In solution 1, the ionization remains constant at all temperatures, but for the other solutions there is a very marked decrease with rise of temperature, and this becomes greater the stronger the solution."

Referring back to Table E, it will be seen that for solution 1 there has been no change in fluidity by the addition of the salt to the water. It is only when the fluidity becomes smaller by the addition of the salt that there is the diminishing

of the ratio with rise of temperature, so that the

diminution of

Ca

C.

Са

may be due to two causes, the re-combi

nation of the ions and the retarding influence of the molar viscosity; and the latter influence may be felt in much weaker solutions than one has generally supposed.

I have at the begininng of this paper briefly referred to the work of Kohlrausch in connexion with the zero conductivity of dilute solutions. His conclusions were arrived at from a study of the temperature variations of conductivity of dilute solutions between 0° and 34° C. The question of a lower limit to the conductivity has since been attacked by Bousfield and Lowry, who show that the conductivity of dilute solutions and the viscosity of water tend towards the same limiting temperature, and over the range of temperature (from 5° to 34° C.) the temperature variations can be expressed by the same kind of curve. They, however, doubt the existence of the zero at the point indicated by Professor Kohlrausch and Professor Lyle and myself.

Neither Kohlrausch nor Bousfield and Lowry used Slotte's form of equation to represent their results, and it has been shown by Thorpe and Rodger† that this is the one which gives the best values, where a wide range of temperature is involved, for viscosity; and it is probably the best form of equation to use in connexion with conductivity results.

Kunz conducted some low-temperature experiments with strong sulphuric-acid solutions and solutions of other substances, and decided that in these cases no zero conductivity existed at the temperature supposed.

Kohlrausch § has recently studied the temperature variations of ionic mobilities, and has here introduced the idea that ions in solutions are surrounded by watery atmospheres carried along with them, and the resistance the ions *Bousfield & Lowry, Roy. Soc. Proc. p. 42, June 19, 1902, †Thorpe & Rodger, Phil. Trans. 1894.

Kunz, Compt. Rend. vol. cxxxv. p. 788 (1902).

§ Kohlrausch, Sitz. Ber d. Berlin. Akad. p. 572 (1902).

Phil. Mag. S. 6. Vol. 7. No. 41. May 1904.

2 L

have to overcome is mainly friction between this atmosphere and the solvent water.

In a later paper, Kohlrausch has explained more fully his previous paper, and has sketched out the new view of the mechanics of electrolysis, according to which the moving ion carries with it a mass of adhering solvent, and the electrical resistance of an ion is a frictional resistance which increases. with the dimensions of the atmosphere surrounding it.

One of the conclusions he arrives at is that the resistance of an ion expressed in mechanical units must be of the same order of magnitude as the mechanical resistance of a molecule of the solvent.

The velocity of the ions will depend on the viscosity of the medium through which they pass, and on the size of the ionic atmosphere; and the conductivity of the solution will depend on the viscosity of the medium, the size of the ionic atmosphere, and the fraction of dissociated ions in solution.

The lithium ion, which moves very slowly, may be considered as the centre of a larger atmosphere than the fastly moving chlorine ion, and probably with rise of temperature the atmospheres will approach the same size, as Kohlrausch has observed that with rise of temperature the velocity of the ions tend to become equal.

If we examine our weakest solution, we shall see that here the conductivity does not keep pace with the molar fluidity as the temperature is raised, although there is no combination of the ions. It looks as if the atmosphere around the negative ion is increasing rapidly, while that around the positive ion remains constant or diminishes slowly.

C

F

If for our solution at infinite dilution we take the value

= 1.048 at 18° C., and consider the atmosphere of the Cl ion to have unit radius, that of the Li ion will have √2 times this radius at the temperature 18° C., because the Cl ion moves twice as fast, and the retardation will depend on the square of the radius.

Now at 100°946.

C
F

The ratio has decreased in the

proportion 1.048 946 or 1·11 to 1. This decrease is due to the increase of the atmosphere of the ions, and as these atmospheres tend to become equal, if we assume that at 100° they are equal, the radius of each will be 1:29, so that the Cl has increased its atmosphere by 29, and the Li has diminished its by 12.

*Kohlrausch, Roy. Soc. Proc. Feb. 17, 1903.

In solution 1, there is no combination of the ions as the temperature rises, so that here, too, any change in the value C

F

of will indicate a change in the ionic atmospheres. Here also the decrease is from 111 to 1. For this solution, F has the same values as for the solution just considered, but C has smaller values throughout because there are fewer carriers of electricity, although the atmospheres work out to have the same values as in the other solution throughout.

Solution 2 has the same increase for the radii of the atmospheres if the ionization coefficients in our table have the physical meaning which that name should imply, but here the atmospheres at any particular temperature are smaller by one per cent. If, however, the atmospheres remain of constant radius for all solutions, at any particular temperature, the figures under 2 in the table of ionization coefficients must be increased by 1 per cent. throughout to have their physical meaning.

Fig. 3.

For the stronger solutions the ratio 111 to 1 remains throughout, but there is nothing to indicate how the ionic atmospheres vary with the concentration.

In the present paper I have confined myself to the temperature variations of fluidity and conductivity. The fluidityconcentration and conductivity-concentration isothermals are also interesting, but I shall not discuss them here. There are indications, however, that all the isothermals will cut the axis of zero conductivity and fluidity at the same point representing concentrations of about 16 normal. I intend making further experiments on strong solutions to test this point.

The curves connecting the variables F and n, and C and n, are given in figs. 3 and 4 respectively, each curve being the isothermal for the temperature indicated on it.

I wish, in conclusion, to thank Professor Thomson for the interest he has taken in this investigation.

Cavendish Laboratory,

March 14, 1904.

LIX. On the Analysis of Bright Spectrum Lines. By JAMES BARNES, M.A., 1851 Exhibition Scholar, Fellow of Johns Hopkins University*.

[Plates XXV. & XXVI. ¡

TT T is well-known that a change is produced in the wavelength and distribution of light in the lines of the spectrum of metallic vapours and gases when different external conditions are introduced. In most cases these changes were first observed and measured by means of the Rowland grating. Recently, however, these effects have become more readily observable through interference methods, in which the interference-bands are produced with large differences in the paths of the rays.

Michelsont, by aid of his interferometer, resolved the important lines in the radiations of some vapours and gases rendered luminous in vacuum-tubes, and he has studied these radiations in a magnetic field. With his echelon spectroscope he has investigated the same subjects. Fabry and Perot + with their interferometer have investigated the radiations from vapours in the electric are and in vacuum-tubes, and have applied their method for an exact determination of the wave-length of some of the lines in the spectrum of the iron are and of the dark lines in the sun's spectrum. Lummer §, also by an interference method, has studied the same radiations, particularly those from mercury, and has separated its prominent lines into many components.

When one compares the results of these investigations the agreement is not very satisfactory. Not only do the number and intensity of the components differ, but the distances. between the components do not agree.

The work presented in this paper was undertaken at the suggestion of Professor Ames. The objects of the work were: to study interferometer methods; to obtain, if possible, more consistent results as to the constitution of the lines; and to determine the changes produced in the components under various conditions. Michelson remarks in one of the papers cited :-"Still, in many cases, the range of visibility due to slight variations in the conditions shows that the behaviour of each substance must be carefully studied under all possible

Communicated by Professor J. S. Ames.

+ Phil. Mag. [5] xxxi. p. 338 (1891); xxxiv. p. 280 (1892).

↑ Ann. de Chim. et Phys. xii. p. 459 (1897); xvi. pp. 115 & 289 (1899) ; Astrophys. Journ. ix. p. 87 (1899).

$ Verhdlgn. d. D. Phys. Ges. iii. p. 85 (1901); Phys. Zeit. [3] viii. p. 172 (1902).