The Astrophysical Journal Supplement Series, 57:63-76,1985 January © 1985. The American Astronomical Society. All rights reserved. Printed in U.S.A. AN EMPIRICAL Hy LUMINOSITY CALIBRATION FOR CLASS V-III STARS 1985ApJS...57...63M Christopher G. Millward and Gordon A. H. Walker Geophysics and Astronomy Department, University of British Columbia Received 1984 March 21; accepted 1984 June 1 ABSTRACT High signal-to-noise Reticon spectra for 87 members of eight open clusters and associations, together with 37 stars having rehable parallaxes (early A-type stars with reliable trigonometric parallaxes, eclipsing binaries, and visual binaries), have been used to calibrate the W(Ry)-Mv relation for spectral types O to early A of luminosity classes III-V. The new calibration has a mean probable dispersion of ± 0.28 mag. The distance modulus of the Pleiades is 5.54 + 0.06 mag, which is in excellent agreement with other, recent determinations, as are the distance moduli for all the calibrating clusters. The use of visual binary parallaxes implies a Hyades distance modulus of ~ 3.0 which is significantly smaller than the Hanson value of 3.30 mag. Although no spectral type corrections are necessary, stellar evolution probably affects the construction of the new calibration, and special care should be taken when determining distance moduli from slightly evolved cluster sequences or for individual stars. Systematic departures from the calibration may be present for stars with F sin / > 220-250 km s-1. Significant residuals are found between our values of W(Hy) and those of Petrie in the range 1-13 À equivalent width, which are due in part to systematic errors in Petrie’s JF(Hy) measures. Our distance modulus of 11.11 mag for NGC 2244 is in excellent agreement with the photometric distance. The new calibration is compared to other early-type star calibrations. It is 1.2 mag brighter than Petrie’s Hy calibration at spectral type 06 and 0.7 mag brighter at A3. For types B1 and earlier, the new calibration averages 0.4 mag brighter than the Balona and Cramp ton Hy calibration. There is generally good agreement with the Blaauw MK calibration, although the latter is 0.4 mag brighter at spectral type B0. The Crawford Hß calibration is up to 0.5 mag brighter for the earlier spectral types and 0.4 mag fainter for later types. Subject headings: clxisttrs: associations—clusters: open — line profiles — stars: early-type — stars: luminosities I. INTRODUCTION Petrie (1950, 1953) produced two calibrations, one for B Since the pioneering work of Adams and Joy (1923), it has stars, the other for A stars. There were systematic differences been recognized that the appearance of various spectral fea- between his luminosities and those from other calibrations tures correlate with stellar luminosity. Williams (1929), Anger such as Johnson and Iriarte (1958) and Blaauw (1963). (1931), and Gunther (1933), among others, were the first to Unfortunately, no OB stars have sufficiently accurate trigo- use the strength of the hydrogen Balmer lines to predict nometric parallaxes to provide a basis for direct luminosity luminosity. Because of the difficulty, at the time, in deriving calibration, and indirect methods must be used. Petrie (1965) quantitative line strengths, luminosity calibrations took a vi- published a single, revised, JF(Hy) calibration covering spec- sual form culminating in the MKK classification scheme of tral types O to early A. The form of the relation was defined Morgan, Keenan, and Kellman (1943). The rather large by fitting a number of early-type galactic cluster W(Hy)-V0 luminosity dispersion of ~ 1 mag for most spectral types in sequences and tying the V0 scale to luminosity using A stars this system is due in part to the discrete nature of the with rehable trigonometric parallaxes. This procedure made classification scheme. the calibration independent of the Hyades distance. The new The equivalent widths of the Balmer lines are dominated by calibration compared better with others than the earlier ver- Stark broadening which, being a function of electron pressure, sions, but there were still important differences with other is directly related to surface gravity. There is also a depen- calibrations. dence on effective temperature. Both of these effects imply Significant residuals continue to exist between the various that, for a main-sequence star, there should be a unique Hy and Hß calibrations, as well as between the Hy and MK relationship between the strength of the Balmer lines and calibrations (especially for spectral types 09-B2), and these luminosity. have been discussed by Walbom (1972), Walker and Hodge With the advent of semiautomatic devices to measure line (1968), Balona and Crampton (1974; hereafter BC), Crawford strengths from photographic spectra, Petrie and collaborators (1978), Turner (1976), Eggen (1977), and Shobbrook (1983a, at Victoria initiated a program to measure the strength of Hy. b\ It is conveniently located for photographic work, is relatively In this paper we report on a reinvestigation of the uncontaminated by line blends, and does not suffer to the JF(Hy)-himinosity relation using high signal-to-noise (S/N) same degree as Hß from emission. Reticon spectra. The high S/N of the digital spectra is a 63 © American Astronomical Society • Provided by the NASA Astrophysics Data System 64 MILLWARD AND WALKER particular advantage since it simplifies the setting of accurate 3. Secondary component certain from present data, litera- continua, as well as helping to compensate for line blends and ture, or both. AK= +0.2 mag. the effects of duplicity. We have followed Petrie’s technique In double-lined spectroscopic binaries where Am, the which he used in his 1965 calibration, except that we have not magnitude difference between the two components, was known included supergiants. They will be the subject of another or estimated, corrections to the IL(Hy) were made following 1985ApJS...57...63M paper. Petrie (1953), with the exception that, /, the secondary-to- primary light ratio, was restricted to / > 0.4 mag instead of II. OBSERVATIONS AND DATA REDUCTION / > 0.5 mag. One hundred seven high S/N, Reticon spectra of 96 mem- The projected rotational velocities were measured from the bers of the Pleiades, a Per, Orion OBI, Cepheus OB3, Cygnus FWHM of the He H A4387, He i 4471, and Mg il A4481 OB7, NGC 2244, NGC 2264, and the UMa moving cluster, as fines. Twenty three standards (Slettebak 1954,1956; Slettebak well as 57 spectra of 36 early-type stars with rehable paral- and Howard 1955) were observed for the He u À4387 and laxes (early A-type stars with reliable trigonometric parallaxes, He i A4471 fines, and 35 for the Mg n A4481 fine. The “new” eclipsing binaries, and visual binaries), were obtained over a Slettebak et al. (1975) system was not used since only 10 stars 14 month period from 1981 September to 1982 November were observed in common. with the 32 inch (0.8 m) coudé camera spectrograph of the Figure 1 shows examples of the Reticon spectra. Using the 1.22 m telescope at the DAO. The spectra were recorded with Reticent command language (Pritchet, Mochnacki, and Yang a liquid nitrogen cooled RL 1872 F/30 Reticon (Walker 1972; 1982), the data reduction sequence was as follows: Walker, Johnson, and Yang 1983). All spectra have a recipro- 1. Baseline (dark) subtraction from a data frame. cal dispersion of 40 A mm-1 (0.6 À per diode) and cover the 2. Division of the resulting data frame by a flat field frame. region 4000-5000 À. Since a major goal of this work was to 3. A third-order polynomial fitted to selected points to determine whether the observed scatter in the Hy luminosity define the continuum. relation is dependent on the quality of the spectrograms used, 4. Rectification and application of a dispersion relation. an effort was made to achieve a S/N > 100 in each point since The spectra have not been filtered. Details of the reduction 100 is the maximum obtainable with photographic methods. procedure are given in Walker, Johnson, and Yang (1983). The general range of S/N in the cluster spectra is 120-1000 Third-order polynomials were fitted to selected Fe/Ar fines per point, with the mean value near 350. In a few instances (20) from comparison arc spectra taken during each night’s signal levels of between only 50-90 were obtained. The trigo- observations to determine the spectral dispersion. nometric group has a mean S/N of 1400, while the eclipsing While, in theory, it is possible to define fine profiles and and visual binary groups average 400 and 450, respectively. integrate equivalent widths digitally, we found it more satis- Table 1 lists the cluster and association members observed, factory to draw in the Hy wing profiles by hand on tracings. and Table 2 the parallax stars. In Table 1, column (2) gives the Care was taken to maintain symmetry about the fine centers. HD number, column (3) the spectral type, column (4) the Areas were measured with a planimeter. Following Petrie (1965) and BC, the contribution of the O u A4349 fine was apparent visual magnitude corrected for absorption, VQ, col- umn (5) the measured W(Hy) in angstroms, column (6) the estimated and subtracted. This was a relatively easy task given measured projected rotational velocity, V sin /, in kilometers the high S/N of the spectra. The high S/N also facilitated the per second, column (7) the signal-to-noise level of the spectra, continuum and fine wing placement. While it is difficult to and column (8) the number of spectra. Table 2 is similar, quantify the procedures employed in this aspect of the work, it except for column (4), which gives the absolute magnitude, is fair to say the interpretive aspect was considerably reduced as compared to using photographic data, and consistent re- Mv, based on our calibration, and column (6), which gives the Pleiades distance modulus, m — M.