1:
2: %\documentclass[manuscript]{aastex}
3: \documentclass[12pt,preprint]{aastex}
4: %\documentclass[preprint2]{aastex}
5:
6: %New commands
7: \newcommand{\kms}{km~s$^{-1}$}
8:
9: \begin{document}
10:
11: \title{Planetary systems around close binary stars: the case of the very
12: dusty, Sun-like, spectroscopic binary BD+20~307}
13:
14: \author{B. Zuckerman\altaffilmark{1,2}, Francis C.\
15: Fekel\altaffilmark{3}, Michael H.\ Williamson\altaffilmark{3}, Gregory
16: W.\ Henry\altaffilmark{3}, M. P. Muno\altaffilmark{4}}
17:
18: \altaffiltext{1}{Department of Physics \& Astronomy, University of
19: California, Los Angeles CA 90095}
20: \altaffiltext{2}{UCLA Center for Astrobiology}
21: \altaffiltext{3}{Center of Excellence in Information Systems, Tennessee
22: State University, 3500 John A.\ Merritt Blvd., Box 9501, Nashville, TN
23: 37209}
24: \altaffiltext{4}{Space Radiation Laboratory, California Institute
25: of Technology, Pasadena, CA 91125-4700}
26:
27:
28: \begin{abstract}
29:
30: Field star BD+20~307 is the dustiest known main sequence star, based on
31: the fraction of its bolometric luminosity, $\sim$4\%, that is emitted at
32: infrared wavelengths. The particles that carry this large IR luminosity
33: are unusually warm, comparable to the temperature of the zodiacal dust in
34: the solar system, and their existence is likely to be a consequence of a
35: fairly recent collision of large objects such as planets or planetary
36: embryos. Thus, the age of BD+20~307 is potentially of interest in
37: constraining the
38: era of terrestrial planet formation. The present project was initiated
39: with an attempt to derive this age using the
40: {\it Chandra X-ray Observatory} to measure the X-ray flux of BD+20~307 in
41: conjunction with extensive photometric and spectroscopic monitoring
42: observations from Fairborn Observatory. However, the recent realization
43: that BD+20~307 is a short period, double-line, spectroscopic binary whose
44: components have very different lithium abundances, vitiates standard
45: methods of age determination. We find the system to be metal-poor; this,
46: combined with its measured lithium abundances, indicates that BD+20~307
47: may be several to many Gyr old. BD+20~307 affords astronomy a rare peek
48: into a
49: mature planetary system in orbit around a close binary star (because such
50: systems are not amenable to study by the precision radial velocity
51: technique).
52:
53: \end{abstract}
54:
55: \keywords{planetary
56: systems:formation$-$stars:individual(BD+20~307)$-$stars:rotation$-$X-rays:
57: stars
58: }
59:
60:
61: \section{INTRODUCTION}
62:
63: Hundreds of main sequence stars are known with orbiting dusty debris
64: disks. The dust in most of these systems is cold and appears to be
65: analogous to dust in the Sun's Kuiper Belt. Only a small percentage of
66: debris
67: disks contain warm dust analogous to that in the Sun's asteroid belt and
68: zodiacal cloud (e.g., Su et al. 2006; Rhee et al 2008; Trilling et
69: al. 2008; Smith et al 2008).
70: For stars that are at least 100 Myr old, the warm dust luminosity
71: (L$_{IR}$/L$_{bol}$) is typically a few times 10$^{-4}$ or less. Smith et
72: al (2008) discuss and review the significance of this warm, mid-infrared,
73: emission in the context of the Sun's zodiacal dust cloud. They note that
74: {\it Spitzer Infrared Observatory} photometry (Hines et al 2006, Bryden et
75: al 2006) indicates that warm dust is detected at only 2$\pm$2\% of the
76: observed systems. A more recent, more comprehensive, {\it Spitzer} survey
77: (Trilling et al. 2008) yielded a similar small percentage of warm
78: excess emission stars in a sample of 213 F- and G-type stars. And a
79: comparable small percentage also applies to warm excess
80: A-type stars of age $>$400 Myr (Su et al. 2006) which, as will be seen
81: below, is the age range of interest in the present paper. In all of these
82: surveys and in other published {\it Spitzer} surveys, the typical
83: mid-infrared
84: luminosity is sufficiently small that the warm dust could be produced by
85: collisions of asteroids; for example, in a massive asteroid belt perturbed
86: by the gravitational field of a nearby giant planet.
87:
88: In contrast, Rhee et al (2008) discuss the dustiest main sequence stars
89: currently known: Pleiad HD23514 and, dustiest of all, the G0 field
90: dwarf BD+20~307 (HIP8920; Song et al. 2005, Weinberger 2008, Weinberger et
91: al. 2008). The dust luminosity at these two stars is orders of magnitude
92: larger than at other mid-infrared excess main sequence stars and
93: collisions of asteroids seem quite inadequate to explain the data. A
94: plausible model to account for so much dust so close to
95: these stars involves a fairly recent collision of two planets or planetary
96: embryos in the terrestrial planet zone (Rhee et al. 2008).
97:
98: BD+20~307 was originally presumed to be a single star, and
99: a measurement of its (presumably youthful) age would have constrained the
100: duration
101: of the era during which major collisional events occur
102: in youthful planetary systems. Using photospheric lithium content,
103: Galactic space motion, and an upper limit to the X-ray flux
104: measured with the ROSAT All-Sky Survey, Song et al. (2005)
105: estimated the age of BD+20~307 to be $\sim$300 Myr. To refine this
106: estimate, we obtained X-ray measurements with the {\it Chandra X-ray
107: Observatory} and hundreds of photometric optical
108: observations to measure the star's rotation period and age, employing the
109: technique of gyrochronology (Barnes 2007). While we were carrying out
110: these observations, Weinberger (2008) obtained the surprising result that
111: BD+20~307 is a short period ($\sim$3.5 day) spectroscopic binary of two
112: nearly identical stars. The reason this was a surprise is that the Song
113: et al. (2005) echelle spectrum showed no evidence of two stars; by a low
114: probability ($<$10\%) quirk of fate, their 2004 epoch spectrum was obtained
115: at a time when the two stars had nearly the same radial velocities.
116:
117: While {\it Spitzer} surveys for debris disks around stars have often
118: explicitly omitted binaries, Trilling et al. (2007) found small quantities
119: of cool dust around a substantial fraction of binary stars. However, by
120: analogy with the discussion of single stars in the first paragraph above,
121: dust around such binary stars can readily be accounted for by collisions
122: of Kuiper Belt objects.
123:
124: Upon learning of the Weinberger (2008) result, we initiated a program to
125: monitor the radial velocity of BD+20~307. Our various observations are
126: described in the next Section. This is followed by a discussion of the
127: age and some other aspects of this exceptionally dusty binary star, which
128: points to the existence of rocky planets in the terrestrial planet region.
129:
130:
131: \section{OBSERVATIONS}
132:
133: \subsection{X-rays}
134:
135: We observed BD+20~307 with the {\it Chandra X-ray Observatory} Advanced
136: CCD Spectrometer Spectroscopic array (ACIS-S; Weisskopf et al. 2002)
137: starting on 2007 September 25 at 23:34:20 (UT), for a total of 4919 s.
138: We reduced the event lists using the Chandra Interactive Analysis of
139: Observations (CIAO) software version 3.4 and the calibration database
140: version 3.4.1. We filtered the data using the standard techniques
141: described on the web site of the {\it Chandra X-ray Center} (CXC).
142:
143: We then made images of the events in the full bandpass of
144: 0.5--8.0~keV, a soft 0.5--2.0~keV band to provide sensitivity to
145: foreground sources, and a hard 4--8 keV band to provide sensitivity to
146: highly-absorbed sources. We searched these for point sources using
147: the wavelet-based algorithm {\tt wavdetect} (Freeman et al. 2002).
148: Images were made for two spatial scales: one 1024$\times$1024 pixel image
149: covering the central portion of the field at the 0.5\arcsec\
150: resolution of the detector, and one image at 2\arcsec\ resolution that
151: covered all of the active detectors. We identified a total of 30 sources.
152: The majority of these, 26, were detected strongly in the
153: 0.5--2~keV band. Only 4 sources were detected in the full band but not
154: in the soft band, and no sources were detected exclusively in the hard
155: band. Three of the brightest sources had counterparts in the 2MASS
156: catalog, so we used these to register the absolute astrometry for the
157: observation. We found that the X-ray sources were offset by an
158: average of $-$0.24$\pm$0.17\arcsec\ in right ascension, and
159: $+$0.36$\pm$0.17\arcsec\ in declination, and applied this shift to the
160: locations of the X-ray sources.
161:
162: We found an X-ray source coincident with BD+20~307 at 01$^h$ 54$^m$
163: 50\fs37, +21\degr\ 18\arcmin\ 22\farcs3, with an uncertainty of
164: 0.4\arcsec according to Equation 5 of Hong et al. (2005).
165: We extracted photometry and light curves for the source using the {\tt
166: acis\_extract} routine.\footnote{{\tt www.astro.psu.edu/xray/docs/TARA/}}
167: Events for the source were
168: extracted from a circle with radius of 2.5 pixels, which corresponds
169: to 90\% of the PSF at 1.4 keV. Background was derived from an annular
170: 25-pixel region, excluding a circle of 5 pixels around the point source.
171: We found 158$\pm$12 net counts from BD+20~307, all of which arrived in
172: the 0.5--2.0 keV band. In total, only 4
173: sources in the 340 arcmin$^2$ field-of-view had $>$100 net counts, so
174: we conclude that this X-ray source is BD+20~307.
175:
176: Next, we binned the events over energy to obtain spectra of the
177: source and background, and obtained the detector response function and
178: effective area using standard CIAO tools. We modeled the spectrum as a
179: thermal plasma\footnote{{\tt http://hea-www.harvard.edu/APEC}} using
180: XSPEC version 12.2.1 (Arnold 1996). We used the Cash statistic
181: (Cash 1979) to identify a best fit and 1$\sigma$ uncertainty on the
182: model parameters, and found a temperature
183: $kT$$=$$0.61^{+0.06}_{-0.02}$~keV
184: and a flux of $1.1\times10^{-13}$ erg cm$^{-2}$ s$^{-1}$ (0.5--2.0 keV).
185: Finally, we tested for variability by comparing the arrival times of the
186: photons to a uniform distribution using the Kolmogorov-Smirnov statistic,
187: and found that the data were consistent with a constant flux at the 76\%
188: confidence level.
189:
190:
191: \subsection{Optical Photometry}
192:
193:
194: We used the T12 0.8m Automatic Photometric Telescope (APT) located at
195: Fairborn Observatory in Arizona to collect a total of 449 nightly
196: brightness measurements of BD+20~307 between 2005 February and 2007
197: December. Such measurements of solar-type stars often reveal brightness
198: variations caused by cool, dark photospheric spots as they are carried
199: into and out of view by the star's rotation (e.g., Radick et al. 1987).
200: Our APTs have proven to be very successful at measuring stellar rotation
201: periods, even for stars with low photometric amplitudes of a few
202: hundredths of a magnitude or less (e.g., see Henry et al. 1995).
203:
204: We programmed the T12 APT to observe BD+20~307 differentially with respect
205: to three comparison stars. Its two-channel precision photometer observed
206: each star simultaneously in the standard Str\"omgren $b$ and $y$ passbands.
207: The three comparison stars, designated A, B, and C, respectively, are
208: HD~12354 ($V=6.61$, $B-V=0.34$, F1~IV--V), HD~12846 ($V=6.89$, $B-V=0.66$,
209: G2~V), and HD~10088 ($V=7.89, B-V=0.31$, F0~IV), while BD+20~307 ($V=8.98,
210: B-V=0.56$, F9~V) is designated as star D. The observations were reduced
211: to form the six differential magnitudes D$-$A, D$-$B, D$-$C, C$-$A, C$-$B,
212: and B$-$A. To increase the precision of our measurements, we combined
213: the Str\"omgren $b$ and $y$ differential magnitudes into a single $(b+y)/2$
214: passband for each of the six differential magnitudes. For additional
215: information on the telescope, photometer, observing procedures, data
216: reduction techniques, and photometric precision, see Henry (1999) and
217: Eaton et al. (2003).
218:
219: The individual reduced and standardized $(b+y)/2$ differential magnitudes
220: of BD+20~307 are listed in Table~1, where they will be available for
221: future studies such as long-term spot evolution or for comparison with other
222: activity indicators in BD+20~307. To improve our measurement precision
223: still further, we averaged the three D$-$A, D$-$B, and D$-$C differential
224: magnitudes of BD+20~307 into a single value representing the difference in
225: brightness between BD+20~307 and the mean of the three comparison stars
226: ($D-(A+B+C)/3$), which we refer to as the ensemble average. The standard
227: deviations of the three comparison star differentials C$-$A, C$-$B, and
228: B$-$A from Table~1 all lie between 0.001 and 0.002 mag, which is typical of
229: the measurement precision with the T12 APT. Therefore, all three
230: comparison stars are constant from night to night to the limit of our
231: precision.
232:
233: The 449 ensemble differential magnitudes of BD+20~307 in our $(b+y)/2$
234: passband are plotted in the top panel of Figure~1, where they fall into
235: four separate observing seasons. No significant periodicity was found when
236: analyzing the observing seasons separately or the data set as a whole.
237: Therefore, any photospheric starspots on either component of BD+20~307 must
238: have lifetimes of only a few weeks at most. Indeed, a small portion of the
239: third observing season, marked by the pair of vertical lines in the top
240: panel of Figure~1 and shown with an expanded abscissa in the second panel,
241: does appear to reveal a few cycles of low-amplitude but coherent brightness
242: variability. The brightest and faintest observations of each cycle are
243: connected with straight line segments to guide the eye; the interpolation
244: of approximately one and a half missing cycles is shown by the dotted line
245: segments.
246:
247: The third panel of Figure~1 shows the power spectrum of the data from
248: the second panel, computed with the method of Van{\'{\i}}{\v c}ek (1971).
249: This
250: period-finding technique uses least-squares to fit the data with sine
251: curves over a range of trial periods and measures the resulting reduction
252: in the variance of the data for each trial frequency. The strongest
253: peak occurs at a frequency of $0.2833 \pm 0.0020$ day$^{-1}$, corresponding
254: to a period of $3.530 \pm 0.025$ days. The lower peak is a one-day alias
255: of this period. Analysis of small sections of the fourth season lightcurve
256: provides support for the 3.53-day period, yielding less precise period
257: values between 3.28 and 3.44 days.
258:
259: The bottom panel of Figure~1 shows the data from the second panel phased
260: with the 3.53-day photometric period and a time of brightness minimum of
261: JD $2,454,003.18 \pm 0.04$ days, computed from a least-squares sine fit
262: to the phase curve. The sine fit gives a peak-to-peak amplitude of
263: $0.0066 \pm 0.0005$ mag and an rms to the fit of 0.0014 mag, which
264: approximates the typical precision of our photometric measurements.
265:
266: \subsection{Optical Spectroscopy}
267:
268: From 2008 January through March we obtained 28 spectrograms of
269: BD$+$20~307 with the Tennessee State University 2 m automatic
270: spectroscopic telescope (AST), fiber-fed echelle spectrograph, and a 2048
271: x 4096 SITe ST-002A CCD. The echelle spectrograms have 21 orders,
272: covering the wavelength range 4920--7100~\AA\ with an average resolution
273: of 0.17~\AA. The typical signal-to-noise ratio of these observations is
274: $\sim$10. Eaton \& Williamson (2004, 2007) have given a more extensive
275: description of the telescope and spectrograph, which are operated at
276: Fairborn Observatory.
277:
278: For the AST spectra, lines in approximately 170 regions, centered on the
279: rest wavelengths (Moore et al. 1966) of relatively strong, mostly Fe~I
280: lines that were not extensively blended with other nearby strong
281: features, were measured. Lines at the ends of each echelle order were
282: excluded because of their lower signal-to-noise ratios. The wavelength
283: scale of each spectrum was initially determined from Th-Ar comparison
284: spectra obtained at the beginning and end of the night, and that scale
285: was refined with the use of the telluric O$_2$ lines near 6900\AA, which
286: are in each stellar spectrum. A Gaussian function was fitted to the
287: profile of each component. If lines of the two components were blended, a
288: double Gaussian was used to fit the combined profile. The difference
289: between the observed wavelength and that given in the solar line list of
290: Moore et al (1966) was used to compute the radial velocity, and a
291: heliocentric correction was applied. The final mean velocity for each
292: component is given in Table~2. Our unpublished velocities of several IAU
293: standard solar-type stars indicate that the Fairborn Observatory
294: velocities have a small zero-point offset of $-$0.3 km~s$^{-1}$ relative
295: to the velocities of Scarfe et al. (1990).
296:
297: There is no confusion concerning component identification because the
298: line strengths are different enough, as discussed in the next paragraph,
299: so that the primary and secondary can be identified visually. With our
300: photometric period adopted and 0.3 km~s$^{-1}$ added to each Fairborn
301: Observatory velocity, initial orbital elements for the primary were
302: computed with BISP (Wolfe et al. 1967), a program that implements a
303: slightly modified version of the Wilsing-Russell method. The orbit was
304: then refined with SB1 (Barker et al. 1967), a program that uses
305: differential corrections. An orbit for the secondary velocities was also
306: computed. Zero weight was assigned to the velocities of the nine
307: observations that had blended components with velocity separations less
308: than 30 km~s$^{-1}$. Based on the variances of the two solutions, weights
309: of 1.0 and 0.9 were assigned to the primary and secondary velocities,
310: respectively, of the 19 other spectrograms. An orbital solution of both
311: components together was obtained with SB2, a modified version of SB1.
312: Because the eccentricity of this solution is extremely small, 0.004 $\pm$
313: 0.004, a circular orbit was computed with SB2C (D. Barlow 1998, private
314: communication), which also uses differential corrections to determine the
315: elements. For this solution the weights of the velocities were the same
316: as those previously assigned. The tests of Lucy \& Sweeney (1971)
317: indicate that the circular orbit solution is to be prefered. The orbital
318: phases of the observations and velocity residuals to the final computed
319: curves are given in Table~2, and the orbital elements and related
320: parameters are listed in Table~3. For a circular orbit the element {\it
321: T}, a time of periastron passage, is undefined. So, as recommended by
322: Batten et al. (1989), {\it T$_0$}, a time of maximum velocity for the
323: primary, is given instead. In Figure~2 the observed velocities are
324: compared with the computed velocity curves. Zero phase is a time of
325: maximum velocity of the primary.
326:
327: The components are similar in spectral type, as indicated by the best fit
328: effective temperatures and gravities that were determined by Weinberger
329: (2008). Thus, for the two components the iron lines, which dominate the
330: spectrum, and those of other elements except for lithium, which is
331: discussed in \S3, are similar in strength. To determine the magnitude
332: difference between the two components, we followed the spectroscopic
333: method of Petrie (1939). For spectra with well separated components we
334: measured the ratio of the equivalent widths of over 100 Fe~I primary and
335: secondary lines. The average of the equivalent width ratio of the
336: secondary to primary components is 0.78 $\pm$ 0.01 and corresponds to the
337: luminosity ratio of the stars. This is then converted into a very modest
338: magnitude difference of 0.27. Given the wavelength range of our spectra,
339: we adopt this value as the $V$ magnitude difference. Combined with the
340: total $V$ mag of 8.98 listed by {\it Hipparcos}, this results in $V_A$ =
341: 9.61 mag and $V_B$ = 9.88.
342:
343: Weinberger (2008) compared synthetic spectra, computed with solar
344: abundances, to a spectrum of BD$+$20~307 with well separated components
345: and determined the effective temperatures of the two stars. The two
346: components differed by just 250 K, and the average of the two effective
347: temperatures is 6375 K, which corresponds to a $B-V$ color of 0.48
348: (Flower 1996). However, the observed combined $B-V$ color of the system
349: is 0.56 (Perryman et al. 1997). This suggests either that the star has
350: significant extinction in the visual region or the components are metal
351: poor. The first possibility appears to be unlikely. At a distance of
352: only 102 pc (van Leeuwen 2007), chances are small that sufficient
353: interstellar matter lies between the Sun and BD$+$20~307 to produce a
354: color excess of 0.08 mag (Welsh et al. 1998; Lallement et al. 2003; Cox
355: 2000; B. Welsh 2008, personal communication). Also, little, if
356: any, circumstellar extinction from the collision that produced the large
357: infrared excess is expected. Collisions among the dust particles rapidly
358: damp out the components of motion not in the orbital plane. For
359: BD$+$20~307 the plane containing the collision material is presumably the
360: same as that of the binary orbit, which has an inclination of about
361: 31\arcdeg\ and so is much closer to face-on ($i$ = 0\arcdeg) than
362: edge-on. In addition, the spectral energy distribution of BD$+$20~307 is
363: such that very little (hot) dust exists within a few tenths of an AU of
364: the central stars (Song et al. 2005, Weinberger et al. 2008). Thus, the
365: picture of this system is of the close binary components shining out
366: through the central hole in the dust ring distribution.
367:
368: The metal poor alternative appears to be a more attractive explanation of
369: the apparently discrepant B-V indicated in the previous paragraph.
370: Synthetic spectra with solar abundances require a higher effective
371: temperature then do metal poor stars to produce a given (weak) line
372: strength. To investigate whether the system is metal poor, we estimated
373: the iron abundance for the components of BD$+$20~307. We compared the
374: combined average equivalent width of components A and B, determined from
375: the same set of iron lines as those used for the luminosity ratio, with
376: the average iron equivalent widths of several single dwarf stars,
377: including $\chi$~Her, $\beta$~Com, and HR~7560, which have similar $B-V$
378: values and known iron abundances. The spectra of these comparison stars
379: were obtained with the same telescope, spectrograph, and detector as
380: those of BD$+$20~307. The average equivalent widths of HR~7560 and
381: $\beta$~Com, which have essentially solar iron abundances (Taylor 2005),
382: are 30\% larger, but the average equivalent width of $\chi$~Her, which
383: has [Fe/H] = $-$0.43 (Taylor 2003), is identical to that of BD~$+$20~307.
384: Thus, we conclude that BD~$+$20~307 has a similar iron abundance and is
385: metal poor compared to the Sun. Given that dust particles are made of
386: refractory metals, it does seem remarkable that the dustiest main
387: sequence star presently known to astronomy appears to be of low
388: metallicity.
389:
390: \section{DISCUSSION}
391:
392: The initial motivation for the research reported in this paper was
393: determination of the age of BD+20~307 and the implications of its very
394: dusty circumstellar disk for the formation of terrestrial planets around
395: sun-like stars. As such, we
396: initially obtained the X-ray and optical photometry data described in
397: Sections 2.1 and 2.2. However, the discovery that BD+20~307 is a close
398: binary star (SB2; Weinberger 2008) greatly altered our interpretation
399: of these data and motivated the spectroscopic observations discussed in
400: Section 2.3. In the present Section we consider some properties of
401: BD+20~307 derived from the data and their relatationship to its age.
402:
403: Assuming BD+20~307 is a single star, from a fit to its spectral energy
404: distribution, Song et al (2005) derived a radius = 1.25 R$_\sun$ based on
405: the Hipparcos parallax of 10.88 mas, and
406: temperature 6000~K. Because of its low metallicity ([Fe/H] = $-$0.43,
407: Section 2.3) the Hipparcos B$-$V = 0.56 should be corrected to a solar
408: metallicity B$-$V index according to an expression given in Gray (1994);
409: the result is B$-$V = 0.607, corresponding to a temperature of $\sim$5900
410: K. With this effective temperature, the monochromatic Hipparcos V mag
411: (= 8.98), and the Hipparcos parallax, we derive a single star radius of 1.3
412: R$_\sun$, in good
413: agreement with the results of Song et al. Van Leeuwen (2007) revised
414: the Hipparcos parallax to 9.82 mas, which corresponds to a single star
415: radius of 1.44 R$_\sun$. With the simplifying
416: assumption that the two stars in BD+20~307 are of equal magnitude, then
417: their representative radii would each be 1 R$_\sun$. (More precise
418: values for the radii, appropriate for the relative magnitudes given
419: in \S2.3, can easily be derived.) Conversion of T and R to
420: stellar mass depends on the uncertain stellar age (see below). If, the
421: mass of the primary is 1 M$_\sun$, then the orbital inclination $i =
422: 31\arcdeg$.
423:
424: Our orbital period for a circular orbit (Table~3) agrees well with the
425: period of 3.448 days derived by Weinberger (2008) for an assumed circular
426: orbit. However, her systemic radial velocity, $\gamma$ = $-$8.4
427: km~s$^{-1}$, does not agree within the respective errors with our $-$12.43
428: km~s$^{-1}$. From Weinberger's discussion, the quoted error on her value
429: for $\gamma$ would be about 0.6 km~s$^{-1}$. Then, the $>$6 $\sigma$
430: discrepancy in radial velocities must be attributed to either an
431: underestimate of measurement error or a variable radial velocity or both.
432: To investigate the possibility of a variable radial velocity, we divided
433: our Table 3 data set into two halves each 22 days long. The result was
434: $\gamma$ = $-$12.4 km~s$^{-1}$ for each half, with estimated uncertainties
435: of 0.1 and 0.2 km~s$^{-1}$, for the first and second half, respectively.
436: Thus, if the systemic radial velocity of BD+20~307 is variable, then it
437: changed by about 4 km~s$^{-1}$ over the 117 day interval between the
438: 2007 October observations of Weinberger and the first half of our data set
439: but then essentially did not change over the 22 days that separated the
440: two halves of our data. While this pattern seems quite peculiar, it
441: cannot be ruled out a priori. We note that in September 2004, Song et al.
442: (2005) measured a systemic radial velocity of $-$11$\pm$1 km~s$^{-1}$.
443: Tokovinin et al. (2006) investigated spectroscopic binaries for the
444: presence of tertiary companions. For a system with orbital period of 3.4
445: days, the
446: probability is $\sim$70\% that a third star will also be present. The
447: period distribution is such that, depending on mass
448: ratios, if a third star at BD+20~307 exists, then it might be detected
449: either through adaptive optics imaging or monitoring of the
450: systemic radial velocity.
451:
452: Our observed orbital period of 3.42 days is in reasonable agreement with
453: our (less precisely) determined photometric period of 3.53 days, as would
454: be expected for a system with synchronized rotation and revolution. As
455: noted by Weinberger (2008 and references therein), for solar mass stars
456: and an orbital period of 3.4 days, this synchronization would occur in
457: $\sim$100 Myr which is likely to be substantially less than the age of
458: BD+20~307 (see below). Quasi-sinusoidal lightcurves are typical of
459: spot-induced variability in magnetically active stars (e.g., Henry et al.
460: 1995; Henry \& Winn 2008). The low photometric amplitude implies that the
461: difference in spot filling factor over the observable hemisphere during a
462: full rotation period is less than 1\%. The moderately small inclination
463: of the rotation (orbital) axis, $\sim$30$\arcdeg$, would be consistent
464: with the low photometric amplitude of the lightcurve. Mid-infrared
465: interferometric observations, for example with the Keck Nuller
466: Interferometer, might confirm or deny whether the debris
467: disk around BD+20~307 lies in the orbital plane of the binary.
468:
469: For main sequence stars with outer convection zones, both rotation rate and
470: associated stellar activity (e.g., X-ray flux, optical emission lines) can
471: be used to derive stellar ages (see, for example, Gaidos 1998; Zuckerman \&
472: Song 2004; Barnes 2007). While the rotation rate of BD+20~307 is
473: controlled by its orbital period and not by its age, it is nonetheless
474: worthwhile to characterize the binary's X-ray luminosity.
475:
476: With the effective single-star radius and temperature calculated in the
477: second paragraph of the present Section and the Chandra-measured X-ray
478: flux in the $0.5-2.0$ keV band quoted in Section 2.1, we calculate a
479: fractional X-ray luminosity (L$_x$/L$_{bol}$) of $1.6\times10^{-5}$. This
480: may be compared to the upper limit to the ROSAT measured X-ray flux over a
481: similar energy band quoted by Song et al (2005), i.e., $1.26\times10^{-5}$.
482: Thus, it appears that the (presumably somewhat variable) X-ray flux from
483: BD+20~307 was just below the ROSAT detection limit. The high X-ray flux
484: and the rapid rotation period of 3.4 days are consistent with the fact that
485: this system is a close, synchronized binary.
486:
487: In the absence of rotation and activity age indicators, one can instead
488: turn to lithium abundance, Galactic space motion (UVW), metallicity, and
489: placement on evolutionary tracks on an HR diagram. With the lithium
490: absorption feature equivalent width (EW) given by Weinberger (2008) and
491: our observed ratio of the two stellar continua (= 0.78) at 6708 \AA\, we
492: derive an EW of 62 m\AA\ for the primary. This would be consistent with
493: an age perhaps as young as stars in the UMa or Hyades groups (i.e.,
494: $\sim$500 Myr, e.g., Zuckerman et al. 2006), but also consistent with
495: early G-type stars in the 2 Gyr old NGC 752, IC 4651, and NGC 3680 open
496: clusters (Sestito et al. 2004). Indeed, this EW is even similar to that
497: of early G-type stars in the 6$-$8 Gyr old NGC 188 (Randich et al. 2003).
498: Then there is the secondary to consider, for which the 3 $\sigma$ EW is
499: $<$14 m\AA\ (Weinberger 2008). Such small EW have been measured for some
500: early G-type stars as old as $\sim$5 Gyr, for example M67 (Sestito \&
501: Randich (2005) and the Sun, and perhaps occasionally also as young as 2
502: Gyr (NGC 752, Sestito et al. 2004). Thus, one can conclude that the
503: lithium lines in the two stars are probably most consistent with an age of
504: a few Gyr or more.
505:
506: With the van Leeuwen (2007) parallax, the Tycho~2 proper motion, and a
507: radial velocity of -12.4 km s$^{-1}$, we calculate the Galactic space
508: motion of BD+20~307 with respect to the Sun (UVW = -3.3, -24.5, +4.7 km
509: s$^{-1}$). This UVW is more similar to the UVW of moving groups with ages
510: $\leq$100 Myr than it is to the UVW of older moving groups such as Ursa
511: Majoris and the Hyades (see, e.g., Table 7 in Zuckerman \& Song 2004).
512: While relatively few stars of age several Gyr or more have such small
513: absolute values of all three space motion components, it is of course
514: possible that BD+20~307 is one such star.
515:
516: The low metallicity of BD+20~307 suggests an old age. Even the 5 open
517: clusters with ages between 2 and 8 Gyr listed by Sestito \& Randich (2005)
518: all have metallicities not much different than solar.\footnote{Lambda Boo
519: stars have low abundances of refractory elements and dusty/gaseous disks
520: and are believed to be young, age $\leq$few 100 Myr. Their spectral
521: types, late-B to early-F, are substantially earlier than that of
522: BD+20~307. Thus, BD+20~307 is very unlikely to be a (young) $\lambda$
523: Boo type star.} With the
524: evolutionary tracks presented by Nordstrom et al. (2004; Fig. 12, [Fe/H =
525: -0.50] panel), and the van Leeuwen (2007) parallax, BD+20~307 looks to be
526: many Gyr
527: old. However, the parallax error is sufficiently large, 13\%, that,
528: within the uncertainties, a much
529: younger age is consistent with these tracks. A very accurate
530: measurement of the parallax of BD+20~307 would be worthwhile.
531:
532: In summary, various methods taken together seem to imply a BD+20~307 age
533: of least a few Gyr, and possibly considerably more. It is unlikely that
534: the extraordinary quantity of warm dust at BD+20~307 has any direct
535: relationship with the origin of terrestrial planets.
536:
537: \section{CONCLUSIONS}
538:
539: Until new techniques are available to image or otherwise detect planets
540: with masses and semi-major axes similar to those of the terrestrial planets
541: of our solar system, we must rely on more indirect observations and models.
542: One promising indirect approach is detection of dusty debris associated
543: with terrestrial planet formation (e.g., Gorlova et al. 2007; Currie et al.
544: 2007; Rhee et al. 2008; Lisse et al. 2008). To delineate the full extent
545: of the era of terrestrial planet formation it is necessary to identify main
546: sequence stars with copious quantities of warm dust grains and ages of tens
547: to many hundreds of millions of years. Thus, the goal of our research
548: project was to establish the age of BD+20~307 which, along with HD 23514
549: (an F6 member of the Pleiades), are the only two stars known with age
550: $\geq$100 Myr and strong indications of the existence of orbiting
551: terrestrial planets (Rhee et al. 2008).
552:
553: Based on the findings of our study, it now appears likely that BD+20~307
554: is an old binary star and its copious quantity of warm orbiting dust has
555: no direct relationship with the era of planet formation. Rhee et al.
556: (2008) present a moderately detailed discussion of the origin and fate of
557: the dust at HD 23514 and BD+20~307. Given the large quantity of dust at
558: BD+20~307, its short lifetime in orbit, and its concentration to a narrow
559: range of warm temperatures (Weinberger et al. 2008), it is hard to escape
560: the conclusion that something with the mass of a terrestrial planet was
561: involved in a catastrophic collision, independent of whether that
562: collision took place between two isolated objects or inside some sort of
563: massive asteroid belt. A remarkable aspect of BD+20~307 is that this
564: collision apparently has taken place so late in its history. Perhaps, as
565: suggested by the discussion of possible systemic radial velocity
566: variations in Section 3, there is an unseen stellar or brown dwarf object
567: present whose gravitational influence recently destabilized the orbits of
568: rocky objects of planetary mass.
569:
570: In any event, whatever its age, if the massive amounts of dust at
571: BD+20~307 do point toward the presence of terrestrial planets, then this
572: represents the first known example of planets of any mass in orbit
573: around close binary stars -- because the precision radial velocity
574: technique is not sensitive to planets in such systems.
575:
576: Because of its only modest sensitivity, IRAS was quite limited in the distance
577: out to which it could detect warm dust at low luminosity, main sequence, stars
578: of K- and M-type. Should they exist, such very dusty, low luminosity, stars
579: will be detectable by the Wide-Field Infrared Survey Explorer (WISE)
580: mission. A substantial sample of very dusty stars of
581: known age could delineate the era of terrestrial planet formation while
582: perhaps revealing the existence of additional remarkable stars like
583: BD+20~307.
584:
585:
586: \acknowledgments
587: We thank Inseok Song for a helpful conversation and Alycia Weinberger for
588: calling the binary nature of BD+20~307 to our attention before
589: publication. We thank the second referee for constructive comments.
590: This research was supported in part by NASA through a Chandra Observatory
591: award to UCLA. Automated Astronomy at Tennessee State University is
592: supported
593: by NASA, NSF, Tennessee State University, and the State of
594: Tennessee through its Centers of Excellence program.
595:
596: %\clearpage
597: \begin{thebibliography}{}
598:
599: \bibitem[]{587} Arnaud, K. A., 1996, Astronomical Data
600: Analysis Software and Systems V, eds. Jacoby G. and Barnes J., p17, ASP
601: Conf. Series volume 101
602:
603: \bibitem[]{591} Barker, E. S., Evans, D. S. \& Laing, J. D. 1967, R. Obs.
604: Bull., 130
605:
606: \bibitem[]{594} Barnes, S. A. 2007, ApJ 669, 1167
607:
608: \bibitem[]{596} Batten, A. H., Fletcher, J. M. \& MacCarthy, D. G. 1989,
609: Publ. Dom. Astrophys. Obs., 17, 1
610:
611: \bibitem[]{547} Bryden, G. et al. 2006, \apj, 636, 1098
612:
613: \bibitem[]{599} Cash, W., 1979, ApJ 228, 939
614:
615: \bibitem[]{ } Cox, A.N., editor, 2000, Allen's Astrophysical Quantities
616: (New York: Springer-Verlag), 527
617:
618: \bibitem[]{601} Currie, T., Kenyon, S. J., Rieke, G., Balog, Z. \& Bromley, B.C.
619: 2007, ApJ 663, L105
620:
621: \bibitem[]{604}
622: Eaton, J. A., Henry, G. W. \& Fekel, F. C. 2003, in The Future of Small
623: Telescopes in the New Millennium, Volume II - The Telescopes We Use, ed.
624: T. D. Oswalt (Dordrecht: Kluwer), 189
625:
626: \bibitem[]{609} Eaton, J. A., \& Williamson, M. H. 2004, SPIE, 5496, 710
627:
628: \bibitem[]{611} Eaton, J. A., \& Williamson, M. H. 2007, PASP, 119, 886
629:
630: \bibitem[]{ } Flower, P. J. 1996, \apj, 469, 355
631:
632: \bibitem[]{613} Freeman, P. E., Kashyap, V.,
633: Rosner, R. \& Lamb, D. Q. 2002, \apjs, 138, 185
634:
635: \bibitem[]{616} Gaidos, E. J. 1998, PASP 110, 1259
636:
637: \bibitem[]{620} Gorlova, N. et al. 2007, ApJ 670, 516
638:
639: \bibitem[]{622} Gray, D. 1994, PASP 106, 1248
640:
641: \bibitem[]{624} Henry, G. W. 1999, \pasp, 111, 845
642:
643: \bibitem[]{626}
644: Henry, G. W., Fekel, F. C. \& Hall, D. S. 1995, \aj, 110, 2926
645:
646: \bibitem[]{629} Henry, G. W. \& Winn, J. N. 2008, \aj, 135, 68
647:
648: \bibitem[]{579} Hines, D. et al. 2006, \apj, 638, 1070
649:
650: \bibitem[]{631} Hong, J., van den Berg, M., Schlegel, E.
651: M., Grindlay, J. E., Koenig, X., Laycock, S. \& Zhao, P. 2005, \apj, 635,
652: 907
653:
654: \bibitem[]{ } Lallement, R., Welsh, B., Vergely, J., Crifo, F. \& Sfeir, D.
655: 2003, A\&A, 411, 447
656:
657: \bibitem[]{ } Lisse, C. M., Chen, C. H., Wyatt, M. C. \& Morlok, A. 2008,
658: \apj, 673, 1106
659:
660: \bibitem[]{635} Lucy, L. B. \& Sweeney, M. A. 1971, \aj, 76, 544
661:
662: \bibitem[]{637} Moore, C. E., Minnaert, M. G. J. \& Houtgast, J. 1966, The
663: Solar Spectrum 2935 \AA\ to 8770 \AA, National Bureau of Standards
664: Monograph 61 (Washington, D. C.: U. S. Government Printing Office)
665:
666: \bibitem[]{641} Nordstrom, B. et al. 2004, A\&A 418, 989
667:
668: \bibitem[]{ }
669: Perryman, M. A. C., et al. 1997, The {\it Hipparcos} and Tycho Catalogues
670: (ESA SP-1200; Noordwijk: ESA)
671:
672: \bibitem[]{ }
673: Petrie, R. M. 1939, Publ. Dominion Astrophys. Obs. Victoria, 7, 205
674:
675: \bibitem[]{643}
676: Radick, R. R., Thompson, D. T., Lockwood, G. W., Duncan, D. K. \&
677: Baggett, W. E. 1987, \apj, 321, 459
678:
679: \bibitem[]{647} Randich, S., Sestito, P. \& Pallavicini, R. 2003, A\&A 399, 133
680:
681: \bibitem[]{649} Rhee, J. H., Song, I. \& Zuckerman B. 2008, ApJ 675, 777
682:
683: \bibitem[]{654} Scarfe, C. D., Batten, A. H. \& Fletcher, J. M. 1990, Publ.
684: Dominion Astrophy. Obs., 18, 21
685:
686: \bibitem[]{657} Sestito, P. \& Randich S. 2005, A\&A 442, 615
687:
688: \bibitem[]{659} Sestito, P., Randich, S. \& Pallavicini, R. 2004, A\&A 426, 809
689:
690: \bibitem[]{608} Smith, R., Wyatt, M.C. \& Dent, W. R. F. 2008, A\&A in press
691: (also, astroph 0804.4580)
692:
693: \bibitem[]{663} Song, I., Zuckerman, B., Weinberger, A.J. \& Becklin, E. E.
694: 2005, Nature 436, 363
695:
696: \bibitem[]{614} Su et al. \apj, 653, 675
697:
698: \bibitem[]{666} Taylor, B. J. 2003, \aap, 398, 731
699:
700: \bibitem[]{ } Taylor, B. J. 2005, \apjs, 161, 444
701:
702: \bibitem[]{668} Tokovinin, A., Thomas, S., Sterzik, M. \& Udry, S. 2006, A\&A
703: 450, 681
704:
705: \bibitem[]{621} Trilling, D. E. et al. 2007, \apj, 658, 1289
706:
707: \bibitem[]{623} Trilling, D. E. et al. 2008, \apj, 674, 1086
708:
709: \bibitem[]{671} Van{\'{\i}}{\v c}ek, P. 1971, \apss, 12, 10
710:
711: \bibitem[]{ } van Leeuwen, F. 2007, A\&A, 474, 653
712:
713: \bibitem[]{673} Weinberger A. 2008, ApJ, 679, L41
714:
715: \bibitem[]{675} Weinberger A. et al. 2008, submitted to ApJ
716:
717: \bibitem[]{677} Weisskopf, M. C., Brinkman, B.,
718: Canizares, C., Garmire, G., Murray, S. \& van Speybroeck, L. P. 2002,
719: \pasp, 114, 1
720:
721: \bibitem[]{ } Welsh, B., Crifo, F. \& Lallement, R. 1998, A\&A, 333, 101
722:
723: \bibitem[]{681} Wolfe, R. H., Horak, H. G. \& Storer, N. W. 1967, in Modern
724: Astrophysics, ed. M. Hack (New York: Gordon \& Breach), 251
725:
726: \bibitem[]{684} Zuckerman, B., Bessell, M.S., Song, I. \& Kim, S. 2006, ApJ
727: 649, L115
728:
729: \bibitem[]{687} Zuckerman, B. \& Song I. 2004, ARAA, 42, 685
730:
731: \end{thebibliography}{}
732:
733: \clearpage
734: \begin{deluxetable}{lrrrrrr}
735: \tabletypesize{\normalsize}
736: \tablecaption{Individual Photometric Observations of BD+20~307}
737: \tablewidth{0pt}
738: \tablehead{
739: \colhead{Hel. Julian Date} & \colhead{D$-$A} & \colhead{D$-$B} & \colhead{D$-$C} & \colhead{C$-$A} & \colhead {C$-$B} & \colhead {B$-$A} \\
740: \colhead{(HJD $-$ 2,400,000)} & \colhead{(mag)} & \colhead{(mag)} & \colhead{(mag)} & \colhead{(mag)} & \colhead{(mag)} & \colhead{(mag)}
741: }
742: \startdata
743: 53,402.5866 & 2.4388 & 2.0525 & 1.1812 & 1.2576 & 0.8712 & 0.3864 \\
744: 53,402.5922 & 2.4401 & 2.0518 & 1.1816 & 1.2585 & 0.8703 & 0.3881 \\
745: 53,404.5932 & 2.4366 & 2.0495 & 1.1767 & 1.2598 & 0.8728 & 0.3871 \\
746: 53,405.5882 & 2.4385 & 2.0507 & 1.1799 & 1.2586 & 0.8708 & 0.3878 \\
747: 53,405.5939 & 2.4363 & 2.0516 & 1.1766 & 1.2598 & 0.8750 & 0.3848 \\
748: \enddata
749:
750: \tablecomments{Differential magnitudes were measured in the
751: Str\"omgren $(b+y)/2$ passband. This table is available in its
752: entirety in machine-readable form in the online journal. A portion
753: is shown here for guidance regarding its form and content.}
754: \end{deluxetable}
755:
756: \clearpage
757: \begin{deluxetable}{lcrrrr}
758: \tablenum{2}
759: \tablewidth{0pt}
760: \tablecaption{RADIAL VELOCITIES OF BD +20 307}
761: \tablehead{ \colhead{Hel. Julian Date} & \colhead {} & \colhead{$V_A$} &
762: \colhead{($O-C$)$_A$} & \colhead{$V_B$} & \colhead{($O-C$)$_B$} \\
763: \colhead{(HJD $-$ 2,400,000)} & \colhead {Phase} &
764: \colhead{(km~s$^{-1}$)} &
765: \colhead{(km~s$^{-1}$)} &\colhead{(km~s$^{-1}$)} &\colhead{(km~s$^{-1}$)}
766: }
767: \startdata
768: 54,495.585 & 0.842 & 10.8 & $-$0.4 & $-$37.1 & 0.7 \\
769: 54,497.586 & 0.427 & $-$51.4 & $-$0.2 & 29.8 & 0.6 \\
770: 54,498.586\tablenotemark{a} & 0.720 & $-$23.4 & $-$2.8 & $-$4.8 & $-$1.2
771: \\
772: 54,498.690\tablenotemark{a} & 0.750 & $-$12.9 & $-$0.5 & $-$12.9 &
773: $-$0.5 \\
774: 54,499.586 & 0.012 & 30.6 & $-$0.1 & $-$58.7 & 0.0 \\
775: 54,499.690 & 0.042 & 28.4 & $-$0.9 & $-$57.2 & 0.0 \\
776: 54,502.690 & 0.919 & 26.1 & 0.7 & $-$52.6 & 0.4 \\
777: 54,505.630\tablenotemark{a} & 0.779 & $-$3.4 & 1.2 & $-$20.0 & 0.9 \\
778: 54,505.679\tablenotemark{a} & 0.793 & $-$1.6 & $-$0.8 & $-$24.6 & 0.3 \\
779: 54,506.630 & 0.071 & 26.2 & $-$0.3 & $-$54.3 & $-$0.1 \\
780: 54,506.679 & 0.086 & 24.9 & 0.3 & $-$52.2 & 0.0 \\
781: 54,507.590 & 0.352 & $-$38.2 & 0.1 & 14.7 & $-$0.7 \\
782: 54,508.590 & 0.645 & $-$38.0 & 1.0 & 16.6 & 0.5 \\
783: 54,510.591\tablenotemark{a} & 0.230 & $-$3.3 & 3.6 & $-$22.2 & $-$3.9
784: \\
785: 54,514.593 & 0.400 & $-$47.7 & $-$0.4 & 25.3 & 0.2 \\
786: 54,516.594 & 0.985 & 30.1 & $-$0.5 & $-$59.4 & $-$0.8 \\
787: 54,516.658 & 0.004 & 30.7 & $-$0.1 & $-$59.6 & $-$0.8 \\
788: 54,517.594\tablenotemark{a} & 0.277 & $-$20.4 & $-$0.6 & $-$5.7 &
789: $-$1.2 \\
790: 54,517.658\tablenotemark{a} & 0.296 & $-$23.4 & 1.3 & 1.4 &
791: 0.6 \\
792: 54,522.597\tablenotemark{a} & 0.740 & $-$12.4 & 2.7 & $-$12.4 &
793: $-$2.9 \\
794: 54,523.597 & 0.032 & 29.9 & 0.0 & $-$58.1 & $-$0.2 \\
795: 54,523.648 & 0.047 & 29.8 & 0.9 & $-$55.8 & 1.0 \\
796: 54,524.598 & 0.325 & $-$31.6 & 0.5 & 8.8 & 0.2 \\
797: 54,532.602 & 0.665 & $-$34.2 & 0.1 & 10.7 & $-$0.4 \\
798: 54,533.604 & 0.958 & 29.9 & 0.6 & $-$56.0 & 1.2 \\
799: 54,534.604\tablenotemark{a} & 0.251 & $-$11.8 & 0.8 & $-$11.8 &
800: 0.4 \\
801: 54,535.604 & 0.543 & $-$54.9 & $-$0.8 & 31.6 & $-$0.7 \\
802: 54,540.602 & 0.004 & 29.9 & $-$0.9 & $-$60.0 & $-$1.2 \\
803: \enddata
804: \tablenotetext{a}{Spectrum has blended components, velocities given zero
805: weight}
806: \end{deluxetable}
807:
808: \clearpage
809: \begin{deluxetable}{lc}
810: \tablenum{3}
811: \tablewidth{0pt}
812: \tablecaption{ORBITAL ELEMENTS AND RELATED PARAMETERS OF BD +20 307}
813: \tablehead{ \colhead{Parameter} & \colhead {Value } \\
814: }
815: \startdata
816: $P$ (days) & 3.42015 $\pm$ 0.00067 \\
817: $T_0$ (HJD) & 2,454,506.3855 $\pm$ 0.0028 \\
818: $\gamma$ (km~s$^{-1}$) & $-$12.43 $\pm$ 0.11 \\
819: $K_A$ (km~s$^{-1}$) & 43.20 $\pm$ 0.18 \\
820: $K_B$ (km~s$^{-1}$) & 46.39 $\pm$ 0.18 \\
821: $e$ & 0.0 (adopted) \\
822: $M_A$~sin$^3$~$i$ ($M_{\sun}$) & 0.1322 $\pm$ 0.0012 \\
823: $M_B$~sin$^3$~$i$ ($M_{\sun}$) & 0.1231 $\pm$ 0.0011 \\
824: $a_A$~sin~$i$ (10$^6$ km) & 2.0317 $\pm$ 0.0083 \\
825: $a_B$~sin~$i$ (10$^6$ km) & 2.1816 $\pm$ 0.0087 \\
826: Standard error of an observation of unit weight (km~s$^{-1}$) & 0.6 \\
827: \enddata
828: \end{deluxetable}
829:
830: \clearpage
831: \begin{figure}[p]
832: \epsscale{0.8}
833: \plotone{f1.eps}
834: \caption{$Top~Panel$: The ensemble photometric observations of BD+20~307
835: taken in the Str\"omgren $(b+y)/2$ passband with the T12 0.8m APT at
836: Fairborn Observatory over four observing seasons. The error bar in the
837: lower left corner represents the typical 1-$\sigma$
838: uncertainty ($\pm 0.0015$ mag) for a single observation.
839: $Second~Panel$: A
840: portion of the third observing season, marked with the vertical lines in
841: the top panel and plotted with an expanded absissa, shows low-amplitude
842: but coherent variability in BD+20~307. $Third~Panel$: Power spectrum of
843: the data in panel 2 revealing a best period of 3.530 days. $Bottom~Panel$:
844: Plot of the data in the second panel phased with the 3.530-day period and a
845: computed time of minimum. The amplitude of brightness variation is only
846: 0.0066 mag.}
847: \end{figure}
848:
849: \clearpage
850: \begin{figure}[t!]
851: \figurenum{2}
852: \epsscale{0.8}
853: \plotone{f2.eps}
854: \figcaption{Radial velocities (dots = component A, open circles =
855: component B) of BD +20 307 compared with the computed radial
856: velocity curves.
857: Zero phase is a time of maximum velocity of the primary, component A.
858: \label{fig2}}
859: \end{figure}
860:
861:
862: \end{document}
863:
864: