I am compelled to observe that the observations do not always give such concordant results for the cases of the radiation in B and A as shown in the above tables. If the electrodes are not freshly polished, and if the radiation in B is not strong enough (so that fewer ions are produced), then the results obtained are of the type shown by the following

table :

Here one sees the discharge with radiation in B taking place at potentials lying between the potentials obtained for discharges without introduction of ions or without direct radiations, and the potentials measured in exposing the sparking system to direct radiation. But the potentials necessary to produce the discharge were always about 1000 to 1500 volts lower than those measured without employing any means to destroy the retardation.

3. It might be permissible to insert here a few words with a view to a theoretical explanation of these results. There are, in my opinion, mainly two points to be considered which.

so far as I am aware, have not yet been taken into consideration in the various attempts that have been made to explain the phenomena of retardation and of the spark-discharge in general. These two points are (1) The fact discovered by Geitel and by C. T. R. Wilson f, that the air always contains ions, and that there is a continuous production of ions connected naturally with the re combination of these ions. (2) The principle first established by Prof. J. J. Thomson ‡, and also brought forward and strengthened by experiments executed in the Cavendish Laboratory by Prof. Townsend §, viz., the principle of the production of new ions by the collisions of negatively charged corpuscles moving under the influence of strong electromotive forces with the molecules of the gas.

A careful comparison of the results of experiments undertaken from different standpoints, shows that the electromotive force per cm. which is required to give to the negative ions a velocity such that they can produce new ions by collisions with the molecules of the gas, is very nearly the same as the electromotive force required to produce spark-discharge in the gas at the same pressure and with the electrodes at the distance of 1 cm.

Prof. Thomson illustrates this, in the paper mentioned above, by a small table taken from a paper by Skinner (Phil. Mag. [5] 1. 1900). Here X (the potential-gradient per cm. in the positive column) means the above described minimum E.M.F., and p is the pressure. The table is as follows::

Liebig (Phil. Mag. [5] xxiv. p. 106) found the value required for a spark-discharge at a distance of 1 cm. in air

* H. Geitel, Phys. Zeitschrift, ii. p. 116 (1900); J. Elster & H. Geitel, ibid. p. 560 (1901).

+ C. T. R. Wilson, Proc. Roy. Soc. lxviii. p. 151 (1901).

J. J. Thomson, Phil. Mag. [5] 1. p. 278 (1900); ibid. [6] i. p. 361 (1901).

§ J. S. Townsend, June, 1901.

Nature,' August 1900; Phil. Mag. February and

at atmospheric pressure to be 31,000 volts.

X

=40-8. The fact that the values of

p

X

p

This gives

are nearly coinci

dent seems to entitle one to draw the above conclusion. I

hoped to be able to calculate the values of

X

for a wider p range of pressures from the recent observations of Orgler*, but I was prevented from so doing as his numbers and curves do not apply for greater distances of the electrodes than 0.5 and 0.6 cm. respectively. Paschent, also, in his well-known paper, does not give enough observations for a spark-length of 1 cm. at different pressures to enable me to calculate the X values of in a sufficient number of cases.

Starting from the fact already mentioned, that even air at atmospheric pressure always contains ions, and that ions are continuously produced, it seems possible to obtain a fairly clear idea of what happens before a spark passes, as well as what happens during the spark-discharge.

If, for instance, one subjects two electrodes to slowly increasing electromotive forces, there will be formed immediately a very feeble current, as observed by Warburg and other experimenters. This current will, so long as no external agents are acting, remain constant within very wide ranges of the E.M.F.; but its intensity will increase very quickly and tend to a maximum, as soon as the E.M.F. attains the value necessary to give to the ions the velocity required to produce new ions by impact.

It seems to me that the experiments of Kreusler prove this very decidedly. It ought to be observed that in all his experiments (the final ones) the electrodes were exposed to ultra-violet light, but nevertheless when he approached the discharge-potential, a change in the value of the E.M.F. of 17, 17, and 13 per cent. (he used Pt, Cu, and Fe electrodes) corresponded to an increase in the intensity of the current of 337-2, 243-1, and 392 per cent. The theory strongly demands that just as the spark-potentials are approximately the same, so also these last numbers ought to be the same; but it seems that here secondary circumstances of the experiments exert a certain influence.

Let us assume, for instance, the intensity of such a saturation current between the two electrodes to be only * A. Orgler, Ann. d. Physik, i. p. 159 (1900).

+F. Paschen, Wied. Ann. xxxvii. p. 69 (1889).

p. 91.

H. Kreusler, Ver. Phys. Ges. Berlin, 1898, p. 86, especially table,

of the intensity of the maximum current observed by Mr. Kreusler; thus let I (intensity) be 10-10 ampere. Then a simple calculation by means of the formula

gives

I=qe

q=2× 10° approximately,

where q equals the number of ions per unit-volume. Comparing this value with Loschmidt's number, we see that only about the 10-11 part of the molecules become ionized. If no external ionizing agents are acting, then the transformation of the 20 ions contained in unit volume according to C. T. R. Wilson (if I is greater then, of course, there is a correspondingly greater number of ions) into about 2.109 ions must take place by collisions. The time which is necessary for this transformation is what Prof. Warburg calls the period of retardation. It is obvious that this time is considerably shortened if we produce by radiation a new set of ions whose number is large compared with the number of ions originally present, which new ions are also put in motion, thus producing still more new ions by collisions with the molecules of the gas.

It will be easily seen that the above considerations explain fairly well Prof. Jaumann's results.

The following is a new definition of Maxwell's "electric strength" of a gas based on this view of the nature of the spark-discharge:-"The electric strength of a gas at a pressure p is defined as the electric intensity required to give to the negative ions a velocity sufficient to enable them to produce other ions by collisions with the molecules of the gas. According to this view, the determinations of the sparkpotential under the action of radiation made by Prof. Warburg and his pupils are to be regarded as giving the normal sparkpotential, as Prof. Warburg maintains. It follows also that when working without radiation one ought to obtain the same normal potential, provided sufficient time is allowed for the electric intensity to act.

I hope soon to be able to publish further experimental results in support of these views.

The spark itself appears to amount practically to a short circuit between the electrodes. In addition to the production of ions by collisions, the following causes help to explain this :

:

1. The ionization resulting from the high temperature of the spark.

2. The presence of hot metal vapour.

3. The emission of cathode rays by the cathode due to the influence of the ultra-violet light given out by the spark.

In conclusion I wish to say that my heartiest thanks are due to Prof. J. J. Thomson for the kind and liberal hospitality with which he received me at his laboratory, and for the continuous interest he has taken in my work.

Cavendish Laboratory, Cambridge.

XXIX. The Anomalous Dispersion of Cyanin. By Privatdocent Dr. A. PFLÜGER, University of Bonn, Germany*.

several papers Wood† has communicated a new method of making prisms of solid cyanin, and also a repetition of the measurements of the dispersion-curve, as I made and used them several years ago for the proof of the Ketteler-Helmholtz dispersion-formula ‡.

Using my photographic method, Wood finds that cyanin has a strong absorption-band in the ultra-violet, beginning at the wave-length λ=372 up. He says that in this part of the spectrum it makes the measurements of the refractive indices impossible, since the strong absorption prevents any impres sion on the photographic plate, even with a five-hours' exposure. He continues §: "Pflüger found no traces of this band, and gives values for the refractive index within its limits. It seemed at first that the reason of this might be found in the difference in the optical properties of fused cyanin and that obtained by the evaporation of an alcoholic solution, but we have found that films prepared in the same way as those by Pflüger show the band also."

My measurements of the refractive index [after the manner accurately explained [] in this part of the spectrum were made on photographs, some taken with an exposure of 25 minutes, others with 40 minutes. The plates show plainly the double image of the iron lines used for the purpose of the measure

ments.

Furthermore, I have made photographs of the whole absorption-spectrum, which show plainly the absorption-band in the visible part of the spectrum, but not the faintest trace of an absorption in the ultra-violet. In these experiments the light of an iron spark passed through a quartz plate

* Communicated by the Author.

Wood and Magnusson, Phil. Mag. [5] xlvi. pp. 380-386; [6] i. pp. 36-45, January 1901. Wood, Phil. Mag. June 1901, pp. 624–627. Pflüger, Wied. Ann. lvi. pp. 412-432; lxv. pp. 173–228. Wied. Ann. lxv. p. 199.

Phil. Mag. Jan. 1901, p. 41.