Bornemann (Ann. Chem. Pharm. 187.7, 212) prove that the complete conversion of the iodine monochloride illto iodic acid requires 10 parts of water to 1 part of the iodine. The extent of the decomposition would seem to depend on the ratio of the amount of the products of decomposition to the amount still undecom-posed. This seems evident from-the fact observed by Hannay that if the liquid after exhibiting no further change be filtered from the pre-cipitated iodine or if that substance be removed by the addition of carbon bisnlphide further decomposition ensues and if the iodine be constantly removed as it separates out the whole of the monochloride is gradually but slowly decomposed in the manner indicated by the above equation.The formation of the ethene chloriodide under the conditions of Dr. Simpson's reaction is readily explained by its inso-lubility in water. At the moment of its formation it separates out, mid is thus removed from the in3uence of the uncombined iodine chloride. I am indebted to Dr. Maxwell Simpson for the sample of the chlor-iodide employed in my observations. After washing with dilute potash solution and drying over phosphorus pentoxide if boiled almost constantly between 139.1 and 140.1". Bar. 759.3 mm. rt = 50, f = 30. Two observations of specific gravity gave (1) 2.13363 at 155BA and (2) 2.13329 a t 15*4SA compared with water a t .PA. Reduced to OA the specific gravity becomes (1) 2.16440 and (2) 2.16437. Nean 2.16439, compared with water at 4A.Ethene chloriodide solidifies to a dhite crystdlline mass in a mixture of snow and hydrochloric acid. Other observations on record are-Corrected and reduced boiling point 14O*lA. Boiling point. Specific gravity. Simpson . . . . . 140-143" 2'151 a t 0" Maumenk . . . . 146 a t 753mrn. 2.39 a t 20" The following observations were made in Dilatomater C. I n t.he water-bath : These results may be expressed by the formula-3003.647 + 8.739 9% + 0.001 1'79 221* + 0.000 013 492t3, by means of which the numbers in the last columns of the above tables are calculated. Dividing through by the first term and correcting for the expansion of the glass (.0000247> we obtain-V = 1 + 0.000 936 917 6t + 0.000 000-415 129P + O*OOO 000 004 501t3, which represents with sufficient accuracy the expansion of ethene chloriodide between 0" and its boiling point.The following table gives the true volume of this liquid for every 5* between these limits :-Etlzene Dichloride C,H4Cl = CH,Cl.CH,CI. The product was washed wit'h water decanted shaken with oil of vitriol again decanted, digested with aolid causbic potash snd distilled. The entire quantity boiled between 83.0 and 83*3" the greater portion appearing to come over at 83.2". Bar. 753.9 mm. n = 20" t = 20" ; corrected and reduced boiling point 83.5*. A determination of the specific gravity of ethene dichloride at OA, compared with water at 4A gave 1.28082. Other observers have found for the specific gravity and boiling point of this liquid :-Regnault 82.5" at 765 mm.1.256 at 12" Liebig. . 82.4" - 1.247 , 18 Pierre 84-92" , 761.88 1.28034 , 0 Haagen . . . . . . . . . . 85" 1-2562 , 20 Prepared by the action of chlorine upon ethene. Dumas . - 85" , 770 -Observations with Dilatometer C In the wnter-bath gave-These observations lead to the formula-2947.000 + 3.325 195t + 0,002 351 185P + 0-000 028 308t3, which gives the numbers in the third column of the above table. Dividing through by the first term and correcting for the expansion of the glass (-0000247) the formula becomes-1 + 0.001 153 032t + 0.000 000 825 693t2 + *OOO 000 009 625t3, by means of which the following table showing the volume of ethene dichloride at every 5A between OA and 85* is calculated : These observations afford the formula-3001.810 + 3.763 &t + 0.006 431 7t2 + O*OOO 014 52t3, which as the aboPe comparison shows very accurately represents the observed volumes.I have made n second series of ,observations upon a sample of ethi-dene chloride obtained by the actian of phosphorus pentachloride on aldehyde :-CzH,O + PC] = C,H,Cl + POCIS. The product was washed with water dried and distilled. It boiled constantly between 59.0 and 59.6" (uncorr.). Bar. 761.0" mm. Cor-rected and reduced boiling point 59.2". Two determinations of its vaponr-density made in different tubes, gave the following results :-I. 11. Weight of liquid . 0.1322 gram 0.2468 gram Volume of vapour . . . . 84.20 C.C. 115.49 C.C. Barometer. . 762.2 mm. 732.8 mm. Height of column. . 393.0 , 229.4 ,, Temperature.. 100.1" 99.0 THE MOLECULAR WEIGHTS OF SUBSTANCES ETC. 185 -Obseryed. I. 11. Calculated. - Vapour-density 49.29 49.10 49.34 Two determinations of specific gravity gave-I 1.1897 at 10.05A I1 1.1863 , 1 2 ~ 2 4 ~ compared with water at the same temperature respectively. become-These numbers reduced t o O* and compared with water at 4* I. 1.2050 11 1.2048 Mean . . . . . . . . . . . . 1.2049 Observat6ons with Dilatometer C in the water-bath gave-Calculated. These numbers afford the formula-11. 3003.400 + 3.799 05t + 0.004 734 65t2 + 0.000 032 338t2, which gives the results seen! in the third column of the foregoing table. On dividing through by Che first term in each of the two formula, and correcting for the expansion of the glass we obtain-I.1 + 0.001 278 4% + 0.000 002 173 58t2 + 0.000 000 004 89P. 11. 1 + 0-001 289 62t + 0.000 001 607 67t2 + OaOOO (300 010 806P. These formulae give closely accordant results as will be evident; from the following table which shows the relative volumes of ethidene chloride at every 5A between OA and 60A 186 11. THORPE ON THE RELATION BETWEEN The mean formula is-1 + OsOO1 284 02t + 0.000 001 890 62t2 + O*OOO 000 007 848t3. Pierre (Ann. ~7ainz. Phys. [3] 31 118) found that the expansion of monochlorinated ethyl chloride between 0" and 61.3" may be repre-sented by the expression-1 + 0.001 290 717 95t + O*OOOOOO 118 334 518t2 -+ O*OOO 000021 339432t3, which affords very different numbers from that deduced from my observations :-20".a". 60". Pierre . . . . 102603 105319 108249 Thorpe 102631 105460 108522 There can be little doubt that the ethidene chloride of Pierre was very far from being pure. The high specific gravity and boiling point which he observed would indicate the presence of higher chlorinated products in the material used by him indeed the method of prepam-tion which he eriiployed could hardly fail to yield a mixture of such bodies. Acetyl Chloride CzH30C1 = CH3COCl. I prepared this liqnid by the action of phosphorus trichloride upon glacial acetic acid in the manner first described by BBchamp (Jahr. 1856 427). The reaction may be represented by the following equations :-I. 3CCHa0 + PCI3 = POjH3 + 3C2HSOCl. 11. 4CzHa02 + 2PC& = PzO3.HzO + 4CzH3OC1 + 2HC1.+ 3C,H,OC1 + 3HC1. 111. ~ C ~ H ~ O Z + 2?c&j = p,o THE MOLECULAAR WEXGHTS OF SUBSTANCES ETC. 187 The first equation is given in several modern text-books ; the second is that of Bbchamp ; t h third is identical rnutatis mzitalzdis with the reaction formulated by Gal for the mode of formation of acetyl br omi d e . A mixture of 90 grams of glacial acetic acid (99 per cent.) and 69 grams of phosphorus trichloride being the proportions demanded by Equation I was gently heated in R flask connected with an upright condenser for several hours and then subjected t o distillatmion. A large quantity of hydrochloric acid gas was evolved and the acetyl chloride obtained weighed 54 grams. According to Equation 1 no hydrochloric acid should have been evolved and the yield of acetyl chloride should have been 118 grams.The amount of the chloride if formed according to Equation 11 would be 77 grams ; and 59 grams if according to Equation 111. Hence it would seem that Gal's equa-tion correctly represents the mode in which acetyl chloride is formed by the action of phosphorus trichloride upon glacial acetic acid. The following experiments undertaken at my suggestion by Mr. John Muir definitely settle this point. Dehydrated acetic acid prepared with great care and freshly disfilled phosphorus chloride were allowed to act upon each other in an apparatus so arranged that the whole of the hydrochloric acid formed in the reaction might be retained. In the first experiment-Phosphorus trichloride . . . . . . Acetic acid . . . .. . . . . . . . . . . . 15.13 grams 20.00 ,, being the proportions required by Equation I were mixed together. At the end of the reaction 6.03 grams of hydrochloric acid and 13.2 grams of acetyl chloride were obtained. Equation I11 gives 6.04 grams hydrochloric acid gas and 12.96 grams of the chloride. The materials were then mixed in the proportion demanded by Equation 111 6.1 grams of acetic acid being mixed with 9.3 grams of the phosphorus chloride. The amount of acetyl chloride obtained was 7.8 grams Equation IIT gives 7.97 grams. In this experiment it was noticed that the two liquids when first mixed were withmt visible action on one another but that when heated to about 60" the mixture became turbid and a white precipitate was formed on raising the temperature to about 80 or 85" the contents of the distilling flask solidified.In all the former experiments a considerable quantity of acetic anhydride boiling between 130 and 135' was obtained. This may have been formed by the dehydrating action of the phosphorus trioxide upon the excess of acetic acid present or by the mutual action of the acetyl chloride and acetic acid C2HsOC1 + C,H4O2 = czH30} C,H,O 0 + HCl 188 THORPE ON THE RELATION BETWEEN Observed. Kanonnikoff and Saytzeff have actually obtained the anhydride in this manner ( A m . Chem. Pharm. 185 192). The acetyl chloride obtained in the various experiments was repeatedly distilled until no further alteration in its boiling point was perceptible. Two observa-tions made with different thermometers and at different times gave the following results :-I.T = 50.25" r~ = 35 t = 17.5 Bar. 746.1 Corr. and red. b.p. 50.90" 11. T = 50.43P n = 0 t = 0 Bar. 746-3 , , 50-93" Mean . . . . 50.92" or 50.73'. The freshly distilled product was analysed by breaking bulbs con-taining known quantities of the chloride in. water adding a slight excess of pure precipitated calcium carbonate warming and determin-ing the chlorine by decinormal silver nitrate solution and potassium chromate solution. I. 0.7634 gram chloride required 97.3 C.C. silver solution. 11. 1.0009 2 ) 1 127.9 ) ' 9 Pound. Calculated. I. 11. C1 45.20 4.5020 45-31 A determination of the specific qavity of acetyl chloride gave At 1.12221 at 10.22' compared with water of the same temperature-OA the specific gravity becomes 1.13773 compared with water at 4'.Gerhardt 55" boiling point 1.125 at 11" Kopp 5-5-56' , 1.1072 at 16' or 1.1305 at 0". Both samples were made by Gerhardt's method i.e. by the action of phosphorus pentachloride or phosphoryl chloride upon an acetate. The last traces of the phosphoryl trichloride are removed with great difficulty and acetic anhydride is simultaneously formed in the reaction. The difference of 4' observed in our determinations of the boiling point of acetyl chloride is doubtless due to this cause. The following observations of the rate of expansion were made with Dilatometer E in the water-bath :-Other observations on record are-Calculated. A. THE MOLECULAR WEIGHTS OF SUBSTAXCES ETC. 189 These results may be represented by the formula-2892.84 + 3.895 076t + 0.005 909 78t2 + 0.000 035 ll$t3, which affords the numbers in the third column of the above table.of the glass (.00002303) we obtain the formula-Dividing through by tho first term and correcting for the expansion V = 1 + 0.001 369 4% + 0.000 002 073 9t2 + O*OOO 000 012 185fS, by the aid of which t,he following table showing the relative volume of acetyl chloride at every 5" between OA and 55" is calculated:-Observations on the rate of expansion of acetyl chloride have already been made by Kopp (Ann. Chem. Pharm. 95 307) who has expressed his results by the following formula :-V = 1 + 0.001 315 4t + 0.000 003 370 6t2, which gives somewhat lower numbers than that deduced from my observat,ions -15". 30". 45".Kopp 102049 104250 1C6602 Thorpe 102089 104301 106661 T&hloracetyl ChEoride CC1,CO.CI. This liquid was obtained by the action of phosphorus trichloride on trichloracetic acid in the manner described by Gal (Juhr. 1873 536) the trichloracetic acid being 'prepared by the oxidation of chloral hydrate according to the very convenient method of Clermont (Jahr., 1872 495). The yield of the chloride obtained by Gal's method is very small and the product even after repeated distillations per-sistently retains small quantities of phosphorus trichloride that em-ployed in the observations contained about 1.3 per cent. of the tri-chloride. It boiled constantly between 116.3 and 117.3". rz = 73 t 30". Bar. 758 mm. This agrees with the number found by Hubner and also by Gal.Corr. and red. boiling point, 118*. Two determinations of the sp. gr. of this liquid gave-I . . . . . . . . . . 1.6305 at 15*37* I1 . . . . . . . . . . 1.6291 , 16.20A VOL. XXXVII. 190 THORPE ON THE RELATION BETWEEN 0 VOO 10.75 20 -80 30 -96 42 *08 compared with water at the same temperatures. Since the specific gravity of phosphorus trichloride at 0" is 1.613 t,he small quantity of this liquid present in the preparation could exercise but an insigai-ficnnt influence on the specific gravity. The above numbers reduced to 0" and compared with water at 4" become-1.6565 1.6563 1.6564 I . . . 11 ., Mean The following observations of the expansion of trichloracetyl chloride were made with Dilatometer IY in the water-bath :-' ChZoraZ C2C130H = CCl,.COH.A quantity of this liquid prepared from the hydrate by treatment with strong oil of vitriol and lime was distilled. It boiled between (36" and 97.4". It was again distilled and the fraction which came over between 96-7' and 96.9" which was by far the main portion, was collected separately. n = 62 t 28. Bar. 766.8 rnm. Corrected and reduced b. p. 97.2". A second preparation made from another sample of chloral hydrate, after treatment with lime boiled at 96.4" (corr. and reduced). The specific gravity of'the first sample was found to be 1.52813 a t 9 ~ 4 3 ~ compared with water a t the same temperature. That of the second made by a different bottle and different balance mas 1.52939 at 20-2.jA compared with water a t @.These reduced to OA and compared with water at 4A become respectively-I 1.5439 I1 1.5466 2879.2 28'79 *4 35 -68 2996 *5 2996 *7 2906 -1 2905 *9 45 *83 3032 -4 3032 *2 2936.2 2936 *O 54 *23 3060 -9 3061 *6 2965 -0 2965 -2 62 3 7 3091 *1 3090.4 I am disposeif to regard the first number as moTe nearly correct as it was made upon a larger qnantiby of the liquid obtained from a much larger preparation. Giving i t therefore twice the weight of the other the mean becomes 1.54480. Other observations on record are :-Liebig. 94" 1.502 a t 18". Kopp . . The following observations of the rate of expansion were made in 98-1-99" at 745.9 1:4903 a t 22.2" = 1.5183 a t 0". the water-bath in Dilatometer E :-A. 1 Observed. 1 Calculated. I A. 1 Observed. 1 Calculated.No observations were possible beyond 65" owing to the appearance These observations may be represented by the formda-of gas-bubbles in the liquid. 2879.369 + 3.064 OO6t + 0.008 000 75f - 0.000 046 369t3, which on dividing through by the first term and correcting for the expansion of the glass (-00002303) becomes-V = 1 + 0.001 087 15t + 0.000 002 803 058t2 - 0.000 000 016 O&, 0 192 THORPE ON THE RELATION BETWEEN by means of which the following table is calculated. relative volumes of chloral at every 5A between OA and 60A :-It shows the The rate of expansion of chloral has already been observed by Kopp (Ann. Chem. i'lzarnz. 95 307) who has expressed his results by the formula V = 1 + O*OOO 954St - 0.000 002 213 9i2 + 0.000 000 056 939 2t3, This gives a much slower rafe of expansion than is shown by my observations as will be evident from the following comparison :-20".40". 60". Kopp . . . . . . . . . . . . 10187 10385 10615 Thorpe 10226 10462 10717 Pentachlorethane CCl,. CHC1,. I prepared this liquid (which was first obtained by Regnault by the prolonged action of chlorine on ethyl chloride) by the reaction dis-covered by Paterno (Ja,hr. 22 505) viz. by the action of phos-phorus pentachloride upon chloral. 190 grams of the pentachloride was added in small portions a t a time to 113 grams of chloral and the mixture boiled for some hours in a flask connected with an up-right condenser. It was then distilled and all coming over below 170" was treated with water dried by calcium chloride and redistilled.Apparently no phosphoryl chloride was formed as little or no liquid came over in the first distillation below 125". The yield was much less than that indicated by the equation given by Paterno. The de-hydrated liquid boiled between 157.S" and 158.8" (uncorr.) the greater portion coming over between 158.0' aad 158.5". t 14 1% 125. Bar. 759.9. The specific gravity was found to be 1.69263 at 10*15A compared with water a t 4A. Paterno found that pentachlorethane boils a t 158" and has a sp. gr. of 1.71 a t 0" and 1-69 a t 13". The same substance obtained by Pierre by the action of chlorine on ethene dichloride boiled a t 153.8" and had a sp. gr. of 1.66267 a t 0" (Jahr. 1 685). Corrected and reduced b. p. 159.1A. At OA compared with water a t 4A it is 1.70893 THE MOLECULAR WEIGHTS OF SUBSTANCES ETC.193 110*14 117.19 133.38 146.32 The following observations of the rate of expansion of pentachlor-ethane was made in Uilatometer D. In the water-bath. :.-These results lead to the formula-3278.618 + 3.041 605t - 0.000 309 073t' + 0*000017 316 7ta, which on dividing through by the first term and correcting for the expansion of the glass (*00002130) becomes-V = 1 + 0.000 949 00% - O*OOO 000 074 503t2- + 0.000 000 005 280t3, by means of which the following table is calculated :-Pierre found that the rate of expansion of pentachlorethane might be represented by the expressions-1 + 0.000 899 04t + O*OOO 002 457 77t2 - 0.000 000 012 865.P (between 0" and 75"), and 1 + 0.000 973 39t + 0.000 000 025 77t2 + 0.000 000 006 364t3 (between 75" and 148.3").These formulae give a much higher rdte of expansion than that deduced from my observations as will be evident from the following comparison :-0". 50". 100". 150". Pierre . . . . . . . . 100000 104949 110396 116807 Thorpe 100000 104772 1099G 115874 Prom the close agreement of my nnmkers for the specific gravity and boiling point of this body with those obtained by Paterno 3 am indined to believe that Pierre's preparation which boiled 5" lower, was mixed with less highly chlorinated products produced by the incomplete action of the chlorine on the Dutch liquid ; these would tend t o diminish the boiling point and specific gravity and t o increase the rate of expansion. Methene Chloride CH,Cl,. Prepared by the action of .zinc-dust and ammonia upon chloroform, as described by Perkin (Chem.Neeus 18 106). Although a consider-able quantity of chloroform was worked up but a very small yield of the chloride was obtained. This after very careful rectification through a long tube fitted with Linnemann's fractionation arrangement boiled constantly between 41 and 42". Bar. 751.8. 12 = 0 t = 0. Cor-rected and reduced boiling point 41.6A. Butlerow who first obtained this compound in the pure s t a b by the action ef chlorine upon methene iodide (Awu. Chern Skarm. 111 242) found that i t boiled between 40 and 41" Perkin observed 40.442". The quantity of the liquid at my disposal was too small to enable me to determine its specific gravity by the bottle in the usual manner.The dilatometer in which the observation of the expansion was to be made was therefore weighed before and after the introduction of the chloride. This occupied-The weight of the liquid taken was 4.6043 grams. at 11.69" 2961.7 unit-volumes, and at 12.04 2963.1 77 Value of 1 unit-volume = 0.015578 gram of mercury at 0" THE MOLECULAR WEIGHTS OF SUBSTANCES ETC. 1'35 Observed. Calculated. I A. 1 Observed. The weight of water a t 0" contained in the dilatometer at the two points would therefore be-Calculated. Since as the obserrations of the thermal expansion show 1 vol. of methene chloride at 0" occupies 1.01554 vols. at 11.69 and 1.01602 vols. a t 12*04" the specific gravity of the liquid at 0" compared with water a t 4" ie-These results may be expressed by the formula-2915.942 + 3.747 O56t + 0.007 888 09t2 - 0.000 004 074t3, Dividing through by the first term and correcting for the expansion of the glass (.00002303) we obtain V = 1 + 0.001 308 05t + 0.000 002 735t2 - 0.000 000 001 33t3, by means of which the following table showing the relative volume of methene chloride a t every SA between O4 and 45* is calculated:----THORPE ON THE RELATION BETWEEN Calculated.Chloroform CHCl3. About 500 grams of the purest commercial chloroform were fre-quently agitated during several days with successive quantities of water in order t o remove traces of ethyl alcohol. The decanted liqiiid was then digested for about a week with concentrated oil of vitriol, and after separation shaken with recently healed potassium carbonate and distilled.It began to boil at 61*2" the temperature quickly rose to 61*3" and then slowly to 61.4" at which point fully three-fourths of the liquid came over. Corrected and reduced boiling point 61*20A. Bar. 757.mm. ?L = 0 t = 0. Two determinations of sp. gr. at Oh gave-I 1.32660 I1 1.52653 Mean 1.52657 compared wit,h water at O*; compared with water at 4A the sp. gr. of chloroform at O* is 1.52637. Among the previous observations of the sp. gr. and boiling point of chloroform may be mentioned-SP. gr. Regnault . 61.0" -Pierre 63.5 at 772.52 mm. 63.0" at 760 mm. 1*52:23 at 0". The older determinations of Swan Soubeiran and Mialhe and Gregory are not given as it does not appear that any special pains were taken to free the chloroform employed by them from accompany-ing alcohol.Two series.of observations of the rate of expansion were made. The following readings were taken with Dilatometer B in the water-bath :-by means of which the numbers in the third column are calculated THE MOLECULAR KEIGHTS OF SUBSTANCES ETC. 197 The second series made with Dilatometer C gave-According to Pierre (Ann. Chim. Phys. 631 33 199) the expansion of chloroform between 0" and 62.7" may be represented by the formula-V = 1 + 0.001 107 15t + 0.000 004 664 73t2 - 0.000 000 017 433t3, which gives a much slower rate of expansion than the formula deduced from my experiments-15". 30'. 45". 60". Pierre 101760 103694 105768 107946 Thorpe . . . . . . 1018T3 103843 105931 10815 198 THORPE ON THE RELATION BETWEEN A.0 .oo 10 *21 19.11 30.66 Chloropicrin C(NOp) Cl,. Observed. I Calculated. A. Observed. Calculated. A quantit,y of the pure liquid obtained by beating picric acid with bleaching powder as directed by Hofmann (Jahr. 19 494) boiled constantly between 111.5 and 111.6" the greater portion coming over between 111.55 and 111.58" n = 9" t = 30". Corr. and red. boiling point = 111.91A. A determination of its specific gravity gave 1.69247 a t OA compared wit,h water at the same temperature ; compared with water at 4& it is 1.69225. Other observers have found for the boiling point of chloropicrin-Bar. 751.9 mm. Hofmann 112" Cossa 112.8" a t '743 mm. The following observations were made with Dilatometer N in the water-bath :-Observed.3651 -7 3694.3 3734 -7 Calculated. These observations may be represented by the formula-3378.99 + 3.632 970t + 0.001 480 70t' + 0.000 026 430t3, which gives the numbers in the last columns of the preceding tables. of the glass (*00002553) this expression becomes :-V = 1 + 0.001 100 7 t + 0.000 000 465 757P + 0.000 000 007 833t3, by the aid of which the following table showing the relative volume of chloropicrin a t every 5" between OA and l l O A is calculated:-Dividing through by the first term and correcting for the expansio Carbon Tetrachloride CCl+ About 300 gra.ms of this liquid obtained by careful distillation from a large quantity of the commercially pure product were digested with freshly heated and roughly powdered calcium chloride for a couple of days.On redistillation the liquid boiled absolutely constankly at 76*47" t 17" TL = 20". Corrected and boiling point = 76.74A. A determination of the vapour-density of this body gave the follow-ing results :-Bar. 754.3. Weight of liquid Volume of vapour 109.80 C.C. Temperature 99.50 Barometer . . . . . . . . . . . . . . . . 746.9 mm. Height of column 283.4 m. 0.3380 gram Found 76.85 Calculated. . 76.73 Two determinations of sp. gr. at OA gave-I. 1.63215 11. 1-63216 compared with water at O h com,pared with water a t 4A the sp. gr. of carbon tetrachloride at OA is 1.63195. Other observers have found-BoiLng point. XpecXc gravity. Regnault . . . . . . 76.5 at 760 -Pierre . . . . . . . . '78.1 , 748.3 1.6298 at 0" Riche .. . . . . . . . . 77.0 ? 1.567 , 12" Hofmaiin 77.0 P -Two series of observations of the thermal expansion of this liqui 200 THORPE ON THE RELATION BETMTEEN were made. gave :-The first set made with Dilatometer B in the water-bath Observed. The experimental numbers may be expressed by the formula-2894.08 + 3.414 139t + 0.002 154 48t2 + 0.000 036 147ts. The second series made with Dilatometer C gave the following results :-The observed volumes lead to the fornula-2988.95 + 3.540 382t + 0-001 609 23t2 + 0.000 043 045ts. Dividing through by the first term and correcting for the expansion of the glass these formulce become respectively-I. V = 1 + 0.001 205 20t + 0.000 000 724 52t2 + 0.000 000 012 509t3. 11. V = 1 + 0.001 209 19t + 0.000 000 567 65t2 + 0.000 000 014448t3.The mean formula is-V = 1 + 0.001 207 1% + 0.000 000 671 09t2 + 0.000 000 013 478t", by the aid of which the following table showing the relative volume of carbon tetrachloride at every 5* between OA and 80A is calcu-lated : THE MOLECULAR MTIGHTS OF SUBSTANCES ETC. 201 According to Pierre the expansion of carbon tetrachloride from 0 to its boiling point may be expressed by the forrnula-V = 1 + 0.001 183 84t + O*OOO 000 898 881t2 + 0.000 000 013 513t", which gives very concordant results with that deduced from my observations :-20". 40°. 60". 75". Pierre 102414 104966 107719 109955 Thorpe 102434 104996 107750 109982 Horn represents the expansion of carbon tetrachloride between 30" and 160" by the formula-V = 1 + 0.001 067 188 3t + 0.000 003 565 14t2 - O*OOO 000 014 942 811" + 0.000 000 000 085 182 318P.which unquestionably fails to indicate the true rate at temperatures below the ordinary boiling point of the liquid :-30' 70" Hirn 103489 108909 Thorpe. . 103695 109220 Bromoform CHBr,. A quantity of this liquid obtained partly from Kahlbaum of Berlin, and partly from Schuchardt of Gtirlitz was distilled digested with calcium chloride and again distilled. The liquid boiled constantly between 149.2 and 150*0" the greater portion coming over betweer 149.6 and 149.7". t = 21" = 120. Bar. 761.6 mm. Corr. and red. b. p. = 151.2*. Bromoform solidifies at 2.5' in lustrous crystalline plates-a fact appnren tl y hit herto unnoticed. A determination of the specific gravity of bromoform gave 2.81185 at 8.56A compared with water at the same temperature its sp.gr. a t 0" (on the assumption that it remained liquid at that temperature and contracted regularly) would be 3.83413 compared with water at PA 202 THORPE ON THE RELATION BETWEEN 3287 -1 3327.1 3359 *5 Other observers have found for the specific gravity and boiling point of bromoform :-Cahours 152" ? Bar. 2.9 at 12" Lowig - 2-13 ? Schmidt 149-150 2.775 at 14.5" Bolas and Groves . 144-146 -The following observations were made with Dilatometer D in the water-bath -328'7.6 45.70 3390 *2 3389 -7 3327.1 62-31 3442 *8 3pP2 *9 3359 *2 A. 1 Observed 1 Calculat,ed. 1 A. 1 Observed. Calculated, - -- -- I!-Observed. These results lead to the formda-3250.042 + 2.989 60t + 0.000 839 85ta + 0.000 013 826t3, from which on dividing through by the first term and correcting for the expansion of the glass we obtain-V = 1 + 0.000 941 16t + 0.000 000 278 OOf + 0.000 000 004 259t3, by means of which the numbers in the following table showing the relative volumes of bromoform at every 5A between O* and 150* are calculated : Trichl oro b ro mrne t hane C C 1,B r .This compound was prepared according to the method by which it was first obtained by Paterno vie. by the action of bromine upon chloroform . CHCI + Br2 = CC13Br + HBr. The mat'erials were heated together in sealed tubes for several d n p to a temperature of about 210" the tubes being occasionally opened to allow of the escape of the hydrobromic acid gas. On each occasion the liquid was distilled and all coming over below 95" was reheated.The distilled product containing a slight excess of bromine was treated with an aqueous solution of potash and dried over calcium cbloride. On fractionation by far the greater portion of the liquid boiled betweeu 102 and 103" ; n = 100" ; t = 18". Bar. 755.5 mm. Corrected and reduced boiling point = 104.07". Trichlorobrommethane when freshly prepared is a colourless liquid possessing a pleasant aromatic smell ; on exposure to light it gradually becomes brown owing to the liberation of a small quantity of bromine. Two determinations of its specific gravity gave-2.03674 a t 8*41A 2.03492 , 9.3lA compared with water at the same temperature. for the specific gravity a t 0" compared with water at 4"-These numbers give I 2.05494 I1 2.05498 Mean 2.0549 204 THORPE ON THE RELATION BETWEEN Observed.Paterno found the boiling point of this compound to be 104*3" and its specific gravity a t 0' (compared with water at the same tempera-ture ?) to be 2.058. These numbers agree closely with those of my own determinations. Friedel and Si1r-a (Jahr. 18'72 300) found 103-104" at 752 mm. ; sp. gr. 2.063 at O" and 2.016 a t 25". The following observations of the rate of expansion were made with Dilntometer N in the water-bath :- These observations may be represented by the formula-,342899 -+ 3.623 57t + 0.002 156 213t2 + 0.000 020 032t3, which affo~ds the numbers in the last columns of the above tables. Dividing through by the first term and correcting for the expansion of the glass (*00002553) this formula becomes-V = 1 + 0.001 082 31t + O*OOO 000 655 82t2 + 0.000 000 005 858 2t3, by the aid of which the following table showing the volume of tri-chlorobrommethane at every 5* between Oh and 10.5* is calculated.Propiorzitrile C2H,. CN. About 200 grams of this liquid obhained from Kahlbaum of Berlin, was treated with dilute hydrochloric acid dried and distilled and the portion boiling between 96.4 and 96.6" (uncorr.) was collected sepa-rately it formed fully nineteen-twentiet,hs of the whole quantity ; n = 37.5 ; t = 12.0. Bar. 751.6 mm. Corrected and reduced boiling point 97.08A. That the liquid was pure will be evident from the following deter-mination of its vapour-density :-Weight of liquid 0.0512 gram Volume of vapour .. . . . . . . . . 80.78 C.C. Temperature 99%" Barometer 749.1 mm. Mercury column. . 482.0 ,, Its specific gravity was found to be 0.79375 a t 7.36" compared with Other observations on record are-water a t 4* a t O" compared wit)h water a t 4" i t is 0*80101. Limpricht . . . . 97-96" Gaut,ier. . 96.7' Engler . . . . . . . . 98.1" (from barium snlphovinate and potassium , . . . . . . . . 97%-98*0" (by action of phosphorus pent-The older determinations are not given as they were evidently made upon impure material contaminated probably with the isomeric ethFl carbamine. No determination of the specific gravity of the pure sub-stance seems to have been hitherto made. The following observations of the expansion by heat were made in Dilatometer N in the water-bath :-cyanide) oxide on propionamide).These observations may be represented by the formula-VOL. XXXVII. 206 THORPE ON THE RELATION BETWEEN 3395.037 + 4.082 285t + 0.006 824 85t2 + 0.000 015 700P, which affords the numbers in the third column of the above table. of the glass (*00002553) this expression becomes-Dividing through by the first term and correcting for the expansion V = 1 -+ 0.001 227 75t + 0.000 002 040 64t2 + 0.000 000 004 675t3, which affords the following table showing the relative volumes of propionitrile a t every jA between OA and loo*. Epich.ZorKydrin C3H50C1 = CH2Cl.CH.0.CH2. The constitution of epichlorhydrin has been variously represented In the former expression it is seen that the oxygen is directly con-nected with two of the carbon atoms; whereas in the latter it is present as hydroxyl.The experiments of Darmstadter leave however, very little doubt that the first formula correctly represents the struc-ture of this compound. As a knowledge of the specific volume of epichlorhydrin would probably serve to throw additional light on the matter I have prepared a quantity of this liqnid and have determined its boiling point specific gravity and thermal expansion. A large quantity of dichlorhydrin was first prepared by Berthelot's method as modified by Reboul (Jnhr. 12 456) and this was con-verted into epichlorhydrin by treatment with potash. The epichlor-hydrin boiled constantly between 116.75 and 117*00" the greater portion coming over at 116.9".Bar. 762.2 mm. Corrected and reduced boiling point llCi.56*. *n = 10 ; t 19" THE MOLECULAR WEIGHTS OF SUBSTAKCES ETC. 207 Two de5erminations of specific gravity gave-I . . . . . . . . . . JI . . . . compared with water at the same temperature. compared with water at 44 these numbers became-1.19108 at 10.154 1.18783 ,) 13*07A Reduced to OA aud I . . ,. . . . . 1.20316 I1 . . . . . . . . 1.20310 1.20313 Mean . . . . . . . . Other observations on record are-Darmstiidter . . . . 117" at 755.5 1.204 at 0" Rebod . . . . . . . . . . 118-119" The following observations were made with Dilatometer N in the 1.194 ,) 11". water-bath :-Observed. --_ These results may be expressed by the formula-3380.336 + 3.384 6245 + 0.002 009 22t2 + O*OOO 022 445 tZ3.Dividing through by the first term and correcting for the expansion of the glass (-00002533) this expression becomes-V = 1 + 0.001 026 Got + 0.000 000 619 74t2 + 0.000 000 006 G57t3, by means of which the following t8able showing the relative volume of epichlorhydrin at every !jA between OA and 120A is calculated : 205 On the publication of Tollens and Henninger's memoir " On the Preparation of Allyl Alcohol " (Ann. Chem. Yharm. 166 134) Kopp remarked in a note appended to the paper that a comparison of the specific volume of +,he alcohol with that of its isomeride acetone would doubtless throw light on the relation of the specific volume of a body to its molecular strncture. It is definitely established that allyl alcohol made from glycerin and oxalic acid or obtained by the action of sodium on dichlorhydrin, has the constitution-CH, II CH I H = C - OH, whereas the constitution of acetone or dimethylketone is-CH3 I c = o ! CH3.Adopting Kopp's values the calculated specific volumes of these liquids are-Allyl alcohol 73.8 Acetone 78.2 the difference being of course due to the variation in the mode of combination of the oxygen (see p. 142). Tollens determined the rate of expansion of allyl a,lcohol by est,imat-ing its specific gravity a t different temperatures by means of th THE hIOLECULhR IYEIGHTS OF SCBSTASCES ETC. 209 pyknometer and he found that its volume a t to might be expressed with approximate accuracy by the formula-V = 1 + 0.000879t + 0*000002Gt2, the volume a t 0" being 1.The boiling point of the allyl alcohol was >j7O and its specific gravity at 0" compared with water a t the same temperature was 0.8709. Hence its specific gravity a t the boiling point 0.8709 58 - = 0.7848 and its specific volume- = '13.90 agree-Wi848 was l.lU973 ing almost exactly with the calculated value. The accuracy of this determination of the specific volume of allyl alcohol was called in question by H. L. Buff. Buff was of opinion that the specific volume of carbon in combination is not invariable a s Kopp supposes and he found experimental evidence to support this proposition in the relatively high specific volumes of certain ally1 derivatives notably of diallyl in which moreover he assumed the existence of dyad carbon.I n fact Buff concluded that the specific volume of an element was a function of or was related to its com-bining value and as he assumed that carbon had a variable combining value he assumed too ihat it had a variable specific volume Buff was led to redetermine the specific volume of allyl alcohol and in a pre-liminary nobe published in the Ber. for July 1876 he states that he had actually found the value to be considerably higher than the calcu-lated number in accordance with his h-j-pothesis. The number he gives vie. 74.6 was regarded by him as provisional and probably too low as he was of opinion that the alcohol which served for his experi-ments was incompletely dehydrated his expressed intention to repeat the determination on a larger qnantity of tlhe anhydrous alcohol was frustrated by his death in the following year.AT the question was still open I suggested to Mr. James Monck-mann a former student in the laboratory of the Yorkshire College, that he should prepare a quantity of allyl alcohol by Tollens's method from glycerin and oxalic acid with a view of determining its specific gravity and thermal expansion. About t kilo. of the alcohol was boiled for some hours with caustic soda distilled and digested for about a fortnight with a concentrated aqueous solution of acid sodium sulphite to remove any admixed acrolein. After separating the sul-phite solution the alcohol was treated with recent17 ignited potassium carbonate and repeatedly digested with quicklime (the alcohol being distilled off the lime after each digestion) until its specific gravity waB constant.The last four determinations of specific gravity were 210 A. 1 Observed. THORPE ON THE RELATION BETWEEN Calculated. I 0.8574 111 . . . . . . 0.8573 I V . . 0.8573 II 0.8575 Temp. 15*. The mean is 0.85738 at 15* compared with water a t the same tempe-rature. This number agrees exactly with that of Dittmar and Steuart (Pi-oc. Phil. Soc. Glasgow 1875-76 a) viz. 0.8576 a t 15" compared with water at 15.5. Tollens found (1) 0.8709 a t Oo and (2) 0.86045 a t 13" compared with water at 0" ; horn his interpolation formula the specific gravity at 15" compared with water at 15" would be (1) 0.8597 and (2) 0.8594. Ally1 dcohol retains traces of water with extraordinary tenacity a fact already noticed by Tollens and Dittmar.Repeated and prolonged heatment with freshly heated quicklime is required to dehydrake it completely. Altogether our alcohol stood for 38 days over six successive quantities of lime before its specific gravity became constant. As the dehydration proceeded we noticed the re-markable rise in the boiling point observed by Dittmar and Steuart. After the first treafment with lime the alcohol boiled at 92" after the third it boiled a t 94*5" and finally boiled constantly between 95.9" and 96*3" the greater portion coming over between 96.0" and 96.3. I) = 51 t = 18.5". Corrected and reduced boiling point 96-6A. A determination of the vapour-density afiorded the following Barometer 0.766 mm. data. :-Weight of liquid Volume of vapour 930 C.C.Temperature 99.8" Barometer . . . . . . . . . . . . . . . . i53 mm. 0.0954 gram Height of column 413 7, Found 28-85 Calculated. . 28.93 The following observations of the rate of expansion were made with Dilatometer C in the water-bath:-36-06 1 3102.7 3102 -7 These results may be expressed by the formula-3022.531 + 3.003 52t + 0.001 813 lot2 + 0.000 037 13W. On dividing through by the first term and correcting for the expan-sion of the glass (.0000247) we obtain- . V = 1 + 0.000 993 71t + 0.000 000 599 861t2 + 0.000 000 012 285t3, by means of which the following table showing the relative volume of allyl alcohol at every 5A between OA arid looA is calculated:- The above observations show that allyl alcohol is considerably more expansible than Tollens's formula would indicate-20".4.0". 60". SO". 96.6". Tollens . . . . . . 101862 103932 106220 108696 110916 T h o r p 102058 104230 196579 109163 111529 The specific gravity of allyl alcohol at 0" compared with water at 0" is 0.87063; compared with water a t 4A it is 0.8699; a t 9 6 . P it is 0.7800 ; hence its specific volume is ____ = 74-19. 57.87 0.7800 The volume thus obtained agrees so nearly with that calculated by means of Kopp's values that we must conclude that allyl alcohol fails to afford any evidence that the specific volume of carbon is variable 212 THORPE ON THE RELATION BETTVZEN Observed. *- I Acetone C,H,O = CH,.CO.CHJ. A quantity of this liquid obtained from Kahlbaum in Berlin was purified by converting it into the acid sodium sulphite compound.It boiled absolutely constantly between 55.78 and 55.80". 'YL = 0 t = 0. Barometer 728.2 mm. Corrected and reduced boiling point 5 6 ~ 5 3 ~ . A determination of vapour-density gave the following numbers :-Weight of liquid . . . . . . Volume of vapour Temperature . . . . . . . . . . 99.4" Barometer . . . . . . . . . . . . 742.5 mm. Height of column 487.0 ,, 0.0735 gram. 61.6 C.C. Found. . 28-35 Calculated. . . 28.93 Two determinations of specific gravity pve-1. 0.80755 a t 10.22* 11. 0.80636 at 11*27* compared with water at the same temperature. Reducing these re-sults by means of the formula given on p. 213 they give for the specific gravity at OA-I. 0.81858 11. 0.81858 compared with water at 4*.Other observations on record are-Calculated. Liebig Kopp Linnemann . . . . 7 Kopp found the boiling which agrees exactly with tension observations. - point of acetone to be 56.3" at 763 mm., Regnault's number derived from vapour- 54 -79 A series of observations made with Dilatometer C in the water-bath afforded the following numbers :- THE MOLECULAR WEIGHTS OF SUBSTASCES ETC. 213 These numbers afford the formula-3046,624 + 4.046 6285 + 0*009 116 88t2 - 0.000 001 113t9, which gives the results contained in the third column of t,he above table. Dividing through by the first term and correcting for the expansion of the glass (*0000247) we obtain-which gives the following table showing the relative volume of acetone a t every 5A between OA and 60A.V = 1 + 0.001 352 93t + 0.000 003 024 26t' - 0.000 000 000 29t3, The rate of expansion of acetone has already been determined by Kopp (Pogg. Am. 72 1 and 223) who found that it might be expressed by the formula-V = 1 + 0.001 348 1t + 0.000 002 609i2 + 0.000 000 011 559 2t3, which gives results in close agreement with that deduced from my observations as the following comparison shows :-15". 30". 45". Kopp 102085 104310 106700 Thorpe . . . . . . 102082 104303 106666 Heptane C7H, = CH,(CH,),CH3. I made several trials to procure this body in a state of sufficient purity to warrant the attempt to determine its rate of thermal expan-sion by the fractional distillation of so-called " cracked '' paraffin oil, but the results were not very satisfactory.No two samples of the hydrocarbon although boiling at approximately constant and almost identical temperatures had the same specific gravity or the same rate of expansion. My friend Scharleulmer sent me from time to time various specimens of heptane separated with all possible care from petroleum but the specific gravities and rates of expansion of these products were equally discordant and I came t o the conclusion that i t was impossible to separate the pure hydrwarbon from its congeners by the ordinary processes of fractional distillation. The discovery that the exudation fram the nut-pine or Digger pin 214 THORPE ON THE RELATION BETWEEN (Pinus sabiniana) a tree indigenous to California constitutes an abundant source of almost absolutely pure heptane has enabled me to obtain a more satisfactory result and it is with this vegetable hep-tane that all the determinations herein given have been made.For details of the mode of separation and purification of the hydrocarbon I may refer to my paper " On Heptane from Pinus sabiniaita " (Chem. SOC. J. June 1879). The first observation gave 98.27" a t 755% mm. ; the second 97.97' at 746.9 mm. ; column in both cases entirely immersed. Corrected and reduced boiling points 98*42A and 98*UA. The hep tane boiled absolutely constantly. Two determinations of vapour-density gave-I. 11. Weight of liquid,. . 0.1445 gram 0.0769 gram Volume of vapour . . 90.39 C.C. 69.77 C.C. Temperature 99-90" 99.60" Barometer 757.6 mm. 7'49.0 mm. Height of column .. 387.8 , 493-4 ,, Pound. Calculated I. 11. 49.90 50.07 49-94 Three determinations of specific gravity made with different bottles gave the following results :-I . . 0.68848 at 14*9BA 1 1 . . 0-68855 a t 14.87 111 0.68859 at 14-87 compared with water at the same temperature. Reducing these numbers by the formula given below the specific gravity af OA com-pared with water at 4A becomes-I . . 070048 11 . . . . . . 0.70046 111 . . . . . . 0.70049 0.70048 Mean . . _ . _ Two series of determinations of the rate of expansion were made. The first i n Dilatometer D gave the following data:- From these numbers we obtain the expression-2879.302 + 3.432 8853 + 0.002 478 31t2 + 0.000 039 64t3, by means of which the numbers in the third column of the foregoing table are calculated.Dividing through by the first terms md correcting for the expansion of the glass of the dilatometers (0.0000213 and 0*00002303) %he formulae become respectively-I. V = 1 + 0.001 205 17t + 0.000 001 338 41t2 + 0.000 000 099 701t39 and 11. V = 1 + 0.001 215 29t + 0.000 000 888 19t2 + 0.000 000 013 787t". The mean formula is-V = 1 -+ 0*001.210 23t + 0.000 001 113 3.P + O*OOO 000011 74t3, by the aid of which the following table showing the relative volume of'heptane at every 5 l between OA and loo* is calculated :- 216 THORPE ON THE RELATION BETWEEN A. j Observed. Ethyl-nmyl or Dirnethylbutylmetkane C,H, = (CH3),CH(CH2),CH3. I am indebted to Mr. Harry Grimshaw for the sample of this hydrocarbon which served for my observations.I t was obtained by the action of sodium on a mixture of the bromides of ethyl and nmyl, and was purified with very great care. F o r its analysis see CJzem. Xoc. J. 18'13 300. It boiled constantly between 90.5" and 91" under a pressure of 762.3 mm. Two determinations of its specific gravity a t OA afforded the follow-ing results :-I 0.69691 PI 0-69692 Corrected and reduced boiling point 9 0 ~ 3 ~ . Calculated. compared with water at 4A. Grimshaw found that ethyl-amyl boils a t go" and that its specific gravity was 0,6833 a t 18.4" compared with water a t the same temperature. This at OA and reduced t o a vacuum becomes 0.6990 compared with water a t 4b. Two series of observations of the rate of expansion mere made. The first in Dilatometer B gave the following numbers :- These numbers lead to the formula-2967.443 + 3.590 505t + 0.004 001 19tz + 0.000 033 309 6ts, which gives the numbers in the third coIumn of the above table THE MOLECULAR WEIGHTS OF SUBSTAXCES ETC.2 17 Dividing through by the first terms respectively and correcting for the expansion of the glass we obtain-I. V = 1 + 0.001 244 2t + 0.000 001 008 12t2 + 0.000 000 014 5572, and 11. V = 1 + 0.001 234 7t + 0.000 001 378 25t2 + 0.000 000 011 258t3. The mean formula is-V = 1 -+ 0.001 2394t + 0~00000119318t2 + 0~000000013058t3, by means of which the following table showing the relative volume of ethyl-amyl at every 5" between 0" and 95A is calculated :-A. Octane CeH, = CH,( CH,),CH,. I am indebted to Professor Schorlemmer for the sample of this liquid which has served fGr my observations.It was obtained from methyl-hexyl-carbino1 and after rectification over sodium boiled between 125.3 and 186.8". n = 23 t = 19. Bar. 760.8 mm. Cor-rected and reduced boiling point t= 125.46A. A determination of vapour-density afforded the following numbers :-Weight of liquid. . Volume of vapour 65.33 C.C. Barometer 740.2 mm. Height of column 503.3 ,, 0.0786 gram Temperature 99.3" Pound 56.68 Calculated 56.88 Two determinations of specific gravity at 0" gave the following I 0.71882 11 0.71885 Mean . 0.71883 results :-compared with water at 4" 218 THORPE ON THE RELATION BETWEEN The following obselwitions of the rate of expansion of octane were made with Dilatometer D.In the water-bath :- These results may be expressed by the formula-3180.483 + 3,694 92t + 0.000 514 93t2 + 0.000 041 l67t3, which gives the numbers contained in the last columns of the above tables. Dividing through by the first term and correcting for the expansion of the glass (*0000213) we obtain the following expression as repre-senting the expansion of octane between OA and its boiling point. V = 1 + 0.001 1E3 04t -+ O*OOO 000 186 648t2 + O*OOO 000 012 947t3, by means of which the following table showing the relative rolnmo of octane at every 5h between Oh and 125A is calculated :-A. THE MOLECULAR F'EIGHTS OF SUBSTAKCES ETC. 219 A. Observed. 0 .oo 2887.1 9 3'0 2919 -9 19.60 2954 1 2988.3 29*19 I Di-isobuty Z CSH, = (CE3)2CH(CH2),CH(CH,),.Prepared by Mr. W. C. Williams of the Owens College Man-Chester by the action of sodium on isobutyl iodide. The hydro-carbon thus obtained was treated with a mixture of concentrated sulphuric and nitric acids washed with water dried by potash and distilled over sodium. It boiled between 108.2 and 108.7". n=0. Bar. 748.2 mm. Williams observed 108-108*3" at 745 mm. on the same preparation (Chern. SOC. J. March 1879). That the hydrocarbon was of a high degree of purity will be evident from the following determination of Corrected and reduced boiling point = 108*534. Calculated. A. Observed. Calculated. vapour-density :-Weight of liquid . . . . . . . . . . . . . . 0.1449 gram Volume of vapour . . . . . . . . . . . . 85.30 C.C. Temperature .. . . . . . . . . . . . . . . 99.1" Barometer . . . . . . . . . . . . . . . . . . 734.9 mm. Height of mercury . . . . . . . . . . . . 389.7 ,, Found. . 56-90 Calculated. . 56.88 Two determinations of specific gravity a t OA compared with water at 44 gave-I . . . . . . . . . . 0.71109 I1 . . . . . . . . . . 0.71111 Mean 0.71110 Other observers have found-Kolbe 108 ? 0.694 at 18" Kopp . . . . . . 108.5 a t 747.5 mm. 0.7135 , 0 Wurtz . . . . . 106 ? 0,7057 , 0 Williams . 108.2 , 74.5 mm. 0.7085 , 0 7 7 . . 108-108.3 a t 745 mm. 0.7091 , 0 The following observations of the rate of expansion were made in In the water-bath :-Dilatomet.er E These results lead to the expression-2886.980 + 3.389 27t + 0.001 715 4.P + 0.000 040 85t3, by means of which the numbers in the last columns of the above tables are calculated.Dividing through by the first term and correcting for the expansion of the glass (*00002303) the above formula becomes-V = 1 + 0.001 197 O l f + 0.000 000 621 221t2 + 0.000 000 014 166t3, by the aid of which the following table showing the relative volume of di-isohutyl at every 5A between OA and 108*53* is calculated. The rate of expansion of di-isobutyl obtained by the electrolysis of potassium valerianate has already been determined by Kopp who has expressed his observations by the formula-V= 1 + 0.001 212 .tit + 0.000 000 279 3t2 + 0.000 000 016 297t3. This give numbers in close accordance with those furnished by the formula deduced from my observations. Mr. W. Carleton-Williams has also published a number of deter-minations of the specific gravity of di-isobutyl at different temperatures (Chem.SOC. J. 1879 125) but the rate of expansion calculated fro THE MOLECULAR WEIGHTS OF SUBSTANCES ETC. 221 Observed. Calculated. ~-~ --. ~-6A:96 1 2818.6 1 8818 - 8 1 52 *' 5 6 14 91 2837'3 2837 -1 69 -54 31 *65 2877 *9 2877 '9 84 -70 his observations is considerably less than that found by Kopp and myself. 25'. 50". 75". 100". Kopp 10306 10633 10995 11404 Thorpe 10305 10633 10993 11401 Williams - 10593 - 11322 Observed. AniZine C,H,.NH,. The sample of aniline which served for my observations boiled absolutely constantly at 181.5". 9% = 64" t = 65". Bar. 733.2. Cor-rected and reduced b. p. 183*7*. Its specific gravity was found to be 1,02763 at 11.63" compared with water at 4A; this at OA and compared with water at 4A becomes 1.03790.Other observers have found- These numbers may be represented by the expression :-VOL. XXXVII. Q 2502.383 + 2.353 77t + 0.000 712 265t2 + 0.000 009 695 7t3 222 THORPE ON THE RELATION BETWEEN which on dividing through by the first term and correcting for t h e expansion of the glass (.00002303) becomes-V = 1 + O*OOO 862 9% + 0.000 000 273 509t2 + 0.000 000 003 465 W, which affords the numbers in the following table showing the relative volume of aniline at every jA between OA and its boiling point. The rate of expansion of aniline has already been determined by Kopp who has expressed his results by the formula-1 + 0.000 817 3t + 0.000 000 919 lot2 + 0.000 000 000 627 84P.which gives slightly lower numbers than that afforded by the expres-sion deduced from my observations :-0°- 50". 100". 150". Kopp 10000 10433 10975 11454 Thorpe . . . . ,. 10000 10441 10925 11475 CH2=C - PicoZitae C6H,N = - CH\N (Schiff).* CH = CH - CHN The picoline which served for may experiments was a portion of a large quantity prepared by the late Dr. Anderson. A platinum deter-mination made by Dr. Ramsay (to whom I am indebted for the sample) showed it to be pure. * CAem. Xoc. J. 1871 403 THE NOLECULAR WEIGHTS OF SUBSTANCES ETC. 223 A. Observed. --___---0 '00 2982 -7 22 -39 3017.6 24 -25 3052.1 Calculated. A. Observed. Calculated. --_-2_- Nitrogen Tetroxide Nz04 or NOz. A special interest attaches to the knowledge of the specific volume of nitrogen tetroxide by reason of the light which it is calculated to throw on the constitution of this substance.Moreover as the group NOz plays an important part in many chemical compounds such know-ledge would presumably tend to elucidate the cause of the widely different volumes which nitrogen seems to possess in its various combi-nations. A quantity of the liquid was therefore prepared by heating carefully dried lead nitrate in a hard glass retort and a slow current of dry oxygen was passed through the stroiigly cooled product to insure the oxidation of any trioxide which might be present. The tetroxide was then distilled in an apparatus composed wholly of glass. It boiled completely between 21.67" and 22*20" the greater portion coming over between 21.7 and 21.95"; n = 22" t = 13.5".Bar. 760.0 mm. Corrected and reduced boiling point 21.64A. Its specific gravity was found to be 1.4903 at Oh compared with water at 4A. Dulong observed 1.461 (temperature not given) and the boiling point 28" (Schweig. J. 18 17'7). The boiling point is given as 22" in Watts's Dictionavy 4 '76 ; the authority is not stated. The following observations of the volume at different temperatures were made in Dilatometer D : THE MOLECULAR WEIGHTS OF SUBSTANCES ETC. 225 Observed. These observations may be represented by the formula-3304.594 + 5.187 52t - 0.01.3 228f + 0.000 071 175t3, which on dividing through by the first term and correcting for the expansion of the glass (*0000213) becomes-by means of which the following table showing the relative volumes of nitrogen tetroxide at every 5A between OA and 25* is caicu-lated :-V = 1 + 0.001 591t - 0.000 003 970 15t2 + 0.000 000 215 3t3, The thermal expansion of this liquid has already been studied by Drion (Ann. Ch,im. Phys. [3] 56 5 ) more particularly at temperatures above its ordinary boiling point but our results cannot be strictly compared as Drion's numbers are not corrected for the expansion of the glass. In a subseqnent communication I propose to give the results of my observations on the volatile chlorides of silicon titanium and tin and of certain arsenic phosphorus vanadium sulphur and chromium derivatives; and finally to discuss the entire series and to indicate the bearing of the conclusions to which they lead on this question of the relation of the molecular weight of a body to its specific gravity as a liquid