;82 



NA TURE 



[February i8, 1892 



"solvent," metal will mix in all proportions with either of the 

 other "immiscible metals" separately, the ternary mixture 

 proJuced either forms a "real" ternary alloy, not separating 

 into two different mixtures on standing (molten), or else an 

 " ideal " alloy incapable of existing, and immediately separating 

 into two different ternary mixtures ; e.^. one chiefly containing 

 lead with some of the tin and a little zinc, the other mainly 

 consisting of zinc with the rest of the tin and a little lead. The 

 two points representing these two mixtures lie on opposite sides 

 of the point indicating the total mass of the three metals origin- 

 ally used, this third point lying somewhere on the straight 

 line connecting the other two "conjugate points," or "tie- 

 line." By employing a series of mixtures containing gradually 

 increasing proportions of thi third "solvent" metal, two 

 branches of a curve are gradually traced out representing the 

 respective loci of the pairs of conjugate points ; these two 

 branches of the "critical curve "tend to meet at some point, 

 termed the "limiting point," where the tie-lines vanish. 



A large number of experiments were made to determine the 

 conditions necessary to obtain the nearest possible approxima- 

 tions to the positions of truly conjugate pairs of points ; former 

 experiments having indicated that the compositions of the two 

 alloys formed might be influenced by the relative proportions of 

 the two immiscible metals, in such fashion that points slightly 

 varying in position along one branch might be obtained as 

 conjugates to a given point on the other branch under different 

 conditions. It was found that such variations entirely dis- 

 appeared with thorough admixture, which is more readily 

 effected when the proportions used are such as to give rise to 

 approximately equal quantities of the two alloys than when 

 they are such that one alloy is formed to a much larger extent 

 than the other. Observations with non-metallic analogous 

 fluids (chloroform, water, glacial acetic acid) always gave 

 sharply concordant values, thorough intermixture (by shaking 

 vigorously in a stoppered bottle) being much more easy than 

 with metals melted in a crucible and simply stirred vigorously. 



Plotting the curves with the heavier immiscible fluid (lead, 

 chloroform, bismuth) at the left hand corner of the base of the 

 triangle, the lighter one (zinc, water) at the right hand corner, 

 and the solvent fluid (tin, acetic acid) at the apex of the triangle, 

 it was found that the tie-lines with chloroform-water-acetic 

 acid, always sloped downwards to the left, the angle of slope 

 continually increasing. The right hand branch of the critical 

 curve rises to a maximum elevation, and then descends again to 

 the limiting point, whilst the left hand branch continually 

 ascends to that point. At the limiting point the chloroform 

 and water are in the proportions 2CHCI3, SHgO ; the acetic 

 acid present being less the higher the temperature, i.e. the criti- 

 cal curve for a higher temperature lying inside that for a 

 lower one. At positions near the limiting point the mixture is 

 extremely senshive to temperature-variation ; a few tenths of a 

 degree will often make all the difference as to whether a single 

 homogeneous fluid results (a "real" mixture), or two different 

 ones, the point indicating the mixture in the first case lying just 

 outside the critical curve, and in the second case inside it. 



Analogous results were obtained with the metallic mixtures ; 

 with lead-zinc-tin mixtures, the lower ties slope to the left, the 

 upper ones to the right, the limiting points also lying on the 

 right and below the highest point of the critical curve ; at this 

 point the lead and zinc are nearly in the proportion PbZug. 

 Two points, one in the left hand branch, the other in the right, 

 are indicated by the configurations of the tie-lines, corresponding 

 approximately with the definite atomic compounds SnPbg and 

 SnZn4 (when aluminium is used instead of zinc, similar con- 

 figurations are developed, corresponding with SnPba and SnAl4). 

 When silver is employed instead of tin, all the ties slope to the 

 left ; irregularities of outline (bulges) are noticed, caused by the 

 formation of definite atomic compounds, AgZng and Ag4Zn5 : 

 these bulges are still more pronounced when bismuth is used 

 instead of lead (bismulh-zinc-silver curve). With both silver-lead- 

 zinc and silver-bismuth-zinc curves, the limiting points lie to the 

 lejt of and below the highest point of the critical curve ; with 

 both tin-lead-zinc and tin-bismuth-zinc curves, to the right and 

 below the highest points. 



In any given curve, substitution of bismuth for zinc depresses 

 the curve, so that the curve with lead uniformly lies outside that 

 with bismuth. The curve for a higher temperature always lies 

 inside that for a lower temperature, the effect of temperature- 

 variation being very much less marked at the lower parts of the 

 curve, than at points lying anywhere near the limiting point. 



NO. I 164, VOL. 45] 



February 4. — "On the Thermal Conductivities of Crystals 

 and other Bad Conductors." By Charles H. Lees, M.Sc, tl?e 

 Owens College, Manchester, 



The author commences by referring to Kundt's discovery, that 

 the metals stand in the same order as conductors and as to the 

 velocity of propagation of light through them, and mentions that 

 his experiments were originally intended to furnish data for a 

 similar comparison for crystals, but that their object has been 

 extended. 



After some preliminary experiments, he adopted the "divided 

 bar " method, which consists in placing a disk of the material 

 the conductivity of which is required, between the ends of two 

 bars of metal placed coaxially, heating one end of the combina- 

 tion, and observing, by means of thermo-junctions applied to the 

 bars, the distribution of temperature along them — first, with the 

 disk in position ; second, with the bars in contact without the 

 disk. When the conductivity of the bar is known, these 

 observations suffice to determine that of the disk. 



The ends of the bars which came in contact with the disks 

 were amalgamated, as this was found to be the best method of 

 securing good contacts These bars were suspended horizontally 

 in a frame, and, by means of screws, set accurately in the required 

 position. 



The conductivity of the bar was determined by the method 

 due to Forbes of determining the loss of heat from the surface by 

 allowing the bar to cool and observing the rate of change of 

 temperature, and then observing the steady distribution of 

 temperature along the bar when heated at one end. 



The author finds it to be 0*27 C.G. S. unit, and to increase 

 slightly with temperature. 



The disks used were of the same diameter as the bar, and 

 were of various thicknesses, in order to make the distribution 

 of temperature throughout the bars nearly the same in each 

 case. 



The following are the results obtained. No relation of the 

 kind found by Kundt for metals seems to hold for the crystals 

 experimented on : — 



C.G.S. units. 



Crown glass 00024 



Flint glass 0'0020 



Rock salt ... 0'0I4 



Chemical Society, January 21. — Prof. W. A. Tilden 

 F. R. S., Vice-President, in the chair. — The following papers 

 were read : — The estimation of oxygen dissolved in water, by 

 M. A Adams. The author describes an apparatus in which the 

 estimation of dissolved oxygen in water by Schiitzenberger's 

 method maybe carried on so as entirely to avoid loss of oxygen 

 by diffusion. The results obtained by this method are liable to 

 differ according to the rate at which the determinations are 

 effected, higher results being obtained when the operations are 

 quickly performed. — The luminosity of coal-gas flames, by V. B. 

 Lewes. The author has quantitatively studied the actions which 

 occur in luminous gas flames, and also those which lead to loss 

 of luminosity in the flame of a Bunsen burner. He considers 

 that the most accurate method of dividing a luminous hydro- 

 carbon flame into zones, is to regard it as made up of three : — 

 (i) The inner zone, in which the temperature rises from a 



