Non-Periodic Phenomena in Variable Stars IAU Colloquium, Budapest, 1968 THE O'CONNELL EFFECT IN SOME ECLIPSING VARIABLES E. F. MILONE Gettysburg College, Gettysburg, Pennsylvania University of Maryland, Astronomy Program, College Park, Maryland Kitt Peak National Observatory, Tucson, Arizona INTRODUCTION The O'Connell effect, a name Dr. Wesselink and I have given to the phenomenon of unequal light maxima in certain eclipsing binary stars, was formerly called the 'periastron effect'. Although there is no definite known cause, in the majority of cases it cannot be due to any periastron effect. This is clear from the negative correlation found by Mergentaler (1950) between the magnitude of the effect and the orbital eccentricity. A more likely origin lies in clouds or streams of matter existing on or around the Lagrangian surfaces of close binaries with nearly circular orbits. UBV observations of two stars in O'Connell's (1951) list do not contradict the latter hypothesis. PRESENT WORK The systems RT Lacertae and CG Cygni were selected for observation on these grounds: strong O'Connell effect, lack of previous photoelectric photometry, and brightness. The data were gathered with more than the usual care. Careful attention was paid to changes in sky transparency in the following ways: 1) a double pair of "U" observations were placed on the outside of the observing sequence, 2) observations of the comparison star inevitably flanked those of the variable; and 3) three times a night high and low standard stars of matching color were observed to provide accurate primary extinction coefficients. Transformations to the standard UBV system were done in the usual way (Hardie, 1962) using coefficients obtained from the low air-mass observations of standards paired according to contrasting spectral types. The reduction technique, described more fully elsewhere (Milone, 1967), produced both differential magnitudes and colors (in the sense: variable-comparison stars) and UBV values for the comparison. Comparison stars (BD + 34 4216 for CG Cygni and BD + 43 4108 for RT Lacertae) were selected because of similarity in colors to the variables and because air mass differences between the variable and comparison stars never exceeded 0.01 within a ten-hour range in hour angle. These conditions minimized the effects of extinction and transformation coefficient changes on the light curve. As a check on the constancy of each comparison star, at least once per night a near-by check star was observed. The magnitudes and colors of the comparison and check stars are given in Table I. Table I Comparison and Check Stars Star V B-V U-B for RT Lac BD + 43 4108 7.410 +- 0.003 1.355 +- 0.003 1.527 +- 0.005 BD + 43 4109 8.562 0.336 0.112 epsilon_mse in mean difference: +- 0.002 +- 0.002 +- 0.004 for CG Cyg BD + 34 4216 8.969 +- 0.003 0.744 +- 0.002 0.231 +- 0.003 BD + 34 4213 6.636 1.484 1.811 epsilon_mse in mean difference: +- 0.002 +- 0.002 +- 0.003 RT LACERTAE The differential light curve for the 5.07d-period binary RT Lacertae during 1965 is shown in Figure 1. Filled circles are normal points of Kitt Peak observations obtained by the author in October-November, open circles are normal points of Yale observations, x's represent individual Kitt Peak observations contributed by Dr. Douglas Hall (1967b) over a somewhat wider range in time. The light curve is clearly incomplete. Further photoelectric observations are being obtained by Hall (1968), and the author plans simultaneous spectrographic observations. The remarkable features of the present light curve have been reported earlier (Milone, 1967, 1968a, 1968b) and need only be summarized here as: 1) an anomalously blue primary minimum, independently discovered by Hall (1967a) and 2) an apparent change in the magnitude and sign of the difference between maxima from Wachmann's (1935) photographic light curve. A previous radial velocity solution by Joy (1931) yielded masses of 1.90 and 1.00 for the "fainter" and "brighter" components respectively. Entering with the mass ratio of .53 in Kratochvil's (1964) Table II, the limits of the inner contact surface in the direction normal to the line of centers are for the primary: .434 and for the secondary: .318. These exceed by 20% the largest radius derived for either component in this direction by previous workers, viz: Fowler (1920) using Luizet's (1910, 1915) data; Krat and Nekrasova (1936) using Wachmann's light curve. Consequently we cannot assume that the system is a contact binary. Table II Scatter in the light curve maxima of CG Cygni Run Maximum sigma V* sigma B* sigma U* 1965 I 0.015m 0.015m 0.022m II .014 .014 .025 I & II .015 .015 .025 1967 I .011 .010 .023 II .012 .010 .021 I & II .012 .010 .022 1965+1967 I .020 .020 .026 II .016 .019 .027 I & II .019 .020 .027 * sigma refers to the mean standard error of a single differential observation. Fig. 1. Differential V, B-V, and U-R observations of RT Lacertae made in 1965. Filled and unfilled circles are normal points of data obtained, respectively, at KPNO in October-November and at Yale Observatory during September - November. X's are the individual points of D. Hall obtained at KPNO. The mechanism for causing the O'Connell effect in this star, as well as light curve features 1) and 2) is still unknown. Joy's (1931) study precludes the possibility that the hotter star is seen masked by an absorbing cloud at primary minimum. The increase of the O'Connell effect with decreasing wavelength means that clouds of the negative hydrogen ion are not responsible for that effect if it is caused by absorption at maximum I. Further analysis must await completion of the light curve and high-dispersion spectroscopic work. CG CYGNI The differential light curve for the 0.63d-period binary CG Cygni is shown in Figure 2. Here unfilled circles are normals of observations made at Yale in 1965, filled circles are individual October-November 1965 Kitt Peak observations, x's mark individual June 1967 Kitt Peak observations, and +'s indicate individual observations made at the Dyer Observatory of Vanderbilt University, Nashville, Tennessee in July, 1965. All data have been reduced to the UBV system. It will be noticed that although the 1965 observations were made in three different local systems, the data from this year forms a generally more homogeneous set than the combined 1965-1967 Kitt Peak data obtained with the same telescope and filter slide. In particular, over several cycles, the 1967 primary minima are consistently more shallow and have an earlier rise than the 1965 data. This is true for all colors. The scatter in the maxima, computed by differencing phase-adjacent observations, is given in Table II. In all bands the combined-run data show more scatter than either run separately. The scatter in maximum II is slightly less than in maximum I, a feature shared with Yu's (1922) light curve. There is also evidence of shorter term changes. Around phase 0.4P, observations from October 29-30, 1965 are apparently fainter by 0.04m in V, 0.06m in B, and 0.10m in U than corresponding-phase data of October 7-8, 1965. With E_0 = 2422967.4283 and p = O.6311437d (Kukarkin et al., 1958), the dates correspond to cycles 25503 and 25468. Small differences between the color transformation coefficients (cf Table III) calculated for those nights and the mean run values adopted for the two sets of observations cannot account for the magnitude and color differences. The overall appearance of the light curve is also apparently changing with time. Current extrema are listed along with earlier published values in Table IV. The recent minima values have been read from the light curve but those for maxima have been obtained from the mid-maxima values of truncated Fourier series of the form The purpose in obtaining the 1967 data was to fill in the light curve - particularly on the branches of the minima - so that a preliminary solution could be attempted. The apparent depression of maximum I between fall, 1965 and June, 1967 suggests that it is growing fainter with time and that maximum II may be unchanging. Low dispersion spectra taken with a 36" Kitt Peak telescope and Yale's 40" reflector fail to show evidence of emission, although this work should be repeated with greater dispersion on larger instruments. Fig. 2. Differential V, B-V, and U-B observations of CG Cygni made in 1965 and 1967. Filled and unfilled circles are, respectively, individual observations made at Kitt Peak in October-November, 1965 and normal points of observations made at Yale Observatory during 1965; X's mark individual June, 1967 Kitt Peak observations; and +'s are individual July, 1965 observations made at the Dyer Observatory. The dark line through the DV observations is the Fourier representation of the combined 1965-67 Kitt Peak observations outside of eclipse. Table III Transformation coefficients and fitting errors for Kitt Peak Telescope No. 3 photometric system during two nights in 1965 A = Date B = epsilon C = sigma epsilon D = sigma V * Y_0 E = mu F = sigma mu G = sigma(B-V)*(b-y)_0 H = Psi I = sigma Psi J = sigma(U-B)*(u-b)_0 A B C D E F G H I J 10/7-8 +0.003 1.092 0.989 Mean for Run 1: -0.015 +-0.008 +-0.026 1.088 +-0.002 +-0.020 0.967 +-0.006 +-0.017 10/29-30 -0.040 1.118 0.989 Mean for Run 2: -0.026 +-0.006 +-0.029 1.103 +-0.004 +-0.019 0.977 +-0.001 +-0.019 The notation is that of Hardie (1962). sigma_epsilon is the m.s.e. of the run mean, sigma_V,Y_0 is the mean square deviation between tabulated and calculated values of V for standard stars. The other quantities are analogous. At the present time no radial velocity curve exists, and the mass ratio is not known. The orbit is sensibly circular with r = 1.44 X 10^-2 X ((m_1+m_2)/m_Sun)^1/3 a.u. The times of external contact are sharp enough to obtain the sum of the radii: r_1 + r_2 = 0.48 X 10^-2 X ((m_1+m_2)/m_Sun)^1/3 a.u. = 1.2R_Sun for solar masses. If the mass ratio is of the order unity, the stars do not fill their inner lobes. From a spectrogram taken at Yale, it is quite clear that the stars cannot be of early spectral type, but are closer to late G. The origin of the O'Connell effect in this star is not known, but a slowly changing absorption of maximum I, and occasionally minimum I may be taking place. If absorption is the agent, it cannot be due to clouds of H^-. Clearly, the system must be monitored during the coming years for further light curve changes, and at least moderate dispersion spectra secured with a large instrument, preferably simultaneously with the photometry. CONCLUSION The two binary systems discussed here were selected from O'Connell's list of more than 50 stars exhibiting the asymmetry at maximum light. It is most remarkable that both should have undergone a shift in the sign of the effect. It is possible that both systems are not properly members of his list, but it seems more likely that the underlying causes for the O'Connell effect themselves change with time, and that the general case Delta m > 0 is probable only. ACKNOWLEDGEMENTS It is a pleasure to thank Dr. A. J. Wesselink for suggesting the problem of the O'Connell Effect and for many valuable discussions, Mr. E. W. McClurken who provided valuable observing assistance in June, 1967, and Dr. Douglas Hall for his RT Lacertae data. Table IV Extreme of the light curves of CG Cygni Source Band Date max I max II d max (II-I) min I min II d min (I-11) Williams (1922) visual 1921 9.93 m 9.94 m +0.01 m 10.42 m (10.15m) (0.27m) Yu (1922) ptg. 1922 9.219 10.264 +0.045 11.374 10.544 0.830 Milstein and Nicolaev (1940) ptg. ~1936 11.02 11.02 <=0.00 11.78 11.29 0.49 Milone (1966, 7) V 1965 10.124 10.060 -0.064 (10.86) 10.429 (0.43) Milone (1966, 7) B 1965 10.990 10.918 -0.072 (11.78) 11.253 (0.53) Milone (1966, 7) U 1965 11.381 11.301 -0.080 (12.42) 11.604 (0.82) Milone (unpubl.) V 1967 10.151 10.069 -0.082 10.737 10.459 0.276 Milone (unpubl.) B 1967 11.004 10.898 -0.106 11.663 11.255 0.408 Milone (unpubl.) U 1967 11.409 11.301 -0.108 12.212 (11.61) (0.60) Milone (unpubl.) V 1965+1967 10.134 10.063 -0.071 -* 10.444 - Milone (unpubl.) B 1965+1967 10.992 10.917 -0.075 -* 11.258 - Milone (unpubl.) U 1965+1967 11.393 11.303 -0.090 -* 11.60 - * Differences are too extreme. Brackets indicate uncertain values. In addition, the visual (~ 1931) and photographic (~ 1951) light curves of Tsesevich (1954) show no discernible O'Connell Effect. The values cited for the early investigations are in local magnitude systems. This work was begun when the author was a graduate student at Yale, and was carried forth with the help of a Creativity and Research Grant of the Lutheran Church in America and Gettysburg College in 1967-1968, a Gettysburg College Faculty Fellowship in June, 1968, and a summer research participation fellowship at the University of Maryland from June to September, 1968, the help of all of which the author gratefully knowledges. REFERENCES Fowler, M., 1920, Astrophys. J. 52, 257. Hall, D. S., 1967a, private communication. Hall, D. S., 1967b, I. A. U. Information Bulletin on Variable Stars, No. 259, Budapest. (IBVS N°.259) Hall, D. S., 1968, private communication. Hardie, R. H., 1962, Photoelectric Reductions in Astronomical Techniques, Stars and Stellar Systems II, 178. Joy, A. H., 1931, Astrophys. J. 74, 101. Krat, W. and Nekrasova, S., 1936, Acta Astron. Ser. C., 2, 129. Kratochvil, P., 1964, Bull. Astron. Inst. Czech. 15, 165. Kukarkin, B. V., Parenago, P. P., Efremov, Yu. I., and Kholopov, P. N., 1958, General Catalogue of Variable Stars, 2nd ed., Moscow. Mergentaler, J., 1950, Wroclaw Contrib. 4. Milone, E. F., 1966, Astron. J. 71, 864. Milone, E. F., 1967, Thesis, Yale University. Milone, E. F., 1968a, Astron. J. 73, S26. Milone, E. F., 1968b, Astron. J., in press. Milstein, I. P. and Nicolaev, S. P., 1940, Vsesoiuznoe Astron.-Geodetic Soc. N. 6, 9. O'Connell, D. J. K., 1951, Riverview Publ. 2, 85. Tsesevich, V. P., 1954, Odessa Izvestia 4, Part I, 255. Wachmann, A. A., 1935, Astron. Nachr. 255, 367. Williams, A. S., 1922, MN 82, 300. Yu, Ch'ing-Sung, 1923, Astrophys. J. 58, 75.