DISINTEGRATION OF ATOMS BY COSMIC RAYS AND THE POSITIVE ELECTRON*

Blackett and Occhialini, Cambridge

1. Experimental Method

Recently we developed a method by means of which it is possible to make the very fast particles accompanying cosmic radiation photograph themselves or, more precisely, the traces of their own paths in a Wilson chamber^1. In this way photographs of tracks can be obtained much more easily than by the usual method, when the adiabatic expansion is made at random. In fact, only a very small fraction of the photographs obtained by the previous method makes it possible to detect even a single track. The average number of photographs required to obtain one track depends here on the dimensions and orientation of the chamber and on the effective time of expansion (which must not exceed \(1/20\) sec.). From measurements with counters it is known that, in all directions, about 1.5 fast particles pass through \(1\ \text{cm}^2\) in 1 min. This figure is, on the whole, in agreement with the results found with the Wilson chamber. Skobeltsyn^2 succeeded in obtaining a successful photograph with one or several tracks after approximately every ten expansions. In Anderson’s^3 work, 50 exposures yielded only a single track with a range length corresponding to the order of the measured energy. The use of our method gives tracks in 80% of all photographs taken. We intend to give a more detailed account of the technique of this photographic method in a separate article, limiting ourselves here only to the main features.**

* Published in Proc. Roy. Soc. 130, No. A839, 639 (March 1933); translated by V. V. Bobin. The title of the article in the English original is: Some Photographs of the Tracks of Penetrating Radiation. This is excessive modesty on the part of the authors. The true significance of the authors’ work lies in the remarkable discovery of the phenomenon of “showers,” consisting of fragments of atoms. These “showers” constitute a picture of the continuous and ubiquitous disintegration of atoms by cosmic rays.

** Mott, Smith, and Locher^4 found a relation between the number of tracks appearing in the chamber and the number of discharges in the counter, while Johnson, Fleischer, and Street made use of coincidences to produce a flash during photography at the time when the chamber was continuously in operation. On the limits achieved in recent attempts to combine the advantages of the Wilson chamber and Geiger counters, see also the review articles by L. V. Mysovskii in UFN. Translator’s note.

The Wilson chamber is cylindrical (diameter—13 cm, depth—3 cm); it is arranged so that its flat walls (the cover and bottom) are vertical (Fig. 2). Two Geiger–Müller counters (each 10 cm long, 2 cm in diameter) are placed one above the chamber, the other below it. Thus any ray passing in a straight line through both counters will necessarily also pass through the illuminated part of the chamber. Another possible arrangement of the counters is described at the end of the article (see the explanations to the photographs). The counters are connected into a grid circuit of a circuit with cathode tubes, registering exclusively only simultaneous discharges—at once in both counters1.

The piston of the chamber is a light aluminium disk, connected with the base of the chamber by means of a rubber diaphragm. It is allowed exactly such freedom of motion as is needed to produce the required expansion. Before each expansion the piston is in equilibrium: above it is the pressure of the gas in the chamber, close to 1.7 atm, and below it is the same pressure of air. The expansion takes place as the result of an impulse in the input circuit of the tube circuit, causing air to be discharged from under the piston to the outside; the pressure falls to atmospheric. The sequence of events during the whole operation is as follows. In the counters a coincidence of discharges has occurred. Instantly the grid of the tube connected in the input circuit of the amplifier becomes positive. This impulse at the output of the amplifier is sufficient to short-circuit a small electromagnet which until then had held a light armature from “jumping” under the action of a spring. The armature springs back and moves a latch, releasing the outer valve under the piston; expansion occurs. By very careful attention to all the details in designing the various parts it proved possible to reduce to 1/100 sec. the total interval of time elapsing after the coincidence of discharges in the counters and before the end of the expansion. In such a short time the ions, formed by some ionizing particle along its path, diffuse only a very small distance from their place of origin. Thus, in oxygen at an initial pressure of 1.7 atm, tracks with a maximum width of 0.8 mm are obtained in this time. This width, of course, is much greater than the width of those tracks which were formed along the path of particles that flew through the gas later, toward the end of the expansion,—and nevertheless it is small enough to permit very accurate measurements. The observed width is in good agreement with the value computed on the basis of the known diffusion velocity of gases2.

Approximately 1/100 sec. after the expansion, the illumination is switched on for an instant (about 1/30 sec.). This flash is produced by passing a current impulse from a transformer at 4000 V through a capillary mercury lamp.

The entire chamber is placed inside a water-cooled solenoid capable of maintaining in the chamber a uniform magnetic field of 3 thousand gauss. The photography is carried out

simultaneously with two apparatuses. The optical axis of one of them coincides with the axis of the Wilson chamber and at the same time is parallel to the magnetic field; the axis of the other is directed at an angle of \(20^\circ\) to the axis of the first. Pairs of photographs were needed not at all for viewing them in a stereoscope: an angle of \(20^\circ\) would be too large for this purpose. Instead, of the stereoscopic pair of positives, each was inserted into the photographic apparatus by which it had at one time been taken; then the plates were illuminated from behind. Then, in place of the object, two images appeared, together reconstructing a sufficient number of its points in space. It remained to construct wire models of the tracks at natural size. This method—the stereoscopic reprojection and reconstruction of an object in space—is essentially the same as the methods used by Cortisson2, Williamson and Gerrou[^2]. Such use of two cameras, giving at the same time a pair of photographs taken in two different directions, is extremely important, because a single photograph by itself gives very little for the study of tracks.

The mean waiting time from bringing the whole apparatus into readiness until the first coincidence proved to be about 2 min. This does not contradict the observed rate of two coincidences per 1 min.

In this way more than 700 photographs were taken, and in more than 500 of them tracks of particles possessing very great velocity were found.

In many experiments, in order to investigate the interaction of these particles with matter, plates of various metals were placed in their path, across the middle of the chamber. Anderson[^3], for analogous experiments, took a lead plate; we investigated, besides lead, also copper and tungsten. In some quite recent experiments, with the aim of studying the nature of secondary particles arising in various metals, we placed thick metallic screens immediately under the upper counter \((B_1,\ \text{Fig. }2)\).

2. Photographs

About 75% of the successful photographs are similar to one another. On each of them there is a single track, traced by a particle that has passed through both counters. Most of these particles in a magnetic field of strength \(H = 2\) thousand gauss were not deflected by any appreciable amount at all. The minimum deflection that we can regard as different from zero lies somewhat above the limit at which the radius of curvature \(\rho\) becomes greater than 500 cm and, consequently, \(H\rho\) is greater than \(10^6\) gauss·cm. In our case it may be assumed that the kinetic energy \(E\) of the particles considerably exceeds the rest energy:

\[ E \gg m_0 c^3 \approx 0.5 \cdot 10^6 \text{ V-electrons}^*. \]

* See p. 494, the author’s note. The energy \(m_0c^2 = 9\cdot10^{-28}\cdot(3\cdot10^{10})^2\) erg is easily converted into volts on the basis of formula (a) of the following note, where one must put \(m=m_0,\ e=4.77\cdot10^{-10}\) (and multiply the result of the calculation by 300). Translator’s note.

Therefore the expression for the energy of a particle with charge \(e\) will take the simple form: \(E_e = 300\,H\rho\) V-electrons. Taking each of our particles to be an electron, we conclude that in practice we do not observe electrons with velocities exceeding 300 million V. Moreover, in these photographs many curved tracks were also recorded, belonging to electrons of lower kinetic energy*.

Using magnetic fields up to 18 thousand gauss, Millikan, Anderson \(^{3,10}\) and Kunze \(^{11}\) obtained deflections of almost all particles. The results of these two investigations are not entirely consistent with one another, but they give fairly comparable numbers characterizing the frequency of deflections in one direction or another. The distribution appears to be generally continuous up to the very largest values \(H\rho \simeq 7\cdot 10^6\) gauss·cm, corresponding to electron energies of about \(2\cdot 10^9\) V.

In each of the remaining 25% of all successful photographs, either individual isolated paths are recorded which do not pass through both counters, or whole groups of two or more paths, first discovered by Skobeltsyn \(^{4,12}\), and in which we now see an invariable companion of penetrating radiation. The appearance of such a number of tracks in individual photographs is evidently due to various secondary processes occurring as the penetrating radiation passes through matter. A systematic study of the secondary particles formed in this way by means of counters was first undertaken by Rossi \(^{13}\), and then by Johnson and Street \(^{14}\).

The result of the present work which most strikingly attracts attention consists in the discovery of an astonishing variety and complexity of these combinations of paths. Already 18 photographs have been obtained, in each of which more than 8 high-speed particles were recorded, and 4 photographs with a number of tracks exceeding 20.

* Here the kinetic energy \((E)\) of a charged particle is numerically equal to the equivalent work \((eV)\) of a potential electric field on this particle: \(E=mc^2-m_0c^2=eV\). Since \(E \gg m_0c^2\), it may approximately be assumed that \(E=mc^2=eV\), or, measuring \(E\) in absolute units of potential difference (conditionally, for particles with the given charge \(e\)):

\[ E_e=\frac{mc^2}{e}=V. \tag{a} \]

On the other hand, the centripetal force \(mv^2/\rho\), which compels the particle to move along a circular path, is equal in a uniform magnetic field to the Lorentz force \(evH/c\). Therefore \(mv^2/\rho = evH/c\). Noting that the assumption \(E \gg m_0c^2\) is equivalent to asserting that the velocity \(v\) is close to \(c\), we write the last equality in simplified form:

\[ \frac{mc^2}{e}=H\rho. \tag{b} \]

Comparing (b) with (a), we have \(E_e=H\rho=V\) abs. units, or \(E_e=300\,H\rho\) volts. Editor’s note.

** In order to overcome the thickness of the glass walls of the chamber (5 mm) and the brass walls of the counters (1 mm), an electron must possess an energy greater than \(\sim 5\cdot 10^6\) V.

At the end of the article 13 photographs are placed on separate inset sheets, and detailed explanations are given for each image.

Undoubtedly, very extensive investigations will be required before it becomes possible to give any reasonably complete interpretation of those extraordinarily complex atomic processes which are the true cause producing such groups of tracks. In this article an attempt is made to give only a preliminary and predominantly qualitative assessment of some of the most striking phenomena reflected in the photographs, leaving for subsequent reports almost all the details, measurements, etc.

The most remarkable characteristic feature, common to many photographs with a large number of tracks, is that they contain groups of several tracks issuing from some region situated somewhere inside the material surrounding the chamber, and most often diverging downward (Figs. 3 and 4, 5 and 6, 10—15; see at the end of the article).

Sometimes such a group of tracks appears to diverge from a single point; sometimes it is possible to establish the presence of two or more such centers of radiation at once; finally, wild tracks are often encountered which clearly have no relation to the main groups.

In the majority of the paths forming these groups there is no noticeable curvature in a magnetic field of 2 thousand gauss. When such a shower of particles falls into the chamber, it here and there calls additional centers of radiation into being. This is attested by photographs in those cases where a plate was placed across the chamber, inside which new centers also appeared (Figs. 10, 13). Then, at times, one can observe how a particle possessing enormous energy is, as it were, thrown back—in a direction almost opposite to the incident shower (Fig. 13).

3. The nature of the particles in the shower

To take the first step in elucidating these complex phenomena means, first of all, to establish, by identification, the nature of the particles producing the tracks. It is not altogether easy to cope with this, since the data drawn from the photographs and used for conclusions are often contradictory. However, apparently, one must inevitably arrive at that remarkable conclusion, removing the difficulties, which Anderson\(^{15}\) had already reached in deciphering analogous photographs. It consists in the fact that some of the tracks must be ascribed to particles carrying a positive charge, but having a mass negligible in comparison with the mass of the proton.

The most responsible measurement is made in determining the degree of curvature of the path in the magnetic field, because the radius of curvature of the track includes essentially the product \(H\rho\) (energy).

In addition, one can roughly estimate the change in the density of ion-

tion along the track, and in some cases then unambiguously determines the direction of motion of the particles. In exceptionally rare cases a particle stops in the gas inside the chamber, so that it is not difficult to measure the range of the particle; and even when this is not so, the knowledge that the range is greater than a certain length is sometimes of decisive importance.

The density of ionization produced by a fast particle depends only on the charge and velocity of the particle, but does not depend on its mass. However, for a given \(H\rho\) the velocity of the particle itself depends on the mass, and therefore two particles with the same \(H\rho\), but with different masses, will ionize differently*. Consequently, the measurement of \(H\rho\) for some given track, together with the measurement of the ionization along it, in principle makes it possible to determine the mass of the particle.

Fig. 1. Loss of energy in water per \(1\ \mathrm{cm}\) of path, in million volt·g\(^{-1}\) cm\(^3\).

The change of ionization density along the tracks of \(\beta\)-particles has at present been determined experimentally up to values of the ratio \(v/c\) of about \(0.95\), and for larger values can be extrapolated theoretically. For our purposes it is convenient to take the values of the loss of energy of the ionizing particle for each centimeter of the path traversed by it. These values were obtained theoretically by Bethe\(^{16}\). In Fig. 1 are shown curves representing, as functions of \(\log_{10}(H\rho)\), the loss of energy along the track in two cases: for an electron and for a proton.

As we see, there is only a small difference between the ionization produced by the one and the other when \(H\rho\) has values not less than, approximately, \(1.5\cdot 10^{6}\) gauss·cm. In the region of smaller values of \(H\rho\), however, protons, as heavier particles, ionize, of course, incomparably more strongly than electrons. Consequently, if for some track the observed value \(H\rho\), say, is less than \(1.0\cdot 10^{6}\) gauss·cm, we have an excellent possibility of deciding whether we are dealing with a mass of the order of the mass

* The relation between the velocity \(v\) and \(H\rho\) is

\[ \frac{v}{c}=H\rho\left[\left(\frac{m_0 c^2}{e}\right)^2+(H\rho)^2\right]^{-\frac12}, \]

where \(m_0\) is the rest mass and \(e\) is the charge in CGS electrostatic absolute units. (The latter formula is obtained simply if the equality \(mv\rho = evH/c\) is solved with respect to the fraction \(v/c\). Of course, here \(m=m_0(1-v^2/c^2)^{-1/2}\). — Translator’s note.)

electron or proton. Thus, by measuring \(\rho\) and knowing \(H\), \(H\rho\) becomes known for one track or another. Knowing, in addition, the range of the particle, one can estimate its mass. Table 1 gives the values of velocities and ranges for protons and \(\alpha\)-particles at various \(H\rho\).

TABLE 1

Relation between \(H\rho\), velocity, and range of protons and \(\alpha\)-particles

According to Anderson’s report, several tracks have been found which must be ascribed to positively charged particles of negligible mass. Anderson gives a detailed description of these photographs, although the photographs themselves are not reproduced. In one of them the direction of motion can be unambiguously inferred from the change in the curvature of the path after passing through a lead plate. In another photograph, two tracks emerging from the plate are curved in opposite directions. In a third, two particles leave the plate, deviating to the side toward which positive charges would deviate. The length of the range and the characteristic ionization—all this, together with the preceding, gives Anderson grounds to assert that before us are positively charged particles with a mass considerably smaller than the mass of the proton.

To determine the sign of the charge of a particle, it is necessary to know in which direction it moved along the track. There are four ways to learn this from photographs:

1. The particle passes through a sufficiently thick metal plate, so that upon emerging from it the particle has managed to lose an appreciable fraction of its energy. Obviously, in this case the motion proceeds from the larger value of \(H\rho\) toward the smaller. Otherwise one would have to admit the existence of an energy gain inside the plate, and this possibility is so improbable that we are entitled to discard it. If, during photography, a very slow particle is encountered, then an opportunity presents itself to detect a change in \(H\rho\), [[unclear: word begins “iz-”]]

...caused by the continuous loss of energy during the passage of the particle through the gas.

1. On the other hand, if the particle serves as the cause of the appearance of some secondary particle with sufficient energy—say, in a collision with a free electron—then the angle between the secondary track and the primary one will indicate the direction of motion of the particle.

2. If a group of tracks diverges from some common point or from some small region of space, then there is a very high probability—though not complete certainty—that every particle of such a group is moving away from this region.

3. If a track is observed in an almost vertical direction, then it is more probable that the particle was moving downward rather than upward. The basis for this last assumption is the indisputable fact that ionization under the action of cosmic radiation increases from depths to heights. However, it is difficult to estimate this probability numerically, since the recurrence rate of such phenomena as that recorded in Fig. 13 is not known; here there is, at the very least, one particle reflected upward.

In the photographs in Figs. 5 and 6, the majority of the particles did not deviate \((H\rho > 10^6)\), but there are nevertheless tracks whose curvature can be measured. One part of the tracks proves to be deflected to one side, the other—to the opposite side*. The outward appearance of the entire group as a whole strongly suggests that the particles diverged from above. If this supposition is correct, then the particles deflected to the left have negative charge, and those deflected to the right have positive charge. There are two particles with positive charge and a value of \(H\rho\) for one of \(0.4\cdot 10^5\) and for the other \(1.5\cdot 10^5\) gauss·cm. If these two tracks belonged to protons, their total range in air would not exceed, respectively, \(0.2\) and \(3\) cm. In reality, however, their length in the chamber reaches \(12\) cm (after reduction to air at normal pressure and temperature). Therefore these two tracks, undoubtedly, cannot be ascribed to protons; they belong to some other particles having a considerably smaller mass.

Similar tracks can also be seen in the photographs in Figs. 3 and 4; one such track is in Fig. 11, another in Fig. 12, and, probably, two tracks are in Fig. 15.

The study of the ionization density along these tracks fully supports these conclusions. It is quite clear at first sight from all these photographs that the ionization density along the positively curved tracks differs very little from the ioniza—

* The curvature of an individual, more or less vertical track will be called positive if it corresponds to the curvature of the path in the motion of a positive charge. Then a track that is part of a shower will have positive curvature only in the case when, from the well-expressed region whence the tracks of the shower diverge, a positive charge would move along it. An analogous rule is also introduced for the term “negative curvature” of tracks.

...with uncurved tracks or tracks having negative curvature. Unfortunately, it is impossible, without special experiments, to make an exact estimate of the magnitude of the ionization—neither the total ionization nor that produced along the primary tracks.* However, even a rough estimate gives values of the ionization density that are the same for all these tracks, as was to be expected for fast electrons. Further, by comparing in appearance some undeflected track with the track of a slow secondary \(\beta\)-particle, for which \(H\rho\) has been measured, i.e., whose energy is known, it becomes clear that any track along which the ionization is approximately three times greater than in the undeflected tracks—any such track can easily be identified (see the explanation to Fig. 13). Consequently, we are entitled to conclude that these tracks, in all probability, cannot be ascribed to protons. If they were protons, the ionization would be 10–100 times more abundant, and the track would look approximately as in the photograph of Fig. 16, where the thick track is an unquestionable proton with \(H\rho \approx 3 \cdot 10^5\) gauss·cm.

The only possible conclusion from these two arguments, based on range and on ionization, is that these tracks belong to positively charged particles with a mass rather comparable with the mass of the electron than with that of the proton.

Of course, one may also imagine the case in which one of these tracks owes its origin to a negative electron moving upward. However, this is only a great rarity: the electron must then pass precisely through the region from which all the other electrons, judging by their tracks, appear to be diverging. It is difficult to estimate the probability of such an event numerically; nevertheless, the presence of these tracks with positive curvature is so frequent and characteristic a phenomenon in the observed showers that, in general, one may boldly and definitively accept the explanation given above.

In showers we have found in all 14 tracks which almost certainly must be attributed to such positive electrons, and several others that are less reliable.

Until stronger fields are used, it will be impossible to find the ratio between the numbers of positive and negative electrons in a shower. There are, however, some grounds for supposing that the numbers of the two are approximately the same, at least in some showers. In the photographs of Figs. 3, 4, 5, and 6, of the total number of tracks whose curvature could be measured, about half are deflected to the right and the same number to the left. Nevertheless, negative particles are in general encountered more often.

Independently of showers, an additional proof of the existence of a positively charged particle with a mass close to

* This can be done by working with \(H_2\) or \(He\) (as was done by Anderson and by the authors of the present paper), or by somewhat slowing the action of the mechanism in such a way as to give the ions time to diffuse independently of one another in different directions.

mass of the electron, is the photograph in Fig. 9. Here a particle is recorded penetrating a 4-mm lead plate. The track is more strongly curved below the plate than above it, so that the particle must have been moving from top to bottom, unless one assumes that, while passing through the plate itself, the particle gains energy from outside; consequently, it has a positive charge. On leaving the plate the value of \(H\rho\) for the particle is equal to \(1.2 \cdot 10^5\) gauss·cm. If it were a proton, then the corresponding range in air would be of the order of 1.5 cm, and it would produce ionization more than 100 times greater than that of a fast electron.

Several tracks were observed in which the secondary tracks, formed in the gas, indicate the direction of motion of the particle penetrating both counters. In most cases the direction indicated by the secondary tracks, as was to be expected, proved to be from top to bottom. But in the photograph of Fig. 16 a track is seen which can be ascribed only to a positive electron, if the direction of the “spur” of the secondary track* is regarded as convincing evidence. In general, however, it is quite possible that such a slow secondary particle (about 35 thousand \(V\)), after some subsequent collision—and more than once—could have deviated from its initial path. This means that the recorded direction of the projection of its track differs from that obtained at the moment of collision. Thus one cannot attach great importance to this one track.

In the photograph of Fig. 7 the secondary track, formed inside the plate, indicates the direction of motion from top to bottom. One of the two lower tracks has positive curvature which, although small, nevertheless considerably exceeds the curvature that could have arisen owing to the motion of the gas.

Annoying as it is, in our experiments the magnetic field was insufficient for deflecting by a measurable amount most of the tracks, both among those which appear single and among those which form part of a shower. As has already been mentioned, so far measurements at stronger fields have been made only in the work of Anderson and of Kunze, and, although these works do not show complete agreement with each other, they both convincingly show that approximately the same number of tracks are deflected in one direction and in the other. The same is true of all those tracks in our photographs which have measurable curvature. Anderson and Kunze found, respectively, 30 and 60% of tracks deflected positively.

It is quite possible that almost all single tracks belong to particles that originated as part of some shower arising somewhere high in the atmosphere or in some material far above the chamber. If this is really so, one should expect that the ratio between the number of positive

* Very thin, but distinct (see the explanation to Fig. 16). Translator’s note.

** Skobeltsyn also observed the existence of such particles \(^{18}\).

and negative particles among all single tracks turns out to be the same as for showers. The fact that the experimental data agree with this suggests that this is perhaps indeed the case.

However, the existence of such photographs as in Fig. 13 proves that fast particles sometimes move from below upward, so that one may say with confidence that some fraction of the single tracks was traced by particles moving upward, not downward. Consequently, one cannot be certain of the sign of the charge of a particle that has left only a simple isolated track on the photograph, if one is guided only by its curvature and cannot resort to other data concerning its direction.

And yet, of course, one may safely assert that the majority of positively curved tracks are in fact formed by positively charged particles passing through the chamber from top to bottom.

It is difficult to agree with Anderson and Kunze that the positively curved tracks belong chiefly to protons. In fact, as has already been mentioned, the most striking feature of our photographs is precisely that the enormous majority of tracks possess almost the same specific ionization. Kunze, considering the particles that described a path with positive curvature to be protons, finds for them a mean energy of about \(4 \cdot 10^5\) V. Protons with this energy (equivalent to \(H\rho \simeq 3.5 \cdot 10^5\) gauss·cm) produce an ionization differing little from the ionization of electrons (of the same energy) (Fig. 1). But if such fast protons were present, then slower protons should also have been present in some quantity, for the descending flux of particles must undoubtedly be quite heterogeneous at the bottom of the atmosphere, whatever it may be at the top. And these slower particles would produce stronger ionization along 1 cm of their path, and the tracks traced by them would appear noticeably heavier. We found four tracks having the character of proton tracks, of which two are on the photograph in Fig. 16 and one in Fig. 15, but in no case can they be assigned to the main group of particles moving upward, and most likely they are connected with some local process of nuclear disintegration. Further, all isolated tracks showing positive curvature have almost the very same specific ionization as the undeflected ones. Such tracks belong partly to negative electrons going from below upward, and partly to positive ones coming from above downward. By a positive electron is meant a particle with a unit positive electron charge and with a mass considerably smaller than the mass of the proton.

Thus it seems justified to conclude that the primary flux of upward-moving particles consists predominantly of positive and negative electrons. Protons are probably also present in some quantity.

4. Recurrence of showers

Showers—these streams of particles—are generated in the process of the disintegration of a nucleus, caused by particles or protons of enormous kinetic energy, with which penetrating radiation is always associated. This assertion sounds very plausible. It will become still more probable if one takes into account the fact that the majority of showers have their beams diverging from above downward.

It is possible that these showers of particles belong to the same class of phenomena as the “bursts,” or “flashes,” of ionization discovered by Hoffmann and studied by Steinke, Schindler, Messerschmidt, and others.^19 These bursts reveal a definite connection with the intensity of cosmic rays, judging from experiments with them in deep mines.

One may hope that in similar ways it will also be possible to study the recurrence of showers—by counting a large number of coincidences. This does not exclude, of course, the entirely conceivable possibility that some part of all shower phenomena occurs owing to spontaneous processes inside the nucleus.

These showers undoubtedly appear very rarely in comparison with isolated tracks—far more rarely than has been directly recorded in the photographs, for the recurrence of the appearance of each given type of track depends on the product of the actual recurrence of this event (the passage of particles), multiplied by the probability that at that same moment both counters will operate at once. This latter probability is very different for showers and for isolated tracks. In our case there is one shower for every thirty simple tracks. However, in reality the ratio of the number of showers crossing the chamber to the number of simple tracks is certainly still considerably smaller. Indeed, if a single particle penetrates the chamber, there is very little chance (only about 1 in 200) that this particle will pass through both counters. At the same time, if an entire shower of many particles cuts through the chamber at once, there is a rather high probability, sometimes perhaps of the order of 1:5, that a coincidence will occur. Thus the ratio of the number of showers to the number of single tracks found in the photographs may be forty times greater than their ratio in reality.

Showers apparently arise indiscriminately in any material surrounding the chamber. From the moment when the chamber was almost completely surrounded by a solenoid with a copper winding, most of the showers began to arise in the copper, but some of the centers of radiation are also found in the glass walls and the chamber cover, in the aluminum of the piston, and in the air of the room. When a plate of lead or copper was inserted into the chamber along its diameter, across the rays, groups of tracks appeared diverging from centers lying inside the plate. A Wolfram plate was also used for several photographs, but the photographs made with it are of no special interest.

It is interesting to make a rough estimate of the recurrence of these showers, by comparing their number per unit time with the total number of atomic nuclei in the material surrounding the chamber. One coincidence is observed every 2 min. and one shower with a number of tracks exceeding 8 approximately every thirty coincidences. In other words, one shower is recorded, on the average, per 1 hour. Almost all showers are emitted from centers located predominantly in one definite part of the solenoid winding above the chamber itself. This part of the winding contains a mass of copper of approximately 10 kg. Assuming that one out of five showers developing in copper produces a pulse in the counters, we find that one shower is produced in 10 kg of copper, on the average, every 10 min. Expressing this in the form of the mean lifetime of copper nuclei, we obtain \(10^{18}\) years.

Moreover, we can also determine a lower limit for the effective cross section of the nucleus of a copper atom in the production of such showers, under the assumption that all incident fast cosmic particles, on entering the nucleus, destroy it. Since in the course of 1 min. 1 cm\(^2\) is penetrated by 1.5 particles, we find the order of magnitude of the effective cross section of the copper nucleus: \(10^{-27}\) cm\(^2\).

One can, in a first approximation, estimate the total ionization that would be produced in an ionization chamber of the type used by Steinke and Schindler, for example, owing to such a shower as is recorded in Figs. 3 and 4. These photographs contain about 20 tracks with an energy more than sufficient to pass straight across an ionization chamber of this type. Assuming that the layer of air equivalent to the thickness of the material of the ionization chamber is 240 cm, and considering that on each centimeter of its path a particle produces 80 ion pairs, we obtain for the total ionization produced a figure of about \(4 \cdot 10^5\) ion pairs. Although this is less than the value given by Messerschmidt\(^{18}\) for the lower limit of ionization “bursts” (\(3 \cdot 10^6\)), it is nevertheless almost certain that in showers there exist far more particles than actually pass through the Wilson chamber. Some of them may have a large mass, a large charge, and, consequently, also a more abundant specific ionization. Thus, Figs. 16 and 17 provide documentary confirmation of the existence of at least several protons entering into the showers.

But there may also exist other particles with an even greater ionizing power, which are created inside the gas in the chamber, only too rarely to become observable.

It must be remembered that there is a certain probability that some ionizing particle, emitted from any center inside the chamber walls or the surrounding material, will be observed in the form of a track. This probability is proportional to the range of the particle and, consequently, inversely proportional to its ionizing power. As a result, particles that ionize na

more ionization, will not be able to propagate in the chamber at all and thus prove unobservable. Therefore there is nothing impossible in particles with such a short range being produced, together with more rapidly penetrating particles, in the process that initiates the shower.

Consequently, it seems very plausible that the shower phenomena in the Wilson chamber and the ionization “bursts” are related to one another, although the latter occur more rarely and give greater ionization than some of the showers observed by us.

The total energy of the particles penetrating the chambers in the photographs of Figs. 3 and 4 is more than \(2 \cdot 10^9\) V; this is what is obtained if one assumes that each of the 20 tracks has an energy of \(100 \cdot 10^6\) V.

5. Mechanism of the Origin of Showers

Almost undoubtedly, showers owe their origin to some process taking place in the interaction between particles, or quanta, with enormous energy and atomic nuclei. Modern theory is capable, at least approximately, of considering the interaction of such particles and quanta with electrons outside the nucleus. Thus, Heisenberg\(^{20}\) has recently collected the theoretical results pertaining to this question in order to compare them with similar experiments carried out for the investigation of penetrating radiation. We intend to study, with the aid of our photographs, the repeatability of the phenomena of formation of secondary particles and quanta, as well as the processes of scattering of the primary ones. In Fig. 7 a secondary particle with an energy of about \(60 \cdot 10^6\) V is photographed, produced under the action of a fast particle.

The effective value of the cross section in a collision of an electron or a proton of very high energy with a free particle having mass \(m_0\) and one electron charge is expressed as follows:

\[ \frac{2\pi e^4}{m_0 c^2 \varepsilon}, \tag{1} \]

where \(\varepsilon\) is the energy initially imparted to the particle at rest.

The direction \(\vartheta\) of emission of the particle is determined by the quantity

\[ \operatorname{tg}^2 \vartheta = \frac{2 m_0 c^2}{\varepsilon}. \tag{2} \]

It is noteworthy that the mass of the bombarding particle does not enter into these expressions. Some collisions, for which these relations are approximately fulfilled, have appeared in the photographs.

From (1) we find that the length of the free path of a particle in lead, sufficient to cause the appearance of a secondary electron with an energy of \(100 \cdot 10^6\) V, must be about 16 cm; our experiments confirm this in order of magnitude. From this figure it is clear that the probability of ejecting two or more ne-

dependent secondary electrons from one and almost the same common point of annihilation. Therefore, when one has to see a trace branching from one point into three or more traces, it is impossible to refrain from the conclusion that here some interaction with the nucleus is taking place. If the incident particle is a quantum, we are entitled to draw the very same conclusion.^10,12

It is obvious that there exist several different processes giving rise to complex shower paths. In a small number of cases this process is quite simple. The incident particle—usually a negative or positive electron—knocks out of an individual nucleus, in all probability, three or more particles. Fig. 17 confirms very clearly that the incident particle ejects from a copper nucleus 2 electrons (both with \(E_e \approx 13 \cdot 10^6\ \mathrm{V}\)) together with one proton. The eruption may also have been accompanied by other particles, but they evidently had too short a range to overcome the thickness of the plate and emerge from it. Fig. 13 gives the picture of two electrons (\(E_e \approx 10 \cdot 10^6\) and \(13 \cdot 10^6\ \mathrm{V}\)) knocked downward out of a lead nucleus, and of two others, with greater energy (\(E_e > 100 \cdot 10^6\ \mathrm{V}\)), knocked upward. It is possible that one of the latter two is the incident particle that explodes the nucleus, and then the other electron is one of the fragments flying upward in the explosion. It is also possible that both upper particles are products of the destruction of the nucleus; then in this case the destruction itself must be attributed to some non-ionizing agent.

However, both these cases are comparatively simple in comparison with the complex picture of abundant showers. In this most typical process there is observed the simultaneous eruption of a certain number of particles flying out with enormous energy. These particles are usually ejected in directions enclosed within a fairly narrow cone, but there are cases (Fig. 12) when this cone is fairly wide. It is quite natural to seek the explanation of the narrow cone of particle spread in the impulse imparted to them at the moment of impact by the incident particle, which possesses exceedingly great energy. It is still impossible to establish the nature of all the particles ejected from the nucleus, but, apparently, among them negative and positive electrons predominate; there are, it is true, some as yet insufficient indications that in a number of cases both types of electrons are knocked out in approximately equal numbers.

The origin of these particles arouses enormous interest; in particular, they are undoubtedly often produced inside material with light and medium atomic weights, since the emitting centers have been found both in air, and in glass, and in aluminum, and in copper. According to the very latest ideas^21 about the structure of the nucleus, there should be no free negative electrons in such light nuclei. Yet there have already been found, at the very least, positive and negative electrons, issuing from an individual pointlike center of radiation in glass, copper, or lead—

(Figs. 12, 11, and 10) and, consequently, in all probability, from a separate nucleus.

There are three possible hypotheses which we may make concerning the appearance of these particles: they may have existed in the destroyed nucleus from the very beginning, even before the collision; they may have existed in the incident particle; finally, they may have arisen during the collision process. In the absence of any independent evidence for the separate existence of particles before the shaking of the nucleus, it is reasonable to accept the last of these three hypotheses. Then, taking into account the well-known difficulties^22 that arise when electrons inside nuclei are treated as independent mechanical objects, the last hypothesis perhaps has an even greater advantage in this sense. Thus, according to this hypothesis, all showers (together with ordinary β-decay) should be imagined as a process of the creation of a particle in the direct sense of this word.

This question is very closely connected with the problem of the structure of the neutron^23. According to the view of the neutron as a composite particle, negative electrons in showers may be obtained by the splitting of each neutron into a negative electron and a proton, but this scheme gives no explanation for the appearance of positive electrons. Moreover, it leads to the expectation of a larger number of proton tracks in the photographs than is actually observed.

There is also another view, that the neutron is a certain indivisible particle and that there are no free negative electrons in light nuclei. In that case it must be said that both negative and positive electrons in the shower arise during the process of destruction; nevertheless, if the law of conservation of electric charge is observed, then positive and negative electrons must arise from the nuclei in equal numbers, for it is hardly possible for many protons to be created because of the colossal energy \((E_e \approx Mc^2 = 940 \cdot 10^6\ \mathrm{V})\)* required for the formation of each proton.

In this way one may imagine how, in the process of destruction of light nuclei, negative and positive electrons are born in pairs. If the mass of the positive electron is the same as that of the negative one, then each act of twin birth requires an expenditure of energy \(2m_0c^2\), i.e. about \(1 \cdot 10^6\ \mathrm{V}\). This is much less than the kinetic energy of translational motion which they generally display in showers. Although the ultimate aim of the experiments should be to obtain such data from which it would be possible to compile an overall balance of the number of particles, mass, and energy, this undoubtedly proves to be an extremely difficult matter. In order to achieve this, in essence

* Another possible explanation, in which the electric charge may be conserved during the process causing the emission of an electron from the nucleus, was proposed by Prof. Dirac: the neutron may transform into a proton, while at the same time a negative electron is born.

it is necessary to obtain a shower that would arise in the gas inside the chamber, so that one could see the tracks of all the particles arising and flying apart after the disintegration of the nucleus. Some of the showers are represented by groups of nearly parallel tracks. This involuntarily suggests Viльson’s conjecture^24 that in electric fields during thunderstorm discharges rapidly “escaping” electrons and protons should appear. But if these particles, which undoubtedly may exist in general, enter the chamber, one must expect the presence of groups of parallel tracks, running far from one another and having the most varied width. For this it is only necessary that they pierce the chamber shortly before expansion or during it. It is, however, characteristic of all showers that the tracks in them, although almost always somewhat divergent, all intersect at one point and always have noticeably one and the same width, indicating that the particles pass through the chamber during a much shorter interval of time than the time of expansion (1/100 sec.). Thus one may be certain that the showers cannot be attributed to groups of Wilson’s “escaping” electrons.

6. Assumptions Concerning the Properties of the Positive Electron

The existence of positive electrons in these showers immediately raises the natural question: why have they until now eluded observation? It is clear that they can possess only a limited lifetime as free particles, since they are not encountered in any substance under normal conditions.

It is quite admissible that they may enter into combination with other elementary particles and form stable nuclei, thereby ceasing to be free. But it seems more acceptable that they disappear in interaction with a negative electron, emitting in the process 2 quanta or more. This latter mechanism is given directly in Dirac’s theory of the electron^25. According to this theory, the quantum states in the region of negative kinetic energy, which had previously presented an insuperable obstacle to physical interpretation, are, with few exceptions, filled by negative electrons. The few unoccupied states behave like ordinary particles with positive kinetic energy and positive charge. Dirac himself thought to identify these “holes” with protons, but this had to be abandoned when it was established that these “holes” must have the same mass as negative electrons^6. There remains the immediate and important task of experimentally determining the mass of the positive electron by accurate measurements of its ionization and \(H\rho\). At present one can only say that the absence of a difference between the ionization of the tracks of negative

…of negative and positive electrons at the same \(H\rho\) became a certainty, and this indirectly serves as temporary proof of the equality of their masses.

According to Dirac’s theory, positive electrons have only a very short average lifetime, until some negative electron “jumps” down from above, with ease, into an unoccupied state*. In this way the “hole” is filled, and there occurs the disappearance of both—the positive and the negative—electrons simultaneously; in this process 2 quanta of energy are emitted**.

We feel obliged to Prof. Dirac not only for a very valuable and repeated discussion of these questions, but also for permission to cite the results of his calculations for determining the actual probability of this process of “annihilation” (disappearance) of electrons. The size of the cross-section of electrons in annihilation (in units of area) is\(^{27}\):

\[ \Phi=\frac{\pi e^{4}}{m_{0}^{2}c^{4}}\,f(\gamma), \tag{3} \]

where

\[ f(\gamma)=\frac{1}{\gamma+1}\left[ \frac{\gamma^{2}+4\gamma+1}{\gamma^{2}-1}\log\left\{\gamma+\sqrt{\gamma^{2}-1}\right\} -\frac{\gamma+3}{\sqrt{\gamma^{2}-1}} \right] \]

and

\[ \gamma=(1-v^{2}/c^{2})^{-\frac12}, \]

while \(v\) is the velocity of the positive electron.

Dirac calculated the following values for the mean free paths in water up to the moment of annihilation. This “annihilation path” \(\lambda=1/n\Phi\), where \(n\) is the number of extranuclear electrons per unit volume.

In the last line of the table are given the real values of the path \(l\) of the electron in water.*** All figures for the values of the energy,

* The words “above” and “down” must of course be understood here in a conditional sense, imagining the energy states graphically plotted along the ordinate, as the levels of a radiolamp staircase. For a free electron the levels are arbitrary, except for generally quite impossible energy values symmetrically located about zero (for which \(|E|<m_{0}c^{2}\)). Translator’s note.

** The “disappearance” of the negative electron is understood here in the sense of its unobservability. In general, according to the very essence of Dirac’s conception, in the normal state of space (regarded as an unusual kind of undiluted space) the entire region of negative energy values is completely occupied at all its points and the region of positive values is not occupied at all. What may be observed are only deviations from this: either the absence of negative electrons in the region where \(E<m_{0}c^{2}\), or their presence in the region where \(E>m_{0}c^{2}\). Translator’s note.

*** According to the preceding, the free path of the positive electron up to the moment of annihilation, called in translation the “annihilation path,” is an abstraction. It is computed under the assumption of an ideal electron gas (electrons completely free, the forces of atomic bonds so negligible that they may even be neglected). However, in the real and most probable case the positive electron, moving in matter, expends its energy on the ionization of the atoms it encounters, i.e. on doing the work of detaching electrons. Therefore almost all positive electrons in fact lose their…

TABLE 2

Annihilation range and real range for a positive electron

smaller than \(2\cdot10^6\) V—the data are experimental\({}^{28}\), the rest theoretical\({}^{21}\).

If the probability that, before annihilation of the positive electron, its energy decreases from the value \(E_1\) to \(E_2\) is denoted by \(W(E_1,E_2)\), then it is easy to see that

\[ \log\left[1-W(E_1,E_2)\right] = \int_{E_1}^{E}\frac{dl}{\lambda}, \]

where \(dl\) is an element of the path of the positive electron, and \(\lambda\) is the annihilation range of particles with energy \(E\).

An approximate numerical integration, carried out with the aid of the data of Table 2, leads to the following value of the probability of annihilation of a positive electron accompanied by a decrease of energy from \(200\cdot10^6\) V to \(10^5\) V:

\[ W(E_1,E_2)=0.36. \]

If one estimates the probability of annihilation of a positive electron in water during a unit of time, then one finds that this probability increases as the energy decreases and reaches a certain constant value of \(2.5\cdot10^9\ \mathrm{sec}^{-1}\) for energy values smaller than \(10^5\) V. Thus these positive electrons, which exist only if they have the specified energy, obey a certain probability law strongly resembling that which holds in the case of radioactive decay. The only difference is that their mean lifetime is proportional to the concentration of negative electrons. In water their lifetime is equal to \(3.6\cdot10^{-10}\) sec. When the behavior of positive electrons has been studied in greater detail, there will be in hand a criterion for testing these predictions by dir—

speed considerably earlier than is required by the Dirac annihilation range. Having lost speed down to approximately \(10^5\) V (see below), it already annihilates with one of the electrons of some electronic shell. The result is one and the same: the emission of very hard quanta, but producing ionization. This “real” range, which under one or another experimental condition can in principle be observed and is observed, is represented in the last row of Table 2. The absence of measurable ionization for annihilation quanta is explained by the fact that they belong to that region of the spectrum (to those “rays”) for which there is no photoeffect, while Compton scattering for an individual quantum is of low probability according to the Klein–Nishina formula. Translator’s note.

of the Dirac theory. Apparently, at present there are no grounds whatever against its validity, while in its favor speaks the fact that the lifetime it predicts for the positive electron is sufficiently large for it to be observed in a Wilson chamber and sufficiently small to explain why the positive electron had not previously been detected by other methods.

It should be possible to find in photographs some data on positive electrons which entered a metal plate but could not emerge from it again, owing to their annihilation while passing through the metal. It is possible that, in observing the Compton recoil of electrons, it will be possible to discover the annihilation spectrum (lying in the region of $\gamma$-rays). According to the Dirac theory, this spectrum should have a lower energy limit of about $0.5 \cdot 10^{6}\ \mathrm{V}$; then, at somewhat higher energy values, there should be a maximum and after it—a gradual decrease of intensity.

Judging from everything, positive electrons can also be obtained by other means—without the aid of cosmic radiation. It may be that the anomalous absorption of $\gamma$-rays by heavy nuclei is connected with the formation of positive electrons, which, upon their annihilation, emit quanta of radiation in return. For this return radiation an energy value has in fact been found experimentally of the same order as that expected for the annihilation spectrum. Moreover, the hypothesis of the existence of positive electrons among the secondary particles produced in neutron bombardment explains one curious phenomenon discovered by Curie and Joliot[^30], namely, that tracks of fast electrons have been found with a curvature indicating the motion of a negative electron in the direction toward the neutron source.

7. Non-ionizing particles and secondary centers of radiation

It is impossible to explain the appearance of showers without connecting them with the existence of streams of some non-ionizing agent. This is supported by the circumstance that individual tracks have been found to which no visible particle producing them corresponded. In order to explain such secondary centers of radiation, it is necessary to postulate the existence, in showers, of some non-ionizing particles or photons (quanta).

When two (or more) radiating centers are clearly visible at the top of the chamber, as in Fig. 3, it is natural to suppose that either one of them is a secondary process excited by the particles of the other, or both are secondary processes with respect to some primary one. In either case one must admit that the particle or quantum producing the secondary process has an astonishingly short mean free path before the next interaction.

It can be seen from the photographs that, when one appears ...

...shower, there is a relatively high probability that some other shower will appear a little below the first; this is rather unexpected, and it would be very interesting if it proved possible to show that these phenomena can, with high probability, be explained by the supposition that the non-ionizing particles are neutrons or quanta.

8. Summary

1. A method has been briefly described by which particles possessing enormous energy can be made to photograph the traces of their own paths in a Wilson chamber (Fig. 2).

1. A picture is drawn of the most striking, characteristic phenomena recorded by this method in some of the 500 successful photographs; the question of the nature of the “showers,” consisting of particles giving in the photographs a combination of several, and even many, paths at once, has been discussed.

2. Consideration of the range, ionization, curvature, and direction of motion of the particles leads to confirmation of the view first expressed by Anderson, namely that there must exist particles with positive charge, but with a mass rather comparable with the mass of the electron than with that of the proton.

3. The question of the mean recurrence of showers has been investigated, as well as their possible relation to the ionization “bursts” observed by Hoffmann, Steinke, and others.

4. The question of the origin of positive and negative electrons in the shower has been analyzed, and the most probable conclusion has been drawn that they are produced during the process of collision.

The subsequent behavior of positive electrons is considered in the light of Dirac’s theory of “holes.”

The plausibility of the existence of non-ionizing particles in the processes that give rise to complex showers has been discussed.

Addendum

After the publication of this work, an important supplementary communication by Chadwick, Blackett, and Occhialini was printed in Nature of April 10, 1933, under the title “New Evidence for the Existence of the Positive Electron.” We reproduce this communication in full.

Anderson’s and Blackett—Occhialini’s* experiments have also made it possible to discover phenomena in the Wilson chamber which, with great

* Anderson, “Science” 76, 238, 1932.
** See the preceding article by Blackett and Occhialini.

convincingly prove the existence of positive electrons—particles with approximately the same mass as the ordinary electron, but carrying a positive elementary charge.

Certain phenomena observed when neutrons pass through matter, and the experiments of Curie and Joliot*, in which the track of an electron moving toward the bombarding neutrons was recorded in a Wilson chamber, led one to suppose that positive electrons might be produced in the process of interaction between neutrons and matter. And indeed, in recent days we have succeeded in photographing phenomena that may be interpreted in this sense.

A polonium preparation and a beryllium screen were placed in the immediate vicinity of the walls of the Wilson chamber. A small lead screen, with a surface area of about \(2.5\ \mathrm{cm}^{2}\) and \(2\ \mathrm{mm}\) thick, was attached to the inner wall of the chamber. This lead screen was therefore photographed at the same time as it was subjected to the action of the radiation coming from the beryllium and consisting of \(\gamma\)-rays and neutrons. The photographs were taken by means of a stereoscopic pair of cameras. During expansion a magnetic field was applied, the strength of which was ordinarily approximately 800 gauss.

The greater part of the tracks appearing in the photographs, judging from the sign of their curvature, are evidently due to negative electrons. But quite a number of cases were also found in which the tracks, having one end either in the lead screen itself or close to it, possess curvature of the opposite sign. One of two things is possible: either these tracks belong to particles carrying a positive charge, or to negative electrons emitted from various remote corners of the chamber and, by some favorable arrangement of the magnetic field, whose tracks happen to end precisely at the lead plate. From the point of view of statistics, the first explanation is of course far more probable, i.e. that the tracks begin in the screen and are therefore traced by particles carrying a positive charge.

A striking proof of this hypothesis was obtained by placing a metal plate across the chamber so as to intercept the paths of the particles. In this way it has so far been possible to obtain only a few photographs good enough for a track with positive curvature, after passing through the plate, to remain in focus from beginning to end. However, even these photographs leave no doubt that the particles are emitted from the lead screen or from a place close to it; consequently, we are dealing here with positively charged particles. In one case the plate consisted of a layer of copper \(0.25\ \mathrm{mm}\) thick. In this case the track on the side of the plate exposed to irradiation had a curvature corresponding to

* Curie et Joliot, L’existence du neutron, Éd. Hermann et Cie, Paris.

value \(H\rho = 12\,700\), while on the protected side \(H\rho = 10\) thousand. This indicates that the particles propagated from the screen through the copper plate, losing a certain amount of energy in the plate. The change in the value of \(H\rho\) while the particles pass through copper is approximately the same as for a negative electron placed under the very same conditions. The characteristic ionization of the particles is also approximately the same as that of a negative electron. These observations agree with the assumption that the mass and magnitude of the charge of the positive particle are the same as those of the negative electron.

How these positive electrons are produced is still unclear: do they arise under the action of the neutrons entering the beryllium radiation, or owing to the \(\gamma\)-radiation accompanying the neutrons? It is to be hoped that further experiments, already under way, will resolve this question.

Cavendish Laboratory.
Cambridge, March 27, 1933.

LITERATURE

1. Blackett and Occhialini, Nature 130, 363, 1932.
2. Skobelsyn, C. R. Acad. Sci., Paris 195, 315, 1932.
3. Anderson, Phys. Rev. 41, 405, 1932.
4. Mott-Smith and Locher, Phys. Rev. 38, 1399, 1931; 39, 833, 1932.
5. Johnson, Fleischer and Street, Phys. Rev. 40, 1018, 1932.
6. У. Ф. Н., X, 1, 1930; XII, 625, 1932. In Russian on circuits that count coincident discharges in two counters, see pp. 635, 639, and 641.
7. Rossi, Nature 125, 636, 1930.
8. Curtiss, Bur. Stand., J. Res. 4, 663, 1930.
9. Williams and Terroux, Proc. Roy. Soc. A. 126, 289, 1930.
10. Millikan and Anderson, Phys. Rev. 40, 325, 1932.
11. Kunze, Zs. Physik 79, 203, 1932.
12. Skobelsyn, Zs. Physik 54, 686, 1929; C. R. Acad. Sci., Paris 194, 118, 1932; Auger and Skobelsyn, C. R. Acad. Sci., Paris 189, 55, 1929.
13. Rossi, Phys. Zs. 33, 304, 1932; Acad. Lincei 15, 734, 1932.
14. Johnson and Street, Phys. Rev. 40, 635, 1932.
15. Anderson, Science 76, 238, 1932.
16. Bethe, Zs. Physik 76, 293, 1932.
17. Blackett, Proc. Roy. Soc. A. 135, 132, 1932.
18. Skobelsyn, C. R. Acad. Sci., Paris 195, 315, 1932.
19. Hoffman, Phys. Zs. 17, 633, 1932; Steinke and Schindler, Zs. Physik 75, 115, 1932; Naturwiss. 26, 491, 1932. Messerschmidt, Zs. Physik 78, 668, 1932.
20. Heisenberg, Ann. d. Physik 13, 430, 1932; Naturwiss. 21, 365, 1932.
21. Heisenberg, Zs. Physik 77, 1, 1932; Iwanenko, Phys. Zs. Soviet Union 1, 820, 1932; Blandel, Phys. Zs. Soviet Union 2, 286, 1932; Perrin, C. R. Acad. Sci., Paris 195, 236, 1932.
22. Bohr, Report of Congress in Rome, 1931; J. Chem. Soc., 349, 1932.
23. Chadwick, Proc. Roy. Soc. A. 136, 693, 1932.
24. C. T. R. Wilson, Proc. Camb. Phil. Soc. 22, 534, 1925; Proc. Phys. Soc. 31, 3D, 1925.
25. Dirac, Proc. Roy. Soc. A. 126, 360, 1930; A. 133, 60, 1931.
26. Weyl, Gruppentheorie und Quantenmechanik, 2nd ed., 234, 1931.
27. Dirac, Proc. Camb. Phil. Soc. 26, 361, 1930.
28. Rutherford, Chadwick and Ellis, Radiation from Radioactive Substances, 43.
29. Gray and Tarrant, Proc. Roy. Soc. A. 136, 662, 1932; Meitner and Hupfield, Naturwiss. 19, 775, 1931; Chao, Phys. Rev. 36, 1510, 1931.

In “UFN” there is a review of these works in the article by M. P. Bronstein, see “UFN,” XII, 1932.

1. Curie et Joliot, Exposés de Physique théorique, 21, 1933.

EXPLANATIONS OF THE PHOTOGRAPHS

All photographs were taken in the space of a magnetic field directed from the camera lens normally to the glass plates of the vertical Wilson chamber (Fig. 2). Consequently, the particle, flying upward, is deflected to the right if charged positively, and to the left if charged negatively. Observing from the side of the camera, there are two possibilities: either to see a positive particle describing a path counterclockwise, or a negative one—clockwise; this is what is recorded in the photographs.

Here the photographs are reproduced so as to give the dimensions of the tracks in the chamber on a scale of 0.66 of the true scale. In all cases the chamber was filled with oxygen; the initial pressure was 1.7 atm*. Between the chamber lid and bottom an electric field was applied with a gradient from 3 to 4 \(V\,cm^{-1}\). The expansion was usually carried out within the range from 1.29 to 1.31 times.

For photography two cameras were used: \(A\)—with its optical axis coinciding with the direction of the magnetic field, and \(B\)—with its axis making a first angle of \(20^\circ\) with it. Of all the original photographs, the first four (Figs. 3, 4, 5, and 6) are stereoscopic pairs. All the others are separate pictures, obtained either in \((A)\) or in \((B)\); the choice from each pair was made depending on the better distinctness of the details.

The counters were usually in positions \(B_1\) and \(B_2\) (Fig. 2), but sometimes the lower counter was moved to \(B_2\) and \(B'_2\). Of all the photographs presented, only two (Figs. 12 and 15) are exposed when the lower counter is in \(B'_2\). The arrangement \((B_1, B_2)\) gives pictures more abundantly covered with tracks, while \((B_1, B'_2)\), on the other hand, permits photographing straight vertical tracks of particles that have not penetrated the plates. The angle between the optical axes of the cameras \((20^\circ)\) is too large for viewing the photographs in a stereoscope and was chosen so as to increase the accuracy of work in reconstructing the tracks in space. By this method, stereoscopic models of the tracks in natural size were made from wire and putty.

In the descriptions of the photographs the following abbreviations have been introduced: \(H\rho = H\rho\) gauss-cm—the product of the magnetic-field intensity by the radius of curvature of the track; \(E_e\)—the energy, equivalent to \(H\rho\) and calculated under the assumption that the mass of the deflected particle is equal to the mass of an electron;

\(MV = 10^6 V\); \((A)\) or \((B)\) denotes a picture taken by camera \(A\) or \(B\).

In some photographs attention is drawn to distortions and blurring of the tracks, occurring because of the motion of masses of gas during expansion (Fig. 10 and Fig. 13 on the right side and Fig. 17 in the center). Such distortion diminishes the accuracy of the observations. However, in general, it may be considered that even a curvature of the order of 600 cm can be discerned when metal plates are present in the chamber, and a curvature of about 200 cm when there are no particles there.

Figs. 3 \((B)\) and 4 \((A)\). A pair of photographs containing about 23 separate tracks. Lead plate, thickness 4 mm. \(H = 2200\) gauss.

A group of tracks in the upper part of the chamber diverges downward and somewhat forward from some region in the copper of the solenoid. A second clearly distinguishable group is located on the right side of the photograph.

A large part of the tracks passed through the lead plate, but there are so many of them that it is difficult to identify each one separately in the two images. Various secondary processes, scattering, etc., take place in the plate.

* The indicated mean scale of the reproduced photographs. In contrast to the English original, where the scale was exactly maintained, comparing the checks attached to the reproductions with the originals gave the following scale values: 0.59 for Fig. 3, 0.60 for Fig. 4, 0.67 for Fig. 5, 0.66 for Figs. 6–11, and 0.67 for Figs. 12–17. Transl. note.

Most of the tracks are almost straight, corresponding to electron energies greater than 100 MV. But in the middle of the upper part of the chamber there are 2 tracks curved to the left, with \(H\rho \simeq 2\cdot 10^5\) and \(E_e \simeq 1\) MV. Almost beyond any doubt these tracks belong to electrons.

There are also 2 tracks, clearly curved to the right, with \(H\rho \simeq 0.7\cdot 10^5\) and \(0.5\cdot 10^5\), i.e., the particles which traced them had energies \(E_e \simeq 20\) and 15 MV. Since these tracks are situated above, having first made one common tangent with the other tracks, they must have been traced by particles with positive charges. And since the ionization along them does not appear different from the ionization along electron tracks, they must belong to particles with mass comparable to the electron mass. The white cluster is probably due in its origin to some still more penetrating particle (a random \(\alpha\)-particle) that passed through the chamber before expansion.

The broad white band on the left side in photograph (B), Fig. 3, is the result of reflection of the illuminating glass cylinder in a drop. It is very difficult to avoid these highlights, but this was managed in photograph (A), Fig. 4.

Fig. 5 (B) and 6 (A). A pair of photographs containing about 16 separate tracks.
\(H = 3100\) gauss.

The center of origin of the shower is somewhat behind—in the solenoid winding. On the left are 2 tracks of negative electrons with \(H\rho \simeq 0.4\cdot 10^5\) and \(1.5\cdot 10^5\) and with energies \(E_e \simeq 12\) and 45 MV.

Some of the remaining tracks are slightly curved—some in one direction, others in the opposite. Most of the almost straight tracks diverge from one common point, but the more strongly curved tracks apparently diverge below from some secondary center of radiation, located lower down.

Fig. 7 (B). \(H = 2200\) gauss. Lead plate, 4 mm thick.

One particle, having too great an energy for the curvature of its path to be measured (\(H\rho > 3\cdot 10^5;\ E_e > 100\) MV), passes through the lead plate and knocks out from it a secondary particle outward (?). Below the plate one track should be considered straight, and the other curved in the direction corresponding to a positive electron with \(E_e \simeq 60\) MV.

Fig. 8 (A). \(H = 2200\) gauss. Lead plate, 4 mm thick.

Two particles, one of which penetrates the plate, traversing in the plate a path almost straight, i.e. having \(E_e > 100\) MV, after it is deflected in the negative direction, already having an energy \(E_e\) of about 30 MV. It is obvious that the energy loss is too great for it to be justified by normal absorption. Thus, Anderson found for normal absorption an energy loss of the order of 85 MV per 1 cm path in lead.

Fig. 9 (B). \(H = 2100\) gauss. Lead plate, 4 mm thick.

A track recording the deflection of a particle by the plate; the curvature of the path is positive, and greater below than above. The direction of motion is, consequently, downward, and therefore the charge is positive. Above the plate \(E_e \simeq 60\) MV, below it \(E_e \simeq 22\) MV. The energy loss is rather large for normal absorption; but it is quite natural if the energy is lost owing to the act of collision that also causes the deflection.

Fig. 10 (A). \(H = 2200\) gauss. Lead plate, 4 mm thick.

A shower consisting of approximately eight tracks penetrates the roof of the chamber; the center of radiation is apparently in the copper solenoid. In the lead plate a secondary center of radiation appears, from which a group, apparently of six tracks, diverges. By detailed study of them by the method of stereoscopic reprojection, although not with complete certainty, it can be established that not one of the tracks forming the primary shower passes through the center of origin of the secondary one. If this is really so, we have a fairly clear indication that the secondary shower in this case arose under the action of some non-ionizing agent forming part of the primary shower. Under the plate one track is curved negatively; the others are almost straight. Above the plate some tracks became blurred and disappeared during expansion.

Fig. 11 (B). \(H = 2200\) gauss.

A shower of four particles enters obliquely into the top of the chamber.

Three more nearly straight tracks diverge from the center in the glass lid. On the left, somewhat above the middle line, there is a distinct track with negative

curvature (to the left), and the last—the second track, lying in one plane with the first and curved to the right—is probably a positive electron. These two tracks do not emerge from the point from which the other three go, but from a point somewhat adjacent to the center of emission.

The spiral track is a negative electron with an energy of about 130 thousand V; it may be due, by its origin, to the photoelectric absorption of a quantum.

We leave several other, disorderly scattered tracks without explanation.

Fig. 12 (A). \(H = 700\) gauss. Copper plate, 6 mm thick.

Seven tracks diverge not from a single point in the glass window of the chamber. The curved track emerges exactly from the same point as the other six; it is very close to reality that all the particles are born at one point. The sign of the curvature indicates a positive charge with \(E_\rho \approx 120\) thousand V.

This track corresponds to the positive electron with the lowest energy observed so far.

There are many other disorderly tracks.

Fig. 13 (A). \(H = 2200\) gauss. Lead plate, 4 mm thick.

From some point inside the lead there diverge downward 2 negative electrons with energies of 13 and 10 MV and 2 straight tracks—upward.

With respect to the latter two, two suppositions may be made: a) either one of them occurred owing to the motion of a particle from above, and the other is a track of a particle thrown from the same point in the opposite direction (from below upward), b) or both of them are the result of the motion of particles upward, i.e. in such a case the splitting process will have to be ascribed to the action of some agent which, in its motion, does not itself produce ionization. In neither supposition is the presence denied of at least one track left by a particle possessing a higher energy and moving upward. According to hypothesis (a), one of the particles may be regarded as an element of a primary shower going to the chamber from above; two other members of this shower also fell on the photograph (from the side). Of interest is the secondary electron, \(E_e \approx 5 \cdot 10^4 V\), arising in the gas. As is seen, such an electron ionizes 2–3 times more strongly than a fast electron. The contrast is striking.

Fig. 14. \(H = 2200\) gauss. Lead plate, 4 mm thick.

A shower consisting of four tracks was formed apparently at the beginning, near the solenoid winding. One of the tracks after the plate was deflected, while in another a secondary track with negative curvature appeared.

Fig. 15. \(H = 2200\) gauss.

A complex collection of curved paths; three have negative and three positive curvature, not counting the still short track below. The emitting region is apparently located in the glass of the lid, in which 5 negative and 2 positive electrons diverge. True, it cannot be asserted with certainty, with respect to one of the tracks with positive curvature, that it was not drawn by a negative particle moving upward.

Fig. 16. \(H = 3\) thousand gauss.

Tracks of two protons and two electrons.

The thick horizontal track has \(H\rho \simeq 3 \cdot 10^5\), and, if it is taken, as is assumed, to belong to a charged particle that has entered from the right, it belongs to a proton. At such a value of \(H\rho\), the proton must have a range length \(l\) of about 26 cm (in air at normal pressure and temperature); it must ionize approximately 100 times more strongly than a fast electron. This track cannot be attributed to an \(\alpha\)-particle, since its range would be only 0.6 cm (Fig. 3 and 4). The thin track with negative curvature, distinguishable in depth on the left side, lies in one plane with the long proton track. However, if one assumes that they came from one and the same point, then doubts arise because of the presence, near the bottom, of a slow secondary particle, which appears to be a positive particle moving upward.

Fig. 17. \(H = 2200\) gauss.

Copper plate, 6 mm thick.

In all probability, this photograph should be read as follows. The incident ray is a particle splitting the copper nucleus with the emission of two electrons and one heavier particle: either a proton or an \(\alpha\)-particle. The energy of one

curvature (to the left), and in the middle another track, lying in the same plane as the first ones and curved to the right—probably a positive electron. These two tracks do not emerge from the point from which the other three proceed, but from some neighboring center of radiation.

The spiral track is a negative electron with an energy of about 130 thousand V; it may owe its origin to the photoelectric absorption of a quantum.

Several other, irregularly scattered tracks we leave out of consideration.

Fig. 22 [[unclear: caption text largely illegible]] X-rays [[unclear: several words]] plate of metal 3 mm thick.

Here are visible various [[unclear: tracks/rows]], but not [[unclear: several words]] with the conditions of the experiment [[unclear: several words]]. The curvatures of the tracks emerge exactly from the same point, which is very close to reality, so that all the particles were born at one point. The sign of the curvature indicates a positive charge with \(H\rho \geq 1.5 \times 10^5\ \mathrm{V}\).

This kind of photograph [[unclear: several words]] a positive electron with considerable probability [[unclear: remainder of line illegible]].

[[unclear: final italic line illegible]]

electron of about 12 and another of about 14 MV. Unfortunately, the tracks were strongly distorted near the plate owing to vortices in the gas. One more track apparently belongs to a particle thrown directly backward, but the distortion is too great to make this certain. The following tracks in three further photographs from Fig. 17 probably belong to particles [[unclear: distorted phrase]]. It is quite probable that such particles actually also arise in the phenomena shown in Fig. 13 and in many others. Only they did not possess sufficient velocity and range to escape from the plate. One may conceive of the case in which a proton emerges from the plate and, on leaving it, has just such an energy that, without going beyond the limits of the chamber, it is compelled to stop in the gas. The probability of this phenomenon is negligible; nevertheless, Fig. 17 records this exceedingly rare case.