EXPERIMENTAL STUDY OF THE NATURE OF COSMIC RAYS

L. Mysovskii, Leningrad

The question of the nature of cosmic rays, in connection with recent works in which individual impulses of penetrating radiation have been studied, has once again attracted the attention of broad scientific circles. In order to assess critically the results of the most recent works in this field, we shall briefly consider the ideas concerning the nature of cosmic rays that prevailed in physics until the very latest time, and examine the arguments that were advanced in their defense. Since, in their penetrating power, cosmic rays come closest to the gamma rays of radioactive elements, from the very moment of the discovery of cosmic rays by Hess in 1911 they were recognized as a variety of radiant energy. However, it proved impossible to set up direct experiments that could demonstrate the quantum nature of cosmic rays. Still less success, up to the most recent time, attended the attempts made by some investigators to prove the opposite hypothesis, that of the corpuscular structure of these rays. But if there was no direct evidence in favor of the hypothesis of the quantum nature of cosmic rays, the number of indirect arguments increased with time, and at one time it could have seemed that the quantum hypothesis might be counted among the reliable achievements of physical science. One of the proofs of the quantum nature of cosmic radiation could have been the absence of deflection in a strong magnetic field.

L. MYSOVSKY

There was no question of creating such a magnetic field by laboratory means. Several authors expressed the idea that, if an artificial magnetic field could not be applied to the study of cosmic rays, then, in the case of a corpuscular structure of cosmic radiation, one might expect its deflection in the earth’s magnetic field. The experiments of Mysovsky and Tuwim1 on the study of the direction of cosmic rays, carried out by them on the water tower of the Polytechnic Institute in Leningrad, showed that the intensity of these rays does not depend on azimuth. This circumstance undoubtedly spoke in favor of the quantum nature of cosmic radiation. Millikan and Cameron, comparing intensities at different latitudes, came to the same conclusion concerning the absence of an influence of the earth’s magnetic field on the direction of cosmic rays. Millikan and Cameron2 were so convinced of the quantum nature of cosmic radiation that they already spoke of separate spectral lines. Considering Aston’s curve, obtained on the basis of his work with the mass spectrograph, and using Einstein’s relation on the equivalence of mass and energy \(E = Mc^2\), they, by means of very ingenious reasoning, arrive at the conclusion that the gradual disintegration of atoms, similar to ordinary radioactive decay, cannot furnish the amount of energy that we observe in the cosmic ray. To explain the different absorption coefficients and the associated quanta \(h\nu\) of radiant energy—spectral lines—Millikan and Cameron consider it necessary to suppose the existence in nature not only of the disintegration, but also of the formation of atoms. By the sudden formation from protons of helium, oxygen, silicon, and iron, and by the accompanying loss of part of the mass, transformed into radiant energy (mass defect), the components of cosmic rays observed by them are explained. The views of Millikan and Cameron at the present time already

STUDY OF THE NATURE OF COSMIC RAYS

cannot be regarded as exhaustive, since soon after the publication of the final article by Millikan and Cameron there appeared a very carefully executed work by Regener,^1 which considerably broadened our ideas about the limits of hardness of cosmic rays. This work is so interesting that we consider it necessary to dwell in greater detail both on its experimental part and on the conclusions drawn by Regener. The apparatus used by Regener in his experiments on Lake Constance is shown in Fig. 1.

As the ionization chamber of the recording apparatus, a steel bomb was taken, with walls 1 cm thick and a volume of 39 l. To increase the ionization current, the bomb was filled with carbon dioxide at a pressure of 30 atmospheres. The central electrode of the chamber was connected to a single-fiber electrometer of special design. Before the entire apparatus was immersed in water, the electrometer was charged to 600 V. By switching on the illumination for several seconds every hour, it was possible to obtain and record the image of the electrometer fiber on a photographic plate. The ionization current could be determined from the photograph obtained with an accuracy of up to 0.01 V. The residual current (the chamber’s own radiation), owing to the tinned surface and the careful cleaning and drying of the carbon dioxide gas filling the chamber, was reduced to a mini-

Fig. 1

^1 E. Regener, Naturwiss. 17, 183, 1929.

to the instrument. In Fig. 2 are shown the records of the registering instrument at various depths.

This figure so clearly shows the decrease of the ionization current with depth that further explanations are already superfluous. It is interesting to note what precautions Regener took in order to ensure the correct operation of the apparatus at great depth. From Fig. 1 it is evident that the ionization chamber, with the electrometer attached to it in its upper part, was kept in a vertical position by means of a float. The motion on the surface of the lake did not affect the operation of the electrometer, since the whole apparatus was attached not to the cable lowered from the surface, but to an anchor on the bottom of the lake. The results of the measurements obtained by Regener are already visible from Fig. 2, but in addition these same results are brought together in the following table:

Fig. 2

The most interesting point in Regener’s work is the circumstance that he succeeded in noting a gradual decrease in the intensity of cosmic rays in water down to a depth of 230 m, whereas Millikan and Cameron did not go beyond 70 m. With such penetrating power, even the energy of formation of atoms from protons will be insufficient to explain the hardest part of cosmic radiation. In the formation of atoms of iron, silicon, oxygen, and helium from protons, only part of the mass disappears (mass defect), being converted into radiant energy. To explain the extreme limit of hardness, Regener had to assume that somewhere in nature processes occur in which the entire mass of the proton is wholly converted into radiant energy. Thus the hypothesis of Millikan and Cameron was no longer able to explain the whole complex of cosmic rays. This hypothesis had to be broadened by the introduction of a new component—a new line, dependent on the conversion of protons into radiant energy. Such a broadening, of course, did not in the least shake confidence in the optical nature of cosmic rays. Quite the contrary. The possibility of explaining the extreme limit of hardness of cosmic rays from the point of view of the optical hypothesis could only further increase the trust placed in it. We shall conclude the list of evidence relating to the optical hypothesis of cosmic radiation by dwelling on one more peculiar phenomenon, which until recently was a strong argument in favor of this hypothesis. It is known that

For hard gamma rays the absorption coefficient in different elements is expressed by the formula

\[ \mu=\rho\frac{N}{A}Z\mu_e, \]

where \(\rho\) is the density of the substance, \(\frac{N}{A}\) is the number of atoms in a gram of the substance (\(N\) is the number of atoms in a gram-atom, \(A\) is the atomic weight), \(Z\) is the number of electrons in the atom, and \(\mu_e\) is the absorption coefficient calculated per electron. Thus, on the basis of the above formula, one may say that the absorption of gamma rays by any layer of any substance is proportional to the number of electrons in that layer. The same law of absorption proves to be valid also for cosmic rays. This circumstance alone may serve as an argument in favor of the optical nature of cosmic radiation. An even more convincing argument in the same direction was provided by the study of the absorption of cosmic rays at the boundary between two media, for example air and lead. It turns out that, in passing from air into lead, the absorption coefficient of cosmic rays does not immediately acquire the value that it should have according to the formula given above. For thin layers of lead, not more than \(5–6\) cm, the coefficient assumes a considerably larger value than it should according to the formula, and only after \(7\) cm does absorption proceed correctly, proportionally to the number of electrons per unit volume. The anomalous absorption of cosmic rays in the transition layer was discovered by Topham and confirmed by his pupil Steinke. Myssowsky and Tuwim1 carried out experiments on the absorption of cosmic rays by thin layers of lead in an ice monolith and showed that the anomalous absorption cannot be explained by an admixture, in the cosmic rays, of softer gamma rays from the surrounding space. It remained to suppose that in each beam of cosmic rays there is a certain quantity of softer secondary rays formed during the absorption of the primary

the beam. In passing from air or from ice into lead, the character of the absorption of the soft secondary rays changes sharply. Whereas in air and in water the absorption even of soft secondary rays occurs chiefly by scattering (the Compton effect), in lead, for these same rays, “true absorption” (the photo-effect) predominates. As long as our knowledge of the nature of cosmic rays was limited to the facts cited here, there existed almost complete certainty that cosmic rays are quanta of radiant energy. The quantum hypothesis not only explained quite satisfactorily all the known properties of cosmic rays, but also made it possible, as we have already seen, by using Einstein’s relation on the equivalence of mass and energy, to construct hypotheses about their origin as well. The question was already one of individual spectral lines and of the maximum energy of the quantum. On the basis of Regener’s data, the limiting energy of a cosmic-ray quantum had to be greater than 500 million volts. We shall see, however, that the new hypotheses on the nature of cosmic rays, based on new data, although they give for the maximum energy of the cosmic ray almost the same value, nevertheless each individual ray, insofar as it can be observed, turns out to be not a quantum but a corpuscle.

The Composition of Cosmic Radiation According to the Views of D. V. Skobeltsyn

D. V. Skobeltsyn’s views on the nature of cosmic rays still do not differ too radically from the earlier ones. He still ascribes an optical nature to the primary rays and believes that the “ultra-beta rays” which he has had occasion to observe are secondary electrons liberated by cosmic-ray quanta. At present it is possible to distinguish individual pulses of cosmic rays in two ways: 1) by means of a Wilson chamber placed in a magnetic field, and 2) by means of two Geiger and Mül-

L. MYSOVSKY

Skobeltsyn worked according to the first of the indicated methods.1 In an article on the improvement of methods for observing alpha- and beta-particles2 I have already said that Skobeltsyn, studying electrons produced by gamma rays in a Wilson chamber and applying a magnetic field when photographing these tracks, noticed that a rather small percentage of the tracks retain their straightness even in the presence of a field of strength 1500 gauss. Skobeltsyn regards these electrons as Compton electrons produced by quanta of cosmic rays. In the above-mentioned article there is also reproduced one of Skobeltsyn’s stereoscopic photographs, on which the straight-line path of an electron is clearly visible among the curvilinear paths of Compton electrons from gamma rays. Here we shall draw attention to one feature noted by Skobeltsyn in the photographs at his disposal. Of the 613 stereoscopic photographs which he obtained in a magnetic field, electrons with “very great velocity” occur in 27. Among these latter, in three photographs the straight-line tracks proved to be double, and in one there was even observed a group consisting of three straight-line tracks. Such a large percentage of almost parallel tracks cannot be explained by chance. It must be supposed that tracks belonging to one and the same group emerge from one common center. This supposition was tested by Skobeltsyn with the aid of Pulfrich’s stereo-comparator and was fully confirmed. In Fig. 3 one of the photographs is reproduced, with two straight-line tracks which even to the eye appear to issue from a common center.

The data obtained with the aid of the stereo-comparator show that these two tracks can indeed be regarded as issuing from one common center, if it is assumed that they underwent a slight deflection of \(2^\circ\)–\(3^\circ\) while passing through the glass cover of the chamber.

How, then, can the occurrence of such groups of fast electrons be explained? Skobeltsyn, in his article, suggests that here we are dealing with the splitting of an atom by a quantum of cosmic rays. In that case, instead of the path of an electron, one may expect the path of a fast H-particle. This assumption seems to be confirmed upon careful examination of Fig. 3. The lower straight-line path has a denser distribution of ions than the upper one, and denser than that of the neighboring Compton electrons. Unfortunately, it is not possible to reproduce another picture, found in Skobeltsyn’s article and likewise containing two straight-line paths. The point is that one of these paths is so weakly expressed in the print that

Fig. 3

a secondary reproduction would probably have ended in failure and would have raised only doubts as to the presence of such a path. It must be said, however, that the last two paths just mentioned do not meet at a single point. Skobeltsyn explains this by a deflection that occurred in the glass cover of the chamber or in the copper cylinder on which the wire of the coil, serving to create the magnetic field, was wound.

Although Skobeltsyn still remains on the ground of the quantum hypothesis concerning the nature of cosmic rays, nevertheless his views represent a considerable shift toward the corpuscular theory. According to Skobeltsyn’s views, almost all the ionization caused by cosmic rays is due to electrons of limiting velocity—ultra-beta rays. He supports his considerations by calculating the number of

ions formed by fast electrons in the Wilson chamber. As the basis for these calculations Skobeltsyn takes the total number of his photographs—613, the total number of straight-line tracks—32, the number of ions formed along 1 cm of path at the limiting velocity of beta particles—40, the duration of the phenomenon in the Wilson chamber—0.02 sec, and, finally, the volume of the illuminated part of the chamber. As a result of the calculation he obtains that in 1 cm³ about one ion is formed in 1 sec. (According to Millikan and Cameron 1, 4 J.) If, however, one takes into account that the observations were made by him on the first floor of the laboratory of the Leningrad Polytechnic Institute and, consequently, part of the radiation was absorbed by the walls, it follows that all ionization from cosmic radiation is caused exclusively by “ultra-beta rays.” To explain the anomalous absorption in the transition layer between two substances Skobeltsyn assumes that the absorption of the secondary electrons of ultra-beta rays depends on the atomic number of the element. Greater absorption in a medium with a high atomic number results in a lower ionization intensity inside this medium. Fig. 4, taken from Skobeltsyn’s article, illustrates this circumstance.

Fig. 4

If the upper curve corresponds to absorption in air, and the lower one in lead, then the transition at some depth from air to lead will be represented by a dotted line, showing that for thin layers of lead the absorption coefficient must have a somewhat greater value than in lead itself. It should be noted that this entire explanation is of a purely formal character and is in no way connected with the physical essence of the phenomenon.

APPLICATION OF THE NEW COUNTER OF H. GEIGER AND W. MÜLLER TO THE OBSERVATION OF INDIVIDUAL IMPULSES OF COSMIC RAYS

In the article already mentioned by us on the improvement of methods for observing alpha and beta particles, the structural features and properties of the new Geiger and Müller counter were briefly described. It consists of a small cylindrical ionization chamber, along the axis of which there is stretched, as the electrode, an oxide-coated iron wire. Here we shall dwell in more detail on the counter that Geiger and Müller1 used for observing individual impulses of cosmic rays. The length of the tube was 17 cm, the internal diameter 3 cm, the thickness of the tube walls 1 mm. The sensitivity of the counter, which was determined by the number of discharges per 1 minute from 1 mg of radium placed at a distance of 1 m, was 5500. The residual current in the laboratory consisted of 150 separate discharges. When the counter was surrounded by an iron armor 20 cm thick, the number of these discharges decreased to 40 per minute. The iron armor was at first open in its upper part to allow free access of cosmic rays. Gradually the upper part was covered with iron sheets of area \(60 \times 90\ \mathrm{cm}^2\). The inner electrode was connected in the usual manner to a low-sensitivity electro-

Fig. 5

meter. Figure 5 shows the course of the absorption curve of cosmic rays in iron, obtained by Geiger and Müller with the aid of this simple apparatus.

The sharp decline of the absorption curve depends on radioactive impurities present in the laboratory walls and emitting gamma rays. The accuracy of measurements of this kind is determined by the number of discharges falling on a single point. On the curve presented, each point corresponds to about 2000 discharges. Obtaining the data for the whole curve required from 6 to 7 hours. This time can be reduced severalfold by taking a still larger counter and using a photographic method for recording the deflections. In order to check the constancy of the counter readings, Geiger and Müller observed it for 14 days, and it turned out that, within the limits of unavoidable statistical fluctuations, the counter worked in exactly the same way throughout this entire time. We shall not dwell on the further work of Geiger and Müller, since it concerns chiefly the study of various properties of the counter as a function of various changes in its construction and in the dimensions of its individual parts. Let us note here only one further advantage of the cylindrical counter: small fluctuations in the pressure of the air inside the counter are not reflected in the quality or constancy of its operation. Since the pressure inside the counter is closely connected with the potential applied to its casing, the same may be said of this potential. The Geiger–Müller counter just described was filled with dry air at 50 mm of mercury, and a potential of 1200 V was applied to it; it gave the same readings when the potential fluctuated within 50 V. The only thing that hinders the direct application of the new counter to the study of cosmic rays is the impossibility of separating the pulses due in their origin to cosmic radiation from all the others, which are inevitably present in it and together constitute a considerable part of all observed discharges. In Skobeltsyn’s apparatus, as we have seen, the identification of the tracks of fast electro-

...took place thanks to the presence of a magnetic field. In the case of Geiger and Müller counters, the problem of separating cosmic-ray pulses was an extremely acute one, although it was solved, albeit with some difficulty, in the work of W. Bothe and W. Kolhörster. We shall now turn to a detailed description of this work, which is in every respect remarkable and interesting.

The Work of W. Bothe and W. Kolhörster

The basic idea adopted by Bothe and Kolhörster1 in their work consisted in the use of two cylindrical counters to select individual rays passing through both counters simultaneously. The principal part of their apparatus is shown schematically in Fig. 6.

Fig. 6

cylindrical counters, for selecting individual rays passing through both counters simultaneously. The principal part of their apparatus is shown schematically in Fig. 6.

The internal diameter of the counters \(Z_1\) and \(Z_2\) was 5 cm, their length 10 cm, and the wall thickness (zinc) 1 mm. Along the axis of the tubes, between ebonite stoppers, there was stretched an oxidized wire—

block. The gas pressure inside the counters was from 4 to 6 cm of mercury. Both counters were placed in a stand \(M\), which made it possible to insert between them absorbing layers up to 45 mm thick. On the sides of the counters there were two lead screens \(BB\). These screens were designed in such a way that any particle which, owing to scattering (and not rectilinearly), might pass from one counter into the other would traverse a greater mass of material than along the rectilinear path through the absorbing layer \(A\). This entire apparatus was enclosed in armor of 5 cm of iron plus 6 cm of lead. It is interesting to note that the lead used was 150 years old, and special experiments showed that it was free of gamma rays. A potential of 1300 V was applied to the lining of the cylindrical capacitors, and this remained unchanged throughout the entire work (about 3 months).

For recording coincidences (simultaneous passages of a ray through both counters), the deflections of the electrometers connected with the counters were photographed on a film moving at a speed of 1 cm per second. Two opposite deflections were counted as coincident if they were separated by no more than 0.01 cm, which corresponded in time to \(1/100\) sec. From the total number of coincidences obtained in this way it was necessary to subtract the number of accidental coincidences. If, over a segment of film of length \(l\), the first counter gave \(N_1\) deflections and the second \(N_2\), then the probability of accidental coincidences will be proportional to the product of \(N_1\) and \(N_2\) and to the interval of film over which we count deflections as coincident, that is \(2 \cdot 0.01\) cm (on both sides, which gives the factor 2). Moreover, for given \(N_1\) and \(N_2\), the number of accidental coincidences will be inversely proportional to the entire length \(l\) under consideration.

Hence we have the formula for calculating \(N = \frac{2 \cdot 0.01}{l} N_1 N_2\). As an example, Bothe and Kolhörster give a table (see Table I on p. 15).

To explain it, let us consider the case given in the first line. Here \(\frac{2 \cdot 0.01}{l} N_1 N_2 = \frac{2 \cdot 0.01 \cdot 310 \cdot 176}{391} = 2.8\). Subtracting 2.8 from the total number of coincidences, equal to 10, we obtain

STUDY OF THE NATURE OF COSMIC RAYS

the number 7.2, given in the table as the number of systematic coincidences caused by cosmic rays. Even on examining this table it is clear that the number of coincidences is so large that it involuntarily suggests the presence of corpuscular rays. Experiments with an absorber apparently support this proposition still more.

Table I

First floor. Film 14

The initial experiments were carried out by Bothe and Kolhörster on the first floor of the main building of the Physikalisch-Technische Reichsanstalt in Berlin. The use even of 4 cm of lead as an absorber scarcely decreased the number of coincidences, while a thicker absorber would not fit into the apparatus. Since the entire apparatus had already been constructed, they had to limit themselves to a thickness of 4 cm, only gold was used instead of lead. The results of these measurements are brought together in the second table. In view of the peculiar character of the observations, we also give this table here (see Table II on the following page).

Calculating from the data of this table, Bothe and Kolhörster obtain:

\[ \frac{\mu}{\rho}=(3.2\pm 0.9)\cdot 10^{-3}\ \mathrm{cm}^2/\mathrm{g}. \]

L. MYSOVSKY

TABLE II

First floor

Decrease in the number of coincidences in 216 min. ($\pm$ mean error): $5.4 \pm 26.8 = (1.5 \pm 7.4)\%$.

Such a value of the absorption coefficient still further confirmed Bothe and Kolhörster in their corpuscular nature of cosmic rays. Indeed, this coefficient comes so close to the coefficient for primary rays that the idea of the participation of secondary rays in the coincidences falls away of itself. As we see, however, from the second table the calculated mean error $\pm 26.8$ exceeds in absolute value the measured difference 5.4. In order to strengthen the effect of the absorber Bothe and Kolhörster made observations in the attic of the same building, both shields having been removed from the counters. Since the absorption in the building and in the shields was rather large (it corresponded to 3 m of water), the number of discharges, owing to the increase in intensity, rose in both counters. In addition, in the composition of cosmic rays under the new conditions of the experiment the soft component acquired great importance.

cosmic radiation. All these circumstances made it possible better to observe the influence of an absorbing screen placed between the counters. We shall not reproduce in full the third table of Bothe and Kolhörster, since it is entirely analogous to Table II, but shall include from it only the last line.

Table III

Cherdak

Decrease in the number of coincidences in 360 min: \(242.8 \pm 41.6 = (24.6 \pm 4.2)\%\)

From the last table it is clear that under the new conditions the observational errors \(\pm 41.6\) are already smaller than the observed difference 242.8. The calculation of the absorption coefficient in this case gave the value

\[ \frac{\mu}{\rho} = (3.5 \pm 0.5)\cdot 10^{-3}\ \text{cm}^2/\text{g} \]

which is still closer and more accurately corresponds to the known absorption coefficient of cosmic rays. Such agreement, of course, strengthened still further the confidence of Bothe and Kolhörster in the corpuscular nature of cosmic radiation.

From the work described it is clear what experimental difficulties Bothe and Kolhörster had to overcome in order to make it possible to count the number of coincident indications of two cylindrical counters. These difficulties are also taken into account by the authors themselves. Here

why they do not confine themselves to the results obtained, but set up a further series of control experiments that make it possible to regard the results obtained with greater confidence. We shall dwell on only some of these experiments.

In a mine at a depth of 406 m, two cylindrical counters were placed directly one above the other. It turned out that over the course of an entire hour there was not a single coincidence in the readings of these two counters, since cosmic rays could no longer penetrate through a thickness of 400 m, while the residual discharges were so rare that they did not even give accidental coincidences.

Table IV

Attic. Counters one above the other, without absorber

It was shown that, when the distance between the two counters was increased, the number of coincidences decreased in accordance with the decrease of the solid angle. The numerical data of this experiment are given in Table IV.

If both counters are placed next to each other, the number of coincidences proves to be half that of the case when they are directly one above the other. All this shows that the intensity of the radiation causing coincident discharges in both counters is distributed in space in the same way as in cosmic rays.

Also of interest is the experiment carried out by Bothe and Kolhörster on the study of coincidences caused by the gamma rays of radium C. The general arrangement of the counters is shown in Fig. 7.

Both counters were taken of considerably smaller dimensions (diameter 2 cm, length 4 cm) in order, as far as possible,

to reduce the influence of cosmic rays. In addition, they were placed side by side in a horizontal direction, and not one above the other, which, as we have seen, also reduces the number of coincidences caused by cosmic rays. To facilitate the passage of the comparatively slow electrons of one counter into the other, the halves facing each other were made of aluminum

Fig. 7

foil, only \(0.03\) mm thick. Both counters \(Z_1\) and \(Z_2\) were placed in a glass tube, which was evacuated to a pressure of \(4\) cm of mercury. In addition, in the same glass tube there were placed two rails along which thin screens could slide. It was enough to tilt the glass

Fig. 8

tube for a screen to move and position itself between the counters. Tilting the glass tube in the opposite direction moved this screen into the part of the tube not occupied by the counters. This entire apparatus was placed in the already described double armor, in which an opening had been made for the beam of gamma rays falling on the counters horizontally and per-

perpendicularly to their axes. The results of the experiment are shown as a curve in Fig. 8.

The absorption coefficient calculated from this curve agrees with the mean absorption coefficient obtained by other methods for secondary electrons produced by the gamma rays of radium C.

This experiment is interesting also from another point of view: it shows that gamma rays cannot cause coincidences of discharges in counters. Only secondary electrons produce such discharges. Consequently, in counters with thick walls, coincidences of discharges cannot be affected either by the radioactivity of surrounding objects or even by the radioactivity of the walls themselves. Both these factors can increase only the number of accidental coincidences, not the systematic ones.

Fig. 9

In addition to special experiments, Bothe and Kolhörster support their conclusions with a whole series of arguments and calculations. We shall briefly present only the most interesting ones. First of all, these investigators examine the question of the possibility of the appearance of simultaneous discharges in both counters from an ultra-gamma ray. Fig. 9 illustrates this assumption.

A gamma ray, passing through the shield, liberates a Compton electron in it. Since in the case of an ultra-gamma ray the path of the electron almost coincides with the direction of the ray, the electron enters the first counter but does not reach the second. The same ray liberates in the absorber a second electron, which then enters the second counter. It is obvious that in such a process the discharges in the counters will coincide. However, if one assumes that the mechanism of ionization by ultra-gamma rays is similar to the mechanism of ionization by ordinary gamma rays, then such an occurrence of two electrons from one and the same ultra-gamma ray is improbable. Bothe and Kolhörster

they mention yet another possible explanation of the coincidences they observed. One may suppose that the mechanism of ionization by ultra-gamma rays, owing to the great energy of their quantum, becomes similar to the mechanism of ionization by beta rays. But they reject such an explanation as one taken ad hoc.

Only after control experiments and discussion of various possibilities do Bothe and Kolhörster decide definitively to assert that the coincidences they observed are due to corpuscular rays, and they proceed to a description of the properties of these rays. Plotting absorption points referred to the number of electrons per unit volume of lead, they show that the course of these points agrees with the course of the curve obtained for lead by Steinke. The circles in the figure correspond to the coincident readings of both counters; the crosses—to the readings of the lower counter after correction for its own radiation and the radiation of the laboratory. In the transition layer two curves are given. The dotted line corresponds to absorption in the transition layer from air into lead; the heavy line—to absorption in lead alone. As is seen from the figure, the points of Bothe and Kolhörster lie very well on the curve in its lower part. Matters are somewhat worse at the beginning of the curve, but one can nevertheless note that the absorber in the attic causes a sharp anomalous drop. In the experiments on the first floor this phenomenon is not observed, since the further course of the curve refers to rays that have already passed through the walls of the laboratory and the pan. This result once again emphasizes the identity of the rays observed by Bothe and Kolhörster with cosmic rays.

Fig. 10

After drawing the conclusion about the corpuscular nature

primary cosmic rays, it was natural to pass to definite notions about the nature of these rays. As we have already seen earlier, many investigators proposed, in order to explain cosmic radiation, the hypothesis of cosmic electrons of high velocity. This hypothesis agrees sufficiently well with the data of the experiments of Bothe and Kolhörster. Indeed, these data are not contradicted by the negative results obtained by investigators who attempted to determine the concentration of negative charge from cosmic electrons on an insulated body. According to the calculation of Bothe and Kolhörster, through \(1\ \mathrm{cm}^2\) in \(1\tfrac{1}{2}\) min. there passes no more than one cosmic electron. If one also takes into account the negligible absorption of cosmic rays, then it is quite understandable that there is no point in hoping to detect cosmic electrons by such a method as the simple accumulation of charge on an insulator. Cosmic electrons should have no influence either on the preservation of the negative charge of the earth, since the positive vertical current in the atmosphere exceeds the current from cosmic electrons by \(10^5\) times.

Knowing the number of electrons passing through the counters, the volume of the counters, and the mean path length in the counter, it is not difficult to obtain the approximate magnitude of the ionization falling on \(1\ \mathrm{cm}\) of the path of a cosmic electron. This magnitude proves to be equal to 90 ions. This number agrees rather well with the value of 40 ions obtained by extrapolation for fast beta particles. We shall not dwell in more detail on the calculation of all these quantities, since they have all been obtained by extrapolations and cannot serve for a final judgment about the nature of corpuscular rays.

The authors themselves also arrive at the same conclusion. For their part, they consider it necessary to examine one more possible hypothesis—the hypothesis of rapidly moving protons. At such high velocities and energies, the difference in the external manifestations decreases not only between electrons and quanta, but also between protons and electrons. Bothe and Kolhörster point out that for energies

at \(10^9\) V for a proton, the ratio of its velocity to the velocity of light is \(\beta = 0.875\); for an electron, \(\beta = 1 - 1.3 \cdot 10^{-7}\). The mass of the proton doubles, while the mass of the electron reaches the magnitude of the rest mass of the proton. \(H\rho\)—the product of the magnetic-field strength and the radius of curvature—is \(5.6 \cdot 10^6\) for the proton, and \(3.3 \cdot 10^6\) for the electron—values almost identical. The ionizing power of a proton with a high velocity also does not differ greatly from the ionizing power of an electron of the corresponding velocity. Thus, with equal justification, one could imagine cosmic rays as a stream of cosmic protons. This latter hypothesis is still more tempting than the hypothesis of cosmic electrons, but it is less familiar, and probably for this reason Bothe and Kolhörster are rather inclined toward the former. A final solution of the question can be provided only by further experiments on the dependence of the intensity of cosmic rays on azimuth. Such experiments, as we have seen, have already been carried out and have given no indication of the existence of deflection in the Earth’s magnetic field. It is possible, however, that in places of the greatest intensity of this field different results may be obtained. Bothe and Kolhörster write that observations in this direction are already being planned.

Comparison of D. Skobeltsyn’s Work with the Work of W. Bothe and W. Kolhörster

It remains for us now to compare the results of the works described here and to try to draw general conclusions from them. We have seen that both Skobeltsyn’s work and the work of Bothe and Kolhörster reach the limits of accuracy of contemporary experimental technique. However, the experimental difficulties of each of these works are quite distinctive. Let us first dwell on the shortcomings connected with the observation of individual particles in Wilson’s chamber. The most significant of these shortcomings must be considered the smallness of the interval of time during which one can observe the passage of cosmic

rays through the Wilson chamber. According to approximate, but, as Skobeltsyn himself indicates, not very convincing calculations, the interval of time during which the paths of the particles can become visible ranges from 0.02 to 0.03 sec. Frequent expansion of the chamber is hindered by the heating of the solenoid that creates the magnetic field. As a result, the number of individual rays that it proved possible to observe in the Wilson chamber is only 32 in all. The apparatus itself is so complex that there can be no question at all of moving it or of using any absorber. Moreover, the author of the present article has many times had occasion to observe that, in expansions following rapidly one after another, the track of one and the same beta particle appeared and disappeared up to four times. Taking into account that the magnetic field is switched on only for a short interval of time and immediately before the descent of the piston in the chamber, one must reckon with the possibility that the bases of the path may be formed even before the magnetic field is switched on. After the magnetic field is switched on and the piston is lowered, the droplets on the ions increase, the path becomes visible, but, of course, already remains rectilinear. This latter circumstance, like the others, must be taken into account, and the results of the calculations must be approached with caution. But if the Wilson chamber, as a measuring instrument, has many shortcomings, it also has an enormous advantage—namely, the possibility of directly observing the character of the ionization. In many cases, as is known, it is even possible to count the number of ions per 1 cm of path.

The fact that the results obtained with the Wilson chamber concerning the nature of cosmic rays cannot yet by any means be regarded as finally established is indicated, if only, by the appearance of a note by Auger and Skobeltsyn in Comptes rendus.1 In this note the authors give a different interpretation of the parallel paths observed in the chamber. The first interpretation is the splitting of an atom

STUDY OF THE NATURE OF COSMIC RAYS

the quantum of a cosmic ray has already been abandoned, since such a splitting, if it occurs at all, must be encountered much more rarely. Instead, another hypothesis is advanced: that the quantum of a primary cosmic ray releases successively two Compton electrons. However, this second hypothesis, as we have already seen, was analyzed by Bothe and Kolhörster and appears to be perhaps even less probable than the first. Meanwhile Auger and Skobeltsyn see in it the main proof of the quantum nature of the primary rays. From this polemic it is not difficult to conclude that we are still indeed far from a final solution of the question of the nature of primary cosmic rays.

In the apparatus of Bothe and Kolhörster there are almost none of those shortcomings which we pointed out in Skobeltsyn’s apparatus. The counters can operate for hours and even days without interruption. The number of counts is extremely large. In the work by Bothe and Kolhörster described by us, there were about 90 thousand of them. It is comparatively easy to use various absorbing screens. The operating mode of the counters—the deflections of the electrometer threads—changes almost not at all with time. But, on the other hand, any ray, almost independently of its origin and intensity, corresponds to one and the same deflection of the electrometer. The author of the present article now has at his disposal several different Geiger and Müller counters, and it must be said that at times it is even vexing to see how similarly such counters respond both to cosmic rays, and to the gamma rays of radium C, and to the radioactivity of walls, and finally to the radiation of their own inner surface. If in Skobeltsyn’s apparatus, when fast electrons are identified by a short-term switching-on of the magnetic field, doubt is raised by the possibility that the track was formed before the magnetic field was switched on, then in the apparatus of Bothe and Kolhörster, when cosmic radiation is identified, the presence of accidental coincidences serves as the same sort of disturbing circumstance. In Table I we have already seen that with any appreciable absorption the error begins to exceed the quantity being measured.

All the difficulties we have described here lead to the conclusion that, for the time being, it is still impossible to give an exact answer to the question of the nature of cosmic radiation. Articles appear almost continuously in which it is proved with ever greater persuasiveness that, if quanta, electrons, and protons possess the energy of a cosmic ray, then their properties in passing through matter become almost identical. It is by no means excluded, for example, that a quantum of a cosmic ray, possessing a large mass, ionizes uniformly along its path, just as we formerly observed only for beta particles. It is also possible that cosmic radiation consists partly of quanta, partly of electrons and protons. The differences among these kinds of cosmic rays must be so subtle that they cannot yet be established by means of the methods proposed. But one thing can be noted with complete certainty: we have gained the ability to distinguish and observe individual impulses of cosmic rays. In comparison with the method of observing and measuring the total ionization, this is undoubtedly an enormous step forward in the study of the nature of cosmic rays. And we shall hardly be mistaken if we express confidence that, by the simultaneous application of the two different methods we have described, by the mutual verification of the results obtained by one method and by the other, it will soon be possible to resolve the question of the nature and composition of cosmic radiation. Before experimental physics there seems once again to open a broad and interesting field of research, which may indeed lead to an understanding of the process of formation of the atoms of the chemical elements.

1. C. R. 189, p. 55, 1929. ↩↩↩↩↩↩

2. L. Mysovsky. Uspekhi fizich. nauk 9, 574, 1929. ↩↩