New Insights into Cosmic Rays¹

R. A. Millikan and G. H. Cameron.

Cosmic radiation is understood to mean that small part of “penetrating radiation” which is of cosmic origin. The chief purpose of this article is to give a preliminary report on the very latest work, which sheds new light on the properties of these unusual rays and shows that there exist rays still harder than those discovered before—namely, rays capable of penetrating 190 feet of water or about 16.7 feet (5 m) of lead before being completely absorbed.

Since quite recently—only last summer—some leading physicists expressed doubt as to the very existence of any rays whatever having a definitely cosmic origin, and since up to the present time certain observers deserving of unquestioned confidence, such as, for example, Swann² and Hoffmann³,

¹ This article is, with some changes, an address delivered by Millikan at the meeting of the British Association in Leeds on September 2, 1927. Nature, Suppl., Jan. 7, 1928, p. 19.

² Swann (Phys. Rev., 29, 372, 1927) finds that the ionization produced by such rays on the summit of Pike’s Peak amounts to 0.75 ions per cm³ per second, whereas we found at the same place a number close to 5 ions.

³ Hoffmann (Ann. d. Phys., 82, 413, 1927) finds that, on an assumption based on the discovery of Kolhörster (Kolhörster.

Kohlhörster and we, in estimating the intensity of cosmic radiation, if it exists, sometimes differ by a factor of 8–10; therefore our first task will be to give a very brief account of the arguments that existed at the time of these experiments, and then to see how these arguments are supplemented by new data.

This order of presentation has the further advantage that it will serve as an excellent illustration of that slow, step-by-step developing process by means of which science moves forward. In this process each experimenter bases himself on the past and, if he succeeds, makes, in comparison with his predecessors, a small movement forward. In this way mankind, in its understanding of nature, has at last felt itself at the summit of power, while at the same time not being conscious of those separate moments in which the beginnings of this development are hidden. From the time of Greek mythology down to the present, very few such discoveries have been made which appeared in full growth in the mind of a single person.

EARLIER WORKS ON COSMIC RAYS

The beginning of the study of strongly “penetrating radiations” near the earth’s surface was laid in 1903, when these rays were discovered and so named by McLennan2, Rutherford3, and their collaborators, who found that the rate of discharge of an electroscope could be considerably reduced if the electroscope was surrounded by a whole series of lead screens many centimeters thick; thus it was shown that the rays existing in the atmosphere possess the capacity to—

…by the ability to penetrate through such thick screens, in accordance with which they were named “penetrating radiation.”

The next important step was made by the Swiss physicist Gockel2, who in 1910 was the first to ascend with a shielded electroscope in a balloon to an altitude of about 4,500 m and found, contrary to expectation, that at this altitude the radiation is considerably stronger than at the earth’s surface; this fact at once compelled recognition that, if not all the radiation, then at least a part of it has an extraterrestrial origin, and that it reaches the earth’s atmosphere from above; this idea had been expressed by Richardson3 as early as 1906.

During the following four years Hess4 in Austria and Kolhörster5 in Germany made new flights similar to Gockel’s flight, verified his results, and devoted greater attention to the quantitative side of the matter. Kolhörster made observations up to an ascent height of 9,000 m and found that the rate of discharge decreases slowly up to approximately 1,000 m and then increases, reaching at an altitude of 9,000 m a value seven times that observed at the earth’s surface, or, more precisely, 80 ions more than at the surface, as Kolhörster indicates in his paper.

The war temporarily halted further investigations, but in the autumn of 1921 and the spring of 1922 Millikan and Bowen6 made the next step, launching balloons with self-recording electroscopes; an altitude of about 15,500 m was reached, i.e., more than nine tenths of the distance to the upper layers of the atmosphere, if this height is estimated by the amount of air remaining below. These flights made it possible to verify the results obtained by European ob—

observers, and showed that up to this altitude the rate of discharge increases, although the values of this rate obtained in the new observations proved to be much smaller than those calculated from the former observations up to an altitude of 9,000 m; this shows that the “penetrating rays,” if they come from somewhere above, are in fact far more penetrating than had hitherto been supposed. But if the rays come from above, then the ionization inside a hermetically sealed electroscope must increase according to an exponential law, i.e., in a geometric progression, as one approaches the upper layer of the atmosphere; and these ascents to great altitude had, and still have, especially great significance. They give a quite definite and reliable idea of the upper limit of the coefficient of absorption of the rays entering the atmosphere, provided only that such rays actually exist.

However, the fact that the total discharge of the electroscopes on these flights amounted to only about one quarter of what could have been expected if one accepted the absorption coefficients calculated, on the basis of the hypothesis of cosmic rays, from the data of Hess and Kolhörster—this fact indicates that the cause of the phenomenon is somewhat different. Until very recently, the increasing rate of discharge with altitude was the only phenomenon on which the hypothesis concerning the cosmic origin of the rays rested. Other assumptions, however, were also possible, and indeed were put forward; it was supposed, for example, that certain radioactive particles of unknown origin are distributed in the upper regions of the atmosphere. Such an assumption could have been tested directly by means of direct measurements of the absorption coefficient of the penetrating rays, rather than by attempting to calculate these coefficients, as had been done earlier, on the basis of the assumption that the rays enter the atmosphere from above. If the rays were of radioactive origin, they could hardly be appreciably harder than the rays of known radioactive substances, such as, for example, thorium \(D\) or radium \(C\).

The next step was taken in the summer of 1923, when Kolhörster2 in Europe and Millikan and Otis3 in America, independently of one another, made the first direct measurements of absorption with materials other than atmospheric air—the former in Alpine glaciers and in shallow bodies of water lying at sea level, and the latter in thick lead screens on the summit of Pikes Peak—all with the aim of shedding light on the possible origin of the penetrating rays.

As a result of his experiments in the glaciers, Kolhörster finds the absorption coefficient to be equal to 0.25 per meter of water, or approximately half the value previously found, i.e. 0.55, thereby eliminating the discrepancy between the values found during his flights and in our balloon experiments. In his article devoted to the description of this work, he states that his experiments definitely indicate the existence of γ-rays with an absorption coefficient amounting to about one-tenth of the absorption coefficient of the hardest known γ-rays (4.1 per 1 m of water4), but he speaks with great caution about their origin. After considering various possibilities, he says that “recently one has inclined more and more to the view that the penetrating rays constitute a phenomenon whose origin should be sought in cosmic space”5.

On the other hand, Millikan and Otis, on the basis of their new data on absorption on Pikes Peak, concluded that if any of the penetrating rays detected on the Peak had a cosmic origin, they would have had to be more penetrating or

less intense than would correspond even to the reduced values found by Kolhörster, namely 2 ions at sea level and an absorption coefficient of 0.25 per 1 m of water. The mean radiation coefficient that they found on Pike’s Peak turned out to be only slightly smaller than in theory \(D\), and the greater part of the radiation was probably of local origin. In these experiments they found new evidence for the existence of rays of cosmic origin. Indeed, up to 1925, in no country, as far as can be judged from the literature, was there certainty that the existence of rays of cosmic origin had been proved. The increase of ionization in shielded vessels almost up to 15 km was an undoubted fact, so that Kolhörster’s experiments in glaciers were favorable for a cosmic interpretation; however, the possibilities of contamination of glaciers by radioactive substances are by no means small, and moreover the very irregular form of glaciers, as well as their proximity to the earth, were not suitable for precise work with measurements of absorption coefficients.

Further, Hoffmann\(^{1}\) in Germany, with his exceptionally delicate technique, spoke out in 1925 against the existence of rays of cosmic origin. Swann\(^{2}\) in America likewise advanced the assertion that the work of himself and his collaborators, concerning ionization in vessels at pressures up to 75 atm., is incompatible with a cosmic interpretation of the penetrating radiation.

Observations in mountain lakes.

However, in 1925 Millikan and Cameron obtained, from their point of view, indisputable proof that the penetrating radiation is of cosmic origin. Indeed, this radiation proved to be weaker and more

\(^{1}\) Hoffmann, Phys. ZS, 26, 40, 669 (1925).

\(^{2}\) Downey, Phys. Rev., 20, 186 (1922). Fruth, Phys. Rev., 22, 109 (1923).

penetrating than followed from the previous estimates: the ionizing power at sea level turns out to be equal to only 1.4 ions in \(1 \text{ cm}^3\) per sec., and the absorption coefficient becomes equal to only 0.18 per \(1 \text{ m}\) of water; this radiation has a definite spectral distribution, the greatest wavelengths, calculated by the formula of Compton (A. H. Compton), lying at about \(\lambda = 0.00063 \ \mathring{A}\), and the smallest at about \(\lambda = 0.00038 \ \mathring{A}\). The latter is only one-thirtieth of the wavelength of the hardest \(\gamma\)-rays.

These experiments consisted in immersing insulated electroscopes in deep lakes situated at great altitudes and surrounded by snow; in one particular case, in Lake Muir (altitude 3590 m), it was found that the ionization gradually diminishes with depth, from 13.3 ions in \(1 \text{ cm}^3\) per sec. at the surface to 3.6 ions at a depth of 60 feet (18 m); below this point the sensitivity of the instruments then used did not make it possible to trace any further decrease of the ionization. Thus here, for the first time, a zero was definitely obtained on an electroscope, i.e. the influence of all external radiations was completely excluded, and the results, consequently, began to show that it would be quite possible to determine with certainty the absolute value of the penetrating radiation.

This experiment, in the part of it which we have so far described, proves either the existence at the surface of the lake of penetrating radiation of such hardness that it proves capable of passing to a depth of 18 m before being completely absorbed, or else some other very strange distribution of radioactivity in the lake water.

Soon after this, when we had to make the same observations at another depth of a snow lake situated 450 km to the south and 2060 m lower, we obtained the same curve, with only the difference that each reading corresponded to a shift—

lower exactly by 6 feet upward. But 6 feet of water, in its absorptive capacity, provided that the law of absorption by masses holds, turns out to be exactly equivalent to the layer of atmosphere lying between the heights of 3590 m and 1530 m.

This experiment, supplemented by other similar discoveries, thus definitely proves three propositions.

The first is that the effect on Lake Muir is not dependent on any radioactivity distributed in the water in some special way. The second is that the source of the rays is not at all located in the layer of atmosphere between the two heights, for this layer acts as an absorbing medium having exactly the absorption it ought to have in the case where the rays pass wholly upward. And finally the third is that in different localities, separated from one another by 450 km, the rays act exactly the same at the same heights.

These facts, together with further observations made before and after this, which showed that, within the limits of our experimental errors, the rays arrive here in equal measure from all directions of the sky, and supplemented finally by the circumstance that the observed coefficient of absorption and the total ionization under the action of cosmic rays at the height of Lake Muir satisfactorily predict the results obtained during the balloon flight to a height of 15.5 km,—all this constitutes excellent and indisputable proof that these rays do not originate in our atmosphere, at any rate not in the layers below nine-tenths of it, and therefore may with full justification be called “cosmic rays”—the most descriptive and most appropriate name devised for that part of the rays which reach us from the sky. We shall see how indisputable this proof will appear after our new results have been presented.

These results were obtained from two groups of experiments, one of which was carried out in the Andean mountains in Bolivia at altitudes of 4600 m in the autumn of 1926, and the other—in the lakes Arrowhead and Gem in California in the summer of 1927.

Penetrating Radiation in the Andean Mountains.

The experiments in the Andean mountains pursued four main aims: 1) To see whether the curves of the dependence of ionization on altitude, obtained in the lakes of the southern and northern hemispheres, would coincide with one another. These curves were especially sensitive in very high-lying lakes accessible in the Andean mountains, so that the spectral distribution could be investigated more thoroughly than in 1925. If the curves of the northern and southern hemispheres coincided, then very important data would thereby be obtained against the supposition of the possibility of the penetrating rays arising under the action of fast β-rays upon the very upper layer of our atmosphere, i.e. that only hypothesis which sees the source of these rays in the last tenth part of the air surrounding the earth. For such β-rays one might expect the influence of the earth’s magnetic field, so that over the poles there should be stronger radiation than over the equator. At 17° south latitude we would be completely protected from such a polar influence, especially if we could place ourselves in suitable basins in the high mountains. 2) To subject to verification the hypothesis of Wilson (C. T. R. Wilson), according to which these rays arise owing to the action upon the earth’s atmosphere of electrons possessing velocities of many millions of volts and produced during thunderstorms. Lakes situated in corresponding basins in the Andean mountains would be fully protected from such influences. Likewise, a comparison of rays observed in thunderstorm localities with rays observed in extensive regions such as California, which are comparatively free from thunderstorm phenomena, could furnish verifying data on this point. 3) In order, when determining,

as was indicated above, from the zero readings of the new electroscopes, to obtain new data for checking the value of ionization obtained by us, which is produced by cosmic rays at sea level; this value still varies within wide limits in the results of different observers. 4) To place ourselves in the corresponding basins or valleys in very high mountains, where the rays are three or four times more intense than at sea level, and to carry out there more reliable experiments on the direct action of cosmic rays—in particular, to see whether the Milky Way is more or less effective in sending these rays to the earth than other parts of the sky.

To all these four points we obtained, despite the mishaps with two electroscopes, a satisfactory and definite answer.

As regards (1), on the surface of Lake Titicaca (alt. 3,822 m) we obtained data very much in agreement with the results obtained on Lake Muir in California. Likewise, on Lake Migvilla near Caracoles in Bolivia (alt. 4,500 m), we obtained data which satisfactorily fit the extrapolated curve for Lake Muir. Thus, if on the curves of the dependence of ionization on altitude there are any differences depending on geographical position, they remain beyond the limits of our present observational technique.

As regards (2), Lake Migvilla is a small lake, surrounded on all sides by mountains several thousand feet high. It is apparently quite well shielded from rays having their origin in thunderstorms anywhere on earth. Further, far from the shores of Central America, we carried out a long series of observations in the radiotelegraph room on the ship on one of the nights when lightning flashes were occurring along the shore; we compared these results with observations made on the Californian coast, which is almost entirely free from thunderstorms, and in doing so detected not the slightest differences.

Thus Wilson’s hypothesis is quite definitely ruled out.

By (3) we determined the zero positions of two of our electroscopes, lowering them to a sufficient depth, and then, on board ship, made careful observations at sea level on all routes from Mollendo and Peru to Los Angeles. We found in the readings at sea level no changes whatever with geographic position, and detected only insignificant differences between the ionizations in the different instruments, although their walls were of different material and their volumes were in the ratio of almost \(1:2\). The mean magnitude of the ionization at sea level, thus directly observed, proved to be only a few tenths of an ion above the mean magnitude of the ionization at sea level due to cosmic rays, obtained by means of the two curves of our preceding report. These values were: \(1.4\) for electroscope No. 1, \(1.6\) for electroscope No. 3, and \(1.5\)—the mean value, which is therefore approximately verified, though still inexact (see below), since the ionization due to radioactive substances in the air over the ocean must be very small. The chief uncertainty in this value \(1.5\) for the ionization at sea level lay in the determination of the capacitances of the electroscopes and in the unknown influence of the walls of the electroscopes. We shall give an account of this latter influence below.

As regards (4), we undertook two series of long observations, each of which lasted three days, at an altitude of 15,400 feet (\(4\,620\ \text{m}\)) in a deep valley, from which the Milky Way was visible for 5–6 hours and then disappeared from view for the next 6 hours. The intensity of the cosmic rays that entered our electroscopes in this valley was equal to \(3.6\). We were unable to discover any difference at all in our readings either when the Milky Way was overhead or when it was not visible. Our error in the mean values of these readings could scarcely have been more than \(0.1\) ion.

Even in the event that we had made twice as large an error, we could ultimately have concluded that the Milky Way does not exert upon cosmic rays an influence of the kind that our instruments could at present detect. And this means that the rays coming from the Milky Way, in their intensity, will not differ by more than 6% from the intensity of the rays coming from that part of the sky which is at right angles to the Milky Way. This is in agreement with our earlier, less accurate measurements, and also with the recent very careful work at sea level by Hoffmann and Steinke2, who found no influence at all of direction upon cosmic rays; but it does not agree with the results reported by Büttner3 and Kolhörster4.

Be that as it may, the present work was carried out under the most favorable conditions possible. It seemed extremely important to obtain indisputable data concerning the direction of cosmic rays, since no reliable conclusion could be drawn about the origin of the rays until the place from which the rays emanate had been found. Up to now this place has not been established, so that in the near future it was necessary to carry out more sensitive experiments on this question.

Observations in mountain lakes in California

The subject of a new series of experiments in Arrowhead and Gem lakes, begun at the beginning of 1927, was the repetition of the experiments with still more sensitive electroscopes and an increase in the accuracy of determining the constants of the electroscopes, so that greater precision could be introduced into the work with cosmic rays and the whole problem placed on a more rigorous quantitative basis.

As was indicated earlier, the various observers were still far from knowing the absolute magnitude of the ionization, although a considerable group among us believes that the ionization at sea level lies between one and two ions. However, this can hardly be called quantitative agreement. Indeed, it could not have been expected, since no observers, except us, have so far been able to determine the zero position of their instruments, so that all ionization values, except ours, may be regarded rather as estimates than as measurements. Further, in our own values an error of 10% was possible—and perhaps even less—in determining the capacities of the electroscopes.

As for the mean absorption coefficients, there is now known agreement between Kolhörster’s values and ours, but, besides us, no one has found the inhomogeneity of the rays, although the latest results of Hoffmann and Steinke confirm our discoveries and the supposition that in the mixture of cosmic rays there may exist rays even harder than those found by us. We found that the latter have an absorption coefficient equal to 0.18 per 1 m of water, which corresponds—if Compton’s equations are used for the calculation—to a wavelength of \(0.00038\ \text{\AA}\) or to a potential of \(32\,600\,000\ \mathrm{V}\). Hoffmann2, in order to explain his latest observations at sea level, admits the existence of components corresponding to a wavelength, calculated in the same way, of \(0.00029\ \text{\AA}\) or to a potential of \(41\,000\,000\ \mathrm{V}\).

In order to determine more accurately the intensities of cosmic rays and to study more distinctly their spectral distribution, in the autumn of 1926 we began to build new electroscopes with much greater sensitivity to cosmic rays; in particular, we wished to carry out tests for the existence of rays still harder than those which could be detected by our instru-

ments of the former sensitivity; on theoretical grounds we could suspect the existence of still harder rays. These electroscopes will be described in detail elsewhere. It will suffice to say here that we can now measure the capacities of our electroscopes down to several thousandths (the number 0.791 electrostatic units is the capacity of the instrument with which we obtained the results given below), and that the sensitivity of the electroscopes to cosmic rays has now been increased eightfold in comparison with the electroscopes we have hitherto used; thus, for example, at sea level we obtained with our electroscope 11 ions produced by cosmic rays instead of 1.4, and on Lake Muir—40 instead of 5.

In experiments with such an electroscope, which we conducted last summer on Lake Gem, the ionization at the surface of the lake was 33.6 per 1 cm³ per sec. and decreased with depth regularly and quite smoothly to the zero value 2.6. But this asymptotic value of the ionization curve as a function of depth was reached only at a depth of 50 m instead of the 16.2 m that had been obtained in our previous experiments on Mount Arrowhead in 1925. This should not be understood as a discrepancy between the two groups of results; it means only that the ordinates of the ionization curve have now increased by a factor of 8 owing to the increase in sensitivity. Nevertheless, the points expressing the dependence of ionization on depth and obtained with the aid of the new electroscope fit the curve much better, i.e. are less scattered, than was the case previously.

Thus, owing to the improvement of the technique, the actual sensitivity proved to have been increased considerably more than eightfold. Consequently, only thanks to this increased sensitivity and accuracy of measurement has the ionization in the layer of water between 16 m and 57 m, which was formerly hidden from observation, now become clearly noticeable.

If one takes into account the absorption of rays by the atmosphere situated above Lake Gem and equivalent to 7.45 m of water, then it may be said that the new experiments have revealed rays possessing the ability to penetrate through 57 m of water or 5 m of lead before being completely absorbed.

The new curve could be analyzed for the study of the spectral distribution much more reliably than in the preceding case. In doing so, we were able to state with great satisfaction that, as a result of analysis by the former method, the part of the radiation investigated at the altitude of Arrowhead has exactly the same coefficient as was obtained in the same region from the preceding curve, namely 0.23 per 1 m of water; the lower part of the curve, however, leads to a coefficient of 0.1 per 1 m of water, so that by this very fact we have discovered rays almost twice as penetrating as those which had been found by us earlier. Calculating as before, we find that the shortest wavelength will now be 0.00021 Å, and the equivalent accelerating potential—59,000,000 V, which considerably exceeds the estimate made by Hoffmann.

Our general curve now extends from the absorption coefficient $\mu = 0.25$ per 1 m of water to $\mu = 0.1$, or, in equivalent wavelengths, calculated as before, from 0.00053 Å to 0.00021 Å, which corresponds to an interval of from one to two octaves. If the calculations are made according to Dirac’s formula2, which is probably more reliable than Compton’s formula, the relative values do not change, but the absolute frequencies or energies increase by 30%.

The ionization by cosmic rays at sea level in this electroscope, reduced to atmospheric pressure, is exactly the same

which was published for electroscope No. 1, namely close to 1.4 ions per cm³ per sec., and the error that may be present here for this electroscope will be less than 1%.

Source of Cosmic Rays.

What, then, can now be said concerning the possible source of these extraordinary rays? Their penetrating power alone, or their frequency, calculated by whatever formula, obviously requires that the rays be connected with changes taking place inside the nucleus itself, since no charges situated outside the nucleus can be associated with energies of such magnitude. The simplest hypothesis will be the one to which we drew attention in the 1925 article, namely, that these rays arise through direct collisions between atomic nuclei and high-speed electrons. It is true that, in the case of light atoms, the simple potential energy of separation of an electron from the nucleus without a greater disappearance of mass than has hitherto been assumed will be insufficient to create rays of such hardness.

It must be taken into account that the strongest of the rays discovered earlier correspond very closely to the change of energy—the loss of mass—accompanying the union of four hydrogen atoms into one helium atom; but the new measurements give rays with practically twice as much energy as this, and only with \(1/15\) of the energy that could be obtained in the complete transformation into radiation of the energy of separation of positive and negative electrons, so that, apparently, there are no direct experimental grounds for supposing that the annihilation of mass occurs in this latter manner.

If, nevertheless, there are widely distributed processes in the universe in which electrons prove to be endowed with energies of many millions of volts (as such a process Wilson names thunderstorm ...

phenomena on the earth), then it will not be difficult to imagine that collisions of such electrons with the nuclei of atoms, however light they may be, will act as a source of the observed rays. We shall not try here to enumerate the possible causes of the appearance of electrons with such high velocities; but if we postulate their existence, then the existence of cosmic rays immediately follows from it as well. In fact, from the best spectroscopic observations astronomers now believe that interstellar space is populated by at least one atom in every cubic inch. Moreover, from the ionized state of the calcium atoms that have been found in interstellar space, Eddington2, by determining the kinetic energy of the atoms and electrons present in this space, estimates the temperature of space at 15,000° C, i.e. a value is obtained greater than the surface temperature of the sun and of most stars.

No radiations of the type considered here, if they arise inside stars, can likewise be imagined, since they obey the law of absorption by mass and, according to the preceding experiments, are wholly absorbed in a layer of water 190 feet thick. Rays must pass through this thickness many times before they reach the outer shell of the star. A mass equivalent to such a thickness will be penetrated by rays that have traversed, in any direction, a distance equal to \(10^9\) light-years, on the assumption that space is strewn with atoms at a density of one per cubic inch. If these considerations are correct, this shows that the directions toward the sun and the stars probably do not differ from other directions as sources of these rays, and all observers agree that, at least in the case of the sun, its direction will be indifferent.

If, however, one postulates electrons with high velocities as the source of these rays, then why not suppose that these high-velocity electrons themselves also

are these rays? How can the existence of short ether waves be admitted at all? The answer will be that, as we know from experience, in the photoelectric process an exchange of energy takes place between ether waves and electrons without loss of it. Equally, according to Ellis’s data, these quantum laws will hold in exactly the same way inside the nucleus as in the external structure of the atom. On the contrary, the transformation of energy into heat takes place chiefly when the energy is associated with a moving electron. In other words, the coefficient of absorption of β-rays will be 100 and more times higher than the coefficient of absorption of ether waves having the very same energy. Thus electrons set in motion by the Compton process, or in some other way, very quickly dissipate the energy imparted to them, and therefore the transfer of energy through space in concentrated form (which corresponds to high penetrating power) must occur when it is in the form of ether waves.

Consequently, it is entirely useless to suppose that the source of the observed rays is the bombardment of air in the outer tenth of the atmosphere by electrons of high velocity and of the same maximum energy. If electrons with such great velocities could enter the upper layer of our atmosphere in sufficient quantity, then they would indeed produce just such rays as we observe; but the difficulty lies in how these electrons could reach us from space.

Thus it is rather ether waves, with their high penetrating power, than electrons with high velocities and comparatively small penetrating power, that could reach our atmosphere in sufficient quantity. This objection would be invalid only in the case when electrons with high velocities arose comparatively close to us, for example, on the Sun or on nearby stars, and, consequently, the distance would be insufficiently great for their energy to be dissipated in the form of heat or transformed into more penetrating rays.

But this latter case is precisely the very case that was excluded, since we could not detect any noticeable influence of the sun on the intensity of this radiation.

Or, if the direction toward the sun differed from any other direction in the sky as a direction in which high-speed β-rays were present in abundance, then, necessarily, it would also be the direction in which cosmic rays would arise in anomalous abundance. Thus, according to this theory of the origin of cosmic rays, it is quite impossible to confine the source of these rays to the upper tenth part of our atmosphere or to astronomically nearby regions.

Cosmic origin.

The preceding considerations concerning the distance from the earth at which the rays originate apply with still greater force to any hypothesis that assumes either a spontaneous or an externally induced change of the nucleus, not connected with electron impacts; for matter in the upper tenth part of our atmosphere, according to any such hypothesis, would have to be endowed with properties wholly unlike those possessed by the matter immediately surrounding us.

But as long as matter in distant regions of the universe can be imagined as endowed with such properties, which are not found on earth, it would be a violation of the principle of minimum hypotheses to assume that the thin ring of matter situated just beyond the limits of the matter immediately surrounding us has properties possessed by matter neither more remote nor nearer. Consequently, we see no possible way to ascribe to the rays any origin other than a cosmic one; but if the Milky Way, as a source of rays, does not differ from other parts of the sky (and our experiments, in fact, proved unable to detect such a difference), then the rays must emanate from regions, dis-

located beyond the Milky Way, i.e., either from spiral nebulae, if they are uniformly distributed over the sky, or from star clusters.

From the results of our experiments of 1927, we can now calculate, for the first time with a sufficient degree of confidence, the total energy per \(1\ \mathrm{cm}^{2}\) per sec. that falls on the upper part of the earth’s atmosphere in the form of cosmic rays. It comes to \(3.1 \cdot 10^{-4}\ \mathrm{erg}\ \mathrm{cm}^{-2}\ \mathrm{sec}^{-1}\), or precisely one tenth of the total energy falling on the earth’s atmosphere in the form of visible and thermal radiation from the stars.

1. ZS. f. Phys., 36, 147, 1926), according to whom the absorption coefficient is a constant quantity and \(\mu_{H2O}=22\cdot10^{-3}\ \text{cm}^{-1}\), and the ionization at sea level is 0.29 in \(\text{cm}^3\) per second. On the other hand, we (Phys. Rev., 28, 851, 1926) found that the absorption coefficient is a variable quantity (i.e., that the rays are heterogeneous) and that the ionization at sea level is 1.4 ions. ↩

2. Eddington, “Stars and Atoms,” p. 69 (Oxford Press, 1927). ↩↩↩↩↩↩↩

3. Büttner, ZS. f. Geophys., 2, 190 (1926). ↩↩↩↩

4. Kolhörster, Naturwissenschaften, 14, 936 (1926). ↩↩↩

5. “Neuerdings neigt man immer der Ansicht zu, die Höhenstrahlung als eine Erscheinung aufzufassen, deren Ursprung im Kosmos zu suchen ist.” And further: “Da für die erstere Auffassung der Höhenstrahlung als einer aus den höheren Atmosphärenschichten stammenden bisher keinerlei direkte Andeutung gefunden wurde, so sprechen die augenblicklichen Verhältnisse mehr zugunsten einer kosmischen Erklärung.” ↩↩

6. Millikan and Bowen, Phys. Rev., 22, 198 (1923) and 27, 353 (1926). ↩