A general classification of novae in terms of integral parameters is proposed. It is shown in particular that the distribution of novae in dependence on the intrinsic luminosity at maximum brightness reveals the existence of two distinct classes, designated as common novae and supernovae. In Section (A) the usefulness of the study of temporary stars in comparison with the study of permanent stars is discussed. Sections (B) and (C) give a tentative classification of novae in terms of integral parameters as well as a brief tabulation of some of the observational data on these parameters. In the course of this discussion a number of interesting problems suggest themselves. Among them the problems of the primary and secondary images of novae and of supernovae and the investigation of the final states of temporary stars are of particular interest. In (D) some approximate relations among the integral parameters are described. In (E) the frequency of occurrence $n\left(M\right)\mathrm{}$ of novae in dependence on their absolute magnitude $M$ is explicitly established and graphically represented. The fundamental significance of the fact is emphasized that the function $n\left(M\right)\mathrm{}$ has two pronounced maxima. In (F) the importance of the fact is analyzed that supernovae preferably appear in those parts of nebulae whose surface brightness is low. Two outstanding cases of supernovae which are located in points of excessively low surface brightness are reproduced in Fig. 3. In Section (G) it is shown that the average frequency of occurrence of supernovae in different types of nebulae does not depend essentially on the type. In Section (H) a number of reasons are advanced which appear to be sufficient to rule out the assumption of collisions between luminous stars as a cause for supernovae. Finally in Section (I) a review is given of the decisive observations and considerations which have resulted in the recognition of supernovae as a separate class of temporary stars.

The variation of the hard cosmic-ray component with zenith angle and the temperature effect suggest that the cosmic mesotrons decay with a lifetime of about 3 × ${10}^{-6\mathrm{}}$ sec. The other phenomena so far considered (variation with height, barometer effect, shape of the energy spectrum, number of secondaries, etc.) do not provide any evidence for the decay, but are not inconsistent with the above assumption. On the contrary a considerably smaller lifetime, less, for instance, than 1 × ${10}^{-6\mathrm{}}$ sec., would be inconsistent with most of the experimental results. There is no evidence that the mesotrons which have been stopped actually disintegrate into an electron and a neutrino.

With Geiger-Müller tubes arranged for fourfold vertical coincidence, a Rossi-Hsiung type of experiment was performed in an aeroplane up to an altitude of 25,000 feet. A sheet of lead 8.1 cm thick placed between the counter tubes and another lead plate of 2.2 cm thickness was alternately changed from the position above all the tubes to a position between the second and third tubes. From the difference between the counting rates in the two cases, it was found that at an altitude of 25,000 feet an average of 2 ionizing penetrating particles (mesotrons) per minute were ejected in the forward direction from a lead plate (38 cm×5.2 cm×2.2 cm) by the action of non-ionizing rays (photons). This value aleads to a cross section for the production of mesotrons by photons which is in at least rough agreement with the theory.

It has been found that the cosmic-ray particles which occur in showers of small angular divergence are much more penetrating than those of the usual type of shower. The particles in such "hard showers" are clearly not electrons and are interpreted as medium and low energy mesotrons. The second maximum of the Rossi curve, which is accentuated for small-angle shower arrangements, can be accounted for in terms of such showers, produced by mesotrons. Hard showers are present also in the first maximum, probably being produced in this case by the soft component. In the first maximum 34 percent of the narrow showers are ascribed to ionizing primaries, as compared to 75 percent in the second maximum. The number of rays per hard shower is found both by counter and cloud chamber experiments to be small, in most cases, two. Present theory can account for the production of such mesotron showers, though some modification must be introduced regarding the limitation in angular spread of the emitted particles. Maier-Leibnitz and others have obtained evidence for the relatively frequent production of slow, absorbable mesotron secondaries. Because of the much higher rate of production of these mesotrons as compared with those of the hard showers, a different mode of formation may be present; speculation on this point is at present premature.

Size-frequency distributions of bursts were obtained from analyses of records made by Carnegie model C cosmic-ray meters, shielded by 12 cm of lead, stationed at different locations. The number of particles per burst ranged from 200 to 10,000. Assuming that the number of particles in a given burst is proportional to the energy of the incident ray, energy distribution curves for the burst-producing radiation were made and compared with the depth-ionization curve obtained underground by Wilson. Because both curves have nearly the same shape, it was concluded that at least the greater part of the burst-producing radiation at sea level consists of penetrating ionizing rays, presumably mesotrons. The value of ${10}^{-4\mathrm{}}$ was found for the creation probability of a burst by a mesotron of about 2×${10}^{10}$ ev energy in a thickness of 12 cm of lead. This leads to a cross section per nuclear particle (proton, neutron) of 2×${10}^{-30\mathrm{}}$ ${\mathrm{cm}}^2$, which value is independent of the energy of the incident particle. The ratio of the burst rates at Huancayo (3350 m elevation above sea level) and Cheltenham (72 m elevation) for different burst magnitudes was found to be constant, within the statistical errors, up to energies of about 1×${10}^{11}$ ev, and likewise for Teoloyucan (2285 m elevation) and Cheltenham. For energies higher than ${10}^{11}$ ev, the corresponding ratios of the burst rates at the high to those at the low altitudes increase rapidly with increasing energy of the burst. In order to account for this effect, it is assumed that some of the largest bursts are created by photons or electrons of energies greater than ${10}^{11}$ ev, a part which becomes predominant at higher elevations. This suggests a possible mechanism for the creation of mesotrons by the soft component of cosmic radiation in the earth's atmosphere.

The behavior of electrons and photons of high energy is discussed in relation to the production of large bursts of cosmic-ray ionization, or Hoffmann Stösse, and the occurrence of nuclear "vaporizations." It is shown that the electrons from the disintegration of mesotrons have an importance equal to that of the cascade electrons. The number and size-frequency distribution of large showers (of a hundred or more rays) from thin and thick pieces of lead at sea level, and the variation with elevation of such showers are well accounted for by the action of electrons and photons from these two sources. It is unnecessary to invoke the direct production of bursts by penetrating rays by means of an explosion process. The behavior of showers of a few rays is likewise well accounted for. However, difficulties are encountered in explaining: (a) the relative numbers of bursts from large thicknesses of iron and lead, and (b) the occurrence of showers, from the air, which have a large number of rays per unit area. (See note added in proof page 261.) The hypothesis is advanced that the showers of heavily ionizing particles, or nuclear vaporizations, are produced by electrons and photons in the same range of energy as those which produce the large bursts. The identification of showers of heavily ionizing particles with Hoffmann Stösse is shown to be untenable. A determination of the absolute number of neutrons in the cosmic radiation at sea level is shown to be consistent with the supposition that these neutrons are produced in the nuclear vaporization process.