in this counter there is no change with time in the properties of the surface of one of the electrodes, which is observed in ordinary counters, since the surface of the jet is constantly being renewed. Thanks to this, the operation of the counter is considerably more stable.

The hydraulic counter described may be used for the registration of individual photoelectrons. By illuminating a water jet with a mercury lamp, one can observe individual discharges similar to those produced by elementary particles. In this way it is possible to count individual electrons torn out by ultraviolet light from the surface of water. In doing so, of course, only electrons torn from the part of the jet situated opposite the wire are counted. By inserting a quartz filter which stops the shortest ultraviolet radiation causing the photoeffect from the surface of water, it is possible to reduce the number of particles considerably. Until now it had been believed that ultraviolet rays with wavelength \(\lambda = 200\,m\mu\) do not in general produce a photoeffect in water. The new method, which for the first time made it possible directly to observe the elementary photoeffect in liquids, showed that water is rather photoactive. This observation permits the assumption that ultraviolet rays contained in the solar spectrum tear photoelectrons out of water and water vapor and, consequently, take part in the creation of electric charges in the atmosphere (Helvet. Physica Acta, 7; 360—367, 514—517, 1934, Naturwiss. 45, 761, 1934).

K. Vulfson,

EXPERIMENTS WITH SLOW NEUTRONS

In studying the induced radioactivity caused in matter by neutrons, Fermi and his collaborators established that for certain elements the intensity of this radioactivity increases by a factor of 10–100 if, between the neutron source (radium emanation + beryllium) and the element in which radioactivity is produced, a substance containing hydrogen (water, paraffin) is placed. At the same time it was established that the effect of the increase in radioactivity occurs only for those elements for which the newly produced radioactive atom is an isotope of the original one (capture of a neutron without emission of a heavy particle). As was shown in an earlier paper by the same authors,\(^{1}\) such substances are the heavy elements, for example Ag, I, Au and a number of others.

The above-described effect was explained by Fermi by the fact that, in passing through a hydrogen-containing substance, neutrons lose part of their energy and are scattered in elastic collisions with protons. The slow neutrons thus arising, unlike the fast ones, are absorbed very strongly by the nuclei of certain elements, which causes the large induced radioactivity observed in these cases.

In subsequent work by a number of different laboratories, the existence of the effect described by Fermi was confirmed (it was even used for studying neutron emission when substances were irradiated with \(\gamma\)-rays). However, all these experiments still provide insufficient material for a complete elucidation of the effect under consideration.

In the most recent work, Westcott and Bjerge\(^{2}\) analyzed the question of the slowing down of neutrons as they pass through water. Although the quantitative data of this work have the character of a rather rough approximation, its qualitative conclusions deserve attention. These authors measured the intensity of neutron-induced radioactivity after their passage through various layers of water. The induced radioactivity was observed in a silver cylinder 4 cm long with a diameter of 3.5 cm, which was cut into two parts along its axis; along one line of the cut the two halves were fastened on hinges. This made it possible to irradiate the silver cylinder in an open state (inner side), after which the cylinder was closed and placed on a thin-walled cylindrical Geiger–Müller counter. Irradiation of the cylinder with neutrons was carried out for one minute. The intensity of the induced radioactivity was measured by the number of counter deflections in 1 min. In this case the measurements began 10 sec after the end of the irradiation of the silver neu-

trons. This time was required for placing the cylinder near the counter. The source of neutrons (radium emanation + beryllium) was placed inside a cylinder of water, the radius of which was gradually increased. The silver cylinder was placed at a fixed distance outside the cylinder with water. Thus the cause of the radioactivity of the silver was fast neutrons that had passed through the water, and slow neutrons emerging from the outer layers of the water. For the dependence of the intensity of the effect on the radius of the cylinder with water, a curve was obtained with a maximum at about 9 cm. Such a course of the curve may be explained by the gradual increase in the number of slow neutrons up to their equilibrium distribution with the primary neutrons, and by the gradual decrease in the intensity of the primary neutron beam as the layer of water increases. In this case the rise of the part of the curve up to the maximum is determined by the absorption of slow neutrons, while the fall of the curve after the maximum is determined by the absorption of primary neutrons. Therefore analysis of this curve makes it possible to draw certain conclusions about the absorption of fast and slow neutrons, namely, to determine approximately the thickness of the layer of water necessary to reduce the number of neutrons in the beam by a factor of two. According to the authors’ data this quantity is about 1.5 cm for slow neutrons and about 7 cm for fast ones. The latter value, it is true, was directly confirmed under good conditions by a direct comparison of the number of neutrons after their passage through a cylindrical layer of water of radius 6 cm and in its absence. In this case measurements of the number of neutrons were made with an ionization chamber filled with helium at 15 atm and connected to a linear amplifier. Using these data, the authors determined the effective cross section for the collision of a neutron with a proton, in which a noticeable loss of neutron energy occurs. It was found to be equal to \(1.5 \cdot 10^{-24}\ \text{cm}^2\).

Placing, in the path of neutrons that had passed through 6 cm of water, various layers of paraffin, the authors were able, from the magnitude of the induced radioactivity in a silver cylinder placed immediately behind the paraffin, to determine the absorption of these slow neutrons in paraffin. From these data they calculated the effective cross section for the collision of a slow neutron with a proton. For the effective cross section they obtained the value \(4 \cdot 10^{-23}\ \text{cm}^2\), 25 times greater than the value for fast neutrons. This shows how much more often interaction with protons occurs for slow neutrons than for fast ones.

The effect of slowing down and scattering of neutrons in collisions with protons is also interesting in that here a considerable part of the slow neutrons is directed at very large angles (up to \(180^\circ\)) to the initial direction of the fast neutrons. This follows from the experiments of Grosse and Agruss\(^3\), who showed that the activity of a silver cylinder caused by neutrons from a source placed inside it increases not only when the cylinder is filled with water, but also when the space outside the cylinder is filled with water. In both cases there is approximately the same increase in the activity of the silver.

Westcott and Bjerge\(^2\) confirmed this experiment. They investigated the dependence of the induced radioactivity caused by neutrons in a silver cylinder (source: radium emanation + beryllium inside the cylinder) on the thickness of the layer of water surrounding this cylinder. It was established that the activity increases strongly with increasing thickness of the water layer, reaching an approximately constant value for a layer of 5 cm. At this, more than a 10-fold increase in activity is obtained.

From the curves of the dependence of the activity of a substance on the thickness of the layer of water placed between it and the neutron source, one can calculate, for thicknesses at which an equilibrium distribution is established between fast and slow neutrons, the relative number of the former and the latter. From this, from the increase of the effect for a given thickness in comparison with the case of absence of water, one can determine the relative probability of capture by a nucleus of fast and slow neutrons. Westcott and Bjerge made these calculations for the curves obtained on silver and found,

that for silver there is an approximately 250-times greater probability for capture by the nucleus of a slow neutron as compared with fast ones.

The fact of the strong increase of the effective nuclear cross section for collisions with slow neutrons is confirmed by more detailed measurements by Dunning, Pegram, Fink, and Mitchell.^4 These authors measured absorption in various substances for slow neutrons obtained from a paraffin sphere of radius 6 cm, at the center of which was placed a neutron source (radium emanation + beryllium). The slow neutrons were detected by those ionizing particles which they knocked out of a layer of lithium placed at the window of an ionization chamber connected to an amplifier. To separate fast neutrons from slow ones, the measurements were carried out once in the presence of a cadmium layer 1 mm thick, and another time without it. In this way the absorption was found only for those slow neutrons which cause the disintegration of lithium and are strongly absorbed by cadmium.

The table given here presents part of the authors’ results. The effective cross sections for collisions of a neutron with a nucleus are given in units of \(10^{-24}\ \mathrm{cm}^2\). The data for fast neutrons are borrowed from a previous work by one of the authors.

Although the data of this table have low accuracy (according to the authors’ estimate, \(\pm 10\%\) for small cross sections and \(\pm 28\%\) for large ones), they show, nevertheless, beyond doubt, that for certain elements, for example Li, B, Ba, Hg, U, the absorption of neutrons increases very strongly when their velocity is decreased.

Effective neutron–nucleus cross section

Data for Li and B had already been obtained earlier by Chadwick and Goldhaber^5 with the aid of an ionization chamber. They gave for the effective cross section a value of the order of \(10^{-21}\ \mathrm{cm}^2\), which at least for B agrees with the data of the table.

LITERATURE

1. Fermi, Amaldi d’Agostino, Rassetti, Segrè, Proc. Roy. Soc. 146, 483, 1934.
2. Westcott a. Bjerge. Proc. Cam. Phil. Soc. 31, 145, 1935.
3. Grosse a. Agruss, Phys. Rev. 47, 91, 1935.
4. Dunning, Pegram, Fink, Mitchell, Phys. Rev. 47, 416, 1935.
5. Chadwick a. Goldhaber, Nature 135, 65, 1935.

L. Groshev