(I) The non-stability of the components in close binary systems has been established in a number of systems, mainly among the eclipsing variables. It cannot be regarded as an unusual phenomenon.

Many eclipsing variables exhibit characteristics that indicate lack of stability. Physically, these systems range from certain short-period dwarfs, which show irregular brightness fluctuations unexplainable by any normal eclipse hypothesis, to systems such as AO Cassiopeiae: class O super-giants whose periods, velocity curves, and light curves have shown remarkable variations. Included are systems having one Wolf-Rayet component and probably one old nova. Many eclipsing binaries show erratic changes of period that are difficult to explain on any concept of a stable system. Indeed, when we consider the physical conditions which must prevail when two stars are located with their surfaces only a few hundred thousand miles from each other, perhaps we should be surprised that the irregularities are not larger.

Those eclipsing binaries whose components are non-stable stars deserve particular attention not only owing to the fact that in such cases the physical properties of the stars may be studied in greater detail, but also since the mechanism of the ejection of gases from the atmospheres of the components may then be established with some certainty.

The binary 31 Cygni is not, in the ordinary sense, an unstable star. Nevertheless, the organizers of this symposium have invited the present paper as a discussion of additional recent evidence of random motions of masses of gas in the atmospheres of the late-type giant components of certain double stars.

Following the papers on instability in binary systems, Dr S. Gaposchkin described a classification scheme that he had devised to describe the eclipsing binaries having emission lines, gas streams, or thick atmospheres. Such stars Gaposchkin calls the ‘camouflaged eclipsing variables’, and divides them into five groups:

1. (1) Typical members are UX Mon and SX Cas: the spectral type of the brighter star is B or A, that of the fainter is G or later. The minima are of equal depth. Balmer emission is present.

2. (2) Prototypes are β Lyr and ν Sgr: only one component can be seen distinctly, but two minima are present in the light curve. The spectroscopic behaviour is most complicated.

3. (3) Prototypes are RT And and YY Gem: the spectral class is G or later. Ca 11 emission is present. The depths of the minima are roughly equal, and the dimensions of the components are about the same.

4. (4) The prototype is DN Ori: both components are so well camouflaged that no definite velocity variation has been observed.

5. (5) Prototypes are V444 Cyg or UX UMa: the bright lines are strong, and their displacements and intensities do not vary in a simple manner with phase.

Our knowledge concerning stars of the T Tauri and UV Ceti types has been considerably enlarged during recent years, owing to the accumulation of a large number of facts. As a result, we may now make certain suggestions regarding the causes of the non-stability of these stars, based on observational data. Such an approach to the solution of the problem is altogether different from that applied heretofore in attempts to develop a theory of certain types of variable stars. The method of construction of models has usually been applied in the past. We do not exclude, of course, the possibility of solving a problem on the basis of some theoretical model. However, the physical phenomena observed in the T Tauri and UV Ceti stars are so unusual that it is first of all necessary to define the nature of these phenomena and their mutual interrelation. Only then will detailed mathematical construction of a definite model be possible.

The most immediate purpose of the study of stellar stability is to discover the sources of incipient instability which must be responsible for the observed variability of a great number of stars.

In the discussion of the papers by Prof. Ambartsumian and Dr Ledoux, Dr E. Schatzman expressed himself as being much impressed by the arguments of Ambartsumian concerning the origin of the non-thermal radiation in the stars. He regarded it as very probable that the explanation of this phenomenon in the UV Ceti and T Tauri stars, and in the variable cometary nebulae, lay in the domain of nuclear processes. However, Schatzman wished to propose a different explanation from that of Ambartsumian for this nuclear phenomenon. He stated that it seems possible that hydromagnetic processes would be able to communicate considerable velocities to atoms in the outer layers of certain stars. During this process, nuclear reactions would take place, and Schatzman thanked Mrs Burbidge for having drawn his attention to the resonance reaction Ne20+H1→Ne21+e+, and to the subsequent formation of neutrons from the reaction Ne21+He4→Mg24+n. These neutrons would lead to the formation of radioactive elements, which are carried out to produce the non-thermal radiation of the variable comet-like nebulae.