MARTIN JOHNSON 1407

STRUVE, O . . . . . . . . . . . 1948b

1949 1950a 1950b 1950c

1954

VAN MAANEN, A . . . . . . . . . 1933 VAN BIESBROECK, G . . . . . . . . 1951 VAN Dm KAMP, P . . . . . . . . . 1940 VAN DEN BOS, W. I-I . . . . . . . . 1956

WILSON, 0. C . . . . . . . . . . 1941 WILSON, R. E . . . . . . . . . . 1953 WOOD, F. B . . . . . . . . . . 1953 ZESSEWITSCH, W . . . . . . . . . 1930

Harvard Centennial Symposia ( = Harv. Obs. • Monographs 7).

M.N. , 109, 487. Harvard Coll. Obs., Astron. News Letter No. 50. Harvard Coll. Obs., Astron. News Letter No. 51. Stellar Evolution, p. 154 etc. (Univ. Press,

Princeton. ) Colloque International d'Astrophysique, Liege,

Mem. in-8 °, Ser. 4, vol. XIV, 236. Prec. K. Ned. Akad. Wetensch., 86, 870. I n HYNEK (1951), p. 425. Publ. Amer. Astron. See., 10, 30. In Vistas in Astronomy (Ed. A. BEER), vo1. 2,

Section 11, p. 1035. (Pergamon Press, London and New York.)

Ap. J., 98, 29. Papers of the Mount Wilson Obs., vol. V I I I . Publ. Univ. Pennsylvania, Astron. Ser., vol. V I I I . Bull. Variable Stars ( V F P A ), Nishni-Novgorod,

2, 55.

Evidence For and Against Binary Hypotheses of the Blue-red Erupting Stars

MARTII~ J O H N S O N

Physics Department, The University, Birmingham

SUMMARY"

The peculiar and erupting stars which exhibit simultaneously a cool oxide spectrum and a hot helium spectrum have been variously regarded, either as binary pairs with blue and red components or as single stars with composite envelopes. The implications and obstacles in each of these hypotheses are here critically examined. A suggestion is made for avoiding one difficulty of the binary hypothesis, by a mechanism in which a prominence erupting into a gaseous ring becomes amplified by suppression, resulting in local instability.

1. N A T U R E OF THE PROBLEM

A STELLAR spectrum dominated by titanium oxide normally denotes the low tem- perature (20000-3000 ° ) at which stability and excitation of that material would occur in thermodynamic equilibrium. Correspondingly, at the other extreme of visible stellar temperatures, the lines of ionized helium must imply at least 20,000 °- 30,000 ° in thermodynamic equilibrium. We shall discuss in this note some old and new explanations of those few stars whose spectrum combines both the coolest and the hottest, the oxide and the helium simultaneously imitating phenomena of such extremely differing temperatures that either two separate bodies or two regions in radical departure from thermodynamic equilibrium must be sought.

It has commonly been supposed that these "combination" spectra imply binaries, one component of each pair being a normal hot B-star and the other a normal cool M-star; great distance from us or small distance between the components could conceal the binary character, and eclipses and orbits could be hidden by orientation and the motions of a common envelope. The present author (1951) calculated some time-constants of fluorescence around these stars in terms of such binary hypothesis; some authors employing the same hypothesis have written "companion star" in inverted commas recognizing the uncertainty; MENZEL, who once propounded the

4I

140S Ev idc lwc f'ol' ~md ~gaiu~t~ binacy hypoti lc~e~ o f t, ilc blm~-v(~d erupti~Lg s~v,~

binary theory, stated in 1.(}46 that he considered a single star of complex structure the more likely explanation.

These stars, all faint and irregularly fluctuating, have been studied mainly in the great American observatories, and our problem is set by the many important papers of SWINGS and STRUVE (1940--1943), MERRILL (1935, 1944, 1950a), DAYNE- GAeOSCHKIN (1946), and others. I t would be impertinent for an outside student to claim a solution of the problem, but comparison between implications of some possible solutions is offered here with tentative further suggestions; they may be appropriate in a volume dedicated to F. J. M. STRATTON, whose researches on novae have taught that minor stellar irregularities may be a clue to the major catastrophe.

2. CLASSIFICATION OF MATERIAL

We first arrange into an order of decreasing eruptiveness six stars, each of which combines the extremes of TiO and He II; in brackets we quote the periodicity of M type found by various observers for the cool spectrum, associated with MERRILL'S phase-lag of one-fifth of the period between permitted and forbidden or nebular lines.

T CBr: nova, 6 mag. range, 1866, 1946 (230 days). Z And: nova, 3 mag. range, 1939, 1946 (650 or 715 days). AX Per and CI Cyg: irregularity 2 mag. range, nebular lines post-nova type. R Aqr : irregularity 2 mag. range, with photographed inner and outer nebulosity

(387 days). R W Hya: irregularity 1 mag. range (370 days).

The following are useful for comparison, each containing some resemblance and significant difference :

(a) Other repetitive novae (I~S Oph, T Pyx, Nova Sge No. 2, U Sco). (b) Observed B and M components of the visual binary ~. Sco which has [Fe II]

nebulosity but no pulsation and no present irregularity. (c) Observed B and M binary without either nebulosity or eruption (o Ceti);

it must be recalled that bright lines of super-thermal excitation are common even in quiescent and solitary M-stars.

(d) Combined B and M spectra with [Fe II] (W Cep, W ¥ Gem). (e) Less directly comparable, but offering clues to one or other feature, are also

~] Car, l~Y Scu, P Cyg shells, tail-enveloped binaries (fl Lyr), prominence binaries (~ Aur), binaries with ring round one component (RW Tau), rings round both com- ponents (UX Mon), rings of differing velocity in same binary (RZ Scu, RY Per), and ringed single stars (7 Cas). Some of these descriptions may come under revision, but have been usefully suggestive.

Apart from those mentioned as observed binaries, a likely companion to Z And has been given definite properties by STRUVE (1944), and MERRILL (1950b) has evaluated a very tentative binary orbit for R Aqr with reservation as to alternatives.

3. CRITIQUE OF BINARY EXPLANATIONS FOR THE TIO/HE I I ERUPTING STARS

(i) Binary systems are commonly recognized by their eclipses and orbital velocities, but neither feature has been conclusively extracted from the complexity of intrinsic light fluctuation and gas velocity fluctuation in these stars. However, in so small a number of stars, favourable orientation of orbital plane to line of sight cannot be statistically demanded for any given fraction of the number. Doppler shifts due to

MARTIN JOHNSON 1409

each component body as a whole will be difficult to disentangle from the large and widely fluctuating motion of gases which undoubtedly erupt from these very dis- turbed stars. The lag between emission and absorption velocities in Mira stars is here reinforced by MERRILL'S one-fifth period lag of forbidden lines, and the periodicities may involve orbital as well as pulsatory motion. Hence lack of detection of orbits may mean their concealment under gas velocity rather than evidence of their non-existence.

(ii) MENZEL found against the binary hypothesis that PAYNE-GAPOSCHKIN'S synchronizing of the fluctuations in the B and M "apparent components" would be an unlikely accident if there were really two stars involved. This objection does not seem insuperable if the stars interact with one another, for example in the manner which we suggest below. An apparent "blue" periodicity could well be an effect of even a steady irradiation upon a fluorescing target whose size alters as the red star pulsates, as suggested by the present author (1951).

(iii) Any binary hypothesis sets a difficult task for theories of simultaneous evolu- tion, if very hot and very cool components are supposed to be traceable to a common origin. The extent of this difficulty cannot be estimated until the origin of all binaries is better understood.

(iv) We consider the most serious obstacle is the difficulty of understanding why these stars show irregularity of light amounting in the most disturbed cases to a nova outbreak. Hence no binary explanation is adequate unless it shows why B and M spectra which are quiescent in separated single stars become unstable when associated. We suggest below one way in which interaction between the components of a blue-red pair of stars might be understandable and thus mitigate this obstacle.

4. CRITIQUE OF THEORIES THAT THE HIGH AND LOW EXCITATIONS ARISE

IN A SINGLE STAR

I t must be admitted at once that He II and oxides are both detected in the Sun, so that coexistence in one body is not inherently impossible. But neither spectrum approaches even remotely to domination of the solar radiation, and He n only appears in its rarefied outer appendages and faintly. Extrapolation to cover the blue-red phenomena might invoke any of the following.

(i) Vastly extended chromospheres do envelope stars of early types, so eruptiveness in He II might denote prominences on a super-solar scale. But prominences of such long duration, magnified to dominate the spectrum, would offer no clue to the accompanying oxide. Chemical compounds in a star of high excitation have been detected in one or two novae, but only as brief transients probably denoting condensa- tion between separating shells. Amplification of prominences by a binary structure we investigate below.

(ii) In two-phase models of red giants with a hot core, surges from the inner structure may occur when the source of energy production is in course of change. But this seems a more likely cause of an isolated nova outbreak on the large scale than the repetitive and minor irregularity observed.

(iii) A purely external cause of irregularity is possible, such as GAt'OSCHKIN'S variation of star due to variation in interstellar nebulosity, for instance through changing rates of accretion. Since nothing is known of the density gradients in interstellar material, this suggestion remains untested.

(iv) One recent model of red giants, in accord with JoY's discovery that the bright

1410 Evid¢m('e for and against bim~ry hypotheses of the I)lue-r(~d ~rul)ting star's

hydrogen lines are located below the absorption layers, might possibly be extended to cover blue-red eruption, MCKELLAR and ODGERS (1952) suggest that these bright, lines are the visibility of the convection zone during opacity changes by pulsation, a recombination spectrum at density far exceeding the chromospheric and differing radically from Be spectra or luminous envelopes. Since helium also has a convection zone, a super-Mira model analogous to that of MCKELLAR and ODGERS might account for the blue-red anomalies if some reason were found for the irregularity superposed on the pulsation.

5. THE POSSIBLE AMPLIFICATION OF B-STAR PROMINENCES BY A COOL COMPANION

Without claiming that the binary hypothesis can be established, either through suggestions that the objections are not insuperable or through the shortcomings in the hypotheses of a single star, it may be relevant to outline briefly a process which might mitigate the most serious obstacle, namely that no reason for the abnormal eruptiveness has been found.

Binary explanations which would escape this objection might be approached by combining the gaseous rings, mentioned before, with quite ordinary prominence activity. We shall proceed to discuss, tentatively, whether rings could cause ampli- fication of the normal small prominences to a more noticeable activity which would show as irregularity of the star's total light. "Rings are very common around binaries, perhaps even universal" (STRUVE, 1950), so that the mechanism which we propose is applicable most readily to binaries, and possible extension to a ringed single star will be mentioned at the end. The evolutionary significance of rings is not fully known: rings in binaries of RY Per or RZ Scu type have been suggestively associated not with extruded material but remnants of a pre-binary state. Ring rotation combined with pulsation, as in BALDWIN'S model of the single star y Cas (1940), will offer a novel situation if extended to the binaries whose tail is extruded by KUIPER'S mechanisms (1941), and it is such a situation which suggests how a small prominence might first be suppressed by the ring and then released on an enhanced scale.

Before elaborating the suggestion, we notice that there are two ways of explaining why a binary combination is more eruptive than the similar two stars in separation, (a) the conditions permitting an oxide spectrum in the one, with a helium spectrum in the other, may happen to arise at the same unique evolutionary stage which also permits internal instability: this awaits a future theory of binary evolution, so we are left to investigate the other alternative, (b) that one of the pair reduces the local or surface stability of the other star. It is not difficult to show that the greatest allowable effects of reflection or absorption of tbe radiation from one component of the binary intercepted by the other are negligible for stability, and that even tidal distortion is unlikely to produce the outbursts of T CBr and Z And. Nor is it necessary that a circulating ring extruded by KUIPER'S mechanism is itself unstable except for spiral expansion and lateral spread. Even if one of the pair pulsates, the effect would be of density maxima in the spiral, soon smoothed by viscous drag.

Consider, however, the intrusion of a local prominence into the circulating ring. Of KUIPER'S tWO models of ring and tail extrusion, which he considered in studying fl Lyr, his "A" process is more relevant to this case, and indeed he gave as possible examples of it T CBr, Z And, and AX Per. HOYLE and LYTTLETON (1943) have also described how, when the less massive component is giant, it supplies the material to the common orbital stream enveloping also the more massive. Suppose that, at

MARTIN ,|OHNSON 1411

suitable phase in the spin and orbit of each star relative to the pulsation maxima of the red component, a prominence instead of rising freely intrudes radially from the B-star into the stream circulating from the M-star around both; it will cause some arrestment of the circulation, which when deprived of tangential velocity becomes subject to accretion into the B-atmosphere for the same reason which allows capture of arrested interstellar nebulosity in the accretions of HOYLE and LYTTLETON. The amount will depend on densities, interaction being small if prominence and stream were too rarefied for much viscous drag in their encounter.

When the prominence arises within the zone covered by the stream, and at such phase of the rotation, revolution, and expanding portion of the pulsation cycle, that it is both prevented from escape and allowed to capture from the stream, then local changes in opacity will follow, and in the ionization of the captured material which initially was less than that of its new environment. This will affect both terms

{(d In T/d In P)rad -- (d In T/d In P)adiab

representing the consequences of abnormal gradients upon stability ; but modifica- tion of this criterion for already moving material will be a difficult task. A corres- ponding effect at greater depths has been invoked by BIERMANN (1939) and others as cause of ordinary novae, but in our case the eruption after suppressed instability can be much slighter because less deep-seated in the star.

The event would be rare, as depending upon optima in so many variables: but these blue-red erupting stars are extremely rare. GERASZMOVIC (1934) and MIss UNDERHILL (1949) have shown how near to instability are B-star photospheres, so prominences can always be expected ; but only optimum timing and orientation will provide the initial suppression in the captured stream and sufficient amplification at subsequent release to cause detectable irregularity of the star. Much calculation would be needed to separate the initial instability, the intermediate stage of inhibiting convective relief, and a final eruption of the stored energy with effects in temperature, expansion, and fluorescence.

The slighter eruptiveness of AX Per, CI Cyg, R Aqr, R W Hya would accordingly be less rare than the explosiveness of T CBr or Z And, and the quiescence of the undoubted blue-red pairs :¢ Sco and o Ceti raise the query as to how small an orbit must be for a stream to develop, and how dense the stream must be for prominences to become amplified when they are suppressed in arresting the stream.

This suggestion, of amplication of sporadic prominences by collision with a KUIPER stream, may offer one instance of how stars quiet when single might become eruptive as a pair ; but even if at best it provides the binary hypothesis with one mechanism requiring prominences only ordinary before their amplification, it does not make that hypothesis exclusive. Although rings have mostly been associated with binaries, some single stars do possess them, and if a prominence were to intrude into an already existing ring from the same star it might undergo suppression and release in amplified form, and the whole account revert to a non-binary situation. Such possibility must await the growing knowledge of how binary and single-star rings evolve.

REFERENCES

BALDWIN, R. B . . . . . . . . . 1940 BIERMA~N, L . . . . . . . . . . 1939 GERASIMOVI~, B. P . . . . . . . . 1934

Ap. J., 92, 82. Z. Astrophys., 18, 344. M.N., 94, 737.

1412 Observational rcsult, s on vaviabh,

H:OYLE, F. and LYTTLETON, [~. A. 1!}43 JOHNSON, M . . . . . . . . . . 1951 KU1PER, G. P . . . . . . . . . . 1941 MCKELLAt~, A. and OD(;ERS, G..J. 1952 MENZEL, D. H . . . . . . . . . 1946 MERRILL, P. W . . . . . . . . . 1935

1944 1950a 1950b

PAYNE-(-]APOSCHKIN, (~ . . . . . . . 1946 STRUVE, O . . . . . . . . . . . 1944

1950 SwINCS, P. and STRUVE, O . . . . . . 1940-43 UNDERHIGL, A. B . . . . . . . . 1949

stars with (.Oml)osite spectra

M.N., 103, 22. M.N. , l U , 490. Ap. J., 03, 133. Publ. Astron. Soc. Pacific, 64, 222. Physiea, 12, 768. Ap. J., 81, 312. Ap. J., 99, 15.

I. Ap. J., 111, 484. II . Ap. J., 112, 514. Ap. J., 104, 357, 362. Rev. Mocl. Phys., 16, 293. Stellar Evolution, 193. Ap. J., 91-98, series of papers. M.N. , 109, 562.

Observational Results on Variable Stars with {~omlmsite Spectra

TCHENG MAo-LIN and M. BLOCH Observatoires de Lyon et de Haute Provence, France

SUMMARY

The stars Z And, T CrB, AG Peg, R Y Scu, BF Cyg, AX Per, and CI Cyg have been observed at the Observatoire de Haute Provence from 1946 to 1952. A brief account is given of the spectra of these stars, which are all composite, but to a greatly varying degree. Spectrophotometric measures have led to the determination of the colour temperatures and electron temperatures of these stars.

THE problems offered by the stars with composite spectra are numerous, in particular those concerned with the variations and the origin of the emission lines, the energy distribution in the continuous spectrmn, and the behaviour of the atmosphere which surrounds them.

In view of the interest presented by these stars we have devoted our efforts since 1946 to observing some of them as continuously as possible at the 0bservatoire de Haute Provence; namely, Z And, T CrB*, AG Peg, RY Scu, BF Cyg, AX Per, and ('~I Cyg.

As a rule we used the one-prism spectrograph in conjunction with the 120-cm telescope. The flint prism of this instrument gave a dispersion of 41 A per mm at 3800 A, of 88 A per mm at 4500 A, of 130 A per mm at 5000 A, and of 318 A per mm at H~; on several occasions we have worked with the four-prism spectrograph at the Cassegrain focus of the 80-cm telescope; its uviol-glass prisms yielded 24 A per mm at 3700 A, 36 A per mm at 4000 A, 61 A per mm at 4500 A.

1. Z ANDROMEDAE

Investigations due to PLASKETT (1928), SWINGS and STRUVE (1940, 1943), and MERRILL (1948) show remarkable changes in the spectrum of this star.

We obtained three plates, namely on 27th June, 1946, 8th August, 1946, and on 27th July, 1948 (TcHENG MAo-LI~, 1949a and 1949b). More than 200 emission features could be identified as belonging to the permitted transitions of H I, He II, He I, N III, N II, C III, Fe II, C II, and to the forbidden lines of 0 III, Ne In, S II,

* Contributions to tile s tudy of T CrB have been made by DUF.*~Y and FEHRENB.~.CH in 1946.