LUDLAM: A SIMPLE FORM OF LANDSBERCER’S APPARATUS. 1193 cx IX.-A Simple Form of Landsberger’s Apparatus foy Determining the Boiling Points of Solutions. By ERNEST BOWMAN LUDLAY, M.Sc. SINCE the appearance of Landsberger’s original paper (Bey., 1898, 31, 458), several modifications of his apparatus for determining molecular weights by the boiling point method have been described. Walker and Lumsden (Trans., 1898, 77, 503) increased the rapidity with which a number of consecutive determinations could be accom- plished by graduating the tube in which the rise of temperature was measured, and, instead of weighing the solvent, measuring its volume, The volumes were reduced to weights by multiplying by the density of the boiling solvent, or, what comes to the same thing, in calculating VOL.LXXXI. 4 K1194 LUDLAM: A SIMPLE FORM OF LANDSBERGER’S APPARATUS the molecular weight from the observations a constant was used, obtained by dividing the usual constant by the density of the boiling solvent. A simplification in the arrangement was effected by Smits (Proc. K. A h d . Wetemch. Amsterdam, 1900, 3, 86), who placed the boiling tube inside the flask in which the solvent was boiled, but did not graduate his tube. What may be considered a distinct improvement was added by C. N. Riiber (Ber., 1901, 34, 1060), namely, a reflux arrangement something like that employed in a fat extractor for returning the condensed vapour again to the flask in which it was generated. This apparatus gives good results, but is somewhat elaborate in design, and seems too fragile to come into general use in the laboratory.About fifteen months ago I re- quired an apparatus which would be easy to work with, and with which good remlts could be ob- tained when only small quantities of the substance, of which the mole- cular weight was being investigated, were available. After devising several forms, the one described below was finally adopted. It com- bines the good points of Walker’s and of Smits’, and in addition possesses features of its own. Unfortunately, it has not been found possible to introduce any reflux arrangement, and at the same time to preserve simplicity. Riiber sacrifices simplicity, and his ad- vantage for specially accurate work is a very real one, inasmuch as the same solvent is used continuously, and consequently any error due to the solvent not being perfectly pure is greatly diminished, and also less of the solvent is required than in the apparatus here described. The solvent is contained in an outer flask, A , of 300 C.C.capacity, and possessing a wide neck provided with a small side tube, 8, through which the solvent is introduced. The flask is fitted with a large cork, through which passes the tube B, about 10 cm. in length and 2.6 cm. in diameter.FOR DETERMINING THE BOILING POINTS OF SOLUTIONS. 1195 In the side of this tube is a small hole, b, t o allow of the passage of the vapour from the boiling liquid into B, and then into C. B is fitted with a cork carrying the innermost tube, C, graduated, and tapering off sharply at the bottom to a small hole.This aperture is made to serve as a valve by means of a little spherical glass bead (u) fused to a platinum wire. The bead is inside the tube, and the wire, passing through the hole, is bent at right angles at a short distance below. I n this way, the bead is allowed just a little up and down motion, but cannot be carried away from the opening by a rush of vapour or liquid, and it allows the stream of vapour to pass easily up into the tube from below, whilst preventing the liquid in the tube from rapidly flowing out into B when the boiling in B is checked. This is all that is required, for the bead need not fit the aperture exactly. The liquid can flow from C into B-which is convenient-but only very slowly. The clip on the tube connected to 8 can be removed to allow access of air when the boiling is stopped and the apparatus allowed to cool.The top part of the tube C is wider than the lower graduated portion, and between the two is an annular trough formed by blowing a bulb, and then, while the glass is still soft, gently pressing the lower tube up into i t and so producing a fold in the glass. Any vapour condensing in the upper part of the tube flows into this trough and is carried away by a side piece to the condenser. By this device, very little condensed vapour is returned to the graduated tube, and the volume of liquid in the tube does not increase so rapidly as it otherwise would do. This trough was the suggestion of Dr. Young, and is a valuable addition to the efficiency of the apparatus. The thermometer is held by a cork which closes the wide neck of this tube, and in the larger forms a side tube was also provided, as shown in the figure, for the introduction of the substance under investigation in the form of pellets, or, in case it was a liquid, by means of a pipette in the usual manner.The side piece was no convenience in the smaller forms of the apparatus owing to the fact that the stem of the thermometer occupied so much space in the narrow tube that there was a danger of the pellet falling into the trough instead of into the graduated tube. It was found in practice to be more convenient and satisfactory rapidly to remove the thermometer, drop in the pellet, and then replace the cork immediately. To carry out a determination, the flask A is half filled with the solvent selected, the tubes B and C and the thermometer are fitted into their places, and the condenser attached to the side tube from C .1196 LUDLAM: A SIMPLE FORM OF LANDSBERGER’S APPARATUS Then the liquid in A is heated by a small, carefully regulated flame and maintained in a state of gentle ebullition.The vapour passes upwards into the neck of the flask, round the tube B, and through the little hole, b, into the space separating B and C. Then it descends to the valve,througE, which it passes upwards into C, and so away to the condenser. I n this way, the innermost tube, in which the increment of temperature is measured, is protected by a double layer of vapour and hot glass from external influences. At the commencement, when all the tubes are cold and the thermo- meter also, a considerable portion of the vapour condenses and accumu- lates in the inner tube C .Through this liquid the vapour passes, the valve causingit to break up into a stream of bubbles which serve the double purpose of heating and thoroughly stirring the liquid. After the vapour has been passing through the apparatus for a few minutes, the tubes get hot, and subsequent condensation in them is slight. Hence the liquid accumulates, only very slowly in the tube C, the actual rate depending on the substance, its boiling point, and heat of vaporisation. I n the case of benzene, t,he tube was usually filled in about half-an-hour or a little longer, whereas with carbon disulphide the increase in volume due to condensation was so slight that it was frequently necessary to add the liquid solvent in order t o obtain suit- able readings, and even then the tube was not full after the vapour had been passing for an hour and a half.It was found that $he tube filled more rapidly when the boiling was vigorous than when it was gentle, provided that, in the latter case, the stream of vapour was not intermittent. The temperature rises very rapidly to within a few tenths of the true boiling point and then gradually attains the constant position. This may take a quarter of an hour from the commencement, the actual time depending on the rapidity of the boiling, and to some extent on the previous history of the thermometer, for if the instru- ment has been recently used for determining the boiling point of the same solvent several times in succession, and has not been used for any other purpose, it very rapidly gives a constant reading.The temperature of the pure solvent having been ascertained in this way, a known weight of the substance under examination is dropped into the boiling solvent, and as it dissolves the temperature rises, attaining a maximum in a few minutes, and then falling very slowly as the volume of the solution increases. When the maximum temperature is observed, a reading of the volume is made. The thermometer is raised above the surface of the liquid by sliding it upwards through the cork, and is not removed from the tube at all. If the boiling flask is touched with the cold hand, sufficientFOR DETERMINING THE BOILING POINTS OF SOLUTIONS. 1197 cooling takes place to stop the stream of bubbles through the graduated tube, and an accurate reading of the volume can be taken.The valve prevents the solution running back into B, the correct volume is easily read off, and the boiling once more allowed to proceed. The whole operation of reading the volume need not take more than twenty seconds. By taking several readings, there is no difficulty in obtaining three or four determinations from one quantity of substance, and practically without disturbing the thermometer. Further, a second pellet of the substance may be added, and, after taking several readings, a third may be introduced. Thus, in one case, using carbon disulphide as solvent, three pellets were introduced and seven readings were obtained without interrupting the boiling for more than a few seconds.I n order to ensure a uniform rate of ebullition, the gas supply was regulated by the device due to Sakurai (Trans., 1892, 61, 995) whereby alteration in gas pressure is minimised. The bench gas tap is opened to the full and is connected to two bunsen burners, one much larger than the other. The large one is kept burning full on, and the small one has its supply of gas further regulated by a screw clip. As the smaller flame is using only a small fraction of the quantity of gas used by the large burner, the latter acts as a most efficient regulator for changes in the gas pressure, which may be consider- able in a laboratory where there is a continual turning off and on of other taps by students. Trials were made with the apparatus in which the dimensions of the parts were varied considerably, but the size and proportions given in the figure were finally segarded as the best for most purposes.A larger size is possibly a little more accurate, owing to the temperature being steadier, but with increase in the size, more of the substance is required, and more of the solvent is used. The accuracy of the volume reading remains proportionally the same whatever the diameter of fhe tube. If the boiling tube C had a diameter much smaller than 1.6 cm., the results obtained were always too high, and this is probably due to the rush of vapour past the thermometer. The vapour is at a lower tempera- ture than the solution, and the thermometer consequently records a temperature slightly below that of the liquid, hence the observed incre- ment is smaller than it ought to be, and the molecular weight calculated from it comes out higher than the true value. The large bulb of the ordinary Beckmann thermometer nearly fills a tube of 1.3 cm.bore. For this reason, a Beckmann thermometer with a particularly small bulb, and graduated only to tenths of a degree, was ordered, but while awaiting its arrival, a cheap paper scale instrument graduated over 50' range in fifths of a degree was wed, and below will be found1198 LUDLAM: A SIMPLE FORM OF LANDSBERGER’S APPARATUS Temp. incre- ment. Volume’ results obtained with it. It is interesting to compare the total cost of this apparatus, using such a thermometer, with the expensive and elaborate contrivances usually considered necessary.The whole of the apparatus as described above can be made by a novice in the art of glass blowing, and, including the thermometer, need only cost a few shillings. Withanaccurate and delicate thermometer having a very small bulb it should be possible to reduce the size of the apparatus so that it occupies no more space in the laboratory than an ordinary melting point apparatus; it could be kept permanently in position, and except for the one weighing, a determination of molecular weight need be no more trouble than a careful melting point determination. Although the apparatus was not designed to give specially accurate results, but rather rapid determination with only slight error, never- theless, comparing the results which have been obtained with it and with the modern Beckmann apparatus both by myself and by students who have had no previous practice with either, it is undoubtedly superior as regards accuracy also.Below are given figures which have been obtained at various times during the last fifteen months, some the result of experiments expressly performed to test the apparatus, the remainder obtained in the ordinary course of laboratory work. The first molecular weight determined by the apparatus was that of propyl benzyl ketone in benzene, and, in addition to reading the volume, the weight of the solvent was also taken : Con- I centra- tion. dl M, Error Per obs. calc. cent. Per 100 Substance, Weight in grams. Propyl benzyl ketone.. 0’260 0’40 13 2 164 162 f1’3 , I I ,, I ,, ( w t .10.79 gms. 1 ,, I 161 1 I , I -0% The internal diameter of the boiling tube in this case was 1.8 cm., and a large bulb Beckmann thermometer which had been set apart for determinations with boiling benzene was used. With a tube of 1.6 cm. diameter, but similar in all other respects, the following results were obtained :FOR DETERMINING THE BOILING POINTS OF SOLUTIONS. 1199 - Error Per cent. f 8 + 10 M'. Error per cent. -___ 93 -1-26 M'. Error per cent. , y .................. Naphthalene ............ Benzophenone ......... Azobenzene ............... >? 9 , ............ ........... Y Y Y , ......... ......... , , .............. 0 *b) - - 0 -55 0.238 - - - Con- centra- tion. Grams M. per 100 , C.C. Weight in grams. Temp. incre- ment. Substance. Volume. --- ' I I- Aniline ..................) ) .................. 0.78 0'10 1.87 1'10 wt. 11 gms. 2.9 C.C. 93 93 With the smallest tube, 1.3 cm. in diameter, a large number of determinations were made with aniline, all of which were much too high. As typical, the result of one of these is given : Wejght in grams. Volume. centra, tion. i Con- Incre- ment. Substance. M. Aniline ................ .I 0 '60 .................. i " ,, 1 $"* 1.4 * 0.80 0.66 0.57 10 C.C. 11'8 6 5 123 119 Another series gave : 5-6 2.23 6'4 1 1.9 wt. 5'63 gms. 1-9 92 97 104 ;; I s: 93 +11 Aniline .................. , y .................. ,) .................. 0'125 2 ) ¶ > * In this case, the thermometer used was the small one mentioned above, graduated The following figures were all obtained by means of the apparatus in fifths of a degree from 50" to 100..as described above in its final form, the solvent being alcohol : - Con- entra. tion. 7.0 6 '4 5% 5 *2 5 *o 3-6 3.5 9.0 7% 7 *2 5.6 5-0 - - M. Weight in grams. Incre- ment. Substance. Volume. 7.5 8.3 9.5 10.2 6 8-2 8.5 6 -2 7'2 7 '6 4.2 4-75 I--- I- 100 99 98 99 139 130 130 188 177 174 179 179 - Aniline ................ 0 *53 * , , ................. y , .................. 1-10 1.00 0.89 0 '82 0 '56 0 '44 0 '42 0-74 0 '68 0.66 0.50 0.45 I * Allowing the solution to cool and then weighing it, the value obtained for M was 92, indicating that the volume reading was not accurate.1200 LUDLAM: A SIMPLE FORM OF LANDSBERGER'S APPARATUS Con- entra- tion. The difficulty in obtaining good determinations of molecular weight increases greatly with.the size of the molecule, and still more if there is an accompanying diminution in solubility. Some substances of this kind were under investigation, and their molecular weights were determined by this method.For example, the additive compounds of dibenzyl ketone and deoxybenzoin with benzylidenem-nitraniline gave the following figures : Ob- M. served '2; -I- 1-7 3-4 2.9 1.7 1.6 2.0 Cz8H,,0,N,. .............. 0.293 0.24 8 -7 1 ,, 10.21 I 10'5 ................. 0.192 0-14 11 *o 459 440) 436 409 404) 436 412 422 12'0 Cz,H,,O,Nz ............... - Error Per cent. + 5 +1 -6 -7 -2 - The last was a most interesting determination, for tbe substance was only sparingly soluble in boiling ;benzene, and the pellet did not dissolve completely.The temperature rose to its maximum value and then remained absolutely constant while the volume was slowly in- creasing. The rise in boiling point was the difference between the boiling points of the pure solvent and of the saturated solution of the substance in it, the latter remaining constant as long as there was solid remaining undissolved. When the volume had increased sufficiently, all the substance dissolved and the readings were taken, the result being quite satisfac- tory. A determination of the same substance in pyridine gave 410, instead of 412 as obtained in this case with benzene. The use of pyridine for molecular weight determinations has been recommended by RoseInnes (Trans,, 1901, 79, 261), but he employed the Beckmann method, and found that owing to the length of time that his corks were subjected t o the action of the pyridine vapour it was necessary to protect them, otherwise they lost their springy nature and shrivelled up into a hard material like wood. Owing to the rapidity of the determination by the method now described, no such difficulty was encountered.The density of boiling pyridine is almost exactly unity, the weight- constant is 29.5, therefore 29.5 was also taken as the volume- constant :FOR DETERMINING THE BOILING POINTS OF SOLUTIONS. 1201 Substance. Deoxybenzoin benzylidene- Succinic acid .................... Metaldehyde ..................... m-nitraniline, C27H,20,N,} Y, ..................... Con. Error Weight. Rise. Volume. centra- Jf. M'. per tion. cent. ------- 0.187 0.185 10.6 1.8 410 422 - 3 0'230 0.40 14 1.7 124 118 + 5 0.11 0.21 12 0.9 131 132 -0'6 0'18 0.30 14'8 1.2 121 ,, -8 Many inorganic salts dissolve in pyridine, but they form compounds with i t in so doing.An interesting observation was made in the case of mercurous chloride. When this salt is boiled with pyridine, it is rapidly decom- posed, giving mercury and mercuric chloride, the latter of which combines with the pyridine, and the compound crystallises out in hard, fine needles. This seems to afford valuable evidence as to the condition of the mercurous chloride molecule. I n Debray's experiments, the mercury was removed from the sphere of action by means of the gold tube cooled by a stream of water. I n Victor Meyer's experiment, it was their different volatility that separated the mercury and mercuric chloride.I n this case, the mercuric chloride is removed by the pyridine, and the dissociation proceeds to completion although the temperature is only 11 ti0. Metaldehyde, so far as I can ascertain, has not been investigated in boiling solutions, but the molecular weight in phenol and thymol has been determined by the cryoscopic method and found to corre- spond to three times the molecular weight of aldehyde, which result is now confirmed by this method, Carbon disulphide was found to be a very convenient solvent for use in this apparatus, and a large number of determinations were made with it. It was first used because it was found necessary to deter- mine the molecular weight of some substances at temperatures lower than that oE boiling benzene or alcohol in order to prevent decomposi- tion.It also possessed the further advantage for our purpose that, like benzene, it was an associating solvent, and would be most likely to give as high a molecular weight for a substance as it was capable of assuming at such a temperature, For instance, Riiber (Be?*., 1901, 34, 1060) obtained the value 264 for the molecular weight of benzoic acid in carbon disulphide, whereas in alcohol he obtained 112, the calculated value for C6H,*C0,H being 122. The ordinary constant for carbon disulphide is 2370 ; dividing by the density at the boiling point (1.225) this gives 1930 as the constant VOL. LXXXI, 4 L1202 for use with volume readings. results obtained with methyl acetoacetate benzylideneaniline : BEWITT AND AULD: THE ACTION OF As an example, we give below the Substance. C,,H,,OJY.., ...... ....... .............,... Con- tion. Weight. ::::: Volume. centra- M. ----- 0.22 14.7 2.26 196 ), 15'0 2-20 177 ), 16.0 2.06 166 ), 16.7 1-98 152 The first of these four determinations took place 35 minutes after introducing the substance. As it did not all dissolve at first it was necessary to wait until sufficient carbon disulphide had collected to dissolve it. The fourth reading of the series was taken 40 minutes later ; decomposition had evidently been steadily taking place. In another series of determinations with the same substance, after 8 minutes the molecular weight was 247. Seven readings were taken in the course of an hour, the last giving a molecular weight 181. Owing to the rapidity with which the solution attains a state of equilibrium, or proceeds on a course of gradual displacement in one direction, and the ease with which the thermometer records the condition of the solution, the apparatus should prove useful in tracing the course of reactions such as the above, UNIVIRSITY COLLEGE, BRISTOL.