The Nuclear Structure of the Atom

E. Rutherford.

INTRODUCTION.

The conception of the nuclear structure of the atom originally arose from attempts to explain the scattering of $\alpha$-particles through large angles when passing through thin layers of matter1. Since $\alpha$-particles possess great mass and high velocity, these considerable deflections were most remarkable; they indicated the existence of very intense electric or magnetic fields within atoms. In order to explain these results it was necessary to suppose2 that the atom consists of a charged massive nucleus of very small dimensions in comparison with the ordinarily accepted value of the atomic diameter. This positively charged nucleus contains the greater part of the mass of the atom and is surrounded, at some distance, by negatively charged electrons distributed in some known manner, their number being equal to the total positive charge of the nucleus. Under such conditions there must exist, near the nucleus, a very intense electric field, and $\alpha$-particles, when encountering individual atoms and passing close to the nucleus, are deflected through considerable angles. Assuming that the electric forces vary inversely as the square of the distance in the region adjoining the nucleus, the author obtained a relation connecting the number of $\alpha$-particles scattered through a given angle with the charge of the nucleus and the energy of the $\alpha$-particle. Under the influence of a central force field, the $\alpha$-particle describes a hyperbolic orbit about the nucleus, and the magnitude of the deflection depends on the degree of its approach to the latter. From the data on the scattering of $\alpha$-particles obtained in this way, the conclusion was drawn that the total charge of the nucleus is approximately equal to $\frac{1}{2} A e$, where $A$ is the atomic weight and $e$ is the elementary quantity of electricity. Geiger and Marsden3 carried out a series of careful experiments to test the validity of this theory and confirmed the greater part of its conclusions. They found that the nuclear

the charge is equal to $\frac{1}{2}Ae$, but owing to the difficulty of the experiment it was possible to determine its magnitude only within 20%. C. G. Darwin1 developed a complete theory of the deflection of the $\alpha$-particle and the nucleus, taking into account the mass of the latter, and showed that the experiments of Geiger and Marsden can be reconciled only with a law of central force inversely proportional to the square of the distance. Thus the nuclear structure of the atom found strong confirmation in the experiments on the scattering of $\alpha$-rays.

Since the atom is electrically neutral, the number of external electrons surrounding the nucleus must be equal to the number of units of positive electricity2 contained in the total charge of the nucleus. It should be noted that, on the basis of the study of the scattering of X-rays by light elements, Barkla3 showed in 1911 that the number of electrons should be approximately equal to half the atomic weight. This was derived from J. J. Thomson’s theory of scattering, in which it is assumed that each external electron in the atom acts as an independent scattering unit.

Thus two entirely different methods gave identical results concerning the number of external electrons in the atom. However, the study of the scattering of $\alpha$-rays showed, in addition, that the positive charge must be concentrated in a massive nucleus of small dimensions. Van den Broek4 expressed the idea that the scattering of $\alpha$-particles by the atom does not contradict the possibility that the charge of the nucleus is equal to the atomic number of the atom, i.e. to the number of the place occupied by the given atom when the elements are arranged in order of increasing atomic weights. The importance of the atomic number for characterizing the properties of the atom was shown by Moseley’s remarkable work5 on the X-ray spectra of the elements. Moseley showed that the frequency of the vibrations corresponding to the lines in the X-ray spectra of the elements depends on the square of a number which changes by one in passing successively from element to element.

This relation can be explained by assuming that the nuclear charge changes by one in passing from atom to atom, and that it is numerically equal to the atomic number. Incidentally, I must emphasize that the great significance of Moseley’s work lies not only in determining the number of possible elements and the position of unknown elements, but also in proving that the properties of atoms are determined by a number which changes by one in passing from atom to atom. This gives a new method for studying the periodic system of ele—

elements, for the atomic number, or the equal nuclear charge, is of greater importance than the atomic weight. In Moseley’s work the frequency of oscillation of the atom is not exactly proportional to \(N\), where \(N\) is the atomic number, but is proportional to \((N-a)^2\), where \(a\) is a constant depending on which of the series of characteristic radiations of the elements (the \(K\) or \(L\) series) is being measured. It is assumed that this constant depends on the number and position of the electrons near the nucleus.

Nuclear charge.

The question of whether the atomic number of an element is indeed a measure of its nuclear charge is so important that every possible method must be applied to resolve it. At present several investigations are being carried out in the Cavendish Laboratory with the aim of testing the accuracy of this relation. The two most direct methods are based on the study of the scattering of fast \(\alpha\)- and \(\beta\)-rays. The first method is being applied by Chadwick, using new techniques; the latter by Crowther. The results obtained so far by Chadwick fully confirm the equality of the atomic number with the nuclear charge within the possible accuracy of the experiment, which in Chadwick’s case is about \(1\%\).

Thus it is clear that we have firm grounds for asserting that the nuclear charge is numerically equal to the atomic number of the element. These results, when compared with the results of Moseley’s work, indicate that the inverse-square law is fulfilled with sufficient accuracy in the region surrounding the nucleus. It is of the highest interest to find the dimensions of this region, since these dimensions will give us definite information about the distances of the inner electrons from the nucleus. From this point of view, a comparison of the scattering of fast and slow \(\beta\)-rays should provide important evidence. The agreement of experiment with theory for the scattering of \(\alpha\)-rays between \(5^\circ\) and \(150^\circ\) indicates that the inverse-square law is obeyed exactly in the case of heavy elements, such as gold, for distances approximately between \(36 \cdot 10^{-12}\) and \(3 \cdot 10^{-12}\) cm from the nucleus. Hence we may conclude that if electrons exist in this region at all, they are present only in small numbers.

An \(\alpha\)-particle in a direct collision with a gold atom (nuclear charge 79) is repelled back at a distance of \(3 \cdot 10^{-12}\) cm. This indicates that the nucleus may be regarded as a point charge even at such small distances. Until faster \(\alpha\)-particles are at our disposal, we cannot, in the case of heavy elements, push the question of the size of the nucleus any further. However, we shall see later that in the case of light atoms, where \(\alpha\)-particles can approach closer to the nucleus, we have greater hope of resolving this question.

It is extremely important to emphasize the great significance of nuclear—

nuclear charge for the characterization of the physical and chemical properties of an element, for, obviously, the number and distribution of the outer electrons, on which most physical and chemical properties depend, are determined by the total charge of the nucleus. It should be anticipated theoretically, and this is confirmed by experiment, that the true mass of the nucleus has only an insignificant influence on the arrangement of the outer electrons and on their vibrations.

Thus it is quite possible to conceive the existence of elements with exactly identical physical and chemical properties, but with different masses. Indeed, for one and the same nuclear charge a certain number of different stable combinations of the units from which the complex nucleus is built is possible1. Thus the dependence of the properties of an atom on its nuclear charge, and not on its mass, gives a rational explanation for the existence of isotopes, whose physical and chemical properties may be completely indistinguishable, while their masses may vary within certain limits. We shall consider this important question in still greater detail later, in the light of new data on the nature of the units from which the nucleus is built.

Thus the problem of the structure of the atom naturally falls into two parts:

1) The structure of the nucleus itself.
2) The distribution and modes of vibration of the outer electrons.

I do not intend today to enter into consideration of the second point, since this is a very extensive subject, on which different opinions may exist. This aspect of the problem was first studied by Bohr and Nicholson, and they made a very significant step forward. Sommerfeld and others have with great success applied Bohr’s general method to explaining the fine structure of spectral lines and those complex vibrations which simple atoms perform in the Stark phenomenon. Recently Langmuir and others have studied the problem of the distribution of the outer electrons from the chemical point of view. They emphasized the importance of admitting a more or less cubical arrangement of the electrons in the atom. There is no doubt that each of these theories has its own definite sphere of useful application; however, our information is still too scanty to reconcile with one another the obvious differences in these theories.

I intend today to discuss in some detail experiments carried out with the aim of shedding light on the structure and stability of the nuclei of certain simple atoms. From the study of radioactivity we know that the nuclei of radioactive elements consist in part of helium nuclei with charge \(2e\). We also have very solid grounds for thinking that

Translator.

the nuclei of atoms contain, along with positively charged particles, also electrons, and that the positive charge of the nucleus is the excess of the total positive charge over the negative. It is interesting to note the very different role played by electrons in the outer layers of the atom and within it. In the first case the electrons are situated by themselves at a certain distance from the nucleus, determined, without doubt, chiefly by the charge of the nucleus and by the interaction of their own fields. Inside the nucleus the electrons form a very close and strong combination with the positively charged units. Since, as we know, precisely outside the nucleus electrons are not in stable equilibrium. While the outer electrons, without doubt, interact with the nucleus as point charges, this is not the case for the electrons inside the nucleus. It must be assumed that, under the influence of the enormous forces within the latter, the electrons are strongly deformed, and that the forces here may be of an entirely different character in comparison with those which may be expected from an undeformed electron outside the atom.

In nuclear theory it has usually been assumed that electric forces and charges play the most important role in determining the structure of the inner and outer parts of the atom. A considerable success of this theory in explaining the principal phenomena is the confirmation of the general validity of this point of view. At the same time, if the electrons and the particles composing the nucleus are in motion, then magnetic fields must arise, which should be taken into account in any complete theory of the atom. In this respect magnetic fields must be regarded rather as a secondary than as a primary factor, notwithstanding the fact that these fields may play an essential role in the conditions of equilibrium of the atom.

Dimensions of the Nucleus.

We have seen that, in the case of atoms with a large nuclear charge, even the fastest \(\alpha\)-particles cannot penetrate into the actual structure of the nucleus, so that we can only estimate the maximum dimensions of the latter. However, in the case of light atoms, when the nuclear charge is small, under a central impact with an \(\alpha\)-particle it approaches the nucleus so closely that we can estimate its dimensions and form some notion of the forces developing here. This is seen best of all in the case of a central impact between an \(\alpha\)-particle and a hydrogen atom. In this case the H-atom is set into such rapid motion that it traverses a path four times greater than the \(\alpha\)-particle colliding with it, and can be detected by the scintillations caused by it on a zinc-sulfide screen1. The author2 showed that these scintillations are caused by hydrogen atoms carrying one positive charge and flying with a velocity which—

which could have been expected from the simple theory of impact, i.e. with a speed 1.6 times greater than the speed of the $\alpha$-particles1. The relation between the number and the speed of these H-atoms is quite different from what could have been expected if the $\alpha$-particle and the H-atom were regarded as point charges for the given distances. As a result of collisions with fast $\alpha$-particles, H-atoms are obtained which possess velocities differing little from one another and which are carried approximately in the direction of the incident $\alpha$-particles. From this it was inferred that the law of inverse proportionality to the square of the distance becomes invalid when the nuclei approach one another to a distance less than $3.10^{-13}$ cm. This may serve as an indication that the nuclei have a size of this order of magnitude, and that the forces between nuclei change very rapidly in magnitude and direction when they approach to distances comparable with the usually accepted diameter of the electron. It was pointed out that in such close collisions between nuclei enormous forces are developed, and that, probably, during the collision the structure of the nuclei undergoes considerable deformation. The fact that the helium nucleus, which may be supposed to consist of four H-atoms and two electrons, withstands this collision is an indication of its highly stable structure. Similar results were observed in collisions between $\alpha$-particles and atoms of nitrogen and oxygen. Here too the atoms that received acceleration appeared to be carried mainly in the direction of the $\alpha$-particles, while the region in which special forces developed had the same order of magnitude as in the case of the collision of an $\alpha$-particle with hydrogen.

There can be no doubt that the space occupied by the nucleus, and the distances at which the force becomes abnormal, increase together

Continued.

with the complexity of the structure of the nucleus. It is to be expected that the nucleus of H must be the simplest of all, and, if it is a positive electron, it must have exceedingly small dimensions in comparison with negative electrons. In collisions between \(\alpha\)-particles and H-atoms, the \(\alpha\)-particle should be regarded as a more complex structure.

The diameter of the nucleus of light atoms, with the exception of hydrogen, is probably of the order of magnitude \(5 \cdot 10^{-13}\) cm, and in a close collision the nuclei come into contact and, perhaps, even penetrate one another. Under such conditions it may be expected that only very stable nuclei will withstand collisions, and thus it is of the highest interest to investigate the possibility of the disintegration of the nucleus.

Particles with a long range, obtained in nitrogen.

In earlier papers, loc. cit., I described the phenomena occurring in close collisions of fast \(\alpha\)-particles with the light atoms of a substance, in order to determine whether the nuclei of certain light atoms can undergo decomposition under the influence of the enormous forces developed in such close collisions. In these papers evidence was given that, when \(\alpha\)-particles pass through dry nitrogen, fast particles arise which, in brightness of scintillations and in range of penetration, very much resemble hydrogen atoms set in motion under the influence of collision with \(\alpha\)-particles. It was further shown that these fast atoms, which appear only in dry nitrogen, but not in oxygen or in carbonic acid, cannot be ascribed to the presence of water vapor or of another substance containing hydrogen, but that they must arise in the collision of \(\alpha\)-particles with nitrogen atoms. The number of such scintillations obtained in nitrogen was small (approximately \(1/2\) of the corresponding number in hydrogen), but it was two or three times greater than the number of natural1 scintillations of the source. The number of scintillations observed in nitrogen was, on average, equal to the number of scintillations that were observed when hydrogen at a pressure of approximately 6 cm was added to oxygen or carbonic acid at normal pressure.

Meanwhile, while the general indications pointed to the fact that these atoms with a long range, arising in nitrogen, are charged hydrogen atoms, the first attempts to determine the mass of these particles by deflection in a strong magnetic field did not give definite results.

On the basis of the data reported in my first work, one can

Transl.

to construct several theories concerning these particles. The calculated range of an atom with one charge, set in motion by collision with \(\alpha\)-particles having a range of \(R\) cm in air, has the following values

Owing to the small number and the weakness of the scintillations under the conditions of these experiments, the range of the fast atoms obtained from nitrogen cannot be measured with sufficient accuracy to make a definite choice among the indicated possibilities. Some of my correspondents have pointed out the probability that these fast atoms are the original \(\alpha\)-particles, which have lost one of their two charges in passing through nitrogen, i.e. atoms with charge 1 and mass 4. However, there are no sufficiently clear reasons why, of all the elements investigated, nitrogen should be the only one in which the passage of fast \(\alpha\)-particles leads to the capture of one electron.

However, if the conditions of the experiment were such that a sufficient number of scintillations could be obtained, then, in essence, it would not be difficult to choose one of these various possibilities by studying the deflection of the fast atoms in a magnetic field. The magnitude of the deflection of charged atoms by a magnetic field perpendicular to the direction of their flight is proportional to \(\frac{e}{mv}\). Assuming that these particles are liberated in a central collision of an \(\alpha\)-particle, it is easy to calculate the relative values of this quantity for different masses. If the value \(\frac{E}{MV}\) for the \(\alpha\)-particle is taken as unity, then the values of \(\frac{e}{mv}\) for atoms with one charge and masses 1, 2, 3, 4 will be respectively 1.25, 0.75, 0.58, 0.50; consequently \(H\)-atoms should be deflected more than the \(\alpha\)-particles under whose influence they arise, whereas atoms with mass 2 or 4 should be deflected less than the original \(\alpha\)-particles.

Upon my arrival in Cambridge I approached this problem in various ways. By selecting an objective with a large aperture, the brightness of the scintillations was increased, and thus counting became easier. Along with this, experiments were carried out with the aim of obtaining a more powerful source of radiation, using the radium that was at my disposal. However, in the end it proved, for reasons which we shall not discuss here, that it is best of all to obtain an active source of radiation in the form of radium C by the method described in

of my first work.1 After a certain number of observations with solid nitrogen compounds—they will be described below—a method was developed for estimating the mass of particles arising from nitrogen in the gaseous state. The use of gas for this purpose had several advantages over the use of solid nitrogen compounds, for not only was the number of scintillations greater, but one could also be certain of the absence of hydrogen or hydrogen compounds.

The arrangement finally adopted is shown in Fig. 1. The chief

Fig. 1.

feature consisted in the use of a wide slit through which the α-particles passed. Experience showed that the ratio of the number of scintillations on the screen arising in the gas to the number of natural scintillations of the source increased rapidly with increasing slit width. For plates placed one mm. apart, this ratio was less than unity, whereas for plates placed 8 mm. apart it had a value of from two to three. Such a change was to be expected theoretically, since the majority of the particles are liberated at an angle to the direction of flight of the α-particles.

The horizontal plates \(A\) and \(B\) were 6.0 cm long and 1.5 cm wide and were placed 8 mm apart. Near one end of them was placed the source \(C\) with an active deposit of radium; near the other, a zinc sulphide screen. The support for the source and the slit were arranged in a rectangular brass box, through which a current of dry air or another gas was continuously passed in order to avoid radioactive contamination. The box was placed between the poles of a large electromagnet,

Trans.

so that the uniform field ran parallel to the plane of the plates and perpendicular to their length. A tip \(D\), 1.2 cm long, was placed between the screen and the end of the plates in order to increase the magnitude of the deflection of the rays emerging from the slit. A screen of cerous zinc \(S\) was placed on a glass plate closing the box at one end; the distance between the source and the screen was 7.4 cm. Oxygen or nitrogen atoms that had been accelerated and had a range of 9 cm could be stopped by an aluminum screen with a stopping power equivalent to 2 cm of air; this screen was placed at the end of the plates \(AB\).

With such a slit width it was not possible to deflect the whole broad beam to the side. The deflection of the radiation emerging from close to the edge of the slit was measured. For this purpose it was important to observe scintillations at a definite point of the screen near \(M\). The method of fixing the microscope was as follows: the source \(C\) was set in its place, and the air was pumped out to a pressure of several cm. Without a field the edge of the beam was determined by the straight line \(PM\), intersecting the screen at \(M\). The microscope was arranged so that the boundary line of the scintillations appeared above the horizontal thread of the microscope.

When the magnet deflecting the rays downward (the so-called \(+\) field) was energized, the path of the extreme \(\alpha\)-particles was represented by the curve \(PLNR\), intersecting the screen at \(N\), so that the boundary line of the scintillations appeared in the field of view shifted downward. When the direction of the field was reversed (the so-called \(-\) field), the path of the limiting \(\alpha\)-particle \(PQRT\) intersected the screen at \(T\), and the band of scintillations appeared shifted upward. The strength of the magnetic field was chosen so that, with the negative field, the scintillations appeared scattered over the whole screen, while with the positive field their boundary was situated directly beneath the horizontal thread. The appearance of the field of view in the microscope in these two cases is shown in Fig. 2, where the dots represent, approximately, the density of distribution of the scintillations. The horizontal boundaries of the field of view were produced by the rectangular aperture of a plate placed in the plane of the crosshairs. The horizontal thread, which crossed the field of view, was visible under the conditions of counting and made it possible, if necessary, to determine the relative number of scintillations of the two halves of the field. Since the number of scintillations under the actual conditions of the experiment with nitrogen was too small for the boundaries of the scintillation region to be directly noted, it was necessary, in order to estimate the deflection of the rays, to determine the ratio of the number of scintillations with the \(+\) field and with the \(-\) field.

Fig. 2.

The position of the microscope and the strength of the magnetic field in most

the experiments were chosen so that this ratio was approximately \(1/3\). Preliminary observations showed that it is sensitive to changes of the field; thus a suitable method is obtained for estimating the relative deflection of the rays.

After the position of the microscope had been fixed, a stream of dry air was begun to be drawn continuously through the apparatus. The absorbing screen was set in position \(E\), in order to stop the atoms \(N\) and \(O\) with a range of 9 cm. Then the number of scintillations was systematically counted for the two directions of the field, and, if necessary, a correction was introduced for the weak radioactive contamination of the screen. The deflection of the unknown rays was compared directly with the deflection of known \(\alpha\)-rays. For this purpose the source and the absorbing screen were removed, and in place of the source there was put an identical plate, covered with a small amount of active deposit of thorium. The \(\alpha\)-particles of thorium \(C\), with a range of 8.6 cm, after traversing a path of 7.4 cm in air, produced bright scintillations on the screen. The ratio of the number of scintillations for \(+\)- and \(-\)-fields was determined as before.

An example of such a comparison is given below. With a current of 4 amperes in the winding of the electromagnet the above-mentioned ratio for the particles from nitrogen was 0.33. The corresponding ratio for the \(\alpha\)-particles of thorium \(C\) was 0.44—with a current of 4 amperes in the winding of the electromagnet, and 0.31—with a current of 5 amperes. Hence it is evident that, on the average, the particles emitted from nitrogen are deflected in the given field more than the \(\alpha\)-particles of thorium \(C\). However, in order to make a quantitative comparison, it is necessary to take into account the diminution of the velocity of the rays in passing through air. The quantity \(\frac{mv}{e}\) for the \(\alpha\)-rays of thorium \(C\) with a range of 8.6 cm is known: it is equal to \(4.38 \cdot 10^5\). Since the rays, before reaching the screen, traverse 7.4 cm in a uniform field in air, the actual deflection corresponds to \(\alpha\)-rays in vacuum for which \(\frac{mv}{e}\) is approximately equal to \(3.7 \cdot 10^5\). If it is assumed that the deflection of the \(\alpha\)-particles at a current of 4.8 amperes is the same as for the nitrogen particles at a current of 4 amperes (ratio of fields 1.17), then it follows that the mean deflection of the nitrogen particles under the given experimental conditions corresponds to rays in vacuum for which the quantity \(\frac{mv}{e}\) is equal to \(3.1 \cdot 10^5\).

If one takes into consideration that the particles arise throughout the whole volume of gas between the plates, that their distribution is unknown, and also that the particles are emitted on the average at an angle to the incident \(\alpha\)-particles, then the experimental data are insufficient to calculate the mean value of \(\frac{mv}{e}\) for any mass whatever of the emitted particles. Apparently, the majority of the particles that produce scintillations arise in the first few centimeters of air adjacent to the source. The actual deflec-

by the given particle in a magnetic field must depend on the distance of the place of its origin from the source. These factors evidently tend to diminish the mean deflection of the particles in comparison with the value that this deflection would have if the particles were emitted with constant velocity from the source itself. Assuming that the correction for the diminution of the velocity of particles with a long range during their passage through air is 10%, we find that the mean value of $\frac{mu}{e}$ is approximately equal to $3.4 \cdot 10^5$. Since the value of $\frac{M V}{E}$ for $\alpha$-particles of radium $C$ is equal to $3.98 \cdot 10^5$, then, under the given experimental conditions, the mean value of $\frac{mu}{e}$ for the nitrogen particles is less than for the $\alpha$-particles which produce them.

From the data given earlier it follows that this can be the case only if the mass of the particles is comparable with the mass of the hydrogen atom, for particles with one charge and mass 2, 3, or 4 must undergo a smaller deflection than $\alpha$-particles. Thus the results of the experiments give very strong evidence that the particles liberated from nitrogen are hydrogen atoms.

However, a much more decisive proof of this can be obtained by comparing the deflection of the nitrogen particles with the deflections of $H$-atoms under identical conditions. For this purpose a mixture of approximately one volume of hydrogen with two volumes of carbon dioxide was collected in the gasometer, and this mixture was drawn through the apparatus instead of air. The proportion of the mixture of the two gases was chosen so that the stopping power of the mixture for $\alpha$-rays was the same as in air. Under these conditions $H$-atoms, like the nitrogen particles, arose throughout the whole volume of the gas, and, probably, the relative distribution of the atoms along the path of the $\alpha$-rays did not differ very much from the distribution of the nitrogen particles. If the nitrogen particles are $H$-atoms, then we should expect that their mean deflection will be approximately the very same as that of the $H$-atoms liberated from the given mixture. A series of careful experiments showed that the ratios of the number of scintillations in $+$ and $-$ fields of equal magnitude in both these cases were so close to one another that it was impossible by experiment to distinguish the two cases from each other. Since in both cases the experiments were carried out under conditions as nearly identical as possible, the equality of the ratios shows that the particles with long range liberated from nitrogen are hydrogen atoms. The possibility that these particles have mass 2, 3, or 4 is definitely excluded.

In a previous work I showed that the particles with long range observed in dry air and in pure nitrogen must arise from the nitrogen atoms themselves. Thus it is clear that some nitrogen atoms are destroyed in collisions with swift $\alpha$-particles, and that swift atoms of positively charged hydrogen are thereby produced.

From this it must be concluded that the charged atom of hydrogen is one of the components of the nitrogen nucleus.

At the time when it had long been known that helium is the product of the continuous transformation of certain radioactive elements, the question of the possibility of destroying the structure of stable atoms by artificial means was in an uncertain state. Here, for the first time, proof was obtained that hydrogen is one of the components of the nitrogen nucleus. It should be borne in mind that the total amount of nitrogen decomposed under the influence of \(\alpha\)-particles is extremely small. Indeed, on average only one \(\alpha\)-particle out of approximately 300,000 can come sufficiently close to the nitrogen nucleus to liberate a hydrogen atom with energy sufficient for it to be detected by scintillations. Even if all the \(\alpha\)-radiation of one gram of radium were absorbed by gaseous nitrogen, the volume of the liberated hydrogen would amount to only \(1/300{,}000\) of the helium formed by its \(\alpha\)-particles, i.e. it would be equal to \(5 \cdot 10^{-4}\) cubic mm per year. It is possible that, upon collision with an \(\alpha\)-particle, hydrogen is capable of being liberated without sufficient velocity for it to be detected by scintillations. If this proves to be the case, then the total amount of decomposed substance may be considerably greater than the value given above.

Experiments with solid compounds of nitrogen.

I shall now briefly report on experiments with solid compounds of nitrogen. Since the liberation of particles from nitrogen is a purely atomic phenomenon, it should be expected that similar particles must also be liberated from nitrogen compounds in an amount proportional to their nitrogen content. To test this and to study the nature of these particles, certain compounds rich in nitrogen were investigated. I used the following substances, prepared with every possible care so as to exclude the presence of hydrogen in any form whatever:

1) boron nitride, kindly prepared for me by W. J. Schuff at the University of Manchester;

2) sodium nitride, titanium nitride, and paracyanogen, kindly prepared for me by Sir William Pope and his assistants.

The apparatus used in these experiments was similar to that shown in Fig. 1, with the only difference that the plates were 4 cm long. The material, in the form of a powder, was spread as uniformly as possible, with the aid of a fine sieve, over an aluminium plate of approximately 2 sq. cm in area. The weight of the aluminium plate was approximately 6 mg per sq. cm; usually from 4 to 5 mg of substance per sq. cm was used. The stopping power of the aluminium plate for \(\alpha\)-particles corresponded to approximately 3–4 cm of air, and the material was taken in such quantity that its stopping power ...

the capacity on the average was the same as for aluminum. In order that the substance should adhere to the plate, a layer of alcohol was first applied, then it was quickly sprinkled with the substance, and the plate was dried.

Experience showed that with such use of alcohol no appreciable contamination with hydrogen is introduced. The screen of gray zinc was placed outside the chamber, fitting closely against the aluminum plate with a stopping power equivalent to 5.2 cm of air, which closed the opening at the end of the brass box. The aluminum plate with the substance was then arranged so that it closed the end of the slit facing the source; precautions were taken not to shake off the substance. The air was pumped out, and the number of scintillations was counted for two positions of the plate:

1) when the substance was turned toward the source, and
2) when the plate was turned over and the substance was turned toward the slit.

In the first case the α-particles entered directly into the substance under investigation; in the latter case the α-particles fell on the substance only after their range had been reduced by approximately half and when their ability to liberate fast atoms had been greatly diminished. This method has the further great advantage that it eliminates the need to introduce a correction for the unequal absorption of the H-particles of the source in different experiments.

Thus it was found that all the nitrogen compounds investigated gave a larger number of scintillations in the first position. The nature of the particles that caused these scintillations was investigated by a method similar to that used in the case of nitrogen (gaseous), and a direct comparison was made between the deflection of these particles and the deflection of H-atoms liberated from a paraffin film placed in the position of the nitrogen compounds. In all experiments it turned out that the particles are deflected in the same way as H-atoms from paraffin, and no traces of particles with mass 2 or 4 could be detected.

For films with the same average stopping power for α-rays, it was easy to calculate, by Bragg’s rule, that the relative stopping power of nitrogen in the compounds would be 0.67 for BN, 0.74 for C₂N₂, 0.40 for titanium nitride; the stopping power for sodium nitride is here taken as unity. Since the ejection of particles with long range from nitrogen is an atomic phenomenon, it should be expected that the number of scintillations, after subtracting the correction for the natural scintillations of the source, should be proportional to the above relative values of the stopping power. Observations with sodium nitride and titanium nitride fully confirmed this, and the number of nitrogen particles with long range corresponded entirely to the reduced numbers and proved to be approximately such as could be expec—

to give from experiments with gaseous nitrogen. On the other hand, boron nitride and paradisian gave \(1 \tfrac{1}{2}\)–2 times as large a number of particles as could have been expected theoretically. In these experiments all precautions were taken to avoid the presence of hydrogen and water vapor. Before use, the aluminum plates were heated in a rolled quartz tube almost to the melting point, in order to drive off hydrogen and other gases. The films under investigation were kept in a desiccator and heated in an electric furnace before being transferred into the apparatus. Several control experiments were made with preparations not containing nitrogen, for example, with pure graphite and silicon, which were kindly prepared for me by Sir William Pope. In both cases the number of scintillations observed when the substance was turned toward the source of \(\alpha\)-rays was indeed smaller than in the case when the plate was reversed. This indicates that some quantity of H-atoms is liberated by the \(\alpha\)-rays from the heated aluminum. Thus the control experiments gave satisfactory results, since they showed that H-atoms did not appear in substances not containing nitrogen; at the same time they showed that H-atoms did not arise in any appreciable quantity from carbon, silicon, or oxygen.

The increased effect in boron nitride and paradisian naturally arouses the suspicion that these preparations contain some amount of hydrogen, although all precautions had been taken to avoid it. In the case of boron nitride it is still unknown whether boron itself emits H-atoms. This circumstance has not yet been investigated. In view of such uncertainty, experiments with solid nitrogen compounds were set aside for a time, and the experiments already described with gaseous nitrogen were carried out.

It is interesting to note that considerable contamination with hydrogen is needed in order to obtain the number of H-atoms that was observed in these compounds. In the case of sodium nitride, per gram of substance there would have to be at least 50 cubic cm of hydrogen. I am inclined to think that the H-atoms liberated by \(\alpha\)-rays from sodium nitride are due chiefly, if not entirely, to nitrogen, while in the case of paradisian part of the effect is probably caused by the presence of hydrogen or some compound of it. I hope subsequently to investigate this question in greater detail.

Atoms with a small range arising in oxygen and nitrogen.

Besides the H-atoms with large range liberated from nitrogen, when \(\alpha\)-particles pass through oxygen and through nitrogen there arise, in much larger number, fast atoms which have a range in air of approximately 9 cm, while the range of the incident \(\alpha\)-particles is 7 cm. The method of determining the range and the number of these

of atoms was set forth in one of the preceding papers1. In it it was shown that these atoms arise when α-particles pass through a gas. Just beyond the range of the α-particles of radium C the scintillations proved brighter than the scintillations caused by H-atoms and more like the scintillations caused by α-particles.

For lack of definite information about the nature of these atoms, it was provisionally assumed that they are oxygen or nitrogen atoms carrying one charge and set in rapid motion under the influence of a close collision with α-particles, since the observed range of these particles was in approximate agreement with the range calculated on these assumptions. At the same time it was pointed out that the coincidence of the ranges of O and N atoms is unexpected, for one should have predicted that the range of N-atoms would be 19% greater than that of O-atoms. That these fast atoms might be fragments of disintegrated atoms seemed possible even then, but until very recently I saw no method for posing this question2.

As soon as the use of a wide slit was successful in resolving the question of the nature of the particles with large range arising in nitrogen, experiments were made, with the same apparatus and by the same method, in order to investigate the nature of the particles with small range arising in O and N.

Let us first determine the relative deflection that may be expected for an O-atom set in motion by a central collision with α-particles. The velocity of the O-atom after the collision is equal to \(^{2}/_{5} V\), where \(V\) is the velocity of the incident α-particle. The quantity

\[ \frac{mv}{e} \]

for an O-atom carrying one charge, as is easy to see, is 3.1 times greater than for the α-particle before the collision. Consequently, an O-atom with one charge should be deflected with greater difficulty than an α-particle; the same will also hold in the case when the O-atom carries two charges.

To verify this, the same apparatus was used as that depicted in Fig. 1. The source was placed at a distance of 7.4 cm from the zinc-sulfide screen; tips 1.2 cm long were used, as before, to increase the deflection of the rays. In the case of oxygen, the scintillations observed on the screen were caused by O-atoms with a small proportion of H-atoms from the source itself. In the case of air, the scintillations on the screen were caused partly by N-atoms, partly by O-atoms, and partly by H-atoms from the source and the nitrogen. The actual number of N-atoms with short range appeared to be smaller than that of O-atoms under analogous conditions.

The position of the microscope was fixed, as before, in such a way as to obtain a suitable ratio for the number of scintillations when the direction of the magnetic field was reversed. This ratio varied with the position of the microscope and, in the actual experiments, fluctuated between 0.2 and 0.4.

First of all, it immediately became clear that the atoms arising in \(O\), instead of being deflected less than the \(\alpha\)-particles, as should have been the case if they were \(O\)-atoms, were deflected more. This circumstance at once excluded the possibility that here we were dealing with oxygen atoms carrying one or two charges. Since helium is emitted in many radioactive transformations, it could be expected that it is one of the components of the light atoms and is liberated in intense collisions. The deflection of the atoms arising from \(O\), however, was considerably greater than that which could have been interpreted in this way. In order to test this point in the conclusions from the experiments made with oxygen, the source of \(\alpha\)-particles (an active deposit of radium) was replaced by a plate which had been subjected to the action of thorium emanation, and the deflection of the \(\alpha\)-rays of thorium \(C\), with a range of 8.6 cm, was investigated. If the \(\alpha\)-particles were emitted from \(O\)-atoms close to the source, they would have had to be deflected like \(\alpha\)-particles with a range of 9 cm; if, however, they arose at the end of the path of the \(\alpha\)-rays, the magnitude of the deflection should not have exceeded that which would be experienced by \(\alpha\)-particles with a range of 7 cm, i.e. this magnitude should have been approximately 9% greater than in the first case. Even if one assumes that the particles were liberated uniformly along the path of the \(\alpha\)-rays and moved in exactly the same direction as the \(\alpha\)-particles incident upon them, the mean magnitude of the deflection should not differ by more than 5% from the mean magnitude of the deflection of the \(\alpha\)-particles of thorium \(C\). If, however, one takes into account that some of the atoms were liberated at an angle to the incident \(\alpha\)-particles, then the mean magnitude of the deflection of the beam should be still smaller and, in all probability, smaller than for the \(\alpha\)-particles of thorium \(C\). In reality, the observed deflection was approximately 20% greater, showing that the hypothesis according to which the atoms arising from \(O\) are charged helium atoms is completely incorrect.

If the atoms from \(O\) were \(H\)-atoms, they would be deflected more than the \(\alpha\)-particles; but then they would have to have a range of 28 cm instead of the 9 cm actually observed. Thus it is clear that these atoms must have a mass intermediate between 1 and 4; from the discussion of the magnitude of the range of these particles and of their deflection, it is clear that these atoms carry two units of charge.

In order to make a more decisive test, the deflection of atoms arising in \(O\) in a positive and in a negative field of a definite magnitude was directly compared with the deflection of \(H\)-atoms from a mixture of hydrogen with carbon dioxide, taken ...

in a ratio of one to two by volume. In order completely to absorb the O-atoms from \(CO_2\), an aluminum sheet was placed before the screen of zinc sulfide, so that the total absorption between the source and the screen would correspond to a value slightly greater than the absorption of 9 cm of air. In both experiments the atoms under investigation arose in the gas between the plates and, probably, their relative distribution along the path of the \(\alpha\)-rays did not differ appreciably in the two cases.

The sought ratios upon reversal of the direction of the field in both experiments proved to be approximately the same, but on the average, from several experiments, it became clear that the \(H\)-atoms are deflected somewhat more than the atoms arising in \(O\). From several experiments the conclusion was drawn that the difference of deflection, on the average, does not exceed \(5\%\), although from the character of the observations it was difficult to fix this difference with any certainty.

On the basis of these data and of the magnitude of the range of the atoms arising in \(O\) and in air, we can derive the mass of the particles liberated from oxygen.

Let \(m\) be the mass of the atom arising in \(O\),
\(u\)—its maximum velocity near the source,
\(E\)—its charge.

Let \(M, V, E\) be the corresponding quantities for incident \(\alpha\)-particles, and \(m', u', e\) the same quantities for \(H\)-atoms liberated near the source.

Taking into account that the velocity of particles arising from \(O\), with a range of 9 cm, is continuously diminished in passing through a layer of oxygen of 7.4 cm between the source and the screen, it is easy to calculate that the mean deflection of these particles is proportional to \(1.14\,\dfrac{E}{mu}\), and not to \(\dfrac{E}{mu}\), as is the case in a vacuum.

In the same way, the deflection of an \(H\)-atom is proportional to

\[ 1.05\,\frac{e}{m'u'}, \]

where in this case the correction for the change of velocity is smaller and is estimated at approximately \(5\%\). Further, we have seen that the experimental results show that atoms arising from \(O\) are deflected by \(5\%\) less than \(H\)-atoms; consequently

\[ 1.14\,\frac{E}{mu}=\frac{1.05}{1.05}\,\frac{e}{m'u'}=1.25\,\frac{E}{MV} \]

or

\[ 1.14MV=1.25mu, \tag{1} \]

since it had been calculated and confirmed experimentally that the deflection of \(H\)-atoms in a magnetic field is 1.25 times greater than the deflection of an \(\alpha\)-particle set in motion (see Art. II, loc. cit.). In exactly the same way in the oppo-

In the preceding paper (p. III) I indicated the reasons by virtue of which it must be considered that the range \(x\) of a mass \(m\), having initial velocity \(u\) and carrying a double charge, is expressed by the formula

\[ \frac{x}{R}=\frac{m}{M}\left(\frac{u}{V}\right)^3, \]

where \(R\) is the range of \(\alpha\)-particles with mass \(M\) and velocity \(V\). Since \(x=9.0\) cm for atoms arising in \(O\) and set in motion in collision with a radium \(C\) \(\alpha\)-particle with a range of 7 cm, then

\[ \frac{x}{R}=1.29, \]

and, putting \(M=4\), we obtain

\[ mu^3=5.16V^3. \tag{2} \]

A formula of this type was derived for the range of an \(H\)-atom, and everything suggests that it may be regarded as quite accurate in general for such a short interval of ranges.

From formulae (1) and (2) we obtain

\[ \begin{aligned} u&=1.19V,\\ m&=3.1. \end{aligned} \]

If one takes into account the difficulty of obtaining exact data, the value 3.1 indicates that the atom has a mass approximately 3, and this value should be considered probable in the further arguments.

When air was taken instead of oxygen, no difference could be detected in the deflection of atoms with a short range in the one case and in the other. Since atoms with a small range arising in air must, by their origin, be chiefly nitrogen, we may conclude that these atoms, liberated when \(\alpha\)-particles pass through nitrogen or oxygen, consist of atoms of mass 3, carrying a double charge and initially ejected with velocity \(1.19V\), where \(V\) is the velocity of the incident \(\alpha\)-particles1.

Apparently it is difficult to avoid the conclusion that these atoms of mass 3 are liberated from atoms of oxygen or nitrogen as a result of an intense collision with \(\alpha\)-particles. Thus it may be supposed that atoms of mass 3 are components of the nuclei of both oxygen and nitrogen. We have already shown earlier that hydrogen is also one of the components of the nitrogen nucleus. Thus it is clear that the nitrogen nucleus can be decomposed in two ways—either by knocking out—

Translator.

by the knocking-out of \(H\)-atoms, or by the expulsion of atoms with mass 3 and with two charges. Further, since these atoms of mass 3 are 5–10 times more numerous than the \(H\)-atoms, it appears that these two forms of disintegration are independent of one another and not simultaneous. Since collisions are very rare, it is extremely improbable that individual atoms should undergo both types of decomposition.

In view of the fact that the particles ejected from \(O\) and \(N\) arise not at the source itself, but along the path of the \(\alpha\)-particles, it is difficult to determine their mass and velocity with the desired accuracy. To circumvent this shortcoming, attempts were made to determine the deflection of \(O\)-atoms liberated from a mica plate placed behind the source. Owing to the presence of hydrogen in the mica, the \(H\)-atoms falling on the screen were so numerous in comparison with the \(O\)-particles, and their deflections under the given experimental conditions were so close, that it was difficult to distinguish these atoms from one another.

The Question of Energy.

In close collisions between an \(\alpha\)-particle and an atom, the laws of conservation of energy and momentum remain valid\(^1\). However, in those cases when the atoms disintegrate, we cannot necessarily expect these laws to hold unless we take into account the change in the energy and momentum of the atom as a result of its disintegration.

In the case when a hydrogen atom is ejected from the nitrogen nucleus, the available data are insufficient for judging this, for we do not know definitely either the velocity of the atoms or the velocity of the \(\alpha\)-particle after the collision.

If our assumption is correct, that atoms of mass 3 are emitted from \(O\) and \(N\) atoms, then it is easy to calculate that as a result of such a disintegration there should be a small gain in energy. In fact, if the mass of the ejected atom is exactly three, and its velocity is \(1.20\,V\), where \(V\) is the velocity of the incident \(\alpha\)-particle, then

\[ \frac{\text{energy of the liberated atom}}{\text{energy of the } \alpha\text{-particle}} = \frac{3 \cdot 1.44}{4} = 1.08 \]

Thus a gain in energy of motion of \(8\%\) is obtained, while we have taken no account at all of the subsequent motion of the disintegrated nucleus and of the \(\alpha\)-particle. This excess energy must be borrowed from the nucleus of nitrogen or oxygen, just as the energy of motion of the \(\alpha\)-particle is borrowed when it is liberated from a radioactive atom.

For the calculation let us consider a central impact between an \(\alpha\)-particle and an atom of mass 3. The velocity of the latter is equal to \(\frac{8}{7}V\), where \(V\) is the velocity of the \(\alpha\)-particle, and its energy amounts to 0.96 of the initial

\(^1\) Rutherford, Phil. Mag., 37, p. 562 (1919).

energy of the $\alpha$-particle. Undoubtedly, in the case of an actual collision with an $O$ or $N$ atom, in which an atom of mass 3 is liberated, the $\alpha$-particle, like the atom of mass 3, is already on its path under the influence of the field of the nucleus. Under such conditions one should expect that not only does the $\alpha$-particle give 0.96 of its energy to the liberated atom, but the latter also acquires an excess of energy, depending on the repelling field of the nucleus.

With our ignorance of the structure of the nucleus and of the nature of the forces in its immediate vicinity, it is undesirable to engage in speculations concerning the mechanism of the collision. However, further information in this field can be obtained by studying the paths of $\alpha$-particles through oxygen or nitrogen by means of the well-known method of C. T. R. Wilson. In a previous paper1 I analyzed Wilson’s photographs, where there is a sudden change in the direction of the path by $43^\circ$ and, in addition, a small branch in the form of a short spur. I showed that the relative lengths of the paths of the $\alpha$-particle and of the spur, roughly speaking, agree with the supposition that this spur was caused by an oxygen atom that received acceleration. This is quite probable, for the facts show that atoms of mass 3, after liberation, fly approximately in the direction of the $\alpha$-particle, and an oblique impact may not even cause the disintegration of the atom.

Recently Dr. Schimizu, in the Cavendish Laboratory, has developed a modification of Wilson’s apparatus in which expansions can be repeated periodically several times a second, so that the paths of several particles can be observed over a sufficient time. Under such conditions both Schimizu and I myself have seen how the forked paths of $\alpha$-particles look, where the lengths of both branches were commensurable with one another. Such observations directly by eye are not sufficiently definite to be treated with great confidence. Therefore Schimizu devised an arrangement for obtaining photographs, so that the paths can be examined in detail at leisure. In this way one may hope to obtain valuable information about the conditions that determine the disintegration of the atom, and about the relative energy imparted to the three systems participating in this disintegration, i.e. to the $\alpha$-particle, to the liberated atom, and to the residual nucleus.

For the present we have no definite information about the energy of the $\alpha$-particle necessary to cause disintegration, but general data indicate that fast $\alpha$-particles with a range of approximately 7 cm exert a greater action than $\alpha$-particles with a range of approximately 4 cm. This is not directly connected with the actual energy necessary to cause the disintegration of the atom itself, but is more likely connected with the impossibility for the slower $\alpha$-particles, under the influence of the repelling field, to approach sufficiently close to the nucleus to cause its disintegration. It is possible that the actual ener—

energy required for the disintegration of the atom is small in comparison with the energy of the \(\alpha\)-particle.

If this is so, then other agents with less energy than an \(\alpha\)-particle may also cause disintegration. For example, a fast electron may approach the nucleus with sufficient energy to cause its disintegration, for it moves in an attractive, and not a repulsive, field, as does an \(\alpha\)-particle. In the same way, penetrating \(\gamma\)-rays may impart sufficient energy to cause disintegration. It is therefore very important to verify whether oxygen, nitrogen, or other elements cannot be decomposed under the action of fast cathode rays obtained in a discharge tube. In the case of oxygen and nitrogen this may be checked simply by observing whether there appears a spectrum closely resembling the spectrum of helium after an intense bombardment of the corresponding substance by electrons. Similar experiments were begun by Mr. Ishida in the Cavendish Laboratory, all possible precautions being taken, in the form of heating the tube to a high temperature, the tube being made of special glass, and the electrodes, in order to remove with certainty the occluded helium which might originally have been present in the substance. Helium had previously been observed in discharge tubes by several investigators and was ascribed to occluded gases liberated by bombardment with cathode rays. Investigation of the actual origin of helium in these cases is extremely difficult. However, the latest improvements in the technique of discharge tubes promise to make it possible to give a definite answer to this important question.

Properties of the New Atom.

We have shown that atoms of mass approximately 3 and with two positive charges are liberated by \(\alpha\)-particles from both nitrogen and oxygen. It is natural to suppose that these atoms are independent units in the structure of both gases. This charged atom, in the course of its flight, is probably only the nucleus of a new atom without external electrons; therefore we may predict that if two negative electrons are added to this new atom, it should, in its physical and chemical properties, be quite identical with helium, but with mass 3 instead of 4. We must expect that the spectra of helium and of this isotope should be very close to one another, but, owing to the appreciable relative difference in masses, the displacement of the lines should be greater than in the case of isotopes of heavy elements such as lead.

It should be recalled that Bourget, Fabry, and Buisson1, on the basis of a study of the width of the lines in the spectrum of nebulae, concluded that this spectrum is produced by an element with an atomic mass, in round numbers, of 2.7 or 3. However, from the modern point of view it is difficult to suppose that the spectrum of the so-called nebulium can be produced by an element with an

nuclear charge of 2, unless one supposes that the spectra under the conditions existing in nebulae are quite different from the spectra observed in laboratories. The possible origin of the nebulium spectrum was discussed by Nicholson¹) on the basis of entirely different lines, and at the present time it is not easy to see how the new atoms arising in oxygen or nitrogen can be connected with the substance of the nebulae.

Since, probably, the greater part of ordinary helium was formed in one way or another in the transformations of radioactive substances, and since these latter, as we know, always give helium of mass 4, there is no reason to expect that an isotope of helium of mass 3 could be found in ordinary helium. It is, however, of the greatest interest to investigate whether this isotope may not be present in those cases where the apparent presence of helium is difficult to connect with radioactive substances; such a case is found, for example, in beryllium, to which Strutt²) drew attention. This supposition is based on the assumption that an atom of mass 3 is stable. The fact that it withstands intense disturbances in its structure upon close collision with an \(\alpha\)-particle indicates that it is a formation not readily destroyed by external forces.

Structure of the Nucleus and Isotopes.

In discussing the possible structure of the elements it is natural to suppose that they are built up, in the final analysis, from hydrogen nuclei and electrons. From this point of view the helium nucleus consists of four hydrogen nuclei and two negative electrons, so that the resultant charge is two. It is usually assumed that the fact that the mass of the helium atom—3.997, if oxygen is taken as 16—is less than the mass of four hydrogen atoms (4.032) is due to the close interaction of the fields in the nucleus, which manifests itself in the fact that this nucleus possesses a smaller electromagnetic mass than the sum of the masses of its individual components. Sommerfeld³) concluded from this fact that the helium nucleus must have a very stable structure with respect to forces tending to destroy it. This conclusion agrees with experiment, for no destruction of helium by fast \(\alpha\)-particles capable of destroying the nuclei of nitrogen and oxygen has ever been observed. In his most recent experiments with isotopes of the ordinary elements, Aston⁴) showed that, within the limits of experimental accuracy, the masses of the investigated isotopes are expressed by whole numbers if the mass of oxygen is taken as 16. The sole excep—

¹) Nicholson. Roy. Astr. Soc., v. 72, No. 1, p. 49 (1911); v. 74, No. 7, p. 623 (1914).

²) Strutt. Proc. Roy. Soc. A., v. 80, p. 572 (1908).

³) A. Sommerfeld. Atombau und Spektrallinien, p. 538, Vieweg und Sohn, 1919 [2nd ed., 1921, pp. 566 ff. Transl.].

⁴) Aston. Phil. Mag. 38, p. 707 (1919); 39, p. 449 and 511 (1920) [Nature, 9 Dec. 1920, p. 408. Transl.].

is hydrogen, whose mass is equal to 1.008, in agreement with the observations of chemists. This does not exclude the possibility that hydrogen is a primary substance from which nuclei are built, but it indicates that either the grouping of hydrogen nuclei and electrons is such that the mean electromagnetic mass is close to unity, or—and this is more probable—that the secondary units of which the atom is chiefly built, i.e. helium or its isotope, have a mass close to an integer if the mass of oxygen is taken as 16.

For the present, experimental observations do not allow us to decide whether the new atom possesses a mass exactly equal to 3; but, by analogy with helium, we may expect that the nucleus of the new atom consists of three \(H\)-nuclei and one electron, and that it has a mass closer to three than does the sum of the individual masses of \(H\) in the free state.

If this supposition is correct, then it seems very plausible that one electron can bind two \(H\)-nuclei, and possibly even one \(H\)-nucleus. If the first supposition is true, then it points to the possibility of the existence of an atom with mass about 2 and with one charge. Such a substance must be regarded as an isotope of hydrogen. The second supposition contains the idea of the possible existence of an atom with mass 1 and a nuclear charge equal to zero. Such a formation appears quite possible. From the modern point of view, a neutral hydrogen atom should be regarded as a nucleus with unit nuclear charge, with which an electron is bound at a certain distance from it, and the spectrum of hydrogen is ascribed to the motion of this latter electron. However, under such conditions it is probable that one electron can combine more closely with an \(H\)-nucleus, forming something like a neutral doublet. Such an atom would possess quite fantastic properties. Its external field would practically have to be equal to zero, except in regions very close to the nucleus; as a result, it would have the ability to pass freely through matter. The existence of such an atom would probably be difficult to detect spectroscopically, and it could not be retained in a closed vessel. On the other hand, it should readily enter into the structure of an atom and either combine with its nucleus or be destroyed by the intense field of the latter, giving rise to a charged \(H\)-atom or to an electron, or to both.

If the existence of such atoms is possible, then they must arise, although probably in small quantity, in an electric discharge through hydrogen, where both electrons and \(H\)-nuclei are present in significant numbers. The author intends to carry out experiments in order to test whether there are any indications of the formation of such atoms under the conditions mentioned.

The existence of such nuclei may also not be limited to mass 1, but it is possible that they exist with mass 2, 3, 4, or more, depending on the possible combinations among doublets. The existence of such atoms appears to be absolutely necessary for explaining the structure of the nucleus of heavy elements. Indeed, unless one assumes the possibility of obtaining charged particles with very great velocities, it is difficult to imagine how any positively charged particle could approach the nucleus of a heavy atom against its intense repulsive field. We have seen that, so far as has been experimentally found, the nuclei of three light atoms are probably units of atomic structure. These three atoms are

\[ \overset{+}{H}_{1}, \qquad \overset{++}{X}_{3}, \qquad \overset{++}{He}_{4}, \]

where the subscripts denote the mass of the element.

In discussing the possible structure of the nuclei of elements, difficulties arise at once, since the numerous combinations of these units with negative electrons may give an element with the required nuclear charge and mass. Given our complete ignorance of the laws obeyed by forces in the vicinity of the nucleus, we have no criterion of stability or of the relative probability of a given theoretical system. With the exception of a few elements which can exist in the gaseous state, the possibility of the existence of isotopes of the elements has not yet been confirmed. When further information is obtained concerning the products of disintegration of other elements, besides the two already investigated, and fuller data concerning the number and masses of isotopes, it will then be possible to derive approximate rules which may serve as guides in seeking ways of forming nuclei from simpler units. For these reasons it appears premature, at the present time, to discuss in any detail the possible structure even of light atoms, and still less of complex atoms. It is of interest, however, to give one example to illustrate the possible modes of formation of isotopes in the case of light elements. This example is based on the idea that sometimes the helium nucleus of mass 4, probably in complex structures, may be replaced by the corresponding nucleus of mass 3, without producing serious disturbances in the stability of the system. In that case the nuclear charge will remain unchanged, while the mass will change by one unit.

Let us consider, for example, the case of lithium with nuclear charge 3 and mass about 7. It is natural to suppose that its nucleus is built from helium, or from its isotope of mass 3, and one binding electron. The three possible combinations are shown in Fig. 3 (p. 219).

From this point of view, the existence, theoretically, of at least three isotopes of lithium with masses 6, 7, and 8 is probable; but even if these combinations are equally stable, the question of their relative abundance in the element lithium on earth depends on a whole series of factors about which we know nothing. Such factors include, for example, the modes of actual formation of such nuclei, the relative number of units, and the probability of their combinations.

Fig. 3.
Schematic combinations for lithium isotopes: mass 6; mass 7 (nuclear charge 3); mass 8.

The experimental results presented in this article confirm the view that hydrogen atoms and atoms of mass 3 are necessary units in the nuclear structure of nitrogen and oxygen. In that case it may be assumed a priori that oxygen is some combination of four helium nuclei of mass 4. Probably mass 3 is a necessary unit in the structure of light atoms in general, but it is also plausible that, as the complexity of the nucleus increases and the electric field correspondingly grows stronger, the structure of mass 3 undergoes reconstruction and tends to pass into a presumably more stable nucleus of mass 4. This may be the reason why helium of mass 4 is always released from radioactive atoms, whereas its isotope of mass 3 arises in the artificial decomposition of light atoms such as oxygen and nitrogen. It has long been known that the atomic weights of many elements can be expressed by the formulae \(4n\) or \(4n+3\), where \(n\) is an integer, and this indicates that atoms of masses 3 and 4 are necessary units in the structure of the nucleus1.

The Structure of the Nuclei of Carbon, Oxygen, and Nitrogen.

In the light of the experiments described above, it is of some interest to set forth certain thoughts—perhaps still immature—concerning possible modes of formation of the atoms listed, on the basis of experimental facts. It should be remembered that only nitrogen gives \(H\)-atoms, whereas carbon and oxygen do not.

Both nitrogen and oxygen give rise to atoms with mass 3. Carbon, however, has not yet been investigated from this point of view. In Fig. 4 the possible structures are indicated, with the masses and charges of the combining units given. Negative electrons are denoted by the symbol —.

Fig. 4.

It is assumed that the carbon nucleus consists of four atoms with mass 3 and charge 2 and of two binding electrons. The transition to nitrogen is characterized by the addition of two \(H\)-atoms with one binding electron. The transition to oxygen is obtained by replacing two \(H\)-atoms with a helium nucleus.

From these schemes it is clear that the chances of a direct collision with one of the four atoms of mass 3 in nitrogen are much greater than the chances of removing one \(H\)-atom, for one should expect that the larger part of the nucleus must shield the \(H\)-atom from a direct collision, and that only a limited region proves accessible for collision. This may illustrate the fact that the number of atoms of mass 3 released under the corresponding conditions is considerably greater than the number of \(H\)-atoms. It should be remembered that the structures depicted have only a purely illustrative significance, and no importance should be attached to particular details of the arrangement.

It is natural to ask about the nature of the residual atom after the disintegration of oxygen and nitrogen, assuming that these residual atoms withstand the collision while passing into a new stage of temporary or permanent equilibrium.

The ejection from nitrogen of one \(H\)-atom with mass 1 and with nuclear charge 1 must decrease the mass by one unit and the nuclear charge likewise by one unit. Thus the residual nucleus must have nuclear charge 6 and mass 13 and, consequently, must be an isotope of carbon. If at the same time a negative electron is also released, then the residual atom will be an isotope of nitrogen.

The ejection from nitrogen of mass 3 with two charges, occurring, probably, quite independently of the liberation of an \(H\)-atom,

reduces the nuclear charge by 2 and the mass by 3. Thus the residual atom must be an isotope of boron with nuclear charge 5 and mass 11. If here too an electron is also ejected, then an isotope of carbon with mass 11 remains. The ejection of mass 3 from oxygen gives mass 13 and nuclear charge 6; this must be an isotope of carbon. Similarly, if an electron is also ejected, an isotope of nitrogen with mass 13 remains. The data available at the present time are wholly insufficient to choose between these alternatives.

The author intends to continue the experiments¹) in order to investigate whether there are indications of the disintegration of atoms other than oxygen and nitrogen. This problem is more difficult in cases where the element cannot conveniently be obtained in the gaseous state, since it is difficult to be certain of the complete absence of hydrogen or to prepare homogeneous thin films of such substances. For these reasons, and also because of the very great strain involved in the difficult operation of counting scintillations, further progress in this direction will probably not be rapid.

I express my gratitude to my assistant G. A. R. Grove for preparing the radioactive sources, and also to J. Chadwick and Dr. Ishida for assistance in counting scintillations in some of the later experiments²).

Translated by E. V. Shpolsky.

¹) Preliminary results of these experiments have already been published (E. Rutherford and J. Chadwick, Nature, 107, p. 41, 1921). For them see also the special abstract in the section “From Current Literature.”
Editor.

²) The present article is a translation of Rutherford’s lecture (Bakerian lecture): Sir E. Rutherford, “Nuclear Constitution of Atoms,” Proceedings of the Royal Society, A, 97, p. 374 (1920).

1. On the basis of these considerations and a whole series of others, Harkins [Phys. Rev. 15, p. 73 (1910)] proposed a structural formula for all the elements. The combinable units for him were electrons and atoms of masses 1, 3, and 4, with nuclear charges, respectively, 1, 1, and 2. To the unit of mass 3 he ascribed one charge, not two. Thus it should be regarded as an isotope of hydrogen, and not of helium. ↩↩↩↩↩↩↩↩↩↩↩↩

2. Mr. C. S. Fulcher (National Research Council, USA), in a letter to me, expressed the conjecture that perhaps these are α-particles. ↩↩↩↩

3. Barkla, Phil. Mag., vol. 21, p. 648, (1911). ↩↩

4. Van den Broek, Phys. ZS., vol. 14, p. 32, (1913). ↩

5. Moseley, Phil. Mag., vol. 26, p. 1024 (1913), vol. 27, p. 703 (1914). ↩