of a solution of potassium permanganate which cut off everything except the extreme blue and violet and a patch in the extreme red. The light was then passed through the seawater, and the track therein was examined through a prism with its refracting edge horizontal. In addition to the blue and violet, the green was also conspicuously present. Yellow and red were present, but were very faint. Specimen F showed the effect best, B and E came next, and then C and D. The fluorescence was certainly present even in the last two specimens, though it was very weak.

Since the incident light did not contain any ultra-violet, the fluorescence should be excited by the blue and violet. This implies an extra absorption of the blue and the violet. A sea whose water shows the fluorescence would thus appear green, not only because there is a return of green light due to fluorescence, but also as there would be less of the blue. and violet returning. As we have seen, that water looks most green which shows the most intense fluorescence, and the water gets more and more towards indigo as the fluorescent material gets less and less.

The transition from the indigo of the deep sea far from land to blue and green may thus be due either to the increase in the amount of suspended matter present, or to the presence of some fluorescent material (most probably organic) which not only causes a return of the green and longer wave-lengths, but also produces a greater absorption of the blue and the violet. The second cause is apparently the more important.

V. Summary.

The paper contains a discussion of Raman's theory of the colour of the sea with observations on the scattering of light by different specimens of sea-water from the Bay of Bengal.

(i.) It is shown that an ocean of pure dust-free water would, owing to the effects of molecular scattering and absorption, return light of an indigo-blue colour. The presence of small quantities of suspended matter would not appreciably affect the colour. With increasing quantities of suspended matter the colour would change to bluish, green, greenish white, and white.

(ii.) The transverse scattering from different specimens of sea-water has been compared with that of dust-free vacuum-distilled water. The water from the deep blue sea scatters light of nearly the same colour as dust-free water, and. the intensity of the transversely scattered light is also of the same order.

(iii.) The intensity of scattering in different directions from different specimens of sea-water has been studied. The effect of dust is negligible so far as scattering against the direction of the incident light is concerned.

(iv). The variation of the dust-content is found to be insufficient to explain the changes of the colour of the sea from place to place. An important reason for the colour changes has been traced to the presence of varying amounts of some fluorescent material present in the sea. It is pointed out that this fluorescence also implies a greater absorption in the blue and violet involving a diminution of intensity at this end of the spectrum in the light returned from the sea.

In conclusion, I would express my most sincere thanks to Professor C. V. Raman for his stimulating interest in the work.

Rangoon,

Feb. 10, 1923.

LIX. The Motion of Electrons in Hydrogen under the action of Crossed Electric and Magnetic Fields. By R. N. CHAUDHURI, M.Sc., Ph.D. (London) *.

IN

[Published by permission of the Radio-Research Board.]

Introduction.

N some previous work (by Richardson and Chaudhuri† ; and Chaudhuri ‡) on the motion of electrons in different gases, it was found that the initial current flowing from the cathode to the anode was diminished by applying a magnetic field perpendicular to the direction of the electric intensity; and the percentage of the residual current was directly proportional to the pressure (at least well up to 10-2 mm.). We drew the conclusion that the residual current was mostly due to the collision of electrons with gas molecules, and the mean free path of the electrons in these gases was found to agree fairly well with that deduced from the kinetic theory of gases. But in the case of hydrogen, the percentage of residual current was at first found to be NOT directly proportional to the pressure, nor did the *Communicated by Prof. O. W. Richardson, F.R.S. + Phil Mag. vol. xlv. p. 337 (1923).

Suprà, p 461.

values at any particular pressure remain constant when larger initial currents were taken. In order to study this effect thoroughly in hydrogen the following investigation was undertaken.

Principle of the Method.

In the present case, the source of electrons was a hot tungsten filament, surrounded by a concentric cylindrical anode. If in a thermionic tube like this, an electric field of potential difference V is applied between the hot wire and the anode, and a magnetic field H acts parallel to the wire and so perpendicular to the direction of electric intensity, then, when H is greater than a certain critical value, the electrons will describe a path round the hot wire in such a way that the maximum distancer from the axis will be given by

H2=

8Vi

+

SV log ra
e/m, r2. log b/a e/m.7.2

(1)

where b, a are the radii of the cylinder and wire respectively, and Vi the average energy of the electrons at the point of

emission.

Taking the values we had in our case, viz, b=1·00 cm., a=005 cm., and V;= 1.5 volts, V, the accelerating potential, =4.5 volts.

Thus, if the magnetic field is gradually increased from 0, the current between the wire and the anode will remain constant up to a certain value of H, then it should fall suddenly to zero as the magnetic field is increased slightly. This will only be possible if all the electrons are coming out with zero or a constant velocity, and when there is no gas present. Thus in a perfect vacuum 141 units of magnetic field will stop all the current between cathode and anode if electrons are coming out with zero velocity, and 160 units if they are coming out with energy corresponding to 15 volts so that if we use a larger magnetic field than this the electrons will have no chance of reaching the anode.

But if, however, there is some gas present, the magnetic field will not stop the current altogether, for a number

of electrons will reach the anode by collision with gas molecules during its passage. Or this residual current might be due to heavy ions formed by electrons combining with gas molecules, and the magnetic field being insufficient, the heavy ions will reach the anode. If the residual current is due to the collisions only, we should be able to get an estimate of the mean free path of an electron in this gas. If, however, the mean free path does not agree with that found otherwise, we can conclude that a certain fraction of the residual current is carried by heavy ions.

Description and Arrangement.

The description of the apparatus used in this experiment. has been given in the previous paper (loc. cit.) page 342. The only change that has been made in this paper is that instead of the big electromagnet used there, we have a pair of Helmholtz coils which gives the desired magnetic field. The tube is placed half-way along the axis of the coils so that the magnetic field at the centre is uniform and exactly parallel to the length of the hot wire. The filament used here is of tungsten (3 cm. length and 01 cm. diameter), and the cylindrical anode of copper of 40 cm. length and 2.0 cm. diameter. They are inserted in the glass tube in such a way that the filament is axial to the anode, and both the ends of the wire are well within the cylinder. The magnetic field is measured by a fluxmeter and checked against the calculated value from the current passing through the coils. The temperature of the filament was found from the Richardson equation

i= AT1e-/T,

and also by measuring the resistance of the filament and comparing with Langmuir's data. Both these methods gave very nearly the same temperature.

Another point of difference in this paper was the use of a less sensitive galvanometer and a micro-ammeter, besides the sensitive one used previously. This was done in order to avoid any mistake that might arise in changing the resistance of the circuit by shunting the sensitive galvanometer for higher values of the initial saturation current. All these galvanometers are arranged in such a way that each could be put into the circuit as desired.

Experimental Results.

After joining the thermionic tube with the pump and PO, bulb, the coils were placed in such a way that the field was parallel to the hot wire. Care was taken to avoid

as much as possible any greased taps in the apparatus, so that any vapour arising from them might not interfere with the results. The gas inside the tube was pumped out and the filament was heated very gradually to 2 amps., and all the time the pump was kept in action. In this way all the occluded gases from the filament were taken out. When the pressure inside the apparatus kept constant for three or four hours, the electrode was bombarded by connecting it and the filament with an induction coil, and the evolved gases were pumped out. When there is no more evolution of gas, the tube is in a proper working condition. Now, hydrogen is let in by heating the palladium tube; the whole tube is filled several times with hydrogen and pumped out again before any readings are taken in this gas. All this time, of course, the liquid air is kept in the trap LL, so that mercury, or grease or any other condensable vapour may not come into the apparatus. The spectrum of the gas was also examined, and when only the hydrogen lines were observed, the readings were taken.

Effect of Condensable Vapour.

If no liquid air is kept in L, it is found that there is a residual current in the presence of the magnetic field (170), even at zero pressure, but as soon as liquid air is introduced, this current falls down, though the pressure inside the apparatus is not visibly decreased. This effect of condensable vapour has been noted in the previous paper. It is well shown in fig. 1 where the broken line is drawn through the observational points taken without the liquid-air trap and the full line through those taken with the liquid air in operation. In this case the two lines are parallel showing the effect of the presence of a constant excess pressure, not recorded by the McLeod gauge, of some vapour when the liquid-air trap is not working.

Effect due to altering the Temperature of the Filament.

It was noticed during some of these investigations that the percentage of the residual current at a given pressure under the same electric and magnetic fields did not keep