STRUCTURE.* Guided by the above facts, I have devised a modified | THE CHEMICAL SIGNIFICANCE OF CRYSTAL method of analysis for the barium group which in repeated trials by myself and my students has been found to be eminently practical and easy. 1. Precipitate the elements of the group as carbonates, wash and dissolve in the least quantity of warm dilute acetic acid in the usual way. 2. Preliminary Testing. (a) Add to 1 cc. of the solution in a test-tube a few drops of hydrochloric acid, then I cc. of a freshly prepared concentrated solution of sodium sulphite. A precipitate is barium sulphite, proving the presence of barium. (b) Add to 1 cc. of the solution a few drops of dilute acetic acid, then I cc. of the sulphite solution. A precipitate is barium or strontium sulphite; in the absence of barium, can only be strontium. (c) If barium and strontium are absent, add to 1 cc. of the solution I cc. of the sulphite. A precipitate is calcium sulphite, easily soluble in acetic acid. 3. Remove barium ion, if present, with dichromate ion, and free the filtrate from dichromate ion by precipitation as carbonate and solution in dilute acetic acid in the usual way. The removal of the dichromate ion is generally not necessary as the reactions are not seriously disguised by it. 4. To 5 cc. of the solution of 3 (or of 1 if barium is absent) add about 3 cc. of the sulphite solution, or enough or complete precipitation. A precipitate may be calcium or strontium sulphite. No precipitate does not indicate their absence. In either case add about 1 cc. of dilute acetic acid, or just enough to dissolve or nearly dissolve the precipitate. Avoid much excess. If much strontium s present it will not dissolve. Heat to boiling, and set aside until the precipitate has mostly subsided. Pour the quid on a filter, leaving the bulk of the precipitate in the test-tube. If the filtrate is cloudy, pass it through the filter again and again until it is perfectly clear. The filtrate will contain most of the calcium and very little strontium. The precipitate is strontium sulphite, but may contain some calcium sulphite. 5. Add to 5 cc. of the filtrate of 4 a little ammonium oxalate solution. A precipitate is calcium oxalate. If there is only a faint cloud it may be due to strontium. In this case, dilute another portion of the filtrate with an equal quantity of water, and add the oxalate. The precipitate must be calcium oxalate, since the strontium concentration has been reduced below the precipitation limit of strontium oxalate. 6. Cover the precipitate of 4 in the test-tube with hydrochloric acid, and heat to boiling. Test on platinum wire in the flame for strontium. If much calcium is present there may be enough in the precipitate to mask the strontium flame. In this case a very sharp distinction can be made as follows:-Use two platinum wires, dipping one in a pure hydrochloric acid solution of a calcium salt and the other in the solution to be tested. Insert the middle of the wires in opposite sides of the Bunsen flame near its base, and draw them slowly forward until the looped ends are in the flame. The sodium (and potassium if dichromate has been used) burns off first and leaves the more persistent calcium and strontium in the flame. A decided difference in the flame colours thus brought side by side is caused by even a trace of strontium. 7. For further confirmation of strontium add calcium sulphate solution to a portion of the solution of 3, and boil. A slowly forming precipitate indicates strontium. The chief advantage of this method lies in the easy and unmistakable identification of calcium in the presence of strontium. University of Nashville, Nashville, Tennessee. Iron and Steel Institute.-The Annual Meeting of the Institute will be held at the Institution of Civil Engineers, Great George Street, Westminster, on Thursday and Friday, May 11th and 12th, 1911, commencing each day 10.30 o'clock a.m. By Prof. WILLIAM J. POPE, M.A., F.R.S (Concluded from p. 152). HAVING shown that the crystalline forms of the elements are in complete harmony with the conception that crystal structures can be homogeneously divided into similar cells of polyhedral shapes approximating closely to the spherical, reference may now be made to some simple compounds, those, namely, in which the molecule consists of two dissimilar atoms. case. The conception of the equilibrium of centred forces which has been shown fertile in the case of the crystalline elements can be immediately applied to the binary compounds; as before, each atom will be represented by forces emanating from a centre, and equilibrium will demand closest packing of the spheres used, just as in the previous The atomic centres will now, however, be of two kinds, and the question arises as to whether the domains of atomic influence to be described about them will be all of the same magnitude or whether two magnitudes of spheres must be employed, one for each element present. This question is difficult to answer by reference to the facts already reviewed above; probably the only indication which the latter afford in this connection is that closestpacking of a considerable variety of different magnitudes would certainly be most unlikely to lead to the close similarity of crystal form observed as between the elements and the binary compounds. A direct answer is, however, provided as the result of investigating the crystalline forms of organic substances, to which reference will presently be made; this investigation has led to the discovery of a definite law which governs the magnitudes of the several kinds of atomic domain concerned in any crystalline compound substance. It is found that the magnitudes of the atomic domains in any crystalline compound are very approximately in the ratio indicated by the fundamental valencies of the corresponding elements. Since the molecules of nearly all the binary compounds which have been crystallographically examined contained in the molecule one atom each of two elements of the same valency, the polyhedral cells from which a crystalline binary compound must be supposed built up are all, in general, of approximately the same magnitude. The fact that most binary compounds, like most elements, crystallise in either the cubic or the hexagonal system, represents one of the simple results of this law of valency volumes. The binary compounds thus, in general, affect crystalline structures which are derived from the cubic or the hexagonal closest-packed assemblage of equal spheres ; one-half of the spheres, selected homogeneously, represent atoms of the one element and the remainder atoms of the second element. The mode in which the necessary homogeneous selection may be made in the cubic assemblage, without altering the values of corresponding dimensions in three rectangular directions, is shown in a model. The crystalline forms of the binary compounds are in accordance with what has been above foreshadowed. Table I. indicates that in geometrical respects the crystalline binary compounds closely resemble the elements; 68.5 per cent of those examined are cubic and 19.5 per cent hexagonal, the remaining 12 per cent crystallising in systems of lower symmetry than these. The axial ratios, a c, of all the hexagonal binary compounds known are stated in Table II.; all approximate closely to the value, a c = 1: 08165, for the model hexagonal closest-packed assemblage of equal spheres. In connection with the elements and binary compounds it is noteworthy that the mode of treatment described * A Discourse delivered at the Royal Institution, April 15th, 1910. appears practically to eliminate molecular aggregation of the atoms as a factor in determining the crystalline structure; that is to say, the distance separating two neighbouring atom centres is the same whether those atoms belong to the same or to different molecules. Another interesting fact is that, whilst the elements and binary compounds for the most part crystallise in the cubic or hexagonal systems, substances of greater molecular complexity rarely crystallise in these highly symmetrical systems; thus, of a great number of organic compounds examined, 2.5 and 4.0 per cent only belong to the cubic and hexagonal crystalline systems respectively (Table I.). This observation is important as one of many indications that the cells into which the crystal structure of a complex compound are partitionable are not, in general, all of the same volume. Further investigation shows that the volumes of the polyhedral cells representing the atomic domains of the several elements present in a complex crystalline compound are governed by the law of valency volumes to which reference has already been made. The correctness of this conclusion concerning the proportionality between the numbers expressing the fundamental valencies of the elements and the volumes of the corresponding spheres of atomic influence has been abundantly verified, not only by the laborious process of working out a large number of cases, but in several other ways which may be more rapidly indicated. The following are illustrations of the latter kind of verification. Table III. states the composition and axial ratios, a: b: c, of a series of four crystalline minerals which differ in composition by the increment, Mg2SiO4; the sums of the valencies of the atoms composing the different molecular aggregates are stated under the heading W. The increment, Mg2SiO4, also occurs as the crystalline forsterite. 2'449: 2277: 2.869 2·429: 2*245: 2·867 Clinohumite .. 1'0803: 1: 5.6588 Values for the increment, Observed. I'0757 I: 1.2601 Calculated 10823 1: 12775 mineral forsterite, of which the axial ratios have been determined. It is evident that the ratio, alb, has approximately the same value of 108 for all four members of the series, and that practically all differences in relative dimensions are expressed by the ratio, c/b. On dividing the valency volume, W, by the corresponding value for c/b in each case, the quotients 117, 121, 123, 124, and 12.7 are obtained respectively for the substances prolectite, chondrodite, humite, clinohumite, and forsterite. The relative dimension, c/b, is thus roughly proportional to the sum of the valencies in this set of minerals. The comparison may, however, be made more accurately by including the changes in both relative dimensions, a/b and c/b, in the calculation in the following manner. The " equivalence parameters" are the rectangular dimensions, x, y, and z, of a rectangular block having the volume W, and are in the ratio of the axial ratios, a b: c. The parameters x and y preserve almost constant values throughout the series, and addition of the increment, Mg2SiO4, leads to a practically constant increase of about 2.86 in the dimension z, on passing from one mineral to the next in the series. The mineral forsterite also gives nearly the same x and y values as before, and its z value, 2.87, is equal to the differences between consecutive pairs of z values in the main series these differences vary between 2.85 and 2.88. The axial ratios and equivalence parameters of forsterite can indeed be calculated with considerable accuracy from the data available for the series of four minerals. The relations here displayed may be rendered more obvious by a series of models (Fig. 5). Rectangular blocks having as the horizontal dimensions the x and y values, and as vertical dimension the value, for forsterite, when superposed upon a similar set of blocks having the corresponding dimensions for prolectite, form a stack exhibiting the equivalence parameters of chondrodite; superposing on this a second set of forsterite blocks leads to a stack showing the equvalence parameters of humite, and on again repeating the operation, a stack with the dimensions of clinohumite results. From the numerical data and the models exhibited it must be regarded as definitely proved that, in this series, the volumes appropriated by the constituent atoms are, in any one member, directly proportional to the valency numbers of the corresponding elements. (Mean value of S=2·970). Experimental determinations of the molecular volumes of a long series of normal paraffins, made on the liquid substances at temperatures at which the materials are in physically similar conditions, are stated in column 4. Since the valency of carbon is four times that of hydrogen it would be anticipated from the crystallographic conclusions four times as large a space for occupation as one hydrogen previously drawn, that each carbon atom should appropriate atom; the quotient of the molecular volume by the valency the same value, S, in the case of all the hydrocarbons. sum or valency volume, W, should consequently lead to The mean value of S, namely, the atomic volume of hydrogen, is thus calculated as 2.970, and that it is constant within very narrow limits is seen on comparing volume 4 and 5, the latter of which states the product of the valency volume, W, by the value 2.970. The simple relation between the atomic volumes of carbon and hydrogen in the liquid normal paraffins indicated in the above table was recently pointed out by Le Bas, and is abundantly confirmed by numerous series of determinations that the law of valency volumes, first enunciated on the in addition to that now quoted. It is thus definitely proved ground of the crystallographic evidence, holds rigidly in the case of these liquid substances. has now been devised by means of which the vast stores of accurate goniometric measurements collected by crystallographers during the past century can be interpreted and that the requisite interpretation has in many cases already been given. Prof. Liveing, in a discourse delivered in this room nineteen years ago, suggested that crystalline forms are the outcome of the accepted principles of mechanics; the aid of these, and of these alone, has been invoked to show that crystalline structures result from the equillibrium of the attractive and repulsive forces radiating from the atomic centres. Sufficient has been said to demonstrate that a method Royal Institution.-A General Monthly Meeting of Members of the Royal Institution was held on the 3rd inst., The Duke of Northumberland, K.G., President, in the Chair. Mr. J. Devonshire, Mr. G. H. Griffin, Mr. W. E. Lawson Johnston, Mrs. Guy Pym, and Mr. A. E. Reed were elected Members. The Chairman reported the decease of Professor J. H. van't Hoff, an Honorary Member of the Royal Institution, and a resolution of condolence with the family was passed, CHEMICAL NEWS, Weight of Normal Litre of Hydrogen Chloride. THE WEIGHT OF A "NORMAL" LITRE OF AND THE ATOMIC WEIGHT OF CHLORINE.* As the result of an exhaustive research on the density of The calculation of the most probable atomic weight to be assigned to this element is further complicated by the slight uncertainty which still exists in the atomic weight of hydrogen. According to Morley if O=16 H = 100762 (Zeit. Phys. Chem., 1896, xx., 1), whilst according to Noyes H = 1.00787 (Journ. Amer. Chem. Soc., 1907, xxix., 1718). In consequence, some of our own results, and Scheuer's, as well as those of Dixon and Edgar (Phil. Trans., 1905, ccv., 169), and Noyes and Weber, lead to two parallel sets of numbers, and it is difficult to say which mean value is the more exact approximation. A choice between the two must clearly be governed by the results of other determinations in which chlorine has been referred more or less indirectly to oxygen; but apart from physico-chemical evidence there are only two investigations which fulfil these conditions, viz., that of Richard and his colleagues (loc. cit.), in which chlorine and oxygen were linked by means of the ratios Ag: AgNO3, Ag: AgCl and AgCI: NH4Cl, and that of Guye and Fluss (Fourn. Chim. Phys., 1908, vi., 732) who directly compared the atomic weights of nitrogen, oxygen, and chlorine by the analysis of nitrosyl chloride. The two values yielded by these researches are:Richard Guye and Fluss CI = 35'456 Richards' value involves a knowledge of the atomic weights of hydrogen and nitrogen, but a slight variation in the former hardly affects the results, and there can be little doubt now that the uncertainty in the latter is very small (Guye and Drouginine, Journ. Chim. Phys., 1910, viii., 473). The method of Guye and Fluss has the advantage of affording a direct comparison between chlorine and oxygen; but the agreement one naturally expects between the results of the two researches is strikingly absent, the values differing by nearly 1 part in 3500, so that no decisive conclusion can be drawn from this evidence. TABLE I.-The Atomic Weight of Chlorine, O=16. Noyes and Weber Gray and Burt 35'459 Edgar (Phil. Trans., 1908, A, ccix., 1) 35.461 35'467 35'469 35'462 35'470 35°463 35'471 Scheuer .. 35'465 35'471 35'468 * A Paper read before the Faraday Society, March 14, 1911, 161 This was the Physico-chemical measurements afford another means of obtaining the atomic weight of chlorine directly in terms of the oxygen standard. By comparing the limiting densities of oxygen and hydrogen chloride the molecular weight of the latter gas can be obtained. The atomic weight of the halogen can then be deduced by subtracting the atomic weight of hydrogen, which in this case does not need to be known with great exactness. method we followed in the latter part of the research already cited, and we obtained for chlorine the value 35'461, but the substitution of Herr Scheuer's density for ours yields the value 35°466. It is true that the concordance here is greater than in the former case, but it can hardly be said to set at rest all doubts on the question at issue. A knowledge of the exact value of the density of hydrogen chloride is hence of great importance in a discussion of the true atomic weight of chlorine. At present there is no reason why the values of Dixon and Edgar, Edgar, ourselves, and Scheuer referred to the higher hydrogen basis, as well as that of Guye and Fluss, should be any less probable than the much lower value of Richards, and that of Noyes and Weber expressed on the lower hydrogen basis, but if the density of hydrogen chloride were established beyond be furnished in favour of one of the two competing sets of doubt within narrow limits of error, strong evidence would values. Before attempting to decide which of the two values for the weight of a normal litre, our own or Scheuer's, is the closer approximation, it is necessary to consider the errors which may have affected both series of results. Discusssion of the Chief Sources of Error in the Our experience in measuring the density of this gas has been in many respects similar to Scheuer's. The individual results agree fairly well among themselves, but not so well as the errors of measurement would lead one to expect. The total probable uncertainty in each individual result is, in Scheuer's case, about 1 part in 5000; in our case it is of the same order, and yet the maximum deviations in the two cases are I part in 1640 and 1 part in 1533 respectively. This can harly be attributed to a vitiation in purity of the gas, for fractionation did not cause a progressive change in density, and, moreover, our volumetric analyses yielded very concordant values. We must therefore conclude that these variations are caused by some disturbing factors, which are inherent in the methods employed. Several possibilities are worthy of consideration, and the more important appear to us to be : (a) Chemical action of the gas on the lubricant used for the stopcocks. (b) Solution of the gas in the lubricant. (c) Adsorption of the gas on the glass surfaces of the density bulbs. a and b.-It is probable that there is no stopcock lubricant which does not dissolve this gas to an appreciable extent. Our earlier experiments were carried out in an apparatus in which the stopcocks were lubricated with a grease compounded of the same ingredients as that used by Scheuer. We found the grease was markedly attacked by the gas, soon becoming dark in colour and in some cases apparently charring. This has also been noticed by Steele and Bagster (Trans. Chem. Soc., 1910, lxc., 2608). Its use was therefore abandoned in favour of a pure paraffin lubricant, and the densities previously obtained were rejected. It is noteworthy that the mean of these densities agrees very closely with Scheuer's results, as Table II. shows, and is in marked contrast to our later mean, viz., 1·63922, obtained in the same bulb, when the rubber grease was replaced by the paraffin lubricant. Without laying too much stress on this point, we are inclined to believe that rubber vaseline grease is unsuitable for use with hydrogen chloride, and partially accounts for the discrepancy between Scheuer's results and our own, Even in presence of a lubricant which is not chemically attacked there is still the possibility of an appreciable error resulting from solution. The tap grease dissolves the gas, and part of this is given off when the density bulb is pumped out. In practice this effect will tend to make the density determinations too high. During the filling and adjustment of the temperature and pressure the grease partially saturates itself and causes the full bulb to weigh more than it should. On exhaustion all or part of the gas in solution is given up and the subsequent weighing approximates closely to the initial one, so that the non-variation in weight of the empty bulb during a series of determinations is no proof of the absence of this error. That solution actually occurs is proved both by Scheuer's results (loc. cit., 1910, p. 601) and our own (loc. cit., p. 1643), but the amount of gas dissolved or given up varies so much with the experimental conditions that it is not possible to apply a constant correction. (c) Adsorption.-The adsorption or condensation of compressible gases on glass surfaces has often been pointed out as a source of error of uncertain magnitude in exact determinatious. Most experimenters have sought to determine its influence by varying the capacity of the bulb in which the gas is weighed. This method has yielded no very definite results for gases beyond showing that the magnitude of the effect must be small (G. Baume, Journ. Chim. Phys., 1908, vi, 1. Guye and Davila, Mém. de la Soc. de Phys. et d'Hist. Nat. Genève, 1908, xxxv., 635). It will generally be admitted that when a given mass of gas is introduced into a glass vessel the pressure exerted by the gas on the walls will, on account of adsorption, be slightly diminished. The greater the ratio of surface to volume the greater will be the diminution of pressure, and conversely the pressure corresponding to the total number of molecules in the mass of gas will only be reached when this ratio is infinitely small. As McBain (Zeit. Phys. Chem., 1909, lxviii., 497) has shown in the case of hydrogen adsorbed by cocoa-nut charcoal, the total quantity of condensed gas exists in two states-(a) a surface film, (b) a solution in the material of the condensing surface. When equilibrium between the gas and its condensing surface has been obtained a constant fraction of the total mass of gas is present in these two states, but the magnitude of this fraction, surface remaining constant, must clearly depend on the concentration of the gas producing adsorption and also on the temperature. In other words, the actual mass adsorbed will vary with the pressure and temperature of the gas in the vessel. In the case of hydrogen and charcoal, McBain has shown that equilibrium between the free gas and that adsorbed as a surface layer is quickly reached, whilst adjustment of equilibrium of the gas in solution proceeds slowly and is only complete after a considerable interval of time. Bearing in mind the possible existence of these two effects, we have investigated the adsorption of HCl on glass surfaces in the following way : The method, of which only a brief account has appeared before, was originated by a suggestion of Prof. W. Ostwald, and is based on the assumption that when gas is displaced a constant pressure from a glass vessel by means of mercury the adsorbed gas remains behind between the mercury and the glass surface, and can afterwards be collected by lowering the mercury so as to produce a Torricellian vacuum. The adsorbed gas given up is then removed, and the process can be repeated until further evolution ceases. The apparatus consisted of two cylinders A and B, of about 300 sq. cm. internal surface, connected with each other through a two-way stopcock c. The cylinders could be filled with mercury from the reservoirs D and E, which were attached to a cord running over the pulleys G and н. This system of adjustment enabled one to transfer the gas from one cylinder to the other at constant pressure. The cylinder в was not used in the measurements, and merely served as a reservoir for the collection of the displaced gas. The measuring cylinder A was surrounded with ice and water contained in a tin bath, and the capillary tube K, which was carefully calibrated, was used for the measurement of the volume of the adsorbed gas. Before the hydrogen chloride entered the apparatus the greatest care was taken to ensure the complete absence of moisture, for both Scheuer and ourselves found that condensation is greatly increased if the vessels are not thoroughly dry. The gas was left in contact with the glass for twenty-four hours, and then displaced into the cylinder B at constant pressure. The tap c was then turned off, and a short length of the mercury thread was frozen in the calibrated capillary by means of solid carbon dioxide contained in the paper vessel L. This was necessary because the gas dissolved in the grease of the tap, and afterwards slowly evolved in a vacuum. The mercury was next lowered so as to produce a vacuum, into which the adsorbed gas escaped, and on allowing mercury |