An eclipsing-binary distance to the Large Magellanic Cloud accurate to two per cent

Abstract

In the era of precision cosmology, it is essential to determine the Hubble constant to an accuracy of three per cent or better. At present, its uncertainty is dominated by the uncertainty in the distance to the Large Magellanic Cloud (LMC), which, being our second-closest galaxy, serves as the best anchor point for the cosmic distance scale. Observations of eclipsing binaries offer a unique opportunity to measure stellar parameters and distances precisely and accurately. The eclipsing-binary method was previously applied to the LMC, but the accuracy of the distance results was lessened by the need to model the bright, early-type systems used in those studies. Here we report determinations of the distances to eight long-period, late-type eclipsing systems in the LMC, composed of cool, giant stars. For these systems, we can accurately measure both the linear and the angular sizes of their components and avoid the most important problems related to the hot, early-type systems. The LMC distance that we derive from these systems (49.97 ± 0.19 (statistical) ± 1.11 (systematic) kiloparsecs) is accurate to 2.2 per cent and provides a firm base for a 3-per-cent determination of the Hubble constant, with prospects for improvement to 2 per cent in the future.

Full text

1
An eclipsing binary distance to the Large Magellanic Cloud
accurate to 2 per cent
G. Pietrzyński1,2, D. Graczyk1, W. Gieren1, I.B. Thompson3, B., Pilecki1,2, A. Udalski2, I.
Soszyński2, S. Kozłowski2, P. Konorski2, K. Suchomska2, G. Bono4,5, P. G. Prada
Moroni6,7, S. Villanova1, N. Nardetto8, F. Bresolin9, R.P. Kudritzki9, J. Storm10, A.
Gallenne1, R. Smolec11, D. Minniti12,13, M. Kubiak2, M. Szymański2, R. Poleski2, Ł.
Wyrzykowski2, K. Ulaczyk2, P. Pietrukowicz2, M. Górski2, P. Karczmarek2
1. Universidad de Concepción, Departamento de Astronomìa, Casilla 160-C, Concepciòn, Chile
2. Warsaw University Observatory, Aleje Ujazdowskie 4, 00-478 Warszawa, Poland
3. Carnegie Observatories, 813 Santa Barbara Street, Pasadena, CA 91101-1292, USA
4. Dipartimento di Fisica Universita’ di Roma Tor Vergata, via della Ricerca Scientifica 1, 00133
Rome, Italy
5. INAF-Osservatorio Astronomico di Roma, Via Frascati 33, 00040 Monte Porzio Catone, Italy
6. Dipartimento di Fisica Universita’ di Pisa, Largo B. Pontecorvo 2, 56127 Pisa, Italy
7. INFN, Sez. Pisa, via E. Fermi 2, 56127 Pisa, Italy
8. Laboratoire Fizeau, UNS/OCA/CNRS UMR6525, Parc Valrose, 06108 Nice Cedex 2, France
9. Institute for Astronomy, 2680 Woodlawn Drive, Honolulu, HI 96822, USA
10. Leibniz Institute for Astrophysics, An der Sternwarte 16, 14482, Postdam, Germany
11. Nicolaus Copernicus Astronomical Centre, Bartycka 18, 00-716 Warszawa, Poland
12. Departamento de Astronomía y Astrofísica, Pontificia Universidad Católica de Chile, Vicuña
Mackenna 4860, Casilla 306, Santiago 22, Chile

2
13. Vatican Observatory, V00120 Vatican City State, Italy
In the era of precision cosmology it is essential to determine the Hubble Constant
with an accuracy of 3% or better1,2. Currently, its uncertainty is dominated by the
uncertainty in the distance to the Large Magellanic Cloud (LMC) which as the
second nearest galaxy serves as the best anchor point of the cosmic distance scale2,3.
Observations of eclipsing binaries offer a unique opportunity to precisely and
accurately measure stellar parameters and distances4,5. The eclipsing binary
method was previously applied to the LMC6,7 but the accuracy of the distance
results was hampered by the need to model the bright, early-type systems used in
these studies. Here, we present distance determinations to eight long-period, late-
type eclipsing systems in the LMC composed of cool giant stars. For such systems
we can accurately measure both the linear and angular sizes of their components
and avoid the most important problems related to the hot early-type systems. Our
LMC distance derived from these systems is demonstrably accurate to 2.2 %
(49.97 ± 0.19 (statistical) ± 1.11 (systematic) kpc) providing a firm base for a 3 %
determination of the Hubble Constant, with prospects for improvement to 2 % in
the future.
The modelling of early-type eclipsing binary systems consisting of hot stars is made
difficult by the problem to obtain accurate flux calibrations for early-type stars, and by
the degeneracy between the stellar effective temperatures and reddening8,9. As a result,
the distances determined from such systems are of limited (~5-10%) accuracy. A better
distance accuracy can be obtained using binary systems composed of cool stars; such
systems among the frequent dwarf stars in the LMC are however too faint for an

3
accurate analysis with present-day telescopes.
The OGLE team has been monitoring some 35 million stars in the field of the LMC for
more than 16 years10. Based on this unique dataset we have detected a dozen extremely
scarce very long period (60 – 772 days) eclipsing binary systems composed of
intermediate-mass late-type giants located in a quiet evolutionary phase on the helium
burning loop11 (see Supplementary Table 1). These well detached systems provide an
opportunity to use the full potential of eclipsing binaries as precise and accurate
distance indicators, and to calibrate the zero point of the cosmic distance scale with an
accuracy of about 2 % 5,12,13.
In order to achieve this goal, we observed 8 of these systems (see Figure 1) over the last
8 years, collecting high-resolution spectra with the MIKE echelle spectrograph at the
6.5-m Magellan Clay telescope at the Las Campanas Observatory, and with the HARPS
spectrograph attached to the 3.6-m telescope of the European Southern Observatory on
La Silla, together with near infrared photometry obtained with the 3.5-m New
Technology Telescope located on La Silla.
The spectroscopic and OGLE V- and I-band photometric observations of the binary
systems were then analysed using the 2007 version of the standard Wilson-Devinney
(WD) code14,15, in an identical manner as in our recent work on a similar system in the
Small Magellanic Cloud9. Realistic errors to the derived parameters of our systems were
obtained from extensive Monte Carlo simulations (see Figure 2). For all observed
eclipsing binaries, their astrophysical parameters were determined with an accuracy of a
few percent (see Supplementary Tables 2-9).
For late-type stars we can use the very accurately calibrated (2 %) relation between their
surface brightness and V-K color to determine their angular sizes from optical (V) and
near-infrared (K) photometry16. From this surface brightness-color relation (SBCR) we

4
can derive angular sizes of the components of our binary systems directly from the
definition of the surface brightness. Therefore the distance can be measured by
combining the angular diameters of the binary components derived in this way with
their corresponding linear dimensions obtained from the analysis of the spectroscopic
and photometric data. The distances measured with this very simple but accurate one-
step method are presented in Supplementary Table 12. The statistical errors of the
distance determinations were calculated adding quadratically the uncertainties on
absolute dimensions, V-K colors, reddening, and the adopted reddening law. The
reddening uncertainty contributes very little (0.4 %) to the total error 17,11. A significant
change of the reddening law (from Rv = 3.1 to 2.7) causes an almost negligible
contribution at the level of 0.3 %. The accuracy of the V-K color for all components of
our eight binary systems is better than 0.014 mag (0.7 %). The resulting statistical errors
in the distances are very close to 1.5 %, and are dominated by the uncertainty in the
absolute dimensions. Calculating a weighted mean from the individual distances to the
eight target eclipsing binary systems, we obtain a mean LMC distance of 49.88 ± 0.13
kpc.
Our distance measurement might be affected by the geometry and depth of the LMC.
Fortunately, the geometry of the LMC is simple and well studied 18. Since nearly all the
eclipsing systems are located very close to the center of the LMC and to the line of
nodes (see Fig. 3) we fitted the distance to the center of the LMC disk plane assuming
its spatial orientation 18. We obtained an LMC barycenter distance of 49.97 ± 0.19 kpc
(see Figure 4), nearly identical to the simple weighted mean value, which shows that the
geometrical structure of the LMC has no significant influence on our present distance
determination.
The systematic uncertainty in our distance measurement comes from the calibration of
the SBCR and the accuracy of the zero points in our photometry. The rms scatter on the

5
current SBCR is 0.03 mag13, which translates to a 2 % accuracy in the respective
angular diameters of the component stars. Since the surface brightness depends only
very weakly on metallicity16,17, this effect contributes to the total error budget at the
level of only 0.3 % 9. Both optical (V) and near-infared (K) photometric zero points are
accurate to 0.01 mag (0.5 %). Combining these contributions quadratically we
determine a total systematic error of 2.1 % in our present LMC distance determination.
The LMC contains significant numbers of different stellar distance indicators, and
being the second closest galaxy to our own offers us a unique opportunity to study these
indicators with the utmost precision. For this reason this galaxy has an impressive
record of several hundred distance measurements which have been carried out over the
years2,3,19. Unfortunately, virtually all LMC distance determinations are dominated by
systematic errors, with each method having its own sources of uncertainties. This
prevents a calculation of the true LMC distance by simply taking the mean of the
reported distances resulting from different techniques. Our present LMC distance
measurement of 49.97 ± 0.19 (statistical) ± 1.11 (systematic) kpc (i.e. a true distance
modulus of 18.493 mag ± 0.008(statistical) ± 0.047 (systematic)) agrees well, within the
combined errors, with the most recent distance determinations to the LMC19. Our purely
empirical method allows us to estimate both statistical and systematic errors in a very
reliable way, which is normally not the case, particularly in distance determinations
relying in part on theoretical predictions of stellar properties and their dependences on
environment. In particular, our result provides a significant improvement over previous
LMC distance determinations made using observations of eclipsing binaries7,20. These
studies were based on early-type systems for which no empirical surface brightness-
color relation is available, so they had to rely on theoretical models to determine the
effective temperature. Our present determination is based on many (8) binary systems
and does not resort to model predictions.

6
The classical approach to derive the Hubble constant consists in deriving an absolute
calibration of the Cepheid Period-Luminosity Relation (CPLR) which is then used to
determine the distances to nearby galaxies containing Type Ia supernovae (SNIa).21
SNIa are excellent standard candles reaching out to the region of unperturbed Hubble
flow once their peak brightnesses are calibrated this way, and provide the most accurate
determination of H0.22 A yet alternative approach to calibrate the CPLR with Cepheids in
the LMC is to calibrate it in our own Milky Way galaxy using Hubble Space Telescope
(HST) parallax measurements of the nearest Cepheids to the Sun23. However, the
resulting CPLR from that approach is less accurate for two reasons: first, the Cepheid
sample with HST parallaxes is very small (ten stars) as compared to the Cepheid sample
in the LMC (2000 stars), which can be used to establish the CPLR once the LMC
distance is known. Second, the average accuracy of the HST Cepheid parallaxes is 8%23
and suffers from systematics which are not completely understood, including Lutz-
Kelker bias24,25. Therefore the currently preferred route to determine the Hubble constant
is clearly the one using the very abundant LMC Cepheid population whose mean
distance is now known, with the result of this work, to 2.2 %. This result reduces the
uncertainty on H0 to a very firmly established 3 %.
We have good reasons to believe that there is significant room to improve on our current
2.2% distance determination to the LMC by improving the calibration of the SBCR for
late-type stars,12,16 which is the dominant source of systematic error in our present
determination. We are currently embarked on such a program, and a distance
determination to the LMC accurate to 1% seems within reach once the SBCR
calibration is refined, with its corresponding effect on improving the accuracy of H0
even further. This is similar to the accuracy of the geometrical distance to the LMC
which is to be delivered by GAIA in some 12 years from now. The eclipsing binary
technique will then likely provide the best opportunity to check on the future GAIA
measurements for possible systematic errors.

7
REFERENCES
1. Komatsu, E., et al. Seven-year Microwave Anisotropy Probe (WMAP)
Observations: Cosmological Interpretation, Astrophys, J. Sup. Ser., 192, 18-65
(2011)
2. Freedman, W.L., Madore, B.F. The Hubble Constant, Annual Review of Astron.
and Astrophys. 48, 673-710 (2010)
3. Schaefer, B. E. A Problem with the Clustering of Recent Measures of the
Distance to the Large Magellanic Cloud Astron. J. 135, 112-119 (2008)
4. Lacy, C.H. Distances to eclipsing binaries: an application of the Barnes-Evans
relation, Astrophys., J., 213, 458-463 (1977)
5. Paczyński, B. Detached eclipsing binaries as primary distance and age indicators
Space Telescope Science Institute Series, The Extragalactic Distance Scale,
edited by M. Livio (Cambridge University Press), p. 273-280 (1997)
6. Guinan, E.F. The Distance to the Large Magellanic Cloud from the Eclipsing
Binary HV 2274, Astrophys., J., 509, L21-24 (1998)
7. Fitzpatrick, E. L., Ribas, I.,Guinan, E. F.,Maloney, F. P., and Claret, A.
Fundamental Properties and Distances of Large Magellanic Cloud Eclipsing
Binaries. IV. HV 5936, Astrophys. J., 587, 685-700 (2003)
8. Groenewegen, M.A.T., Salaris, M. The LMC eclipsing binary HV 2274
revisited, Astrophys, J., 366, 752-764 (2001)

8
9. Graczyk, D., et al. The Araucaria Project. An Accurate Distance to the Late-type
Double-lined Eclipsing Binary OGLE SMC113.3 4007 in the Small Magellanic
Cloud. Astrophys. J., 750, 144-156 (2012)
10. Udalski, A., Soszyński, I., Szymański, M., Kubiak, M., Pietrzyński, G.,
Wyrzykowski, Ł., Szewczyk, O., Ulaczyk, K., Poleski, R. The Optical
Gravitational Lensing Experiment. OGLE-III Photometric Maps of the Large
Magellanic Cloud. Acta Astron., 58, 89-102 (2008)
11. Graczyk, D., Soszyński, I., Poleski, R., Pietrzyński, G., Udalski, A., Szymański,
M.K., Kubiak, M., Wyrzykowski, Ł., Ulaczyk, K. The Optical Gravitational
Lensing Experiment. The OGLE-III Catalog of Variable Stars. XII. Eclipsing
Binary Stars in the Large Magellanic Cloud. Acta Astron., 61, 103-122 (2011)
12. Kruszewski, A., Semeniuk, I. Nearby Hipparcos Eclipsing Binaries for Color-
Surface Brightness Calibration, Acta Astron., 49, 561-575 (1999)
13. Pietrzyński, G. Thompson, I. Graczyk, D., Gieren, W., Udalski, A., Szewczyk,
O., Minniti, D., Kołaczkowski, Z., Bresolin, F., Kudritzki, R.P. The Araucaria
Project. Determination of the Large Magellanic Cloud Distance from Late-Type
Eclipsing Binary Systems. I. OGLE051019.64-685812.3, Astrophys. J., 697,
862-866 (2009)
14. Wilson, R.E., Devinney, E.J. Realization of accurate close-binary light curves:
application to MR Cygni, Astrophys. J. 166, 605-620 (1971)
15. Van Hamme W. & Wilson R.E. Third-Body Parameters from Whole Light and
Velocity Curves, Astrophys. J. 661, 1129-1151 (2007)

9
16. Di Benedetto, G.P. Predicting accurate stellar angular diameters by the near-
infrared surface brightness technique. MNRAS, 357, 174-190 (2005)
17. Thompson, I.B., Kalużny, J., Pych, W., Burley, G. Krzemiński, W., Paczyński,
B., Persson, S.E., Preston, G.W. Cluster AgeS Experiment. The Age and
Distance of the Globular Cluster Centauri Determined from Observations of the
Eclipsing Binary OGLEGC 17. Astron. J., 118, 642-467 (1999)
18. Van der Marel, R.P., Alves, D.R., Hardy, E., Suntzeff, N.B. New Understanding
of Large Magellanic Cloud Structure, Dynamics, and Orbit from Carbon Star
Kinematics. Astron. J., 124, 2639-2663 (2002)
19. Mazzarella, J. M. NED for a New Era. Astron. Soc. Pac. Conf. 376, 153-162
(2007)
20. Bonanos, A.Z., Castro, N., Macri, L.M., Kudritzki, R.P. The Distance to the
Massive Eclipsing Binary LMC-SC1-105 in the Large Magellanic Cloud,
Astrophys. J. Lett. 729, L9-15 (2011)
21. Freedman, W.L., et al. Final Results from the Hubble Space Telescope Key
Project to Measure the Hubble Constant. Astrophys.J., 553, 47-72 (2001)
22. Riess, A.G., Macri, L. Casertano, S., Lampeitl, H., Ferguson, H.C., Filippenko,
A.V., Jha, S.W., Li W., Weidong, Ch.R. A 3% Solution: Determination of the
Hubble Constant with the Hubble Space Telescope and Wide Field Camera 3.
Astrophys. J., 730, 119-137, (2011)
23. Benedict, et al. Hubble Space Telescope Fine Guidance Sensor Parallaxes of
Galactic Cepheid Variable Stars: Period-Luminosity Relations. Astron. J., 133,

10
1810-1827 (2007)
24. Lutz, T.E. & Kelker, D.H., On the Use of Trigonometric Parallaxes for the
Calibration of Luminosity Systems: Theory. PASP, 85, 573-578 (1973)
25. van Leeuwen, F., Feast, M.W., Whitelock, P.A. & Laney, C.D., Cepheid
parallaxes and the Hubble constant. MNRAS, 379, 723-737 (2007)
26. Pojmański, G. The All Sky Automated Survey. Acta Astron., 47, 467-481 (1997)
Acknowledgements We gratefully acknowledge financial support for this work from
the BASAL Centro de Astrofisica y Tecnologias Afines (CATA), Polish Ministry of
Science, the Foundation for Polish Science (FOCUS, TEAM), Polish National Science
Centre, and the GEMINI-CONICYT fund. The OGLE project has received funding
from the European Research Council “Advanced Grant” Program. It is a pleasure to
thank the staff astronomers at Las Campanas and ESO La Silla who provided expert
support in the data acquisition. We thank Jorge F. Gonzalez for making IRAF scripts
rvbina and spbina available to us. We also thank O. Szewczyk and Z. Kołaczkowski for
their help with some of the observations.
Author Information
Correspondence and requests for materials should be addressed to
pietrzyn@astrouw.edu.pl
Author Contributions G.P. photometric and spectroscopic observations and
reductions. D.G., spectroscopic observations, modelling, data analysis, W.G.
Observations and data analysis, I.T. , observations, RV determination, data analysis,
B.P. spectroscopic observations and reductions, RV measurements., A.U. , I.S and S. K.
optical observations and data reductions. P.K., K.S., M.K., M.Sz., R.P., Ł., W., K.U.,

11
P.P., M.G., P.K. observations. G.B., P.G.P.M., N.N, F.B., R.P.K., J.S., A.G. and R.S. Data
analysis. S.V., analysis of the spectra, G.P. and W.G. worked jointly to draft the
manuscript with all authors reviewing and contributing to its final form
Figure 1: Change of the brightness of the binary system OGLE-LMC-ECL-06575
and the orbital motion of its components.

12
a, Main panel, orbital motion of the two binary components in the OGLE-LMC-
ECL-06575 system. Filled and open circles, primary and secondary
components, respectively. The top panel shows the residuals of the fit (see
below): observed radial velocities (O) minus the computed radial velocities (C).
b, Main Panel, the I-band light curve (1200 epochs collected over 16 years) of
the binary system OGLE-LMC-ECL-06575 together with the solution, as
obtained with the Wilson-Devinney code. The top panel shows the residuals of
the observed magnitudes from the computed orbital light curve.
All individual radial velocities were determined by the cross-correlation method
using appropriate template spectra and the MIKE and HARPS spectra, yielding
in all cases velocity accuracies better than 200 m/s (error bars smaller than the
circles in the figure). The orbit (mass ratio, systemic velocity, velocity
amplitudes, eccentricity, and periastron passage), was fitted with a least
squares method to the measured velocities. The resulting parameters are
presented in Supplementary Tables 2-9. The spectroscopic orbits, light curves
and solutions for the remaining systems are of similar quality.

13
Figure 2: Error estimation of the distance for one of our target binary
systems.
The reduced χ2 map for the OGLE-LMC-ECL-15260 system showing the
dependence of the fit goodness to the V-band and I-band light curves on the
distance modulus of the primary component. This map was obtained from
110,000 models computed with the Wilson-Devinney code14,15 within a broad
range of the radii R1 and R
2, the orbital inclination i, the phase shift
φ
, the
secondary’s temperature T2 and the secondary’s albedo A2. In each case the
distance d was calculated from the V-band surface brightness – color (V–K)
relation16 and translated into distance modulus via formula
(m−M)=5×log (d)−5
. The horizontal lines correspond to the standard
deviation limits of the derived distance modulus of 18.509 mag (50.33 kpc),
accordingly from down to up: 1σ, 2σ and 3σ.

14
Figure 3: Location of the observed eclipsing systems in the LMC.
Most of our eight systems (marked as filled circles) are located quite close to
the geometrical center of the LMC and to the line of nodes (marked with the
line), resulting in very small corrections to the individual distances for the
geometrical extension of this galaxy (in all cases smaller than the corresponding
statistical error on the distance determination). The effect of the geometrical
structure of the LMC on the mean LMC distance reported in our Letter is
therefore negligible. The background image has a field of view of 8 x 8 degrees
and is taken from the ASAS wide field sky survey26.

15
Figure 4: Consistency among the distance determinations for the target
binary systems
Distance offsets between our particular eclipsing binary systems and the best
fitted LMC disk plane, plotted against the angular distance of the systems from
the LMC center. The identification of the systems is the same as in Figure 3.
The error bars correspond to one sigma errors. We assumed the model of the
LMC from van der Marel et al 18. We fitted one parameter: the distance to the
center of the LMC (R.A. = 5h 25m 06s, DEC = –69o 47’ 00’’) using a fixed spatial
orientation of the LMC disk: inclination i = 28 deg and a position angle of the
nodes of θ = 128 deg. The resulting distance to the LMC barycenter is 49.97 ±
0.19 kpc, with a reduced χ2 very close to unity.

1
Supplementary Information:
1) Observations and target binary systems
All eclipsing systems studied in this letter, and listed in Supplementary Table 1, were
discovered based on the OGLE-II and OGLE-III data11. Additional V and I band
observations were collected with the Warsaw 1.3 m telescope at Las Campanas
Observatory in the course of the OGLE IV project, and with the 1.3 m telescope at
Cerro Tololo Observatory. Once the preliminary periods were calculated the new
observations were secured mostly during eclipses, which resulted in a very good phase
coverage of all targets. All photometric data were reduced with the image subtraction
technique29 and were calibrated based on the OGLE-III data10. The finding chart for all
systems can be found on the OGLE Project webpage
(ftp://ftp.astrouw.edu.pl/ogle/ogle3/OIII-CVS/lmc/ecl/fcharts)11. The raw OGLE V-
band and I-band light curves are available at ftp://ftp.astrouw.edu.pl/ogle/ogle3/OIII-
CVS/lmc/ecl/phot.
The near infrared photometry was performed with the ESO-La Silla 3.5 m NTT
telescope equipped with the SOFI imager. Each system was observed outside of the
eclipses during at least 5 different nights through J and K filters under photometric
conditions together with a large number (12-16) of photometric standards from the
UKIRT system30. The accuracy of the zero points obtained for every night was about
0.02 mag in both J and K bands. For more details regarding the observations, reduction
and calibrations of the near-infrared data the reader is referred to (29).
High resolution echelle spectra were collected with the Las Campanas Observatory
Magellan Clay 6.5 m telescope and the MIKE echelle spectrograph32, and with the ESO
3.6 m telescope and the HARPS fiber-fed echelle spectrograph33. In the case of the
MIKE observations, a 0.7 arcsec slit was used giving a resolution of about 40,000. The
spectra were reduced with the dedicated pipeline software34. Exposure times ranged
from 1200 sec to 3600 sec depending on observing conditions, and a typical resulting
S/N ratio was between 7 and 30 at a wavelength of 450 nm. The HARPS observations
were obtained at a resolution of 80,000 and a S/N of about 4-10 at 500 nm for
integrations in the range from one minute to half an hour, and were reduced with the
data reduction software developed by the Geneva observatory. Radial velocities were
calculated with the Broadening Function (BF) formalism35 and TODCOR36. We used
templates from a library37. Templates were matched to the estimated mean effective
temperature and gravity of the components. Both determinations agree with each other
within 100 m/s.

2
Id
R.A. (2000)
Dec. (2000)
P [days]
OGLE-LMC-ECL-09660
05h 11m 49s.45
−67°05’45.2
167.6350 ± 0.0016
OGLE-LMC-ECL-10567
05h 14m 01s.89
−68°41’18.2
117.8708 ± 0.0012
OGLE-LMC-ECL-26122
05h14m06s.04
−69°15’56.9
771.7806 ± 0.0048
OGLE-LMC-ECL-09114
05h10m19s.64
−68°58’12.2
214.1707 ± 0.0009
OGLE-LMC-ECL-06575
05h04m32s.87
−69°20’51.0
189.8215 ± 0.0010
OGLE-LMC-ECL-01866
04h52m15s.28
−68°19’10.3
251.0068 ± 0.0043
OGLE-LMC-ECL-03160
04h55m51s.48
−68°13’48.0
150.0198 ± 0.0018
OGLE-LMC-ECL-15260
05h25m25s.66
−69°33’04.5
157.3243 ± 0.0008
Supplementary Table 1. Selected eclipsing binary systems.
The periods reported in this table and in all Supplementary Information are true orbital
periods linked with the observed periods through the following formula:
€
P=P
OBS ×1+
γ
c
$
%
& '
(
)
−1
where γ is the systemic velocity and c velocity of light,
respectively.
2) The essentials of the modelling
The V and I-band light curves were cleaned from obvious outliers. No sigma clipping
was applied. Both light curves and the two radial velocity curves – one per each
component of a system – were analysed simultaneously for each system with the
Wilson-Devinney code version 2007. We will denote by a subscript “1” the primary
component and by a subscript “2” the secondary component of the system. We fitted the
following set of adjustable parameters: the semi-major axis
a
, the mass ratio of the
components
q
, the systemic velocity of the system γ, the apparent orbital period Pobs, the
modified surface potentials Ω1 and Ω2, the secondary’s mean surface temperature T2,
the internal luminosity of the primary component in the two bands L1V, L1I and the
orbital inclination
i
. In case of circular systems we fitted the epoch of the primary
eclipse HJD0 and in some cases the albedo of the secondary A2, while in the cases of
eccentric systems we fitted the phase shift ϕ and additionally the orbital eccentricity e
and the argument of periastron ω. The initial values of adjustable parameters for the
Wilson-Devinney code were found by a trial-and-error procedure.

3
The temperature of the primary T1 (which is important because it scales the temperature
of the secondary component and influences the limb darkening coefficients) was set as
follows. The initial value was set to 5000 K, as this value is a reasonable assumption for
late type giant stars. Then we run the WD code to obtain a preliminary model from
which we obtained the components’ light ratios in the V- and I-band. Additionally we
extrapolated the model to calculate the K-band light ratio using theoretical limb
darkening coefficients in this band. Combining these light ratios with the observed
magnitudes in the V-,
I
- and K-bands, and using the reddenings derived from red clump
stars in the fields containing our eclipsing binaries, we calculated the intrinsic colors
(V–I)0, (V–K)0 of both stars. Using several calibrations between color indices and
effective temperature38,39,40,41 we estimated the temperatures of both components,
especially the primary’s temperature T1. Then we run the WD code again with the new
temperatures. Once updated reddening estimates were obtained we repeated the
procedure. We iterated the temperature determination several times until full
consistency of the model parameters was obtained, especially an agreement between the
distance obtained from the surface brightness calibration and the distance resulting from
bolometric flux scaling.
For most of our systems a correlation between the relative radii of both components can
be observed. This is usual in the case of eclipsing binaries with partial eclipses. Two of
our systems, LMC-ECL-09114 and LMC-ECL-09660, show total eclipses thus this
correlation is unimportant in their case. However for the rest of the sample the influence
of this correlation on the stellar radii and the resulting distance determination had to be
investigated. To this end we first calculated approximate spectroscopic light ratios from
the intensity of absorption lines and compared them to our light curve model
predictions. A disagreement was found only for system LMC-ECL-10567.
Subsequently we performed Monte Carlo simulations to investigate multi-dimensional
parameter space and possible correlations between parameters. Some details of these
simulations are given at the end of this section.
Initially we used a logarithmic limb darkening law in all cases utilizing pre-computed
tables of theoretical limb darkening coefficients (setting LD= -2 in the WD2007 code).
Additionally, we computed models with linear limb darkening (setting LD= -1) and also
models where coefficients of the linear limb darkening law were treated as adjustable
parameters (setting LD = +1). Usually models computed with linear and logarithmic law
limb darkenings result in similar reduced χ2 values. However models with adjusted
coefficients of linear limb darkening usually lead to considerably better fits to the light
curves. Exceptions are the systems LMC-ECL-26122, LMC-ECL-01866 and LMC-
ECL-10567. In the first one no improvement in the reduced χ2 can be seen. In the
second case this procedure leads to a solution with very low coefficients (below zero)
indicating limb brightening. In the last system the fitted coefficients are peculiar:
unacceptably high and larger in the infrared than in optical. In those three cases we
adopted the solution obtained with fixed coefficients of the logarithmic limb darkening
law.
We tested the possible presence of a third light in our light curves and spectra. We could
not detect any additional source of absorption lines in our spectra. We tested our spectra
using the CCF, TODCOR and BF methods with different templates corresponding to a

4
temperature range from 3500 to 7000 K but we failed to detect any third light source
stronger that 1% of the combined signal of the two components of the systems. We also
computed models setting the third light parameter l3 (being a fraction of the total
observed flux) as an adjustable parameter of a solution for all our systems. Only in the
case of LMC-ECL-10567 we found a small contribution of a third light in the optical V-
band light curve, however in the I-band the third light contribution is insignificant. The
third light in this system, if real, might be a faint blue companion or an optical blend
(V~20 mag; I~21.5 mag) e.g. a white dwarf or even a QSO.
After inspection of the absorption lines broadening, which mostly comes from rotational
broadening, we concluded that the rotation velocities of the components are consistent
with both components being synchronized. Thus in all our computations the rotation
parameter F was fixed at 1.0 for both stars of a given eclipsing binary system,
corresponding to the rotation being synchronous with the orbital period.
We estimated statistical errors on the parameters by performing Monte-Carlo
simulations. We calculated a large number of models (usually over one hundred
thousand per system) with input parameters randomly selected from a broad range of
possible values and compared them to the V-band and I-band light curves to compute
the residuals of the model and the resulting reduced χ2. In the Monte-Carlo simulations
we allowed to vary the following parameters : the modified surface potentials Ω1 and
Ω2, the secondary’s mean surface temperature T2, the orbital inclination i, the epoch of
the primary eclipse HJD0 and in some cases, the albedo of the secondary A2. The
spectroscopic parameters during the simulations were kept fixed. The resulting χ2 maps
and parameter correlation maps were used to derive realistic errors on the model
parameters. For example, the radii correlation in the case of eclipsing binaries with
partial eclipses is only marginally affecting our distances because the prime source of
statistical error is the uncertainty of the sum of the radii R1+R2. Although for our
systems, on average, radii are known with an accuracy of 3%, the sum of the radii is
always known with an accuracy better than 2.0%. The largest source of uncertainty in
the radii sum determination turns out to be the correlation between the orbital
inclination and the radii.
Parameter
Primary
Secondary
(m-M) = distance modulus
18.489
18.489
M/M = mass
2.969 ± 0.020
2.988 ± 0.018
R/R = radius
23.75 ± 0.66
43.87 ± 1.14
T =effective temperature
5352 ± 70 K
4677 ± 75 K

5
K = velocity amplitude
35.13 ± 0.06 km/s
34.91 ± 0.08 km/s
e = eccentricity
0.0517 ± 0.0013
ω = periastron passage
212.1 ± 1.5 deg
γ = systemic velocity
286.24 ± 0.04 km/s
P =orbital period
167.6350 ± 0.0016 days
i =inclination
87.81 ± 0.31 deg
a/R = orbit size
232.00 ± 0.32
q = mass ratio
1.0065 ± 0.0027
A = albedo
0.5 (fixed)
xV =linear limb darkening coeff.
0.697 ± 0.098
0.726 ± 0.036
xI = linear limb darkening coeff.
0.450 ± 0.077
0.391 ± 0.039
V (observed magnitude)
17.303
16.799
VI (observed color)
0.959
1.234
VK (observed color)
2.190
2.830
E(B-V) reddening
0.127 ± 0.020
[Fe/H] = metallicity
0.44 ± 0.10
Supplementary Table 2. Astrophysical parameters of the OGLE-LMC-ECL-09660
system
Parameter
Primary
Secondary
(m-M) = distance modulus
18.489
18.491

6
M/M =mass
3.345 ± 0.040
3.183 ± 0.038
R/R = radius
25.6 ± 1.6
36.0 ± 2.0
T =effective temperature
5067 ± 73 K
4704 ± 80 K
K = velocity amplitude
39.31 ± 0.13 km/s
41.32 ± 0.14 km/s
e = eccentricity
0.0 (fixed)
ω = periastron passage
90 (fixed)
γ = systemic velocity
265.10 ± 0.08 km/s
P =orbital period
117.8708 ± 0.0012 days
i = inclination
83.47 ± 0.33 deg
a/R = orbit size
189.13 ± 0.45
q = mass ratio
0.9515 ± 0.0043
A = albedo
0.5 (fixed)
0.100 ± 0.053
xV =linear limb darkening coeff.
not adjusted
not adjusted
xI = linear limb darkening coeff.
not adjusted
not adjusted
l3V = third light
0.0482 ± 0.0305
l3I = third light
0.0036 ± 0.0315
l3K = third light
0.0 (fixed)
V (observed magnitude)
17.374
17.114
VI (observed color)
1.019
1.158
VK (observed color)
2.355
2.730
E(B-V) reddening
0.102 ± 0.020

7
[Fe/H] = metallicity
0.81 ± 0.20
Supplementary Table 3. Astrophysical parameters of the OGLE-LMC-ECL-10567
system
Parameter
Primary
Secondary
(m-M) = distance modulus
18.470
18.468
M/M = mass
3.593 ± 0.055
3.411 ± 0.047
R/R = radius
32.71 ± 0.51
22.99 ± 0.48
T =effective temperature
4989 ± 80 K
4995 ± 81 K
K = velocity amplitude
23.80 ± 0.10 km/s
25.08 ± 0.14 km/s
e = eccentricity
0.4186 ± 0.0019
ω = periastron passage
302.3 ± 0.2 deg
γ = systemic velocity
266.38 ± 0.07 km/s
P =orbital period
771.7806 ± 0.0048 days
i = inclination
88.45 ± 0.04 deg
a/R = orbit size
677.64 ± 2.36
q = mass ratio
0.9491 ± 0.0067
A = albedo
0.5 (fixed)
xV =linear limb darkening coeff.
not adjusted
not adjusted
xI = linear limb darkening coeff.
not adjusted
not adjusted
V (observed magnitude)
17.067
17.827

8
VI (observed color)
1.093
1.088
VK (observed color)
2.561
2.558
E(B-V) reddening
0.140 ± 0.020
[Fe/H] = metallicity
0.15 ± 0.10
Supplementary Table 4. Astrophysical parameters of the OGLE-LMC-ECL-26122
system
Parameter
Primary
Secondary
(m-M) = distance modulus
18.465
18.465
M/M= mass
3.303 ± 0.028
3.208 ± 0.026
R/R = radius
26.18 ± 0.31
18.64 ± 0.30
T =effective temperature
5288 ± 81 K
5470 ± 96 K
K = velocity amplitude
32.76 ± 0.08 km/s
33.37 ± 0.10 km/s
e = eccentricity
0.0393 ± 0.0018
ω = periastron passage
97.1 ± 0.3 deg
γ = systemic velocity
272.04 ± 0.05 km/s
P =orbital period
214.1707 ± 0.0009 days
i =inclination
88.77 ± 0.17 deg
a/R = orbit size
281.38 ± 0.54
q = mass ratio
0.9711 ± 0.0037
A = albedo
0.5 (fixed)

9
xV =linear limb darkening coeff.
0.632 ± 0.068
0.555 ± 0.106
xI = linear limb darkening coeff.
0.323 ± 0.067
0.319 ± 0.082
V (observed magnitude)
17.217
17.783
VI (observed color)
1.059
0.984
VK (observed color)
2.319
2.188
E(B-V) reddening
0.160 ± 0.020
[Fe/H] = metallicity
0.23 ± 0.10
Supplementary Table 5. Astrophysical parameters of the OGLE-LMC-ECL-09114
system.
Parameter
Primary
Secondary
(m-M) = distance modulus
18.497
18.497
M/M = mass
4.152 ± 0.030
3.966 ± 0.032
R/R = radius
39.79 ± 1.35
49.35 ± 1.45
T =effective temperature
4903 ± 72 K
4681 ± 77 K
K = velocity amplitude
36.03 ± 0.09 km/s
37.72 ± 0.07 km/s
e = eccentricity
0.0 (fixed)
ω = periastron passage
90 (fixed)
γ = systemic velocity
274.32 ± 0.05 km/s
P =orbital period
189.8215 ± 0.0010 days

10
i = inclination
82.06 ± 0.13 deg
a/R = orbit size
279.44 ± 0.44
q = mass ratio
0.9552 ± 0.0034
A = albedo
0.5 (fixed)
0.648 ± 0.043
xV =linear limb darkening coeff.
0.789 ± 0.076
0.791 ± 0.068
xI = linear limb darkening coeff.
0.336 ± 0.071
0.337 ± 0.065
V (observed magnitude)
16.642
16.490
VI (observed color)
1.099
1.190
VK (observed color)
2.529
2.775
E(B-V) reddening
0.107 ± 0.020
[Fe/H] = metallicity
0.45 ± 0.10
Supplementary Table 6. Astrophysical parameters of the OGLE-LMC-ECL-06575
system
Parameter
Primary
Secondary
(m-M) = distance modulus
18.496
18.496
M/M = mass
3.575 ± 0.028
3.574 ± 0.038
R/R = radius
28.20 ± 1.06
46.96 ± 0.61
T =effective temperature
5327 ± 72 K
4541 ± 85 K
K = velocity amplitude
33.27 ± 0.14 km/s
33.28 ± 0.05 km/s
e = eccentricity
0.2412 ± 0.0008
ω = periastron passage
15.5 ± 0.3

11
γ = systemic velocity
293.44 ± 0.04 km/s
P =orbital period
251.0068 ± 0.0043 days
i =inclination
83.34 ± 0.10 deg
a/R = orbit size
322.68 ± 0.73
q = mass ratio
0.9997 ± 0.0045
A = albedo
0.5 (fixed)
xV =linear limb darkening coeff.
not adjusted
not adjusted
xI = linear limb darkening coeff.
not adjusted
not adjusted
V (observed magnitude)
16.916
16.842
VI (observed color)
0.952
1.250
VK (observed color)
2.170
2.973
E(B-V) reddening
0.115 ± 0.020
[Fe/H] = metallicity
0.70 ± 0.10
Supplementary Table 7. Astrophysical parameters of the OGLE-LMC-ECL-01866
system.
Parameter
Primary
Secondary
(m-M) = distance modulus
18.505
18.505
M/M = mass
1.792 ± 0.027
1.799 ± 0.028
R/R = radius
16.36 ± 1.06
37.42 ± 0.52
T =effective temperature
4954 ± 83 K
4490 ± 82 K

12
K = velocity amplitude
30.47 ± 0.14 km/s
30.35 ± 0.11 km/s
e = eccentricity
0.0 (fixed)
ω = periastron passage
90 (fixed)
γ = systemic velocity
267.68 ± 0.08 km/s
P =orbital period
150.0198 ± 0.0018 days
i = inclination
83.36 ± 0.57 deg
a/R = orbit size
182.01 ± 0.52
q = mass ratio
1.0039 ± 0.0058
A = albedo
0.5 (fixed)
xV =linear limb darkening coeff.
0.763 ± 0.174
0.796 ± 0.139
xI = linear limb darkening coeff.
0.623 ± 0.112
0.567 ± 0.069
V (observed magnitude)
18.589
17.453
VI (observed color)
1.076
1.308
VK (observed color)
2.542
3.062
E(B-V) reddening
0.123 ± 0.020
[Fe/H] = metallicity
0.48 ± 0.20
Supplementary Table 8. Astrophysical parameters of the OGLE-LMC-ECL-03160
system.
Parameter
Primary
Secondary
(m-M) = distance modulus
18.509
18.509

13
M/M = mass
1.440 ± 0.024
1.426 ± 0.022
R/R = radius
23.51 ± 0.69
42.17 ± 0.33
T =effective temperature
4706 ± 87 K
4320 ± 81 K
K = velocity amplitude
27.67 ± 0.11 km/s
27.93 ± 0.14 km/s
e = eccentricity
0.0 (fixed)
ω = periastron passage
90 (fixed)
γ = systemic velocity
276.66 ± 0.06 km/s
P =orbital period
157.3243 ± 0.0008 days
i =inclination
82.99 ± 0.39 deg
a/R = orbit size
174.25 ± 0.56
q = mass ratio
0.9905 ± 0.0059
A = albedo
0.5 (fixed)
0.530 ± 0.031
xV =linear limb darkening coeff.
0.676 ± 0.092
0.936 ± 0.049
xI = linear limb darkening coeff.
0.598 ± 0.060
0.660 ± 0.031
V (observed magnitude)
18.066
17.390
VI (observed color)
1.149
1.392
VK (observed color)
2.737
3.213
E(B-V) reddening
0.100 ± 0.020
[Fe/H] = metallicity
0.47 ± 0.15
Supplementary Table 9. Astrophysical parameters of the OGLE-LMC-ECL-15260
system.

14
3) Reddening determination and its influence on the distance determination
We used two different techniques to derive the reddening. For all our systems but one
(LMC-ECL-15260) we managed to co-add and disentangle their spectra using the
RaVeSpAn software developed by our group, implementing the method proposed in
(42). The disentangled spectra were renormalized with the luminosity ratio of the
secondary to the primary L2/L1(λ) from our best model. Based on the analysis of the
spectra of individual components we calculated a standard set of atmospheric
parameters (the effective temperature Teff, gravity log g, microturbulance velocity vt and
metallicity [Fe/H]). Synthetic V–I colors were calculated utilizing two calibrations
between atmospheric parameters and the intrinsic color (V–I)o 40,41. Comparing the
colors obtained in this way to the corresponding observed colors, we calculated
reddenings with a typical accuracy of 0.02 mag. In addition we used OGLE-III
photometry10 in order to calculate the mean red clump brightness in the V and I bands in
7x7 arcmin regions centered on our binary systems. In all cases, we were able to
calculate the mean red clump magnitudes with an accuracy better than 0.01 mag.
Following other studies of red clump stars43 we assumed a V-I color of 0.72 (e.g.
median color of RC stars in the LMC) as the color corresponding to the foreground
reddening E(B-V) = 0.075 mag44. Then the reddening was determined from the
observed color of red clump stars using the reddening law45, which is appropriate for the
LMC and was used in all calculations presented in our letter. In order to check our
determinations we also calculated the reddening for three other fields where accurate
reddenings were obtained based on Strömgren photometry46, UBV photometry8 and
spectral analysis22. As can bee seen in Table 1, our determinations agree very well with
the corresponding values of the reddenings reported in the literature. Finally, we
estimated reddenings from the LMC reddening maps47 based on OGLE-III photometry
of RR Lyrae stars. All the determinations are in good agreement (see Table 2). As the
final reddening we adopted the mean from the available determinations.
Field
R.A. (2000)
Dec. (2000)
E(B-V) RC
E(B-V) Lit
Reference
P168.6
5h29m53s
-69°09’23’’
0.151
0.140
Larsen et al. 2000
P169.3
5h34m48s
-69°42’36’’
0.119
0.110
Bonanos et al. 2011
P126.4
5h02m40s
-68°24’21’’
0.119
0.103
Groenewegen &
Salaris (2001)
Supplementary Table 10. Comparison of the reddenings obtained with red clump stars
to the corresponding values reported in the literature, for three fields in the LMC.

15
System
E(B-V) RC
E(B-V) RR Lyr
E(B-V) spec
E(B-V) adopted
OGLE-LMC-ECL-09660
0.126
0.13
0.126
0.127
OGLE-LMC-ECL-10567
0.126
0.106
0.075
0.102
OGLE-LMC-ECL-26122
0.138
0.127
0.155
0.140
OGLE-LMC-ECL-09114
0.139
0.131
0.21
0.160
OGLE-LMC-ECL-06575
0.107
0.101
0.113
0.107
OGLE-LMC-ECL-01866
0.119
0.15
0.075
0.115
OGLE-LMC-ECL-03160
0.126
0.124
0.12
0.123
OGLE-LMC-ECL-15260
0.11
0.091
----
0.100
Supplementary Table 11. Reddening determinations for our target eclipsing systems
based on an analysis of their disentangled spectra, red clump stars, and RR Lyrae stars.

16
Figure S1. The final distances of the studied eclipsing systems calculated as a function
of the adopted reddening. The thick red line is an average relation obtained from all
systems. Points correspond to distances of individual systems identified in the plot
legend. The larger red point denotes our distance determination to LMC. As can be
appreciated from the error bars, a change of the reddening by 0.02 mag (1 σ) causes a
change in the distance by 0.4% only (215 pc). This confirms the previous
conclusions9,17 that our method of distance determination is only very slightly dependent
on the adopted reddening.
4) Distance determination
The surface brightnesses of the components of the studied systems were derived based
on their disentangled (V-K) colors using the calibration
!
Sv=2.656 +1.483 "(V#K)0#0.044 "(V#K)0
2
) 16. The angular diameters were
obtained directly from the definition of the surface brightness (
!
"
[mas]=100.2#(Sv$V0)
).
Finally distances to individual stars were derived combining their angular diameters
obtained this way, and their linear diameters determined from the analysis of the
spectroscopic and photometric data (
d[pc]=9.2984 !R[Rsun ]
!
[mas]
).
In case of models with fixed theoretical limb darkening coefficients the disentangling of
colors was performed in the way described in the Section 2. As a final distance we
adopted the mean distance of both components of a system. The difference in the
distance moduli of the components of the same system is not larger than 0.002 mag in

17
all cases. Such small differences serve as an independent check of consistency and
reliability of our results.
For models with adjusted limb darkening coefficients we employed another strategy to
obtain the individual colors because we cannot extrapolate model predictions for the K-
band in this case (there is no possibility to obtain appropriate limb darkening
coefficients by their adjustment because of the lack of the K-band light curve). We
assumed that both components of a system are at the same distance from us and we
made iterations until we found a K-band components’ light ratio fulfilling our
assumption. Then we computed the disentangled (V–K) colors of the components.
In Supplementary Table 12 we list the individual and mean distances measured for all
studied systems. As can be seen, the obtained distances are in excellent agreement.
System
(m-M) 1
(m-M)2
(m-M) mean
σm-M
Δ(m-M)
OGLE-LMC-ECL-09660
18.489
18.489
18.489
0.025
0.035
OGLE-LMC-ECL-10567
18.489
18.491
18.490
0.027
0.007
OGLE-LMC-ECL-26122
18.470
18.468
18.469
0.025
-0.005
OGLE-LMC-ECL-09114
18.465
18.465
18.465
0.021
-0.004
OGLE-LMC-ECL-06595
18.497
18.497
18.497
0.019
-0.021
OGLE-LMC-ECL-01866
18.496
18.496
18.496
0.028
-0.021
OGLE-LMC-ECL-03160
18.505
18.505
18.505
0.029
-0.012
OGLE-LMC-ECL-15260
18.509
18.509
18.509
0.021
0.004
Supplementary Table 12. Individual distance moduli of the studied eclipsing binary
systems. The symbols 1,2 and mean refer to the individual primary (1) and secondary
(2) components of our eclipsing systems. The fifth column gives the total statistical
uncertainty for the mean distance modulus. The geometrical corrections calculated from
the model18 are given in the last column.

18
System
(m-M)fix
(m-M)fit
Δ(m-M)
OGLE-LMC-ECL-09660
18.490
18.489
-0.001
OGLE-LMC-ECL-10567
18.490
18.506
0.016
OGLE-LMC-ECL-26122
18.469
18.482
0.013
OGLE-LMC-ECL-09114
18.481
18.465
-0.016
OGLE-LMC-ECL-06575
18.52
18.497
-0.023
OGLE-LMC-ECL-01866
18.496
18.496
0.000
OGLE-LMC-ECL-03160
18.511
18.505
-0.006
OGLE-LMC-ECL-15260
18.500
18.509
0.009
Supplementary Table 13. Comparison of the distance moduli of our systems as
determined by fitting limb darkening coefficients (fit), and by using theoretical values
of the limb darkening coefficients (fix). The average of the differences listed in the
third column is -0.00075, which clearly shows that the way the limb darkening is
chosen for our systems has a negligible effect on the final distance determination
presented in our letter.
References
29. Udalski, A. 2003, Acta Astron., 53, 291
30. Hawarden, T. G., Leggett, S. K., Letawsky, M. B., Ballantyne, D. R., & Casali,
M. M. 2001, MNRAS, 325, 563
31. Pietrzynski, G., et al. 2006, ApJ, 642, 216
32. Bernstein, R., Shectman, S. A., Gunnels, S. M., Mochnacki, S. & Athey, A. 2003,
SPIE, 4841, 1694
33. Mayor, M. et al. 2003, The Messenger, 114, 20
34. Kelson, D. D. 2003, PASP, 115, 688
35. Rucinski, S.M. 1992, AJ, 104, 1968

19
36. Mazeh, T., Zucker, S. 1994, ApJ, 420, 806
37. Coelho, P., Barbuy, Melendez, J., Sciavon, R.P., & Castilho, B.V. 2005, A&A, 443,
735
38. di Benedetto, G. P. 1998, A&A, 339, 858
39. Ramirez, I., & Melendez, J. 2005, ApJ, 626, 465
40. Casagrande, L., Ramırez, I., Melendez, J., Bessell, M., &
Asplund, M. 2010, A&A, 512, 54
41. Worthey, G. & Lee, H. 2011, ApJS, 193, 1
42. Gonzales, J.F., Levato, H., 2005, A&A, 448, 283
43. Subramanian, A., Subramanian, S., ApJL, 703, L37
44. Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525
45. Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245
46. Larsen, S.S., Clausen, J.V., Storm, J., 2000, A&A, 364, 455
47. Pejcha, Ondřej; Stanek, K. Z., ApJ, 704, 1730