to be recrystallised several times in order to get it into a state fit for analysis. This process of purification, as well as the subsequent determinations, were undertaken by Dr. L. Marchlewski, whose scrupulous care and manipulative skill afford a sufficient guarantee for the accuracy of the results obtained.

I think it unnecessary to give the details of analysis of more or less pure specimens of the substance, but I may state that the crystals when first prepared gave the following percentages of C, H, and N.

The substance, after being purified by dissolving it in chloroform, adding a large quantity of alcohol, and collecting the crystals deposited on standing, the process being repeated three times, gave as the result of two determinations 11.93 and 11.85 per cent. nitrogen. After three further crystallisations the percentage of nitrogen was found to be 12.07 and 11.87. The four analyses agreeing so well as regards the nitrogen it was assumed that the substance was pure, and it only remained therefore to determine the C and H, the results obtained being as follows:

I. 0·1521 gram substance gave 0.3838 CO2 and 0.0852 H2O.

Taking the mean of the numbers given the composition of methylphyllotaonin would correspond to

In order to purify this substance the same method was adopted as in the case of methylphyllotaonin, i.e., it was dissolved in chloroform

Some time ago Mr. Percy C. Bell, working with a less pure specimen of methylphyllotaonin, found it to contain in 100 parts—

and obtained in crystals on the addition to the solution of several times its volume of alcohol. After this process had been repeated several times two determinations yielded as a mean

The substance having this composition was recrystallised four times in the same way as before, after which analysis led to the following results:

:

I. 0.1407 gram substance gave 0.3571 CO2 and 0.0762 H2O.

Want of material prevented further attempts at purification. It will be seen that methylphyllotaonin and ethylphyllotaonin approach each other very closely in composition, as they also do in appearance and general properties.

Phyllotaonin.

This substance was obtained by saponification in the manner formerly described, partly from methylphyllotaonin, partly from ethylphyllotaonin. After drying at 130° analysis yielded the following results:

I. 01245 gram substance from methylphyllotaonin gave 0:3116 CO2 and 0·0675 H2O,

II. 0·1921 gram substance from ethylphyllotaonin gave 04884 CO2 and 0.1062 H2O.

Mr. W. Joseland made two analyses of phyllotaonin, the results being as follows:—

0.5239 CO, and 0·1093 H2O.

I. 0.2076 gram substance gave
II. 0.2149 gram substance gave 24.10 c.c. N. at 22.5° and 951 mm.

This compound was prepared by dissolving phyllotaonin in boiling glacial acetic acid and allowing to crystallise.

0-1282 gram substance dried at 130° gave 0·3194 CO2 and 0·0715 H2O.

In determining the C and H of this compound, Mr. Joseland obtained the following numbers :

0.1425 gram substance gave 0.3564 CO2 and 0.0754 H2O.

There are several formula with which the analyses of the derivatives of chlorophyll given above more or less closely agree, as will be seen from the following table:—

Found.

C20H19N3O2(OCH3). C40H38N6O5(OCH3). C1H1N6O5(OCH3).

Ethylphyllotaonin.

Calculated.

Found. C.... 69-22

38

C20H19N3O2(OC2H5). C40HзN6O5(ОCH ̧). CâНμÑ6O5(OC ̧H;).

Found.

C20H19N3O(OAc). C40H39N6O(OAc). CaH4N6Os(OAc).

VI. "Thermo-electric Properties of Salt Solutions." By GEORGE FREDERICK EMERY, B.A., LL.M., of the Inner Temple, late Scholar of Trinity College, Cambridge. Communicated by Professor J. J. THOMSON, F.R.S. Received February 8, 1894.

The thermo-electrical properties of solutions have not hitherto received much attention from physicists. If we form a circuit of two substances, one a metallic wire and the other a solution, and arrange it so that the junctions between the metal and the liquid are at different temperatures, we generally find that an electromotive force is developed in the circuit which varies in magnitude nearly in proportion to the difference of temperature between the junctions, and which, in comparison with the ordinary thermo-electromotive forces in metallic circuits, is very considerable. Up to the present time, as far as I am aware, the only extensive measurements of such thermoelectric forces are those of M. Bouty (Journ. de Phys.,' vol. 9). He concludes that for a given difference of temperature between the junctions when the metal is constant the E.M.F. is nearly independent of the nature and strength of the solution (the solution being of some salt of the metal used).

The object of my experiments has been to see how far this is true, and to find out how (in the event of its not being strictly true) the E.M.F. varies with variations both in the strength and in the nature of the solution. The results show that both have considerable influence on the magnitude of the E.M.F.

My method of observation is as follows: the solution to be examined is put in a U-tube (fig. 1, A), in each limb of which is one of the ends of metal, which we may for convenience call electrodes, BB.

Close to each electrode is the bulb of a thermometer, C, graduated to one-fifths of a degree Centigrade. One limb of the U-tube is surrounded by a water-jacket, DD, which is heated by an ordinary circulating arrangement, so that its temperature can be regulated as desired. (See fig. 1.) The E.M.F. between the electrodes is balanced against a branch of a circuit of about 20,000 ohms through which a pair of Leclanché cells are kept closed, and the value of an ohm in E.M.F. is measured against a Clarke cell.

At the commencement of each experiment, when both electrodes were at the same temperature, there was very often a slight E.M.F between them. This was carefully read, and then the water-jacket was gradually heated and the readings of the thermometers were taken as the E.M.F. passed through, values represented by definite numbers of ohms generally differing by 5 or 10. The readings at the highest temperature were very carefully taken, and then the whole was allowed to cool, more readings being taken as the temperature fell. The value of any particular experiment is roughly given by the closeness of the readings for corresponding differences of temperature, while the temperature was rising and falling. In some series of observations the temperature was allowed to vary very slowly, so as