astro-ph0201480/ms.tex
1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: %
3: %  Orbit of HD 115071
4: %
5: %
6: %  Revision history:
7: %
8: %     July 27, 2001:  First draft by Gies
9: %     Jan. 14, 2002:  Revised by Gies (Casleo improvements) 
10: %     Jan. 29, 2002:  Submitted to ApJ, assigned No. 55468
11: %                     astro-ph/0201480 (Password: wcyha)
12: %                     ed. J. W. Liebert apjjl@as.arizona.edu
13: %     Mar.  1, 2002:  Referee's report received
14: %     Mar. 20, 2002:  Revisions by Gies (sent to coauthors)
15: %
16: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
17: 
18: %\documentclass{aastex}
19: %\usepackage{emulateapj5}
20: \documentclass[12pt,preprint]{aastex}
21: %\voffset 0.7truein
22: 
23: \newcommand{\myemail}{gies@chara.gsu.edu}
24: 
25: \shorttitle{Massive Close Binary HD~115071}
26: \shortauthors{Penny et al.}
27: 
28: \begin{document}
29: 
30: \received{2002 January 29}
31: \accepted{}
32: 
33: \title{Tomographic Separation of Composite Spectra. IX. \\ 
34: The Massive Close Binary HD~115071} 
35: 
36: \author{Laura R. Penny\altaffilmark{1}}
37: \affil{Department of Physics and Astronomy \\
38:  College of Charleston \\
39:  Charleston, SC 29424; \\
40:  pennyl@cofc.edu}
41: 
42: \author{Douglas R. Gies\altaffilmark{2}}
43: \affil{Center for High Angular Resolution Astronomy and \\
44:  Department of Physics and Astronomy\\
45:  Georgia State University, Atlanta, GA 30303; \\
46:  gies@chara.gsu.edu}
47: 
48: \author{John H. Wise\altaffilmark{3}}
49: \affil{School of Physics \\
50:  Georgia Institute of Technology \\
51:  Atlanta, GA 30332; \\
52:  jwise@astro.psu.edu}
53: 
54: \author{D. J. Stickland, C. Lloyd}
55: \affil{Rutherford Appleton Laboratory \\
56: Chilton, Didcot, Oxon, OX11 0QX, United Kingdom; \\
57: ds@astro1.bnsc.rl.ac.uk, cxl@ast.star.rl.ac.uk} 
58: 
59: \altaffiltext{1}{Guest Observer,
60: Complejo Astronomico El Leoncito (CASLEO), 
61: San Juan, Argentina} 
62: 
63: \altaffiltext{2}{Guest Observer,
64: Mount Stromlo and Siding Springs Observatories, Australia}
65: 
66: \altaffiltext{3}{Current address: 
67: Department of Astronomy and Astrophysics,
68: Pennsylvania State University, 
69: 532 Davey Laboratory, 
70: University Park, PA 16802} 
71: 
72: \slugcomment{Submitted to ApJ.}
73: \paperid{55468}
74: 
75: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
76: 
77: \begin{abstract}
78: We present the first orbital elements for the massive close binary, 
79: HD~115071, a double-lined spectroscopic 
80: binary in a circular orbit with a period of $2.73135 \pm 0.00003$ 
81: days.  The orbital semiamplitudes indicate a mass ratio of 
82: $M_2/M_1 = 0.58 \pm 0.02$ and yet the stars have similar 
83: luminosities.   We used a Doppler tomography algorithm to 
84: reconstruct the individual component optical spectra, 
85: and we applied well known criteria to 
86: arrive at classifications of O9.5~V and B0.2~III for the 
87: primary and secondary, respectively.   We present models 
88: of the {\it Hipparcos} light curve of the ellipsoidal 
89: variations caused by the tidal distortion of the secondary, 
90: and the best fit model for a Roche-filling secondary 
91: occurs for an inclination of $i=48\fdg7 \pm 2\fdg1$.   
92: The resulting masses are  $11.6\pm1.1 M_\odot$ and $6.7\pm 0.7 M_\odot$
93: for the primary and secondary, respectively, so that both 
94: stars are very overluminous for their mass.   
95: The system is one of only a few known semi-detached, 
96: Algol-type binaries that contain O-stars.  We suggest 
97: that the binary has recently emerged from extensive  
98: mass transfer (possibly through a delayed contact and 
99: common envelope process). 
100: \end{abstract}
101: 
102: \keywords{binaries: spectroscopic  --- stars: early-type ---
103: stars: evolution --- stars: individual (HD~115071)}
104: 
105: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
106: 
107: \section{Introduction}                              % Section 1
108: 
109: The hot, massive star, HD~115071 (V961~Cen, LS~2998, HIP~64737), 
110: is found in the sky close to the open cluster, Stock~16 \citep{tur85}, 
111: and is classified as O9.5~V by \citet{hou75} and B0.5~Vn 
112: by \citet{gar77}.  The star is not a known visual binary 
113: \citep{mas98} but early measurements by spectroscopists 
114: indicated it is radial velocity variable and a
115: probable spectroscopic binary \citep{cru74,con77}.  
116: The proof of its binary 
117: nature came relatively recently in studies by \citet{pen96} and 
118: \citet{how97}.  Both papers presented a cross-correlation 
119: analysis of a single, high dispersion, UV spectrum made 
120: with the {\it International Ultraviolet Explorer Satellite}
121: ({\it IUE}) that demonstrated that the system is in fact 
122: a double-lined binary.   \citet{sti01} measured the 
123: radial velocities of the components in this spectrum and
124: proposed an orbital period of 2.73126~d based upon a 
125: light curve constructed from {\it Hipparcos} photometry.  
126: \citet{llo01} present a model of the light curve, and 
127: they argue that the system has evolved through Case~A 
128: mass transfer (commencing during core H burning of the donor star). 
129: 
130: The details and outcomes of Roche lobe overflow (RLOF) in massive 
131: binaries are still subjects of considerable debate \citep{wel01}, 
132: and thus, the orbital and physical parameters of a system like 
133: HD~115071 are of great interest.  Here we present the first
134: double-lined orbital solution for the binary (\S3) based upon 
135: new high quality optical spectra.  We apply a version of 
136: the Doppler tomography algorithm (which we have used to good 
137: effect with UV spectra in prior papers in this series) 
138: to reconstruct the individual spectra of both components, 
139: from which we determine the spectral classifications, 
140: projected rotational velocities, and flux ratio (\S4). 
141: We also present a light curve analysis constrained by 
142: the spectroscopic results that allows us to estimate  
143: the stellar masses (\S5).  These masses are much lower 
144: than expected, and we discuss the evolutionary implications 
145: in \S6.  
146: 
147: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
148: 
149: \section{Observations and Reductions}               % Section 2
150:   
151: Our spectra were obtained in two observing runs at different sites. 
152: The first set was obtained with the 2.15-m telescope of the 
153: Complejo Astronomico El Leoncito (CASLEO) and REOSC echelle spectrograph
154: (on loan from the Institut d'Astrophysique, Universite de Liege, Belgium) 
155: during the period 1997 March 19 -- 28.  The REOSC spectrograph uses 
156: an echelle grating with 70 grooves~mm$^{-1}$ and blazed at 226434 \AA 
157: ~together with a cross disperser grating of 400 grooves~mm$^{-1}$ 
158: blazed at 4000 \AA .  The detector was a TEK $1024\times 1024$ 
159: CCD with 24$\mu$m square pixels used with a gain of 1.98 e$^-$/ADU 
160: (read noise of 7.4 e$^-$).  We used a 200$\mu$m slit that 
161: corresponds to $2\arcsec$ on the sky.  This arrangement produced 
162: an echellogram from which we extracted 23 orders, spanning the range
163: from 3575 to 5700 \AA ~with a resolving power of 
164: $\lambda / \Delta \lambda = 13000$.   We usually obtained 3 exposures 
165: of 660~s duration that were later co-added in software to 
166: improve the S/N ($\approx 150$ per pixel in the better exposed 
167: portions of the spectrum).   Numerous bias, flat field, dark, and Th-Ar
168: comparison images were obtained each night.    
169: 
170: Our second observing run took place at the 74-inch telescope 
171: at Mount Stromlo Observatory over the period 1998 April 6 -- 14. 
172: These spectra were made with the coude spectrograph using grating C
173: (600 grooves per mm, blazed at 12500\AA ~in first order)
174: in third order with a BG12 order sorting filter.  The detector was a
175: SITe CCD (D14) with 15 $\mu$m square pixels in a $4096\times 2048$
176: format.   This arrangement produced single order spectra 
177: that covered the range 3804 -- 4220 \AA ~with a
178: reciprocal dispersion of 0.10 \AA ~per pixel and a resolution
179: element of 0.30 \AA ~FWHM ($\lambda / \Delta \lambda = 13400$).   
180: Exposure times were usually 45 minutes, and the final spectra 
181: have a typical S/N = 160 per pixel in the continuum.
182:  
183: The spectra were reduced
184: using standard routines in IRAF\footnote{IRAF is distributed by the
185: National Optical Astronomy Observatories, which is operated by
186: the Association of Universities for Research in Astronomy, Inc.,
187: under cooperative agreement with the National Science Foundation.}.
188: The MSO single-order spectra were extracted, calibrated, and 
189: flux rectified with the task {\it doslit}.
190: The CASLEO echelle spectra were traced, extracted, and wavelength 
191: calibrated using the task {\it doecslit}, and the extracted orders 
192: were rectified to a unit continuum by fitting a high order spline 
193: function to line-free regions (using the task {\it continuum}). 
194: Finally the individual orders were linked together with the 
195: task {\it scombine}.   Small amplitude irregularities related to the
196: fitting of the echelle blaze function were evident in the 
197: continuum, and the same residual pattern was seen in all spectra made on a
198: given night.  We were able to remove most of the pattern 
199: by dividing the target spectrum by a correction spectrum formed from 
200: spectra of B-star, $\tau$~Sco, which was also observed each night.  
201: The correction spectrum was a smoothed version of the particular
202: night's $\tau$~Sco spectrum divided by a global average 
203: representation of this star's stellar spectrum.  
204: The spectra from each run were then collected and  
205: transformed onto their respective heliocentric wavelength grids.
206: 
207: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
208: 
209: \section{Radial Velocities and Orbital Elements}    % Section 3
210: 
211: Our procedure for measuring radial velocities in {\it IUE} spectra 
212: \citep{pen97} involves fitting Gaussians to the cross-correlation 
213: functions of the target spectrum with a narrow-lined reference spectrum. 
214: The optical spectra we consider here have many fewer stellar lines
215: and much better S/N than the {\it IUE} spectra, so we revised 
216: our techniques accordingly.  First, we fit each absorption feature 
217: separately rather than fitting the entire spectrum through 
218: one cross-correlation measurement.  Secondly, we made the fit of
219: the composite profiles using spectral templates rather than  
220: Gaussian functions (since the lines have shapes dominated by 
221: linear Stark broadening or rotational broadening and since some
222: lines may contain weak blends).   The templates
223: were formed from spectra we obtained during each run of the 
224: star, HD~57682 (O9~IV; \citet{wal72}).  This star is a reasonable  
225: match in classification to both components in HD~115071 (\S4), 
226: but has narrower lines  ($V \sin i = 33$ km~s$^{-1}$; \citet{pen96}). 
227: The radial velocity of this star was measured by parabolic 
228: fitting of the line cores for lines in the list of \citet{bol78}, 
229: and we found an average radial velocity of $25.0 \pm 0.5$ and 
230: $26.0 \pm 1.1$ km~s$^{-1}$ from the CASLEO and MSO spectra, 
231: respectively.  The averaged template spectra from each run 
232: were shifted by these values to place them in the rest frame.  
233: Next, we artificially broadened each template spectrum 
234: by convolution with a rotational broadening function 
235: to produce profiles that matched the spectral components
236: of HD~115071 in the best separated quadrature spectra.  
237: We also used these resolved profiles to estimate the 
238: line depth ratio between the components.   Once these 
239: fitting parameters were set, we determined the radial velocities
240: of each component for a given line by a least-squares fit of
241: the observed profile with the coaddition of the two template profiles 
242: shifted in wavelength to obtain the best match.  This approach 
243: provided good fits of the observed profiles for all but 
244: two cases (HJD 2,450,529.792 and 2,450,531.755) 
245: where the line depth ratio appeared to be reversed. 
246: 
247: We used this technique to measure radial velocities
248: for the strongest lines in the spectrum, specifically
249: \ion{H}{1} $\lambda\lambda  3835$, 3889, 3970, 4101, 4340, 4861, 
250: \ion{He}{1} $\lambda\lambda 3819$, 4009, 4026, 4121, 4143, 4387, 4471, 4921, 5015, 
251: \ion{He}{2} $\lambda 4686$, and \ion{Si}{4} $\lambda 4089$.
252: There was no evidence of systematic line-to-line 
253: differences in the radial velocity measurements, and so 
254: no line specific corrections were applied. 
255: The radial velocities from all the available lines were averaged 
256: together after deletion of any very discrepant measurements. 
257: Finally, we made small adjustments to these averages based 
258: on measurements of the strong interstellar \ion{Ca}{2} 
259: $\lambda\lambda 3933, 3968$ lines.   
260: An interstellar spectrum was formed by extracting the
261: mean spectrum in the immediate vicinity of each interstellar
262: absorption line.  
263: (We made Gaussian fits of the interstellar \ion{Ca}{2} profiles 
264: in the extracted spectra, and we found the radial velocity 
265: was $-17.0\pm 0.2$ and $-16.0 \pm 0.2$ km~s$^{-1}$ for the
266: mean CASLEO and MSO spectra, respectively.)  
267: We then cross correlated this spectrum with
268: each individual spectrum to measure any small deviations in
269: our wavelength calibration (generally $<3$~km~s$^{-1}$),
270: and these small corrections were applied to the mean velocities. 
271: Table~1 lists the 
272: heliocentric dates of mid-observation, orbital phase, 
273: and for each component, the mean radial velocity, the standard 
274: deviation of the mean, the observed minus calculated residual 
275: from the orbital fit, and the number of lines used in the mean. 
276: Table~1 also gives the radial velocities from the single {\it IUE} 
277: spectrum measured by \citet{sti01} (adjusted for the ISM velocity 
278: on the MSO system).  
279: 
280: \placetable{tab1}      % Table 1 - Radial Velocities
281: 
282: \citet{sti01} and \citet{llo01} found that the {\it Hipparcos} 
283: light curve was best fit with a double sine, ellipsoidal 
284: variation for an orbital period $P=2.73126 \pm 0.00009$~d.  
285: We found that this period also agreed reasonably well 
286: with our radial velocity data.  We used the non-linear, least-squares 
287: fitting program of \citet{mor74} to solve for the period and 
288: other orbital elements for the primary (the more luminous and 
289: massive star) and secondary components separately, and this 
290: yielded period estimates of $2.73149 \pm 0.00007$ and 
291: $2.73138 \pm 0.00015$~d, respectively.  We made one additional 
292: calculation of the period by dividing the difference between 
293: the best fit time of the {\it Hipparcos} photometry maximum and
294: our spectroscopically determined time of quadrature by the closest integral 
295: number of cycles, and this led to a period of $2.73130 \pm 0.00004$~d. 
296: We adopted the error weighted mean of these three estimates for 
297: our working value of the period, $P=2.73135 \pm 0.00003$~d. 
298: 
299: We fixed this period and then fit for the remaining orbital 
300: elements independently for both components.  The fitted 
301: epoch of primary maximum velocity, $T_0$, was the same within 
302: errors for both solutions, and so we applied the mean value 
303: to fits of both components.  Eccentric solutions produced 
304: estimates of eccentricity consistent with a value of zero, 
305: and our final solutions in Table~2 assume circular motion. 
306: The observed and calculated radial velocity curves appear 
307: in Figure~1.  The only major discrepancies occur in the 
308: {\it IUE} measurements (not used in the solution), 
309: both of which are $\approx 38$ km~s$^{-1}$
310: above the predicted curve.  Note that in the case of the primary, 
311: the {\it IUE} velocity falls well above the maximum for the 
312: entire curve, so the mismatch cannot be due to an incorrect 
313: orbital phase for example.   The systematic difference may be related 
314: to line formation at different heights in an expanding atmosphere 
315: or orbital motion about a distant, unseen, tertiary star.
316: 
317: \placetable{tab2}      % Table 2 - Orbital Elements 
318: 
319: \placefigure{fig1}     % Figure 1 - Radial velocity curve
320: 
321: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
322: 
323: \section{Tomographic Reconstruction}                % Section 4
324: 
325: We used the Doppler tomography algorithm described by 
326: \citet{bag94} to reconstruct the individual primary and 
327: secondary spectra independently from the CASLEO and MSO 
328: spectra.  We took the radial velocity shifts for each component 
329: from the orbital solutions in Table~2, then the reconstruction 
330: was run for 50 iterations with a gain of 0.8 (the results 
331: are insensitive to both parameters).  The reconstructed 
332: spectra are plotted in in Figure 2 in a format similar 
333: to that used in the spectral atlas of \citet{wal90}.  The 
334: reconstructions from the MSO spectra are shown just above those from 
335: the CASLEO spectra (in the short wavelength portion of Fig.~2), 
336: and there is good agreement between these two sets of spectra.
337: 
338: \placefigure{fig2}     % Figure 2 - Reconstructed spectra
339: 
340: We compared the reconstructed spectra with the spectrum 
341: standards in the atlas of \citet{wal90} to determine the 
342: spectral classifications of the components.   
343: The strengths of the \ion{He}{1} $\lambda\lambda 4026,4143,4387$ 
344: lines relative to those of \ion{He}{2} $\lambda\lambda 4200, 4541$
345: are all consistent with a spectral type of O9.5 for the primary. 
346: The ratio of the \ion{Si}{4} $\lambda\lambda 4088,4116$ lines
347: to the nearby \ion{He}{1} $\lambda\lambda 4121,4143$ features 
348: indicate a main sequence class, as does the relatively strong 
349: \ion{He}{2} $\lambda 4686$ to \ion{He}{1} $\lambda 4713$ ratio. 
350: Thus, we classify the primary as type 
351: O9.5~V, and we compare its spectrum in Figure~2 to that of 
352: HD~93027, which is given as the standard of this class  
353: in \citet{wal90}. 
354: 
355: The secondary, on the other hand, has features indicating a 
356: cooler temperature and later type.  The ratio of  
357: \ion{Si}{3} $\lambda 4552$ to \ion{Si}{4} $\lambda 4088$
358: has a good match in the interpolated type B0.2 introduced 
359: by \citet{wal90}.   The relative strength of the  
360: \ion{Si}{4} $\lambda\lambda 4088,4116$ lines compared to the 
361: neighboring \ion{He}{1} $\lambda\lambda 4121,4143$ features
362: clearly leads to a luminosity class III.  Figure~2 
363: illustrates the good agreement between the spectrum of the 
364: secondary and that of HD~108639 that \citet{wal90} use as 
365: a standard for type B0.2~III.  The \ion{C}{3} $\lambda\lambda 
366: 4070, 4650$ blends appear to be somewhat weaker in the 
367: secondary's spectrum than in the standard spectrum (evidence, 
368: perhaps, of CNO-processed gas in the secondary's photosphere). 
369: 
370: The two spectral standards, HD~93027 and HD~108639, provided 
371: us with the means to estimate the visual flux ratio, 
372: $r=F_2/F_1$, by matching the line depths in the reconstructed
373: spectra with those in the standards.   This was done by 
374: aligning the reconstructed and standard spectra, 
375: adjusting for differences in the placement of the continuum,
376: Gaussian smoothing of the spectra to eliminate 
377: differences in projected rotational velocity and instrumental 
378: broadening, and then finding a best fit line ratio that 
379: allocates a proportion of flux to each component to 
380: best match the line depths.   We found $r=1.04 \pm 0.06$ and 
381: $1.08\pm 0.08$ for the MSO and CASLEO reconstructions, respectively. 
382: 
383: Finally, we used the profiles in the reconstructed spectra 
384: to estimate the projected rotational velocities of the 
385: components.  We focused on the \ion{Si}{4} $\lambda 4088$
386: profile for this purpose since it represents the strongest 
387: metallic line (intrinsically narrow) in the range covered 
388: by the MSO spectra.   Our procedure involved calculating a 
389: grid of rotational broadening functions for a linear 
390: limb darkening law \citep{wad85,gra92} and then  
391: convolving an observed narrow-lined spectrum with 
392: these broadening functions.   We compared the spectral reconstructions 
393: from the MSO spectra with broadened versions of MSO spectra of
394: the narrow-lined stars HD~53682 (O9~IV) and $\tau$~Sco (B0.2~V).   
395: The best fitting profile matches were made with 
396: $V\sin i = 101\pm 10$ and $132\pm 15$ km~s$^{-1}$ for the 
397: primary and secondary, respectively.   These agree 
398: within errors with estimates from the {\it IUE}
399: observation \citep{pen96,how97,llo01}. 
400: 
401: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
402: 
403: \section{Light Curve Analysis and Masses}           % Section 5
404: 
405: \citet{llo01} presented an analysis of the {\it Hipparcos}
406: light curve \citep{per97}, and here we update their work 
407: by restricting a number of the fitting parameters based upon 
408: the new spectroscopic results.  
409: We used the light curve synthesis code GENSYN \citep{moc72} 
410: to produce model $V$-band differential light curves (almost identical to
411: differential {\it Hipparcos} $Hp$ magnitudes for hot stars). 
412: The orbital parameters were taken from the spectroscopic solution,
413: and the physical parameters were estimated from the spectral
414: classifications of the stars.   
415: We first estimated the stellar temperature 
416: and gravity according to the spectral classification
417: calibration of \citet{how89} for the primary 
418: ($T_{{\rm eff}~1} = 32$~kK, $\log g_1 = 3.9$), 
419: and for the secondary, we used data for comparable stars 
420: in the compilation of \citet{und82} 
421: ($T_{{\rm eff}~2} = 29$~kK, $\log g_2 = 3.6$).
422: We then determined the physical fluxes and limb darkening coefficients 
423: from tables in \citet{kur94} and \citet{wad85}, respectively.
424: We also used the Kurucz flux models to transform our observed 
425: flux ratio based upon the relative line depths into a 
426: $V$-band flux ratio \citep{pen97}.
427: The MSO spectra are centered at 4009~\AA, and the transformation 
428: yields a $V$-band flux ratio, $F_2/F_1 = 1.05 \pm 0.06$. 
429: The comparison of line depths in the CASLEO spectra was made over
430: the available range in the standard spectrum from \citet{wal90} 
431: (centered at 4350~\AA ), and the resulting $V$-band flux ratio
432: is $F_2/F_1 = 1.09 \pm 0.06$.  We used the average value, 
433: $F_2/F_1 = 1.07 \pm 0.06$, in the light curve analysis.  
434: The theoretical and observed flux ratios together 
435: yield an approximate estimate of
436: the ratio of stellar radii, $R_2/R_1 = 1.12 \pm 0.03$. 
437: Each trial run of GENSYN was set by two independent
438: parameters, the system inclination $i$ and secondary's 
439: radius relative to the critical Roche-filling case 
440: (with the primary radius set so that the orbital average 
441: flux ratio matched the observed flux ratio).
442: 
443: The observed light curve (Fig.~3) is a double-sine wave 
444: caused by tidal distortion in the stars.  Since the 
445: stars have similar radii but the secondary has a much lower mass (\S3),
446: the secondary must be much closer to filling its critical 
447: Roche radius, so that the tidal generation of the light curve 
448: is due mainly to the distortion of the secondary. 
449: The amplitude of the photometric variation is proportional 
450: to the degree of tidal distortion (how close the secondary 
451: comes to filling its Roche volume) and to the sine of the 
452: inclination (maximal effect for $i=90^\circ$). 
453: Our first fit of the light curve assumed that the secondary 
454: completely fills its Roche volume, so this solution corresponds 
455: to the case of minimum inclination (and maximum masses).   
456: The best fit for this semi-detached configuration is made with 
457: an inclination, $i=48\fdg7 \pm 2\fdg1$, 
458: and this fit is shown as the solid line in Figure~3.   
459: The error in the inclination results from two sources, 
460: the variation in the $\chi^2$ residuals of the fit with parameter $i$ 
461: and the change in the solution introduced by the uncertainty
462: in the flux ratio.   The root mean square of the residuals from 
463: the best fit is 0.019~mag, which is approximately $1.7\times$
464: larger than the errors quoted in the {\it Hipparcos} catalog, 
465: and so some other kind of photometric variation may exist that  
466: is unrelated to orbital phase. 
467: 
468: \placefigure{fig3}     % Figure 3 - Hipparcos light curve
469: 
470: Note that it is possible to obtain fits 
471: with a lower inclination if the flux ratio constraint is 
472: abandoned.  For example, we found that if we assumed a 
473: contact configuration in which both stars fill their Roche volumes, 
474: then we could make a satisfactory fit of the light curve with 
475: $i=38^\circ$.  However, we rule out this model because it predicts a
476: flux ratio, $F_2/F_1 = 0.52$, that is far below the limits 
477: established from the spectra of the components. 
478: 
479: Models with a smaller secondary and less tidal distortion require 
480: a higher inclination to match the observations (yielding lower masses), 
481: but these solutions are less satisfactory for two reasons.  
482: First, higher inclination solutions generally yield light curves with 
483: less ellipsoidal variation but some evidence of eclipses. 
484: We show one example in Figure~3 for an inclination  
485: $i=60^\circ$ and a secondary volume radius of     
486: $R_2/R_\odot = 5.6$ ($\approx 90\%$ of the critical Roche radius).
487: Eclipses as subtle as those shown in Figure~3 are probably 
488: not ruled out by the {\it Hipparcos} photometry, but 
489: models with $i > 62^\circ$ show eclipses that are clearly 
490: inconsistent with the {\it Hipparcos} light curve.  
491: Secondly, the projected rotational velocities predicted 
492: by underfilling models with synchronous rotation are much smaller 
493: than the observed values.  All the known binaries containing 
494: O-stars with periods this short have circular orbits 
495: \citep{mas98}, and we expect that such close systems 
496: have attained synchronous rotation as well \citep{cla97}.
497: The predicted projected rotational velocities are 
498: $V\sin i = 92$ and 109 km~s$^{-1}$ for the 
499: primary and secondary, respectively, in the Roche-filling model, 
500: in agreement within errors with the observed values (\S4). 
501: However, the match is worse in higher inclination models 
502: ($V\sin i = 81$ and 94 km~s$^{-1}$, respectively, for the 
503: $i=60^\circ$ model illustrated in Fig.~3).   Thus, we prefer 
504: the secondary Roche-filling model, and we list in Table~3 the 
505: corresponding stellar parameters.  
506: The system absolute magnitude in this model 
507: is $M_V = -4.57$, and, for $V=7.94$ and $E(B-V)=0.50$ 
508: \citep{tur85}, we estimate a distance of $1.5\pm0.2$~kpc
509: (smaller than but comparable to the distance of 1.9~kpc for 
510: the cluster Stock~16; \citet{tur85}). 
511: 
512: \placetable{tab3}      % Table 3 - stellar properties 
513: 
514: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
515: 
516: \section{Discussion}                                % Section 6
517: 
518: The first striking result from our analysis is the very low 
519: mass we find for both components.  The stars have temperatures 
520: and luminosities that are associated with masses of 18 and 
521: $15 M_\odot$ for the primary and secondary, respectively, in the 
522: single star evolutionary tracks calculated by \citet{sch92}.
523: (These estimates would be slightly reduced using evolutionary
524: models that include rotation; \citet{heg00}, \citet{mey00}.) 
525: The secondary, in particular, has a luminosity characteristic 
526: of a star more than twice as massive than we find (Table 3).    
527: 
528: The second remarkable fact is that the secondary star 
529: has a spectral classification indicating it has 
530: evolved away from the main sequence.   Thus, HD~115071 presents 
531: the classical ``Algol paradox'' that the lower mass 
532: component is the more evolved one, and we suggest the same 
533: solution of the paradox holds here as well, i.e., that the 
534: evolved component was originally the more massive object, 
535: but suffered significant mass transfer to its neighbor. 
536: 
537: There are only a small number of O-stars that are known 
538: to be members of interacting binaries, and we compare 
539: in Table~4 the properties of the components in HD~115071 
540: with those of the four other known semi-detached binaries 
541: that contain O-type stars \citep{hil87,hh98}.  We excluded from 
542: this list contact or over-contact systems and those binaries 
543: in which both components are evolved \citep{van98}.  All the 
544: systems in Table~4 share a number of common properties: the mass 
545: donor appears as an evolved star, the donor star is overluminous 
546: for its mass, the donor fills its Roche volume, and the mass
547: gainer is a late O-type, main sequence star.
548: It is remarkable that all the donor stars have comparable 
549: luminosity, $\log L/L_\odot \approx 4.5$, despite their wide 
550: range in mass and radius.   Evolutionary models generally 
551: predict that the post-RLOF luminosity of the donor is 
552: comparable to its zero age main sequence (ZAMS) 
553: luminosity \citep{van98,wel01}, 
554: and so these donors probably began life as B0~V stars 
555: with masses in the range 14 -- $20 M_\odot$.  Since the 
556: donors were originally the more massive component, the 
557: gainers were probably also B-type stars that were 
558: promoted to their current O-type status through mass transfer. 
559: It is also curious that no semi-detached systems are known with 
560: primaries earlier than type O8~V.  Either this stage is extremely 
561: rapid in more massive systems or the donor stars take on a 
562: different appearance than they do in Algol-type systems 
563: (perhaps as a O-star plus Wolf-Rayet star binary; \citet{van98}).
564: 
565: \placetable{tab4}      % Table 4 - semi-detached binaries 
566: 
567: Evolutionary models give us some guidance about the initial 
568: masses in HD~115071.  \citet{del94} give a relationship between 
569: the final, post-RLOF mass and the initial ZAMS mass, and this 
570: yields an estimate of $14.8 M_\odot$ for the initial mass 
571: of the donor star.  If we further assume that 50\% of the 
572: donor's mass loss was accreted by gainer and the rest lost 
573: from the system \citep{meu89,del94}, then the original total 
574: mass was $22.4 M_\odot$ and the original gainer mass was $7.6 M_\odot$. 
575: Thus, the system probably began with a relatively low mass ratio, 
576: $M_g/M_d \approx 0.5$. 
577: 
578: The theoretical models of binary evolution by \citet{wel01} offer 
579: some guidance in the interpretation of our results.   
580: \citet{wel01} describe the evolution of several very close systems 
581: that begin RLOF during core H-burning (Case A).   Their 
582: models suggest that a mass reversal similar to what we find 
583: in HD~115071 can occur in Case A, but the resulting systems 
584: generally have a much wider orbit and more extreme mass ratio. 
585: Another possibility is that the system began RLOF after 
586: completion of core H burning (Case B).  The initial period 
587: would have been much larger, but the system then shrunk to 
588: its current dimensions during a common envelope phase 
589: in which the donor's envelope would have been ejected from 
590: the system.  This would explain the current mass and 
591: luminosity of the donor star, but it does not account for 
592: the huge overluminosity of the contemporary primary star, 
593: which is the most overluminous star of any of the gainers 
594: in Table 4.   \citet{wel01} point out one other hybrid 
595: scheme they call ``delayed contact'' in which mass transfer 
596: begins conservatively until the donor develops a convective 
597: envelope and the binary enters the common envelope stage.  
598: This scenario would explain the observed low mass of the 
599: donor and the short orbital period, and the overluminosity 
600: of the gainer would result from compression and/or mixing 
601: related to mass accretion.    
602: 
603: The best fit of the light curve suggests that the secondary 
604: donor star is Roche-filling, and so the system may 
605: still be experiencing active mass transfer.  Observations 
606: of any H$\alpha$ emission \citep{tha97} or IR excess \citep{geh95} 
607: would provide valuable clues about the mass loss and/or 
608: mass transfer processes that might be occurring presently 
609: in this exceptional binary system. 
610: 
611: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
612: 
613: \acknowledgments
614: 
615: We thank the staffs of CASLEO and MSO for their assistance 
616: in making these observations.   We are grateful to Norbert 
617: Langer for comments on the evolutionary state of the binary.  
618: Institutional support for L.R.P.
619: has been provided from the College of Charleston School
620: of Sciences and Mathematics.  Additional support for L.R.P.\ was
621: provided by the South Carolina NASA Space Grant Program and
622: NSF grant AST-9528506.
623: Institutional support for D.R.G. has been provided from the GSU College
624: of Arts and Sciences and from the Research Program Enhancement
625: fund of the Board of Regents of the University System of Georgia,
626: administered through the GSU Office of the Vice President
627: for Research.  We gratefully acknowledge all this support.
628: 
629: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
630: 
631: % References
632: 
633: \clearpage
634: 
635: \begin{thebibliography}{}
636: \bibitem[Bagnuolo et al.(1994)]{bag94}
637:          Bagnuolo, W. G., Jr., Gies, D. R.,
638:          Hahula, M. E., Wiemker, R., \&
639:          Wiggs, M. S. 1994, \apj, 423, 446
640: \bibitem[Bolton \& Rogers(1978)]{bol78}
641:          Bolton, C. T., \& Rogers, G. L. 1978, \apj, 222, 234
642: \bibitem[Claret \& Cunha(1997)]{cla97}
643:          Claret, A., \& Cunha, N. C. S. 1997, \aap, 318, 187
644: \bibitem[Conti et al.(1977)]{con77}
645:          Conti, P. S., Leep, E. M., \& Lorre J. J.  1977, \apj, 214, 759
646: \bibitem[Cruz-Gonz\'{a}lez et al.(1974)]{cru74}
647:          Cruz-Gonz\'{a}lez, C., Recillas-Cruz, E., Costero, R., 
648:          Peimbert, M., \& Torres-Peimbert, S. 1974, 
649:          Revista Mexicana Astr. Astrof., 1, 211
650: \bibitem[de Loore \& Vanbeveren(1994)]{del94}
651:          de Loore, C., \& Vanbeveren, D. 1994, \aap, 292, 463
652: \bibitem[Garrison et al.(1977)]{gar77}
653:          Garrison, R. F., Hiltner, W. A., \& Schild, R. E. 
654:          1977, \apjs, 35, 111
655: \bibitem[Gehrz et al.(1995)]{geh95}
656:          Gehrz, R. D., et al. 1995, \apj, 439, 417
657: \bibitem[Gray(1992)]{gra92}
658:          Gray, D. F. 1992, The Observation and Analysis of Stellar Photospheres
659:          (2nd ed.) (Cambridge: Cambridge Univ. Press) 
660: \bibitem[Harries \& Hilditch(1998)]{hh98}
661:          Harries, T. J., \& Hilditch, R. W. 1998, in 
662:          Boulder-Munich II: Properties of Hot, Luminous Stars 
663:          (ASP Conf.\ Ser.\ Vol.\ 131), ed.\ I. D. Howarth 
664:          (San Francisco: ASP), 401 
665: \bibitem[Harries et al.(1997)]{har97}
666:          Harries, T. J., Hilditch, R. W., \& Hill, G.\ 1997, \mnras, 285, 277
667: \bibitem[Harries et al.(1998)]{har98}
668:          Harries, T. J., Hilditch, R. W., \& Hill, G.\ 1998, \mnras, 295, 386
669: \bibitem[Heger \& Langer(2000)]{heg00}
670:          Heger, A., \& Langer, N. 2000, \apj, 544, 1016
671: \bibitem[Hilditch \& Bell(1987)]{hil87}
672:          Hilditch, R. W., \& Bell, S. A. 1987, \mnras, 229, 529
673: \bibitem[Houk \& Cowley(1975)]{hou75}
674:          Houk, N., \& Cowley, A. P. 1975, 
675:          Catalogue of two dimensional spectral types for the HD stars, Vol. 1
676:          (Ann Arbor: Univ. Michigan)
677: \bibitem[Howarth \& Prinja(1989)]{how89}
678:          Howarth, I. D., \& Prinja, R. K. 1989, \apjs, 69, 527
679: \bibitem[Howarth et al.(1997)]{how97}
680:          Howarth, I. D., Siebert, K. W., Hussain, G. A. J., \& Prinja, R. K. 
681:          1997, \mnras, 284, 265
682: \bibitem[Kurucz(1994)]{kur94}
683:          Kurucz, R.~L.\ 1994, Solar Abundance Model Atmospheres for
684:          0, 1, 2, 4, 8 km/s, Kurucz CD-ROM No.\ 19 (Cambridge, MA:
685:          Smithsonian Astrophysical Obs.)
686: \bibitem[Lloyd \& Stickland(2001)]{llo01}
687:          Lloyd, C., \& Stickland, D. J. 2001, \aap, 370, 1026
688: \bibitem[Lorenz et al.(1994)]{lor94}
689:          Lorenz, R., Mayer, P., \& Drechsel, H. 1994, \aap, 291,185
690: \bibitem[Mason et al.(1998)]{mas98}
691:          Mason, B. D., Gies, D. R., Hartkopf, W. I., Bagnuolo, W. G., Jr.,
692:          ten Brummelaar, T., \& McAlister, H. A. \ 1998, \aj, 115, 821
693: \bibitem[Meurs \& van den Heuvel(1989)]{meu89}
694:          Meurs, E. J. A., \& van den Heuvel, E. P. J. 1989, \aap, 226, 88
695: \bibitem[Meynet \& Maeder(2000)]{mey00}
696:          Meynet, G., \& Maeder, A. 2000, \aap, 361, 101
697: \bibitem[Mochnacki \& Doughty(1972)]{moc72}
698:          Mochnacki, S. W., \& Doughty, N. A. 1972, \mnras, 156, 51
699: \bibitem[Morbey \& Brosterhus(1974)]{mor74}
700:          Morbey, C. L., \& Brosterhus, E. B. 1974, \pasp, 86, 455
701: \bibitem[Penny(1996)]{pen96}
702:          Penny, L. R. 1996, \apj, 463, 737
703: \bibitem[Penny et al.(1997)]{pen97} 
704:          Penny, L. R., Gies, D. R., \& Bagnuolo, W. G., Jr.
705:          1997, \apj, 483, 439
706: \bibitem[Perryman(1997)]{per97} 
707:          Perryman, M. A. C. 1997, The Hipparcos and Tycho
708:          Catalogues, ESA SP-1200 (ESA/ESTEC, Noordwijk)
709: \bibitem[Schaller et al.(1992)]{sch92}
710:          Schaller, G., Schaerer, D., Meynet, G., \& Maeder, A. 1992, 
711:          \aaps, 96, 269
712: \bibitem[Stickland \& Lloyd(2001)]{sti01}
713:          Stickland, D. J., \& Lloyd, C. 2001, Observatory, 121, 1
714: \bibitem[Thaller(1997)]{tha97}
715:          Thaller, M. L. 1997, \apj, 487, 380
716: \bibitem[Turner(1985)]{tur85}
717:          Turner, D. G. 1985, \apj, 292, 148
718: \bibitem[Underhill \& Doazan(1982)]{und82}
719:          Underhill, A., \& Doazan, V. (eds.) 1982, 
720:          B Stars With and Without Emission Lines 
721:          (NASA SP-456) (Washington, DC: NASA)
722: \bibitem[Wade \& Rucinski(1985)]{wad85}
723:          Wade, R.~A., \& Rucinski, S.~M.\ 1985, \aaps, 60, 471
724: \bibitem[Walborn(1972)]{wal72}
725:          Walborn, N. R. 1972, \aj, 77, 312
726: \bibitem[Walborn \& Fitzpatrick(1990)]{wal90}
727:          Walborn, N. R., \& Fitzpatrick, E. L. 1990, \pasp, 102, 379
728: \bibitem[Wellstein et al.(2001)]{wel01}
729:          Wellstein, S., Langer, N., \& Braun, H. 2001, \aap, 369, 939
730: \bibitem[Vanbeveren et al.(1998)]{van98}
731:          Vanbeveren, D., van Rensbergen, W., \& de Loore, C. 1998,
732:          The Brightest Binaries (Dordrecht: Kluwer) 
733: \end{thebibliography}
734: 
735: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
736: 
737: % Figures
738: 
739: \clearpage
740: 
741: % Figure 1
742: \begin{figure}
743: %\plotone{f1.eps}
744: \caption{
745: The radial velocity measurements 
746: ({\it primary -- filled circles; secondary -- open circles}) 
747: and orbital solution ({\it solid lines}) plotted against
748: orbital phase.  Phase zero corresponds to the time of primary 
749: maximum radial velocity. The two plus marks show the {\it IUE}
750: measurements that were not used in the solution.}
751: \label{fig1}
752: \end{figure}
753: 
754: % Figure 2
755: \begin{figure}
756: %\plotone{f2.eps}
757: \caption{
758: A comparison of the reconstructed 
759: MSO spectra ({\it above}) and CASLEO spectra ({\it below}) 
760: of the primary and secondary with spectra of the same classifications 
761: from \citet{wal90}.  All the spectra were Gaussian smoothed 
762: to a nominal resolution of 1.2 \AA ~FWHM for consistent line broadening.}
763: \label{fig2}
764: \end{figure}
765: 
766: % Figure 3
767: \begin{figure}
768: %\plotone{f3.eps}
769: \caption{
770: The {\it Hipparcos} light curve plotted against spectroscopic 
771: orbital phase. The solid line shows the predicted curve for a 
772: secondary Roche-filling model with $i=48\fdg7$, while the dashed line 
773: represents the prediction for an under-filling model with $i=60^\circ$.}
774: \label{fig3}
775: \end{figure}
776: 
777: 
778: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
779: % Tables
780: 
781: \clearpage
782: 
783: % Table 1
784: \begin{deluxetable}{cccccccccc}
785: \tabletypesize{\scriptsize}
786: \tablewidth{0pt}
787: \tablenum{1}
788: \tablecaption{Radial Velocity Measurements \label{tab1}}
789: \tablehead{
790: \colhead{HJD}             &
791: \colhead{Orbital}         &
792: \colhead{$V_1$}           &
793: \colhead{$\sigma_1$}      &
794: \colhead{$(O-C)_1$}       &
795: \colhead{}                &
796: \colhead{$V_2$}           &
797: \colhead{$\sigma_2$}      &
798: \colhead{$(O-C)_2$}       &
799: \colhead{}                \\
800: \colhead{(-2,400,000)} &
801: \colhead{Phase}           &
802: \colhead{(km s$^{-1}$)}   &
803: \colhead{(km s$^{-1}$)}   &
804: \colhead{(km s$^{-1}$)}   &
805: \colhead{$n_1$}           &
806: \colhead{(km s$^{-1}$)}   &
807: \colhead{(km s$^{-1}$)}   &
808: \colhead{(km s$^{-1}$)}   &
809: \colhead{$n_2$}           }
810: \scriptsize
811: \startdata
812: 44487.472 & 0.921 &       \phs 104.4 &\nodata&  \phs 34.0 &\nodata &      $-151.6$ &\nodata&  \phs 41.2 &\nodata \\
813: 50526.858 & 0.057 &    \phn\phs 80.5 & 6.2 & \phn\phs 3.6 &     17 &      $-189.6$ & 2.2 &    \phs 14.5 &     12 \\
814: 50527.780 & 0.394 &         $-117.3$ & 3.5 &   \phn$-4.2$ &     17 &    \phs 131.3 & 6.8 & \phs\phn 5.3 &     17 \\
815: 50528.845 & 0.784 &  \phn\phn $-1.8$ & 5.5 & \phn\phs 1.2 &     17 &  \phn $-47.6$ & 5.1 &    \phs 17.7 &     16 \\
816: 50529.792 & 0.131 &    \phn\phs 52.3 & 2.7 & \phn\phs 3.7 &     13 &      $-160.3$ & 2.1 &  \phn $-5.4$ &     14 \\
817: 50530.777 & 0.491 &         $-122.8$ & 7.1 &    \phs 13.5 &     17 &    \phs 182.1 & 7.8 &    \phs 15.6 &     16 \\
818: 50531.755 & 0.849 &    \phn\phs 32.6 & 6.3 &   \phn$-5.4$ &     17 &      $-148.0$ & 4.6 &      $-11.5$ &     17 \\
819: 50532.783 & 0.226 &  \phn\phn $-0.3$ & 6.0 & \phn\phs 9.4 &     16 &  \phn $-28.2$ & 7.7 &    \phs 25.5 &     16 \\
820: 50533.754 & 0.581 &         $-125.5$ & 4.4 &   \phn$-3.0$ &     17 &    \phs 159.1 & 8.4 &    \phs 16.8 &     17 \\
821: 50535.766 & 0.318 &     \phn $-82.8$ & 4.7 &      $-10.8$ &     15 & \phn\phs 43.4 & 3.1 &      $-11.2$ &     12 \\
822: 50910.125 & 0.378 &     \phn $-98.5$ & 1.7 & \phn\phs 7.3 & \phn 9 &    \phs 107.6 & 2.3 &   \phn$-5.7$ &     10 \\
823: 50911.086 & 0.730 &     \phn $-22.1$ & 2.6 &    \phs 18.1 & \phn 9 &  \phn $-31.6$ & 5.6 &      $-31.0$ &     10 \\
824: 50911.171 & 0.761 &     \phn $-29.4$ & 0.4 &      $-10.4$ & \phn 6 &  \phn $-34.3$ & 5.4 & \phs\phn 3.2 &     10 \\
825: 50916.213 & 0.607 &         $-119.0$ & 4.1 &   \phn$-6.4$ &     10 &    \phs 122.8 & 5.6 &   \phn$-2.3$ &     10 \\
826: 50916.990 & 0.891 &    \phs\phn 58.8 & 2.5 &   \phn$-0.2$ & \phn 9 &      $-166.8$ & 2.4 & \phs\phn 6.2 & \phn 9 \\
827: 50917.116 & 0.938 &    \phs\phn 73.9 & 4.6 &   \phn$-1.5$ &     10 &      $-204.6$ & 3.5 &   \phn$-3.1$ & \phn 9 \\
828: 50917.257 & 0.989 &    \phs\phn 79.8 & 5.4 &   \phn$-3.7$ &     10 &      $-220.4$ & 4.4 &   \phn$-4.8$ &     10 \\
829: 50917.971 & 0.250 &      \phn$-26.8$ & 1.8 &   \phn$-0.1$ & \phn 8 &  \phn $-28.0$ & 4.9 &   \phn$-3.8$ & \phn 9 \\
830: 50918.112 & 0.302 &      \phn$-59.8$ & 3.8 & \phn\phs 2.0 & \phn 8 & \phn\phs 23.6 & 2.7 &      $-13.4$ & \phn 9 \\
831: 50918.249 & 0.352 &         $-105.3$ & 3.2 &      $-13.0$ &     10 & \phn\phs 77.5 & 2.4 &      $-12.4$ & \phn 9 \\
832: \enddata
833: \end{deluxetable}
834: 
835: \newpage
836: 
837: %Table 2 
838: \begin{deluxetable}{lr}
839: \tablewidth{0pc}
840: \tablenum{2}
841: \tablecaption{Circular Orbital Elements  \label{tab2}}
842: \tablehead{
843: \colhead{Element} & 
844: \colhead{Value} }
845: \startdata
846: $P$~(days)           \dotfill        & 2.73135 (3) \\
847: $T_0$ (HJD-2,400,000) \dotfill       & 50734.286 (11) \\
848: $K_1$ (km s$^{-1}$)    \dotfill      & 110.1 (28) \\
849: $K_2$ (km s$^{-1}$)     \dotfill     & 191.4 (48) \\
850: $V_{0~1}$ (km s$^{-1}$) \dotfill     & $-26.4$ (19) \\
851: $V_{0~2}$ (km s$^{-1}$) \dotfill     & $-24.6$ (32) \\
852: $m_1$ sin$^{3}i$ ($M_\odot$)\dotfill & 4.94 (40) \\
853: $m_2$ sin$^{3}i$ ($M_\odot$)\dotfill & 2.84 (26) \\
854: $a_1$ sin $i$ ($R_\odot$) \dotfill   &  5.94 (15) \\
855: $a_2$ sin $i$ ($R_\odot$) \dotfill   & 10.32 (26) \\
856: r.m.s.$_1$ (km s$^{-1}$)  \dotfill   &  8.4 \\
857: r.m.s.$_2$ (km s$^{-1}$)  \dotfill   & 14.2 \\
858: \enddata
859: \tablecomments{Numbers in parentheses give the error in the last digit quoted.}
860: \end{deluxetable}
861: 
862: \newpage
863: 
864: %Table 3 
865: \begin{deluxetable}{lcc}
866: \tablewidth{0pc}
867: \tablenum{3}
868: \tablecaption{Stellar Properties \label{tab3}}
869: \tablehead{
870: \colhead{Property} & 
871: \colhead{Primary} &
872: \colhead{Secondary} }
873: \startdata
874: Spectral Classification  \dotfill          &  O9.5~V       & B0.2~III \\
875: Relative flux $F/F_1$(5470\AA )\dotfill    &  1.0          & $1.07\pm0.06$ \\
876: $V\sin i$ (km s$^{-1}$)        \dotfill    &  $101\pm 10$  & $132\pm 15$ \\
877: $T_{\rm eff}$ (kK)             \dotfill    &  $32 \pm 2$   & $29 \pm 1.5$  \\
878: $M/M_\odot$                    \dotfill    &  $11.6\pm1.1$ & $6.7\pm 0.7$ \\
879: $R/R_\odot$                    \dotfill    &  $6.5\pm 0.2$ & $7.2\pm 0.2$ \\
880: $\log g$                       \dotfill    &$3.88\pm 0.01$ & $3.55\pm 0.01$ \\ 
881: $\log L/L_\odot$               \dotfill    &$4.60\pm 0.14$ & $4.52\pm 0.12$   \\
882: \enddata
883: \end{deluxetable}
884: 
885: \newpage
886: 
887: %Table 4 
888: \begin{deluxetable}{lcllcrccc}
889: \tablewidth{0pc}
890: \tabletypesize{\footnotesize}
891: \tablenum{4}
892: \tablecaption{Semi-Detached OB-Star Binaries \label{tab4}}
893: \tablehead{
894: \colhead{} & 
895: \colhead{$P$} &
896: \colhead{Pri.} &
897: \colhead{Sec.} &
898: \colhead{$M_P$} &
899: \colhead{$M_S$} &
900: \colhead{} &
901: \colhead{} &
902: \colhead{} \\
903: \colhead{Name} & 
904: \colhead{(d)} &
905: \colhead{Type} &
906: \colhead{Type} &
907: \colhead{($M_\odot$)} &
908: \colhead{($M_\odot$)} &
909: \colhead{$\log L_P/L_\odot$} &
910: \colhead{$\log L_S/L_\odot$} &
911: \colhead{Ref.} 
912: }
913: \startdata
914: HD 115071 = V961 Cen       \dotfill & 2.73 & O9.5 V & B0.2 III &  $11.6\pm1.1$ &  $6.7\pm0.7$ &$4.60\pm 0.14$ &$4.52\pm 0.12$& 1 \\
915: HD 209481 = LZ Cep         \dotfill & 3.07 & O8.5   & O9.5     &  $15.1\pm0.4$ &  $6.3\pm0.2$ &$4.90\pm 0.03$ &$4.65\pm 0.03$& 2 \\
916: %HD 35652 = IU Aur          \dotfill & 1.81 & B0.5   & B0.5     &  $14.5\pm0.7$ &  $7.3\pm0.5$ &$4.36\pm 0.04$ &$4.21\pm 0.02$& 2 \\
917: BD$+66^\circ~1521$ = XZ Cep\dotfill & 5.10 & O9.5 V & B1 III   &  $15.8\pm0.4$ &  $6.4\pm0.3$ &$4.58\pm 0.04$ &$4.48\pm 0.03$& 3 \\
918: HD 106871 = AB Cru         \dotfill & 3.41 & O8 V   & B0.5     &  $19.8\pm1.0$ &  $7.0\pm0.7$ &$5.21\pm 0.03$ &$4.58\pm 0.03$& 4 \\
919: HD 190967 = V448 Cyg       \dotfill & 6.52 & O9.5 V & B1 II-Ib &  $25.2\pm0.7$ & $14.0\pm0.7$ &$4.54\pm 0.04$ &$4.66\pm 0.06$& 3 \\
920: \enddata
921: \tablecomments{References: 
922: (1) this paper, 
923: (2) \citet{har98},
924: (3) \citet{har97},
925: (4) \citet{lor94}.}
926: \end{deluxetable}
927: 
928: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
929: % Figures
930: 
931: \clearpage
932: 
933: \setcounter{figure}{0}
934: 
935: \begin{figure}
936: \plotone{f1.eps}
937: \caption{}
938: \end{figure}
939: 
940: \begin{figure}
941: \plotone{f2.eps}
942: \caption{}
943: \end{figure}
944: 
945: \begin{figure}
946: \plotone{f3.eps}
947: \caption{}
948: \end{figure}
949: 
950: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
951: 
952: \end{document}
953: