Content uploaded by David Zurek
Author content
All content in this area was uploaded by David Zurek on Feb 11, 2014
Content may be subject to copyright.
The Astrophysical Journal, 710:332–345, 2010 February 10 doi:10.1088/0004-637X/710/1/332
C
2010. The American Astronomical Society. All rights reserved. Printed in the U.S.A.
A FAR-ULTRAVIOLET SURVEY OF M80: X-RAY SOURCE COUNTERPARTS, STRANGE BLUE
STRAGGLERS, AND THE RECOVERY OF NOVA T SCO
∗
Andrea Dieball
1
,KnoxS.Long
2
, Christian Knigge
1
, Grace S. Thomson
1
, and David R. Zurek
3
1
Department of Physics and Astronomy, University Southampton, SO17 1BJ, UK
2
Space Telescope Science Institute, Baltimore, MD 21218, USA
3
Department of Astrophysics, American Museum of Natural History, New York, NY 10024, USA
Received 2009 May 27; accepted 2009 December 16; published 2010 January 18
ABSTRACT
Using the Advanced Camera for Surveys on Hubble Space Telescope, we have surveyed the far-ultraviolet (FUV)
and near-ultraviolet (NUV) populations in the core region of M80. The color–magnitude diagram (CMD) reveals
large numbers of blue and extreme horizontal branch stars and blue stragglers, as well as ≈60 objects lying in the
region of the CMD where accreting and detached white dwarf binaries are expected. Overall, the blue straggler stars
are the most centrally concentrated population, with their radial distribution suggesting a typical blue straggler mass
of about 1.2 M
. However, counterintuitively, the faint blue stragglers are significantly more centrally concentrated
than the bright ones and a Kolmogorov–Smirnov test suggest only a 3.5% probability that both faint and bright
blue stragglers are drawn from the same distribution. This may suggest that (some) blue stragglers get a kick
during their formation. We have also been able to identify the majority of the known X-ray sources in the core
with FUV bright stars. One of these FUV sources is a likely dwarf nova that was in eruption at the time of the
FUV observations. This object is located at a position consistent with Nova 1860 AD, or T Scorpii. Based on its
position, X-ray and UV characteristics, this system is almost certainly the source of the nova explosion. The radial
distribution of the X-ray sources and of the cataclysmic variable candidates in our sample suggest masses >1 M
.
Key words: binaries: close – globular clusters: individual (M80) – novae, cataclysmic variables – stars: individual
(T Scorpii) – ultraviolet: stars
Online-only material: machine-readable table
1. INTRODUCTION
Far-UV (FUV) observations are an ideal tool to study the
exotic stellar populations that reside in globular clusters (GCs),
such as blue stragglers (BSs), white dwarfs (WDs), cataclysmic
variables (CVs, binaries containing an accreting WD), low-
mass X-ray binaries (LMXBs, binaries containing an accreting
neutron star or black hole), blue and extreme horizontal branch
(BHB and EHB) stars, and blue hook (BHk) stars. Identifying
these exotica in visible light can be extremely difficult, partly
due to the severe crowding of optical images (especially of the
cores of GCs) which are dominated by main sequence (MS)
stars and red giants (RGs), and partly because most exotica are
optically faint. However, all of these exotic populations tend to
be hotter than other cluster members and emit much of their
radiation in the FUV. “Ordinary” cluster stars (MS stars and
RGs) are cooler and considerably fainter at wavelengths less
than 2000 Å. As a result, crowding is generally not a problem
in the FUV. Thus, deep FUV-imaging at high spatial resolution
with Hubble Space Telescope (HST) is an excellent way to detect
and study exotic stellar species.
So far, deep FUV studies have been carried out for only three
clusters: 47 Tuc (Knigge et al. 2002, 2003, 2008), NGC 2808
(Brown et al. 2001; Dieball et al. 2005a; Servillat et al. 2008),
and M15 (Dieball et al. 2005b, 2007). In 47 Tuc, Knigge
et al. (2002) found FUV counterparts for the four Chandra
CV candidates (Grindlay et al. 2001) known at that time
within the FUV field of view. All of these were found to be
∗
Based on observations with the NASA/ESA Hubble Space Telescope,
obtained at the Space Telescope Science Institute, which is operated by the
Association of Universities for Research in Astronomy, Inc. under NASA
contract No. NAS5-26555.
variable FUV excess sources and three of them were later
spectroscopically confirmed as CVs (Knigge et al. 2008).
4
In
NGC 2808, Brown et al. (2001) used FUV and NUV imaging
to uncover a population of subluminous hot horizontal branch
(HB) stars, the BHk stars, and suggested that this population is
the result of a late helium-core flash on the WD cooling curve.
Dieball et al. (2005a) re-analyzed the HST data on NGC 2808
and found numerous BSs, CV candidates, and hot young WDs.
Servillat et al. (2008) found 8 FUV counterparts to the Chandra
X-ray sources in the core of NGC 2808; two of those are close
matches and confirm their CV nature. In M15, Dieball et al.
(2005b) found the FUV counterpart of the LMXB M15 X-2
(White & Angelini 2001), and clearly detected an orbital period
of 22.6 minutes, thus confirming M15 X-2 as an ultracompact
X-ray binary (UCXB), only the third in a GC at that time.
Since then, Zurek et al. (2009) have confirmed the UCXB status
of another X-ray source in the GC NGC 1851, also based on
FUV observations. Dieball et al. (2007) constructed a deep
FUV – NUV color–magnitude diagram (CMD) for M15, which
revealed large numbers of CV and WD candidates, a well-
defined BS and HB sequence, and 41 variable FUV sources,
among them RR Lyrae, Cepheids, SX Phoenicis stars, CVs, and
the well-known LMXB AC 211.
Here we present the results of FUV and NUV imaging
observations with the Advanced Camera for Surveys (ACS)
with the HST of the GC M80. This cluster is one of the densest
in the Galaxy and has a metallicity of [Fe/H] =−1.7dex
(Brocato et al. 1998; Alcaino et al. 1998; Cavallo et al. 2004),
a distance of 10 kpc, and a reddening of E
B−V
= 0.18 mag
4
The fourth likely CV was located outside the field of view of the
spectroscopic observations.
332
No. 1, 2010 A FAR-UV SURVEY OF M80 333
Figure 1. Combined and geometrically corrected master image of all FUV
SBC/F165LP exposures taken from M80’s core region. North is up and east to
the left. The field of view is 35
× 31
. The image is displayed on a logarithmic
intensity scale in order to bring out the fainter sources. The positions of the
X-ray sources found by Heinke et al. (2003) are marked with their 3σ error
circles.
(Harris 1996). Despite being a very dense and compact cluster
(r
core
= 9
corresponding to 0.44 pc at 10 kpc, r
halfmass
= 39
corresponding to 1.89 pc; Harris 1996), M80 is not thought to be
a core-collapsed cluster. Ferraro et al. (1999, 2003) found a large
and centrally concentrated population of BSs, and suggested
that M80 is in a state in which core-collapse is delayed by the
production of an extraordinarily large population of collisional
BSs. Only a few variable sources are known in M80 (Wehlau
et al. 1990; Clement & Walker 1991; Clement et al. 2001). Based
on the periods of the six RR Lyrae known in this cluster, M80
is classified as Oosterhoff type II (Oosterhoff 1939).
M80 is also famous for its historic classical nova T Scorpii,
which was discovered in 1860 by Auwers when it outshone
the entire cluster (Luther 1860; Pogson 1860). Shara & Drissen
(1995)foundaverybluestar5
from the cluster center and
within 1
of their estimated position for the nova, and thus
identified it as the (now quiescent) counterpart of the nova.
Our observations, discussed in Section 4, suggest a different
candidate. Apart from the nova, two new erupting dwarf novae
(DNe) were identified by Shara & Drissen (1995). Based on
50 ks of Chandra observations, Heinke et al. (2003) found 19
X-ray sources within the cluster’s halfmass radius to a limiting
L
0.5−2.5keV
≈ 7 × 10
30
erg s
−1
. They suggested that two of
those were quiescent LMXBs and five others were CVs based on
their X-ray hardness ratio, and that the brightest source detected
might be the X-ray counterpart to the classical nova T Sco.
Our report is structured as follows. In Section 2, we describe
the observations and the data reduction. In Section 3,the
analysis of the FUV – NUV CMD is presented. In Section 4,
we describe our comparison of X-ray to FUV locations and
our identification of the quiescent nova with an object which
appears to have been undergoing a dwarf nova outburst at the
time of our FUV observations. In Section 5, we present the radial
distribution of the various populations and compare them to the
Figure 2. Same as Figure 1, but for the NUV HRC/F250W. Note that the HRC
field of view is somewhat smaller than the SBC field of view with 29
× 25
.
X-ray source distribution. Finally, we summarize our results in
Section 6.
2. OBSERVATIONS AND THE CREATION OF THE
CATALOG
The observations of the core of M80 were carried out with
the ACS on board HST using the FUV F165lp filter in the Solar
Blind Channel (SBC) and the NUV F250W filter in the High
Resolution Channel (HRC) and were made at a single pointing
position. The SBC has a field of view of 35
× 31
with a
pixel size of 0.
034 × 0.
030, whereas the HRC field of view
is slightly smaller with 29
× 26
and a spatial resolution of
0.
028 × 0.
025 pixels. Thus, the observations cover only the
central portions (approximately 1.5 core radii if we adopt a core
radius of 9
;Harris1996) of M80. The FUV observation was
carried out during four consecutive orbits in 2004 September.
To facilitate searches for time variability, the FUV observation
(data set j8y501) comprised 32individual exposures of durations
ranging from 310 to 323 s. The total exposure time was 10232 s.
The NUV observation (data set j8y504) comprised a single orbit
in 2004 October, and resulted in a total exposure of 2384 s,
split into eight individual exposures of 298 s. Dithers were not
utilized to simplify searches for time variability.
Beginning with the data products delivered by STScI, we
created master images of the FUV and NUV data using
multidrizzle running under PyRAF.Themultidrizzle rou-
tines correct the field distortion that exists in the individual flat-
fielded images delivered as part of the standard data products
and combine them into master images for the FUV and NUV.
The combined and geometrically corrected output master im-
ages have a pixel scale of 0.
025 pixel
−1
and are normalized to
1 s exposure time.
The FUV and NUV master images are shown in Figures 1
and 2. As expected, the FUV image is considerably less
crowded than the NUV image. Both images show significant
concentrations of sources toward the cluster core.
334 DIEBALL ET AL. Vol. 710
2.1. FUV and NUV Source Detection
We used daofind (Stetson 1991) running under IRAF
5
to
create initial source lists for the FUV and NUV master images.
We then checked and updated these lists, adding a few faint
stars that were missed by daofind and removing obvious false
detections (e.g., multiple detections of very bright sources, noise
peaks near image edges, etc.). For a detailed description of the
source finding procedure, see Dieball et al. (2007). The resulting
final catalogs contained 3168 FUV and 9875 NUV sources.
2.2. Matching FUV, NUV, and Optical Sources
In order to match the FUV and NUV catalogs, we created a
reference list containing the pixel coordinates of 92 stars that
are clearly visible and well within the fields of both images.
We used the geomap and geoxytran task running under IRAF
to determine the geometrical transformation between the two
catalogs. We allowed for x and y shifts, rotation, and scale
changes in the coordinate transformation. The residual errors
in the transformation were quite small, less than 0.2 pixels
(<10 mas; rms) for the 92 stars.
The FUV field is slightly larger than the NUV field and 2574
FUV sources are located within the NUV field of view. Af-
ter some testing, we adopted a maximum matching tolerance of
2.5 pixels between the FUV and NUV source positions, resulting
in 2345 matches (91% of the possible FUV sources). Follow-
ing the procedure described by Knigge et al. (2002), which is
based both on the number of sources which are matched and
those which are not matched, we can expect 45 (≈ 1.9%) false
matches among these 2345 pairs (but note that this estimate does
not account for the increased source concentration towards the
core).
We also used the Piotto et al. (2002) catalog of M80 to search
for optical counterparts to our FUV sources. The optical data
were obtained using the WFPC2 in 1996, with the PC centered
on the cluster center. As a first step, our FUV image coordinate
system had to be transformed to the PC image system. For that
purpose, we used 31 HB stars as reference objects that could be
easily identified in both the PC F555W and the SBC F165LP
master image. (Note that five out of these 31 stars are not located
within the somewhat smaller field of view of the HRC F250W
exposures.) We allowed for a maximum matching tolerance of
1.1 PC pixels (corresponding to ≈2 pixels on our FUV or NUV
master images) and found a total of 1418 optical matches to
the FUV sources; out of these 1268 are inside the NUV field of
view. We can expect ≈ 40 to be false matches.
2.3. Improving the Absolute Astrometry
Even though multidrizzle corrects the field distortion
of our images, it does not improve their absolute astrometric
accuracy. The world coordinate system (WCS) of the images
provided with the standard data products is based on the original
guide star catalog (GSC1), whose absolute positions are often
only accurate to 1
–2
. This makes matching to external (e.g.,
X-ray) catalogs difficult.
TheusualwaytoimprovetheastrometryinHST images is
to locate one or more stars in an image whose positions are
accurately known in a Tycho-based system, and to update the
astrometric solution of the image. However, because the core of
5
IRAF (Image Reduction and Analysis Facility) is distributed by the
National Astronomy and Optical Observatories, which are operated by AURA,
Inc., under cooperative agreement with the National Science Foundation.
M80 is so crowded and because the SBC covers such a small
region of the sky, we were unable to find appropriate stars in the
master images. We therefore adopted a bootstrap approach be-
ginning with an ACS Wide Field Camera (WFC) F435W image
of M80 (namely HST_10573_03_ACS_WFC_F435W_sci.fits)
obtained from the Hubble Legacy Archive. The WFC has a
field of view of 202
× 202
, and the image covers not only
the core of M80, but also regions around the core where the
density of stars is considerably lower. As a result, we were able
to locate 16 stars from the Second US Naval Observatory CCD
Astrograph Catalog (UCAC2; Zacharias et al. 2004) in the WFC
image. The UCAC2 catalog is tied to the Tycho system and has
an absolute astrometric error of ≈70 milliarcseconds (mas) for
stars brighter than R of 16 mag. Based on the positions of the
UCAC2 stars in this field, we updated the astrometric solution
for the ACS WFC, updating the boresight for the ACS WFC
image by approximately 1.
2. The rms error between positions
of the stars in our UCAC2 sample, which were scattered fairly
uniformly around the core of M80 in the WFC image, was
approximately 0.
2.
We then located 16 (non-saturated) stars that could be easily
identified in both the NUV HRC and the WFC image, and used
their positions to remove the offset and distortion between the
NUV HRC and the WFC image. In doing this, we allowed
for image offsets, rotation, and linear scale changes. The same
procedure and the same stars were used to correct the WCS
of the FUV SBC image. The rms error between the positions
of the stars in the NUV image and the WFC image following
this correction was 12 mas. Unless otherwise noted, all of the
positional information we discuss here has been obtained from
master images with the corrected WCSs. We conservatively
estimate our overall error to be less than 0.
2.
A catalog listing of all our FUV objects is available in the
online version of this journal. For reference, we list only 20
entries in Table 1.
2.4. The Cluster Center
Several estimates for the cluster center exist in the literature
and have been compiled in Table 2. The Shawl & White (1986)
6
estimate is based on smoothed scans of ESO/SRC photographic
plates; the Ferraro et al. (1999) estimate is the average of the
stellar coordinates measured in an HST/WFPC2/PC image of
the cluster core; the Shara & Drissen (1995) estimate is based on
smoothed isophotes created from an HST/WFPC2/PC image of
the core. The Ferraro et al. (1999) and Shara & Drissen (1995)
coordinates are both in the Guide Star Catalog system, while the
Shawl & White (1986) coordinates are based on stellar positions
in the SAO catalog.
Since the Tycho system to which we have tied our data
is superior to both the SAO and the GSC systems, we have
redetermined the cluster center, using our own observations.
The NUV data set, in particular, is eminently suitable for
this purpose, since it contains a sufficiently large number
of stars, yet is not seriously affected by crowding. We thus
estimate the position of the cluster center by maximizing the
number of NUV sources contained in a circular region of radius
r
lim
when the center of this region is varied. For our final
estimate, we adopted r
lim
= 150 pixels (= 3.
75), but other
reasonable choices yield consistent results. The uncertainties on
our center coordinates were estimated by a simple bootstrapping
6
Their estimate for the cluster center was adopted by Djorgovski & Meylan
(1993) and Harris (1996).
No. 1, 2010 A FAR-UV SURVEY OF M80 335
Tab le 1
Catalog of All Sources in Our FUV Field of View
1 2 3 4 5 6 7 8 9 10 11 12 13 14
ID
cat
ID
FUV
R.A. Decl. x
FUV
y
FUV
FUV ΔFUV NUV ΔNUV ID
Piotto
BV Comments
(hh:mm:ss) (deg:mm:ss) (pixels) (pixels) (mag) (mag) (mag) (mag) (mag) (mag)
100 702 16:17:01.593 −22:58:43.05 1384.857 329.680 23.771 0.226 ··· ··· ··· ··· ··· no NUV, outside PC
101 1972 16:17:01.598 −22:58:34.98 1381.465 652.124 24.322 0.306 ··· ··· ··· ··· ··· outside HRC
102 1895 16:17:01.599 −22:58:35.45 1380.791 633.057 23.169 0.142 ··· ··· 2548 20.553 19.674 outside HRC
103 2818 16:17:01.603 −22:58:29.23 1378.636 881.567 17.614 0.011 ··· ··· ··· ··· ··· outside HRC
104 382 16:17:01.603 −22:58:45.19 1379.270 244.316 23.155 0.139 ··· ··· ··· ··· ··· outside HRC, outside PC
105 2932 16:17:01.607 −22:58:28.50 1376.075 910.782 24.201 0.328 ··· ··· ··· ··· ··· outside HRC
106 1400 16:17:01.607 −22:58:38.54 1376.704 509.832 23.023 0.129 21.123 0.030 2157 20.765 20.018 MS
107 1811 16:17:01.608 −22:58:35.97 1376.083 612.429 24.293 0.274 22.038 0.135 ··· ··· ··· MS/RG clump
108 1125 16:17:01.608 −22:58:40.34 1376.593 437.772 16.385 0.006 17.662 0.003 ··· ··· ··· EHB, outside PC
109 430 16:17:01.608 −22:58:44.88 1376.571 256.750 22.817 0.118 ··· ··· ··· ··· ··· outside HRC, outside PC
110 2334 16:17:01.610 −22:58:32.60 1374.507 747.079 22.896 0.234 ··· ··· ··· ··· ··· outside HRC
111 1461 16:17:01.612 −22:58:38.14 1374.040 525.815 24.754 0.333 22.035 0.057 ··· ··· ··· MS/RG clump
112 1639 16:17:01.612 −22:58:37.07 1374.187 568.424 24.108 0.204 ··· ··· ··· ··· ··· no NUV
113 3025 16:17:01.613 −22:58:27.88 1372.837 935.394 23.976 0.227 ··· ··· 3213 21.048 20.070 outside HRC
114 2083 16:17:01.615 −22:58:34.26 1372.089 680.526 24.004 0.250 ··· ··· 2668 21.693 20.476 outside HRC
115 887 16:17:01.615 −22:58:41.80 1372.850 379.650 22.505 0.108 18.622 0.006 ··· ··· ··· MS/RG clump, outside PC
116 1261 16:17:01.620 −22:58:39.48 1369.875 472.383 22.852 0.132 20.557 0.018 ··· ··· ··· MS/RG clump
117 1591 16:17:01.621 −22:58:37.34 1369.040 557.550 23.209 0.142 21.172 0.024 ··· ··· ··· MS/RG clump
118 354 16:17:01.623 −22:58:45.35 1368.608 238.054 23.411 0.164 ··· ··· ··· ··· ··· outside HRC, outside PC
119 1767 16:17:01.625 −22:58:36.24 1366.857 601.537 23.346 0.155 20.354 0.014 ··· ··· ··· MS/RG clump
120 947 16:17:01.626 −22:58:41.42 1366.732 394.774 24.767 0.413
··· ··· ··· ··· ··· No NUV, outside PC
Notes. The first column is the line number in the catalog, followed by the FUV ID number in Column 2. Columns 3–6 give the source position in R.A. and decl. and
image pixel coordinates. Columns 7–10 give the FUV and NUV magnitudes and the corresponding photometric errors as derived from daophot. Column 11 gives
the ID number of the optical counterpart taken from Piotto et al. (2002), followed by the optical magnitudes in Columns 12 and 13. The final Column 14 includes the
source type according to its position in the FUV – NUV and optical CMD and further comments. Only 20 entries are listed.
(This table is available in its entirety in a machine-readable form in the online journal. A portion is shown here for guidance regarding its form and content.)
Tab le 2
Estimates of the Cluster Center
R.A. Decl. Reference
16
h
17
m
02.
s
432 −22
◦
58
34.
62 This paper
16
h
17
m
02.
s
29 −22
◦
58
32.
38 Ferraro et al. (1999)
16
h
17
m
02.
s
48 −22
◦
58
33.
8 Shara & Drissen (1995)
16
h
17
m
02.
s
51 −22
◦
58
30.4
Shawl & White (1986)
method. More specifically, we created 1000 fake NUV catalogs
by sampling with replacement from the actual catalog. We
then estimated the center for each of the bootstrapped mock
catalogs in the same way as for the real data. The standard
deviation of the mock center estimates is then adopted as
the error. In this way, we determined the cluster center at
x
F 250W
= 789± 13 and y
F 250W
= 533± 18, which corresponds
to α = 16
h
17
m
02.
s
432 ± 0.
325, δ =−22
◦
58
34.
62 ± 0.
45 in
our Tycho-based WCS (see Section 2.3).
2.5. Aperture Photometry
Photometry was performed on the combined and geometri-
cally corrected FUV and NUV master images using daophot
(Stetson 1991) running under IRAF. Because of the high den-
sity of objects in the core region of M80 (especially in the
NUV image), we used a small aperture radius of 3 pixels, and a
small sky annulus of 5–7 pixels. We also allowed for a Gaussian
recentering of the input coordinates. For the FUV data, we de-
termined corrections for the finite aperture size and the source
flux contained in the sky annulus from a few bright and isolated
stars in the master image. The procedure is described in more
detail in Dieball et al. (2007). For the NUV data, we use the
same method to determine the sky correction, but adopted the
aperture corrections published by Sirianni et al. (2005). Note
that our FUV aperture correction only reaches out to 60 (SBC)
pixels, since larger apertures are invariably affected by bright
neighbors. However, our curve of growth analysis for the bright,
isolated stars in the FUV image suggests that the additional aper-
ture correction from 60 pixels to infinity is small. By contrast,
Sirianni et al. (2005) give aperture corrections to a maximum
aperture of 0.
5 and suggest an additional correction to infinity
of 0.132 mag that should be added to the derived STMAG in
HRC/F250W. Since we do apply this infinite radius correction
to the NUV data, there could be a slight systematic red bias
in our FUV – NUV colors. All magnitudes are given on the
STMAG system, where
STMAG =−2.5 × log
10
(countrate × PHOTFLAM
× apcorr × skycorr) + ZPT + addcorr.
The correction and conversion factors we used to convert count
rates into fluxes and STMAGs are listed in Table 3.
We estimate the overall completeness limits in our catalog
to FUV ≈ 23 mag and NUV ≈ 21.5 mag. However, since the
detection of sources in the broad point-spread function (PSF)
wings of the bright sources in the FUV is extremely difficult,
the completeness limit is not strictly uniform and considerably
lower near bright FUV sources.
3. THE FUV – NUV CMD
The FUV – NUV CMD of the core region of M80 is shown
in Figures 3 and 4 (left diagrams). For orientation purposes, we
include a theoretical zero-age MS (ZAMS; plotted in blue in
336 DIEBALL ET AL. Vol. 710
Figure 3. Left panel: FUV – NUV CMD of the core region of M80. For orientation purposes, we include a theoretical WD and He WD cooling sequence (violet lines),
a zero-age main sequence (ZAMS, blue line), and a zero-age HB track (ZAHB, cyan line). BHB stars are plotted in cyan, EHB stars in green, BSs in blue, gap sources
(which include CV candidates) in magenta, WD candidates in violet, and AGB stars in red. The FUV bright sources, which are likely AGB manch
´
e stars, are plotted
in magenta. The remaining sources are MS stars and RGs. FUV sources with optical counterparts are plotted in a darker shade of the same color (except for MS stars
and RGs). Right panel: optical CMD of M80. The data were taken from Piotto et al. (2002); only the PC data are plotted. The counterparts to the FUV sources are
plotted in the same color as in the left diagram. See the text for details.
Tab le 3
Conversion and Correction Factors
Camera/Filter PHOTFLAM ZPT ee skycorr addcorr
(erg cm
2
Å
−1
counts
−1
) (mag) (mag)
SBC/F165LP 1.3596913E-16 21.1 0.47 ± 0.02 1.029 ± 0.005 ···
HRC/F250W 4.7564122E-18 21.1 0.655 ± 0.006 1.017 ± 0.003 −0.132 ± 0.002
Notes. The conversion factor PHOTFLAM (Column 2) is needed to convert count rates into fluxes. The zero point ZPT is given in
Column 3, followed by the encircled energy fraction ee = 1/apcorr within a 3 pixel aperture radius (Column 4), and the sky correction
from 5–7 to 50–60 pixel sky annulus (Column 5). The final column gives the additional magnitude correction from a 0.
5 aperture to
infinity, as suggested by Sirianni et al. (2005). See the text for details.
Figure 3), a WD (solid violet line in Figure 3), and a Helium
white dwarf (He WD) cooling sequence (dashed violet line).
For all our synthetic tracks we adopted a distance of 10 kpc,
a reddening of E
B−V
= 0.18 mag and a cluster metallicity of
[Fe/H] −1.7 (Harris 1996). For details on these tracks, see
Dieball et al. (2005a). We also plot a zero-age HB (ZAHB; cyan
line) which was constructed based on the α enhanced BaSTI
ZAHB model for [Fe/H] = 1.62 dex and a mass loss parameter
η = 0.4 (e.g., Cordier et al. 2007). We then used synphot within
IRAF to calculate the corresponding FUV and NUV STMAGs.
As can be seen in Figure 3, the ZAHB and ZAMS appear to
be somewhat brighter and/or redder. Increasing the distance
makes the ZAHB and ZAMS fainter, whereas decreasing the
reddening makes the tracks brighter and bluer. We found that
using a somewhat larger distanceof 11.5 kpc and slightly smaller
reddening of E
B−V
= 0.17 mag for the ZAHB and ZAMS,
plotted as dashed tracks in Figure 3, gives a better fit to our data.
However, we point out that the synthetic tracks are plotted for
orientation purposes, and—given the difficulties in calibrating
the UV data—we do not aim to (re)determine distance and
reddening from our FUV – NUV CMD.
The FUV – NUV CMD clearly contains various stellar
populations, including WD candidates (violet data points in
Figure 3, left diagram), BSs (blue data points), HB stars (plotted
in green and cyan), and asymptotic giant branch (AGB) stars
(red) that are located at the faint and red end of the ZAHB.
The CMD also contains a group of objects between the WD
cooling sequence and the ZAMS. This is the expected location of
WD–MS binaries. We call these objects “gap sources” because
the CMD alone does not allow us to distinguish between mass
exchanging binaries (CVs) and non-interacting WD binaries
(see, e.g., Dieball et al. 2007). The remaining sources (black
data points) in the FUV – NUV CMD are MS stars and RGs. The
MSturnoffisatFUV≈ 22.5 mag and can be recognized by the
No. 1, 2010 A FAR-UV SURVEY OF M80 337
Figure 4. Same as Figure 3, but with the variable sources marked with green crosses and their FUV ID. The most likely counterparts to the X-ray sources are marked
with tilted red crosses and their X-ray source ID. The WD candidate that is located on the rim of the repeller wire shadow in the NUV image, and that might thus
actually be NUV brighter and redder, is marked with a small violet arrow. See the text for details.
sudden increase in source numbers along the ZAMS around that
magnitude. Our CMD reaches approximately 2.5 mag fainter
than the MS turnoff in the FUV.
The optical CMD is shown in the right panel of Figures 3
and 4. The optical counterparts to FUV objects are marked with
the same color as in the FUV – NUV CMD. The location of the
stellar populations in the optical CMD agrees well with what we
expect based on the FUV – NUV CMD, e.g., the counterparts
to the FUV BHB stars are on the optical BHB as well, and also
the location of the EHB, AGB, and BS stars agrees in both the
FUV – NUV and the optical CMD.
In the following subsections, we will discuss the HB, BS,
WD, and gap source populations in more detail. The selection
of stars belonging to the various populations is based on the
FUV – NUV CMD, but the numbers we give for the populations
should not be taken as exact, since the various zones in the
FUV CMD overlap and the discrimination between them can be
difficult.
3.1. The Horizontal Branch in the FUV
Our FUV – NUV CMD contains a significant population of
both EHB and BHB stars located along the bright (FUV <
19 mag) part of the ZAMS. We define stars to be EHB stars if
they have colors at least as blue as the ZAHB at T
eff
≈ 20,000 K
(e.g., Momany et al. 2004), corresponding to FUV – NUV =
−1.0 mag in our CMD. As can be seen in Figure 3, the optical
CMD shows a large gap along the vertical BHB/EHB tail, as
was already noted by Ferraro et al. (1998). This gap appears
in the FUV – NUV CMD as well and occurs around the
“knee” of the ZAMS at FUV – NUV ≈−0.7 mag. In both the
FUV – NUV and the optical CMD, another, optically fainter and
FUV bluer gap is visible. In the FUV – NUV CMD, the bluer gap
occurs in the EHB part of the ZAHB sequence approximately
at FUV – NUV ≈−1.2 mag, corresponding to T
eff
≈ 26,000 K
in our model. Sources with (photometrically) hotter T
eff
are
plotted as stars in both CMDs. We refer to these sources as
EHB2 stars. EHB stars redder than the optical faint/FUV blue
gap are denoted as EHB1 stars (see Section 5). As can be seen,
the bluer EHB2 stars agree very well with the optically fainter
EHB stars. Two of the optical counterparts to the EHB2 stars
are located in the BS region close to the RGB and might be
mismatches.
Figure 5 (top panel) shows a zoom on the HB in our
FUV – NUV CMD. For comparison, we plot the FUV – V
CMD in the bottom panel and mark the location of the gaps
visible in our data, and of the four gaps described in Ferraro
et al. (1998) according to their temperatures along the BaSTI
ZAHB. All gaps are marked with a solid arrow in both the
FUV – V and FUV – NUV CMDs and are denoted as
D
if
identified in our data, and
F
if referring to the gaps discussed
in Ferraro et al. (1998). None of the gaps visible in our CMD
appear at the same temperatures as suggested by Ferraro et al.
(1998), instead we found that the gaps appear to be somewhat
shifted. Ferraro et al. (1998) suggested a temperature of 9500 K
for their gap G0
F
. We see a gap close to this temperature position
only in the FUV – V data at 10,000 K, our gap G0
D
, but this
gap is not visible in our FUV – NUV CMD. G1
F
, at 11,000 K
in Ferraro et al. (1998), cannot be identified in our FUV –
V CMD, but we caution that the low number of BHB stars
between −0.5 > FUV – V > −1.5 prevent a secure detection.
Our FUV – NUV CMD does not indicate a gap in that area.
As already noted, a large gap is visible around FUV – NUV
≈−0.7 and FUV – V ≈−1.4 that we denote G2
D
. According
to the temperatures along the ZAHB, this G2
D
is in between
338 DIEBALL ET AL. Vol. 710
Figure 5. Top panel: FUV – NUV CMD zoomed in on the HB of M80. Bottom panel: FUV – V CMD of the HB. In both panels, the gaps G
F
suggested by Ferraro
et al. (1998) are marked with black arrows according to their temperatures. Their location does not agree with the gaps G
D
, marked with red arrows, in our CMDs,
but instead are slightly shifted. See the text for details.
Ferraro’s G2
F
and G3
F
gap. Also, in our data G3 appears at
somewhat bluer colors and higher temperatures. Table 4 gives
an overview of the temperatures assigned to the gaps in Ferraro
et al. (1998, their Figure 4) and this work, all colors refer to
the corresponding temperatures based on the BaSTI ZAHBs.
Please note that Ferraro et al. (1998) shifted their M80 FUV – V
CMD to match that of M13, and they also used a different FUV
filter (the WFPC2 F160BW). The differences between the exact
temperature location of the gaps is likely due to differences in
the FUV filter,
7
the HB models used (Ferraro et al. used the
Dorman et al. 1993 models, whereas we use the newer BaSTI
models), the parameters assumed for the HB model (distance,
reddening), and also the calibration of the data. As can be seen,
the BaSTI ZAHB fit the FUV – V data somewhat better than the
FUV – NUV data, suggesting that the NUV aperture correction
might be underestimated, rendering the FUV – NUV color too
blue. Keeping this in mind, the temperatures assigned to the
gaps (G0 and G2) match relatively well, except for the bluest
gap G3 which we found at higher T
eff
, more comparable to the
G3 gap in NGC 2808 (see Ferraro et al. 1998, their Table 2).
Based on the WFPC2 data presented by Ferraro et al. (1998),
Momany et al. (2004) suggested that M80 might contain BHk
stars. BHkstars are as blue as the EHB stars, but FUV fainter (see
Brown et al. 2001). If these stars exist in M80, our observations
show they are very rare. Our FUV – NUV CMD shows only
one star that is fainter than the hot end of the ZAHB. This could
either be a somewhat fainter but “normal” EHB star, or a BHk
7
Ferraro et al. (1998) used WFPC2 F160BW filter that has a pivot
wavelength 1522 Å and a bandwidth 449 Å, whereas our FUV data were
obtained with ACS, SBC, F165LP which has a pivot wavelength of 1758 Å
and a bandwidth of 86 Å.
Tab le 4
Horizontal Branch Gaps: Colors and Temperatures
This paper Ferraro et al. (1998)
FUV – NUV FUV – VT
eff
FUV – VT
eff
G0 – 0.059 10000 0.329 9500
G1 – – – −0.389 11000
G2 −0.721 −1.405 14500 −0.743 12000
G3 −1.178 −2.814 25500 −2.019 18000
Note. Gap colors and temperatures in our FUV – NUV and FUV – V CMDs
(Columns 2–4). Column 5 denotes the color corresponding to the temperature
(Column 6) associated with the Ferraro et al. (1998)gaps.
candidate. Unfortunately, we did not find an optical counterpart
for this source. Definite BHk stars have so far only been
found in the most massive GCs, although the number of GCs
surveyed does not allow one to conclude that lower-mass GCs
are incapable of producing BHk stars (see Dieball et al. 2009).
3.2. Blue Stragglers
Figure 3 shows a well-defined trail of stars above the MS
turnoff and around the ZAMS, with a few sources located
slightly to the red of the ZAMS. This is the expected location
of BSs in a CMD if they are produced via the collision or
coalescence of two or more lower-mass MS stars. As they
are more massive than the MS stars, we expect them to be
slightly evolved. We have marked 75 sources as likely BSs; 47
of these have optical counterparts which agree very well with
the expected location in the optical CMD. Some of the optical
counterparts are fainter than the optical MS turnoff. These are
No. 1, 2010 A FAR-UV SURVEY OF M80 339
Tab le 5
Number of HB and Gap Sources and WD and BS Candidates in M15 and M80
Name HB Gap WD BS
Gap
pc
2
WD
pc
2
BS
pc
2
F
BS
HB
Distance [Fe/H] lg(tc) lg(th) M
tot
lg(ρ
c
)
(kpc)
M15 133 48 34 75 26 18.5 40.7 0.564 10.3 −2.26 7.02 9.35 1.19 5.38
M80 117 59 31 75 34 17.3 43.2 0.641 10.0 −1.77 7.73 8.86 0.50 4.76
Notes. Number of HB and Gap sources and WD and BS candidates (Columns 2–5) detected in M15 and M80, and the number of sources
per pc
2
(Columns 6–8). Column 9 gives the BS specific frequency, Columns 10–13 give the cluster distance, metallicity, and the logarithmic
core relaxation time log (tc) and halfmass relaxation time log (th) (Harris 1996), Column 14 the cluster total mass (Gnedin et al. 2002), and
Column 15 the central density, log ρ
c
.
likely BSs with progenitors less massive than the MS turnoff
mass. BSs in GCs are thought to be formed dynamical via
stellar collisions and/or from evolution of primordial binaries.
It is still subject to discussion which is the dominant formation
mechanism, if there is one. Recent studies have found no
correlation of BSs numbers (or frequencies) with the collision
rate, arguing against dynamical formation as the dominating
channel (Piotto et al. 2004; Leigh et al. 2007), but on the other
hand the radial distribution of BSs seems to be bimodal in many
clusters, with a strong central peak, indicating that dynamics
play an important role in the formation of BSs in the cores of
these clusters (e.g., Dalessandro et al. 2008; Mapelli et al. 2006;
Lanzoni et al. 2007; Ferraro et al. 2004).
Ferraro et al. (1999) found an unusual large and centrally
concentrated fraction of BSs in M80 (305 BSs in their WFPC2
data set). They suggest that these BSs are collisionally formed,
and that M80 is currently in a transient dynamical state where
core-collapse is delayed via stellar interactions which led to the
formation of the large number of BSs (but also see Knigge et al.
2009, who suggest that most BSs—even in the core of GCs—are
descended from binary stars, although they do not rule out stellar
dynamics as a key factor in the formation and evolution of the
parent binaries). More recently, Ferraro et al. (2003) compared
six GCs (M3, M80, M10, M13, M92, and NGC 288) and found
that M80 has the largest and most concentrated population of
BSs. We compare our FUV M80 data to our FUV M15 data,
which covered a similar field of view as well, and do not find
a remarkable excess in BSs. In both M80 and M15 we find the
same number of BSs (75), but the two clusters are different in the
sense that M15 is even more massive, more concentrated, and
more metal-poor compared to M80. Scaling with the field size at
the distance of the cluster, the ratio of BS numbers in M80 and
M15 is only slightly above unity and not significantly different
from the ratio obtained for WDs, for example (see Table 5). This,
at first glance, argues against an anomalous enhancement of BS
numbers in M80, at least compared to M15. On the other hand,
if the BS specific frequencies as defined in Ferraro et al. (1999),
F
BS
HB
= N
BS
/N
HB
, are considered, M15 shows a slightly lower
BS specific frequency. This agrees with Piotto et al. (2004)who
found an anticorrelation of BS frequency and cluster mass, and
a (mild) tendency of increasing BS frequency with decreasing
central density. However, allowing for a Poisson error on the
number of stars found within the clusters, the difference between
the F
BS
HB
is 0.08 ± 0.12. Thus, although the difference in F
BS
HB
between M80 and M15 seems to reflect the trend discussed in
Piotto et al. (2004), it is still not statistically significant.
3.3. White Dwarfs
We find ≈ 30 sources that are located close to the WD and He
WD sequences in Figure 3. Most of these are likely to be WDs,
although a few could be CVs or detached WD–MS binaries (as
we will argue in Section 4). The number of expected WDs in
our field of view can be estimated from the number of HB stars
and the relative lifetimes of stars on the HB and the cooling
timescale for WDs. In total, we have 117 HB sources (30 EHB
stars, 80 BHB stars, and 7 FUV bright sources that are likely
AGB manch
´
e stars). In making this comparison, we can only
consider WD candidates above the completeness limits in both
FUV and NUV images. If we therefore restrict our WD sample
to a limiting magnitude of FUV ≈ 22 mag (corresponding to
T
eff
≈ 24,000 K and a cooling age of 2 × 10
7
yr), we find 24
WD candidates in our CMD; which compares very well to the
23 WDs that are expected on the basis of the HB numbers. This
suggests that most, if not all, of our candidates are indeed WDs.
Note that one of our WDs has an unexpectedly bright optical
counterpart with V ≈ 19 mag. However, the NUV counterpart
is located at the rim of the repeller wire shadow in the NUV
image, and might thus actually be brighter. In this case, this
source would be redder, which would shift it into the BS region
in our FUV – NUV CMD. This source is marked with an arrow
in Figure 4 (left diagram).
3.4. Gap Sources—CVs and Other WD Binaries
A number of sources can be seen in the “gap” between the
WD cooling sequences and the ZAMS in Figure 3. As mentioned
earlier, we cannot distinguish between the CV candidates and
the detached WD–MS binaries, so we call these objects the “gap
sources.” How many CV candidates can we expect among the
≈60 gap sources? Detailed theoretical work was done on 47 Tuc
(Di Stefano & Rappaport 1994; Shara & Hurley 2006; Ivanova
et al. 2006), investigating various dynamical and primordial
formation channels for CVs and predicting a few 100 CVs in
this cluster. For the sake of simplicity, we adopt Di Stefano &
Rappaport’s (1994) prediction of 190 active CVs in 47 Tuc, and
scale this number with the capture rate (e.g., Heinke et al. 2003)
to M80. This yields ≈100 CVs that can be expected in M80.
According to Di Stefano & Rappaport (1994), approximately
half of the captures should take place inside the cluster core.
Given that our detection limit corresponds to a white dwarf
temperature of T
eff
≈ 24,000 K (see Section 3.3), we will only
be able to detect relatively bright, long-period CVs above the
period gap (Townsley & Bildsten 2003, their Figures 1 and
2). About 20 of these long-period CVs should exist in 47 Tuc
(Di Stefano & Rappaport 1994, their Figure 3 and Table 5).
Ivanova et al. (2006) suggested that 35–40 CVs should be
detectable in the core of 47 Tuc. Scaling these numbers to M80,
we can expect approximately 10–20 such sources in the core
of M80. Note that the NUV field of view covers ≈1.5 times
the core radius of M80. Thus, the number of objects we find
in the gap region is consistent with the number of predicted
CVs. Five of our gap sources have optical counterparts, all of
which lie blueward of the MS and below the optical MS turnoff.
340 DIEBALL ET AL. Vol. 710
These systems might have relatively massive MS companions
that dominate the optical light. On the other hand, all five of
these sources are located close to the ZAMS in the FUV CMD,
making them BS candidates as well.
3.5. Variable Sources
As noted earlier, only few variable sources are known in M80
(Wehlau et al. 1990; Clement & Walker 1991; Clement et al.
2001), and indeed the region where RR Lyrae stars are expected
is unpopulated in our FUV – NUV CMD. Our FUV observations
cover four consecutive HST orbits, comprising 32 single images,
and we have used these data to search for variable sources.
We found three sources that exhibit convincing evidence of
variability, namely source Nos. 2238, 2324, and 2817. These
sources are flagged as variable in Table 1, and are marked in
Figure 4. Source No. 2238 shows short-term variability (≈55
minutes) and is likely a SX Phoenicis star. Source No. 2324
shows long-term variability and might be a RR Lyrae or a
Cepheid. Source No. 2817 is a peculiar object that shows very
strong variability. A more detailed study on the variable sources
in M80, including their light curves, will be presented in a
forthcoming paper (G. S. Thomson et al. 2010, in preparation).
4. IDENTIFICATION OF X-RAY SOURCES
As noted earlier, M80 was observed with Chandra by Heinke
et al. (2003) who identified 19 discrete sources within the
halfmass radius of the cluster. All but four of these—CX05,
CX08, CX10, and CX19—are in the field of the FUV image.
Since the majority of the X-ray sources are expected to be CVs,
and all are thought to be binaries, identifying their counterparts
at longer wavelengths is important, and our UV images and
source catalogs provide us with an excellent opportunity to
accomplish this task.
Heinke et al. (2003) referenced the X-ray source positions
in M80 to a bright star (HD 146457) in the Tycho catalog
that was roughly 4
from the core of M80, and allowed for
an absolute position error of 2
. This error is quite large, given
the crowding of the core of M80, so we have attempted to find
a more accurate way to register the X-ray sources to our fields.
HD 146457 is not in the ACS image we used to establish an
accurate (Tycho-based) WCS for our UV images of the core
of M80. We therefore compared the positions of the 52 X-ray
sources identified by Heinke et al. (2003) outside of the core,
which presumably are mostly foreground stars and background
quasars, to the ACS WFC 435W field. Eight of these sources are
located within the region covered by the WFC image, and two
of those, J161658.3-225838 and 161659.8-225931, were near
relatively bright stars. The offset between the Heinke positions
and these two stars was approximately 0.
13 in α and 1.
17 in δ,
well within Heinke’s estimated error.
8
After correcting the Heinke et al. (2003) X-ray source
positions by these values, we compared the X-ray sources with
the FUV image of M80. It was immediately apparent that a
number of the X-ray sources had counterparts with the brighter
sources in the field. Of the15 X-ray sources within our FUV field
of view, 6 lie within 1
of a bright FUV source (FUV < 22 mag)
that has no optical counterpart. We then applied a final shift of
0.
1 to optimally align the three X-ray sources with the most
8
Heinke et al. (2003) did not attempt a similar comparison. They corrected
the X-ray positions using HD 16457 and used the Shara & Drissen (1995)
positions for the nova which were based on GSC1.
definite FUV counterparts (CX01, CX03, and CX04). (Note
that these are three of the four brightest X-ray sources, and
that CX02, the one not associated with a bright FUV source,
has a spectrum which suggests it is a quiescent LMXB). The
positions of the X-ray sources on the FUV image after these
corrections are shown in Figure 1, where the circles represent
the 3σ statistical uncertainty in X-ray position as determined by
Heinke et al. (2003). As explained in more detail below, there
are six X-ray sources with bright FUV counterparts, whose
identifications we consider secure. The final rms offset between
the X-ray and FUV positions of these six sources is only 0.
14
after alignment.
We then compared the positions of all X-ray sources to those
of objects in our FUV catalog, using the 3σ statistical uncer-
tainty in X-ray position for each source. Table 6 summarizes the
results of this comparison. For some sources, especially those
which are faint in X-rays, and hence have larger error circles,
multiple sources lie within the 3σ radius. In these cases, we
have listed all of the possible FUV counterparts in the order of
increasing distance from the nominal X-ray position. The four
Chandra sources which were outside the FUV field of view ap-
pear in the table, but only to indicate their improved positions.
4.1. CX01: The Nova T Sco
CX01 is particularly interesting since it is located near the
site of nova T Sco, which is one of only two novae known to
have occurred along the line of sight to a galactic GC. Based
on an analysis of the historical and HST WFPC2 data, Shara
& Drissen (1995) obtained two estimates of the position of the
nova, one based on the offset of the nova from the cluster center,
the other based on offsets from two nearby stars. Based on their
estimates of the nova position, they identified a blue star as the
likely post-nova system. Using the finding chart provided in
Shara & Drissen (1995), we identified this blue star as source
No. 2422 in our UV catalog.
The region containing CX01 and the site of the nova is shown
in Figure6. In order to locate the likely position of thenova in our
frames, we offset the Shara & Drissen (1995) locations for the
nova to our Tycho-based coordinate system using the difference
in theposition of the two astrometric reference stars discussed by
Shara & Drissen (1995). In our coordinate system, the historical
nova position as estimated from the offset to the cluster core is
16
h
17
m
02.
s
80 − 22
◦
58
32.
21 (J2000) and the position estimated
from the two nearby stars is 16
h
17
m
02.
s
84 − 22
◦
58
33.
21.
Shara & Drissen’s (1995) object is located close to both
positions and is indeed very blue. More specifically, it has
FUV = 19.14 ± 0.02 mag and FUV – NUV=−1.65 mag,
which places it slightly on the blue side of the WD sequence in
Figure 3; we have classified it as a hot WD, a region of the CMD
that could indeed contain CVs. However, given the proximity
of the brightest X-ray source, CX01, to the nova position, it
seems highly likely that CX01 is, in fact, associated with the
old nova CV system that produced the 1860 eruption. As shown
in Figure 6, the position of Shara & Drissen’s (1995) object
is inconsistent with that of CX01. Moreover, Shara & Drissen
(1995) had already noted that, at M
B
= +6.8, their source was
about 10 times fainter than canonical old novae.
On the other hand, CX01 has a position that is consistent
with source No. 2129, which is one of the 10 brightest objects
in our FUV catalog, at FUV = 15.44 ± 0.01 mag. Furthermore,
as shown in Figure 4, this is the bluest source in our CMD
(FUV – NUV=−3.80 mag). This is actually unphysically blue
No. 1, 2010 A FAR-UV SURVEY OF M80 341
Tab le 6
Chandra X-ray Source Comparison
1 2 3 4 5 6 7 8 9 10 11 12 13 14
ID
X
R.A. Decl. 3σ Offset ID
FUV
FUV ΔFUV NUV ΔNUV ID
Piotto
BVComments
(hh:mm:ss) (deg:mm:ss) (
)(
) (mag) (mag) (mag) (mag) (mag) (mag)
CX01 16:17:02.817 −22:58:33.92 0.22 0.02 2129 15.444 0.005 19.247 0.008 ··· ··· ··· FUVbright
CX02 16:17:02.580 −22:58:37.73 0.13 0.08 1523 23.736 0.229 21.247 0.029 ··· ··· ··· MS/RG
CX03 16:17:01.600 −22:58:29.20 0.18 0.05 2818 17.614 0.011 ··· ··· ··· ··· ··· Outside HRC
CX04 16:17:02.008 −22:58:34.28 0.23 0.04 2082 19.209 0.022 20.277 0.024 ··· ··· ··· WD
0.22 4790 22.589 0.134 18.748 0.007 2190 16.289 15.198 RG
CX05 16:17:01.711 −22:58:16.59 ··· ··· ··· ··· ··· ··· ··· ··· ··· ··· ···
CX06 16:17:03.573 −22:58:26.55 0.29 0.21 3221 23.656 0.181 21.152 0.031 ··· ··· ··· MS/RG clump
0.29 0.25 3181 23.448 0.162 21.210 0.024 ··· ··· ··· MS/RG clump
CX07 16:17:02.169 −22:58:38.52 0.27 0.12 1387 22.578 0.120 21.869 0.055 ··· ··· ··· gap
0.25 4850 22.926 0.159 20.666 0.020 ··· ··· ··· MS/RG clump
CX08 16:17:01.118 −22:58:30.58 ··· ··· ··· ··· ··· ··· ··· ··· ··· ··· ···
CX09 16:17:02.404 −22:58:33.85 0.44 0.40 2106 18.393 0.015 18.693 0.009 1823 18.510 18.343 BS
CX10 16:17:00.412 −22:58:30.12 ··· ··· ··· ··· ··· ··· ··· ··· ··· ··· ···
CX11 16:17:02.476 −22:58:39.11 0.43 0.28 1352 22.751 0.137 20.424 0.017 1134 19.980 19.278 MS
0.33 1341 22.424 0.112 20.209 0.015 1153 19.821 19.296 MS
0.36 1283 23.065 0.158 20.560 0.033 ··· ··· ··· MS/RG clump
CX12 16:17:02.570 −22:58:46.25 0.43 0.10 214 17.965 0.014 ··· ··· ··· ··· ··· Outside HRC
0.24 232 16.306 0.006 ··· ··· ··· ··· ··· Outside HRC
0.33 251 21.981 0.112 ··· ··· ··· ··· ··· Outside HRC
CX13 16:17:01.759 −22:58:30.54 0.42 0.11 2624 23.397 0.182 22.291 0.240 ··· ··· ··· gap
0.25 2605 24.091 0.264 22.384 0.078 ··· ··· ··· MS/RG clump
CX14 16:17:02.558 −22:58:31.75 0.70 0.30 2414 22.920 0.195 20.882 0.022 1871 20.555 19.802 MS
0.30 2453 22.237 0.108 19.372 0.009 1946 18.991 18.185 RG
0.32 2452 17.661 0.011 17.380 0.004 1880 16.596 16.222 BHB
0.38 2415 22.557 0.125 20.614 0.022 ··· ··· ··· MS/RG clump
0.42 2512 22.452 0.130 20.186 0.018 1978 20.135 19.291 MS
0.64 2428 22.337 0.144 20.313 0.015 1935 19.602 18.618 MS
0.67 2541 23.253 0.346 20.673 0.038 ··· ··· ··· MS/RG clump
CX15 16:17:02.104 −22:58:33.05 0.43 0.15 2269 23.664 0.232 ··· ··· ··· ··· ··· no NUV
0.17 2270 22.868 0.146 20.393 0.017 ··· ··· ··· MS/RG clump
0.17 2294 23.197 0.157 ··· ··· ··· ··· ··· no NUV
CX16 16:17:02.124 −22:58:21.05 0.70 0.05 3967 16.388 0.007 17.210 0.003 3338 18.350 18.285 BHB
0.23 4786 18.383 0.020 21.117 0.048 ··· ··· ··· WD
CX17 16:17:02.224 −22:58:34.95 0.68 0.21 1944 22.971 0.235 22.555 0.158 ··· ··· ··· gap
0.33 2005 22.324 0.157 20.989 0.057 ··· ··· ··· gap
0.45 1952 23.262 0.221 21.275 0.045 ··· ··· ··· MS/RG clump
0.47 1911 23.045 0.224 20.214 0.021 ··· ··· ··· MS/RG clump
0.47 2022 21.579 0.093 18.727 0.007 1975 15.046 13.441 RG
0.51 1849 16.799 0.008 18.082 0.004 ··· ··· ··· EHB
0.52 2050 22.752 0.209 20.713 0.026 ··· ··· ··· MS/RG clump
0.53 4791 22.211 0.116 20.005 0.022 ··· ··· ··· MS/RG clump
0.61 1918 22.741 0.155 20.124 0.014 ··· ··· ··· MS/RG clump
CX18 16:17:02.824 −22:58:37.25 0.68 0.35 1559 23.374 0.221 20.390 0.027 ··· ··· ··· MS/RG clump
0.38 1659 23.318 0.178 20.465 0.016 1013 20.127 19.329 MS
0.45 1601 22.161 0.090 18.628 0.005 999 16.279 15.166 RG
0.46 1531 22.672 0.122 20.249 0.020 923 19.973 19.245 MS
0.64 1607 20.351 0.040 19.881 0.012 ··· ··· ··· BS
CX19 16:17:03.854 −22:58:48.35 ··· ··· ··· ··· ··· ··· ··· ··· ··· ··· ···
Note. The X-ray source ID is given in the first column, followed by our revised positions for the X-ray sources in Columns 2 and 3, the 3σ statistical uncertainty
in Column 4, the angular distance from the nominal position to the FUV object in Column 5, the following Columns 6–14 are as in Table 1.
(i.e., significantly bluer than an infinite temperature blackbody,
which has FUV – NUV =−1.8 mag), suggesting that the object
must have been much brighter during the FUV observations than
during the NUV observations a month later. This is surely the
counterpart to CX01, and its variability is consistent with the
suggestion by Heinke et al. (2003) that this particular source is
a CV. The source is close to Shara & Drissen’s (1995) preferred
position for the nova, based on offsets from nearby stars, and,
since this is the type of object that would be expected to produce
a nova, is a much more viable candidate for the quiescent nova
than the blue object identified by Shara & Drissen (1995). It
is also about 1.5 mag brighter in the NUV than the candidate
described by Shara & Drissen (1995) and therefore closer to the
quiescent magnitude of other novae. Given its position, X-ray
and UV brightness and variability, this source is almost certainly
the true counterpart to T Sco. It clearly merits further study.
4.2. CX04, CX07, CX13, CX16, CX17: Cataclysmic Variables
All of the 15 Chandra sources in the FUV field of view
have at least one possible FUV counterpart among the 3000
objects in our FUV catalog if we adopt the 3σ X-ray error
342 DIEBALL ET AL. Vol. 710
Figure 6. Portion of the FUV (left), NUV (middle) and ACS WFC 435W (right) image where the nova T Sco was seen. Shara & Drissen (1995) suggested that a very
blue star, easily visible in our FUV and NUV images as object 2422, was likely to be the CV responsible for the nova. They also used the historical data to make two
estimates of the location of the nova within the cluster, one based on the location of the cluster core and another based on offsets from nearby stars. These are labeled
in blue. Our astrometry suggests that another very blue object, source No. 2129, is more likely to be the brightest X-ray source in the GC, marked with a red circle.
Given the uncertainties in locating the nova from the historical record, our object 2129 seems a better candidate as the CV which gave rise to the nova.
circle as a criterion for identifying candidate matches. Thus,
it is obvious that one cannot assume that the identifications
suggested in Table 6 are real, without considering the magnitude
of the difference in position or the nature of the object. However,
CVs are expected to be found only among the gap sources (59
objects) and the WD candidates (31 objects) in our FUV – NUV
CMD. Given that three X-ray sources (CX07, CX13, CX17) are
associated with gap sources and a further two (CX04, CX16)
with WD candidates, we regard these identifications as secure.
The FUV counterpart to CX07 is one of the sources identified
with a gap object (No. 1387 in our catalog, with FUV =
22.58 mag). This FUV object lies only 0.
12 from the best
X-ray position. There is no reason to suspect that the other
possible candidate, source no. 4850, which lies in the RG/
MS clump, would have been detected as an X-ray source. As it
happens, however, there is more information in this case. Source
No. 1387 was previously identified by Shara et al. (2005)asa
CV, which they observed to have undergone a DN outburst. (The
other DN they identified in M80 is not within our FUV field of
view.) They suggested that this object, which they called DN1,
was associated with CX17, which is nearby. Our more accurate
X-ray positions make it clear that CX07, and not CX17, is the
X-ray source associated with the dwarf nova.
4.3. Other X-ray Sources with FUV Counterparts
There are seven Chandra sources—CX02, CX06, CX09,
CX11, CX14, CX15, and CX18—that are inside both the FUV
and NUV fields of view but have no obvious counterparts. In
each case, there is at least one object within the 3σ error circle
but the counterparts are neither very bright nor in a region of
the CMD expected to be populated by CVs or UV-bright X-ray
sources. In several of the cases, there are no candidates within
the 1σ error circle, and this makes their association with the
sources listed in Table 6 less likely.
Of these sources, CX02 and CX06 are arguably the most
interesting. Both were identified by Heinke et al. (2003)onthe
basis of their hard X-ray spectra and luminosities as possible
quiescent LMXBs. Such objects also are expected to have high
values of F
X
/F
opt
. The only possible counterparts to these
objects are classified by us as in the MS/RG group, and none
of these is within the 1σ error circle of the X-ray sources. This
result is consistent with the suggestion that they are indeed
quiescent LMXBs.
Two further X-ray sources, CX03 and CX12, are located in
regions where there is no NUV coverage. There is a bright FUV
object associated with each of them, but we cannot classify the
object in our FUV – NUV CMD.
5. RADIAL DISTRIBUTIONS AND MASSES OF THE
STELLAR POPULATIONS
5.1. Radial Distributions
The radial distribution of the various stellar populations that
show up in our CMD, and also of the X-ray sources, are plotted
in Figure 7. As our CMD is limited by the NUV data, we
also present the radial distributions for sources brighter than
21.5 mag in the NUV. This selection affects only the gap sources.
We compare the BS candidates, the gap sources, and EHB and
BHB stars. WD and MS populations are not shown as both
distributions suffer from incompleteness in the FUV especially
in the core region of M80 due to the concentration of bright
stars. Such faint sources are not detectable close to the bright
stars because of the broad wings of the FUV PSF; see Section 2.
In order to assess the statistical significance of the dif-
ferences between the various stellar populations, we applied
Kolmogorov–Smirnov (KS) tests. The KS test measures the
probability that two sample populations are drawn from the
same underlying distribution. Thus the larger the probability,
the more similar the two populations, whereas small probabil-
ities signal significantly different distributions. The number of
sources in the various distributions in the full and magnitude
selected samples are given in Table 7; the results from the KS
tests are presented in Table 8.
Figure 7 shows all of the radial distributions. The BS stars
are clearly the most centrally concentrated population. A strong
concentration of the BSs was already noticed in Ferraro et al.
(1999).
In the magnitude-limited sample (Figure 7, panel (b)), X-ray
and gap sources (which include the CV candidates) and BSs
are the most concentrated populations. In fact, the KS test does
not suggest a strong difference between these three populations
(see Table 8). A strong central concentration of BSs and CVs
is to be expected for two reasons. First, a significant fraction of
No. 1, 2010 A FAR-UV SURVEY OF M80 343
(a)
BS
HB
gap
(b)
BS
EHB
BHB
gap (NUV<21)
CX
(c)
EHB1
EHB2
(d)
bright&blue BS
faint&red BS
Figure 7. Cumulative radial distributions of the various stellar populations that
show up in our FUV – NUV CMD, and of the X-ray sources. We only compare
the BS candidates, gap objects, HB stars, and X-ray sources. Panel (a) shows
the radial distribution of BSs, HB stars, and gap sources. Panel (b) shows the
BHB and EHB populations, the magnitude selected (NUV < 21 mag) gap
sources, the BSs and X-ray sources. Panel (c) compares the EHB1 and EHB2
populations. Panel (d) compares the radial distribution of the bright/blue BSs
and the faint/red BSs. Contrary to our expectation, we see the faint (red) BSs to
be stronger concentrated than the bright (blue) BSs. See the text for the details.
these objects may be formed in stellar dynamical interactions
that preferentially take place in the dense cluster core. Second,
BSs (merged MS stars) and bright CVs (composed of a WD and
near-MS star) are more massive than ordinary cluster members
and will thus sink to the core due to mass segregation.
No significant differences are found between the radial
distributions of the various HB populations.
5.2. The Peculiar Blue Straggler Population: Bright versus
Faint Blue Stragglers
Bright BSs are thought to be more massive than faint BSs
(e.g., Sills et al. 2000). They are also thought to be younger
than the faint BSs; see, e.g., Ferraro et al. (2003, their Figure 4).
Provided that the ages of all the BSs are larger than the cluster
relaxation time t
halfmass
, the bright, massive BSs should thus be
more centrally concentrated than the fainter, less massive BSs
due to mass segregation. We decided to test this hypothesis.
Since bright BSs are also bluer (and hotter) than faint BSs,
we created our bright BS sample by selecting BSs with FUV –
NUV < 0.9 mag, and a corresponding faint BS sample by
selecting BSs with FUV – NUV >0.9 mag. Nearly all of the
FUV bright and blue BSs are also optically brighter than V =
19 mag (28 out of 33 of the bright and blue BSs with optical
counterparts), and most of the FUV faint and red BSs are also
optically fainter than V = 19 mag (11 out of 14 with optical
counterparts).
Figure 7, panel (d), shows the radial distribution of the bright
and faint BSs. Surprisingly, the faint BSs (red line) are more
concentrated than the bright BS (blue line). The KS test suggests
that there is only a 3.5% probability that both the faint and the
bright BSs are drawn from the same parent distribution, see
Table 8. In order to test the sensitivity of this result to the adopted
cluster center, we carried out a Monte Carlo simulation. In each
Tab le 7
Number of Sources in the Various Populations, Both in the Full Samples and
the Magnitude Selected Sample with NUV < 21.5mag
Population All NUV < 21.5
BS 75 75
BHB 80 80
EHB 30 30
EHB1 11 11
EHB2 19 19
Gap 59 13
Tab le 8
KS Probability in % that Populations Have Similar Radial Distributions
Population All NUV < 21.5
BS–gap 0.06 95.1
BS–HB 0.20.2
BS–BHB 0.02 0.02
BS–EHB 28.528.5
BS–CX 21.721.7
Gap–HB 79.97.0
Gap–BHB 77.54.0
Gap–EHB 23.440.2
Gap–CX 0.982.8
HB–CX 0.30.3
BHB–EHB 9.19.1
BHB–CX 0.20.2
EHB–CX 4.44.4
EHB1–EHB2 49.549.5
bBS–fBS 3.53.5
iteration, we shifted the cluster center randomly in line with the
error derived in Section 2.4, computed the corresponding new
distance from the cluster center for each BS and then calculated
the corresponding KS probability for the faint/red and bright/
blue BS radial distribution. The outcome of 100,000 of these
iterations was that 63% of the iterations yielded KS probabilities
below 4.6%, i.e., better than a 2σ level of confidence. Thus,
the marginally significant difference we find between the radial
distributions of these two types of BSs is not very sensitive to
the exact choice of the cluster center.
This result is rather puzzling. A tentative explanation could
be that BSs get a kick at their formation. This could work, since
the bright BSs are younger and have shorter lifespan than the
faint BSs. Thus, if their initial kick takes BSs out to regions
where the relaxation timescale is shorter than the typical age of
faint BSs, but longer than that of bright BSs, the latter will not
have had time to sink (back) to the core. In any event, we urge
others to search for a similar effect in other GCs.
5.3. Mass Estimates
The typical mass of objects belonging to a particular stellar
population can be estimated by analyzing their radial distri-
bution. Assuming that the distributions for all masses can be
approximated by King (1966) models, we can then compare the
distributions of different populations to infer the ratio of their
typical masses. Here, we follow Heinke et al. (2003) and com-
pare our source distributions to “generalized theoretical King
models”:
S(r) =
1+
r
r
c
2
1−3q
2
dr, (1)
where q = M
X
/M
, M
is the mass of the stellar popu-
lation that defines the core radius r
c
, and M
X
is the mass
344 DIEBALL ET AL. Vol. 710
BS
HB
EHB
BHB
CX
gap
bright&blue BS
faint&red BS
(a)
(b)
(c)
(d)
Figure 8. Comparison of the source distribution with theoretical King models
for stellar populations with average masses ranging from 0.4 M
(bottom black
line in each panel) to 2 M
(top black line), in steps of 0.2 M
.BSandHB
populations are plotted in panel (a). As can be seen, the BS population agrees
well with a model of mass 1.2 M
, and the HB stars with masses around 0.6 M
.
Panel (b) shows the EHB and BHB population. EHB stars seem to be slightly
more massive (0.8 M
) than BHB stars (0.6 M
). Panel (c) shows the X-ray and
the magnitude selected gap sources. Both source populations seem to be more
massive than 1 M
. Panel (d) shows the bright vs. faint BSs. Surprisingly, the
faint BSs seem to be more massive (≈ 1.4 M
) than the bright BSs (≈ 1 M
).
See the text for details.
of the source population for which we want to find the
mass.
We adopt a core radius of r
c
= 6.
5 as determined by Ferraro
et al. (1999) from fitting a King model (1996) to their WFPC2
data, and we assume that the core radius is defined by MS
turnoff stars with M
= 0.8 M
. In order to have as much
radial coverage as possible, we have corrected the distribution
to account for the fractional area covered by the actual field of
view of the instrument as a function of radius.
The area corrected models are plotted in Figure 8, with
the radial distributions of the BS, HB, gap, and X-ray source
populations overplotted. To avoid confusion, we compare only
two source populations per panel in Figure 8. BS and HB
distributions are plotted in panel (a). The BS population agrees
well with a model of mass 1.2 M
, while the distribution of HB
stars implies masses around 0.6 M
. Both of these numbers are
reasonable and agree with an average mass estimate based on
the mass distribution along the ZAMS and the ZAHB.
In panel (b), we show the EHB and BHB population. EHB
stars seem to be more massive with ≈ 0.8 M
than BHB
stars (0.6 M
; but recall that the difference between these
distributions is not statistically significant). This is contrary to
the mass distribution along the ZAHB, which suggests that the
BHB stars span a mass range of ≈ 0.63 M
to ≈0.52 M
,
resulting in an average mass of ≈0.58 M
. Stars become less
massive toward the end of the ZAHB, so that the EHB stars, on
average, should have a mass less than ≈ 0.51 M
.
The radial distribution of X-ray sources and of the magnitude-
limited sample of gap sources are plotted in panel (c). Both
source populations seem to be more massive than 1 M
,buta
more accurate estimate is not possible based on our Figure 8.Our
result broadly agrees with Heinke et al. (2003) who suggested an
average mass of 1.2 ± 0.2 M
for the X-ray source population.
The FUV bright versus FUV faint BSs are compared in panel
(d). Based on this plot, it seems that the faint BSs are more
massive (≈ 1.4 M
) than the bright and blue BSs (≈ 1 M
). A
mass estimate based on the mass distribution along the ZAMS
suggests that the faint and red BSs cover a mass range of 0.94–
1.13 M
, resulting in an average mass of ≈ 1.04 M
, whereas
the bright and blue BSs should be more massive, spanning
1.13–1.55 M
with an average of ≈ 1.34 M
.Again,thisis
contrary to the mass estimate based on the radial distribution.
However, all of the estimates based on the radial distributions
only hold if the populations have reached thermal equilibrium.
In the “kick” scenario sketched in the previous section (to
account for the unexpected difference between bright and faint
BS distributions), the bright BSs do not satisfy this condition.
6. CONCLUSIONS
We analyzed deep FUV and NUV images of the core region
of M80. We have astrometrically corrected our master images to
the Tycho-based WCS, and identified 3168 sources in the FUV
master image, of which 2345 have counterparts in the somewhat
smaller NUV master image of M80. We have also found optical
counterparts for 1268 of the sources in our FUV – NUV CMD.
The FUV – NUV CMD shows a rich variety of stellar
populations in M80. Among the objects are 75 BS candidates,
80 BHB and 30 EHB stars, 31 WD candidates, and 59 objects
in the gap between the WD and MS. The numbers of bright
WDs (24) and of gap sources are consistent with theoretical
predictions. The FUV – NUV CMD reveals clear gaps along
the BHB and EHB (at T
eff
≈ 14,500 K and ≈ 25,500 K) which
can also be identified in the optical CMD. M80 does not appear
to have a population of blue hook stars in its core, as only one
possible BHk candidate was found.
Overall, the BS stars are the most centrally concentrated
population, with their radial distribution suggesting a typical
blue straggler mass of about 1.2 M
. Ferraro et al. (1999, 2003)
suggested thatM80 comprises an unusual large and concentrated
population of BS stars, compared to other clusters, and suggest
that M80 is currently in a transient dynamical state where core
collapse is delayed via stellar interactions that formed the large
number of BSs. We compared our FUV M80 data to our FUV
M15 data, which covered a similar field of view, and do not find
a remarkable excess in BSs. However, counterintuitively, we
found that the faint and red BSs are significantly more centrally
concentrated than the bright and blue BSs, with only a 3.5%
probability that faint and bright BSs are drawn from the same
distribution. This result is surprising. One possible explanation
could be that the bright BSs get a kick at their formation which
takes them out to regionswhere therelaxation timescale is longer
than the typical age of bright BSs but shorter than the typical
age of faint BSs. In that case, the bright BSs would not have had
time to settle toward the cluster core.
Finally, we believe we have recovered the object that was
responsible for the Nova 1860 AD, also known as T Scorpii. It
is not the UV bright object identified by Shara & Drissen (1995),
but rather a dwarf nova located at the site of the historical event,
which is today the brightest X-ray object, CX01, in M80; see
Heinke et al. (2003). This object, source No. 2129, is one of
the brightest and the bluest FUV source in our catalog. This
identification has also enabled us to clearly identify the FUV
objects associated with another 5 of the 15 X-ray sources located
in the core of M80. We found that CX04, CX07, CX13, CX16,
No. 1, 2010 A FAR-UV SURVEY OF M80 345
and CX17 are associated with gap sources and WDs. All of
these are likely CVs. Our source No. 1387 coincides with the
dwarf nova DN1 observed by Shara et al. (2005) and is the
counterpart to CX07. For seven X-ray sources (CX02, CX06,
CX09, CX11, CX14, CX15, and CX18) the FUV counterparts
are not obvious. The two remaining X-ray sources, CX03 and
CX12 are located in regions where there is no NUV coverage.
There is a bright FUV object associated with each of them, but
we cannot classify the object in our FUV – NUV CMD. The
radial distributions of the 15 X-ray sources and of the brighter
gap sources (NUV > 21.5 mag) are not statistically different
and suggest masses >1 M
.
This work was supported by NASA through grant GO-10183
from the Space Telescope Science Institute, which is operated
by AURA, Inc., under NASA contract NAS5-26555. A portion
of this work was carried out at the Kavli Institute for Theo-
retical Physics in Santa Barbara, CA, USA. This research was
supported in part by the National Science Foundation under
grant No. PHY05-51164.
REFERENCES
Alcaino, G., Liller, W., Alvarado, F., Kravtsov, V., Ipatov, A., Samus, N., &
Sminrov, O. 1998, AJ, 116, 2415
Brocato, E., Castellani, V., Scotti, G. A., Saviane, I., Piotto, G., & Ferraro, F. R.
1998, A&A, 335, 929
Brown, T. M., Sweigart, A. V., Wayne, B. L., & Hubeny, I. 2001, ApJ, 562, 368
Cavallo, R. M., Suntzeff, N. B., & Pilachowski, C. A. 2004, AJ, 127, 3411
Clement, C. M., & Walker, I. R. 1991, AJ, 101, 1352
Clement, C., et al. 2001, AJ, 122, 2587
Cordier, D., Pietrinferni, A., Cassisi, S., & Salaris, M. 2007, AJ, 133, 468
Dalessandro, E., Lanzoni, B., Ferraro, F. R., Rood, R. T., Milone, A., Piotto, G.,
& Valenti, E. 2008, ApJ, 677, 1069
Dieball, A., Knigge, C., Maccarone, T. J., Long, K., Hannikainen, D., Zurek,
D., & Shara, M. M. 2009, MNRAS, 394, L56
Dieball, A., Knigge, C., Zurek, D., Shara, M. M., & Long, K. 2005a, ApJ, 625,
156
Dieball, A., Knigge, C., Zurek, D., Shara, M. M., & Long, K. 2005b, ApJ, 634,
L105
Dieball, A., Knigge, C., Zurek, D., Shara, M. M., Long, K., Charles, P. A., &
Hannikainen, D. 2007, ApJ, 670, 379
Di Stefano, R., & Rappaport, S. 1994, ApJ, 423, 274
Djorgovski, S., & Meylan, G. 1993, in ASP Conf. Ser. 50, Structure and
Dynamics of Globlular Clusters, ed. S. G. Djorgovski & G. Meylan (San
Francisco, CA: ASP), 325
Dorman, B., Rood, R. T., & O’Connell, R. W. 1993, ApJ, 419, 596
Ferraro, F. R., Beccari, G., Rood, R. T., Bellazzini, M., Sills, A., & Sabbi, E.
2004, ApJ, 603, 127
Ferraro, F. R., Paltrinieri, B., Fusi Pecci, F., Rood, R. R., & Dorman, B.
1998, ApJ, 500, 311
Ferraro, F. R., Paltrinieri, B., Rood, R., & Dorman, B. 1999, ApJ, 522, 983
Ferraro, F. R., Sills, A., Rood, R., Paltrinieri, B., & Buonanno, R. 2003, ApJ,
588, 464
Gnedin, O. Y., Zhao, H. S., Pringle, J. E., Fall, S. M., Livio, M., & Meylan, G.
2002, ApJ, 568, 23
Grindlay, J. E., Heinke, C., Edmonds, P. D., & Murray, S. S. 2001, Science, 292,
2290
Harris, W. E. 1996, AJ, 112, 1487
Heinke, C. O., Grindlay, J. E., Edmonds, P. D., Lloyd, D. A., & Murray, S. S.
2003, ApJ, 598, 516
Ivanova, N., et al. 2006, MNRAS
, 372, 1043
King, I. R. 1966, AJ, 71, 64
Knigge, C., Dieball, A., Ma
´
ız Apell
´
aniz, J., Long, K. S., Zurek, D. R., & Shara,
M. M. 2008, ApJ, 683, 1006
Knigge, C., Leigh, N., & Sills, A. 2009, Nature, 457, 288
Knigge, C., Zurek, D. R., Shara, M. M., & Long, K. S. 2002, ApJ, 579, 752
Knigge, C.,Zurek, D. R., Shara, M. M., Long, K. S., & Gilliland, R. L. 2003, ApJ,
599, 1320
Lanzoni, B., Dalessandro, E., Ferraro, F. R., Mancini, C., Beccari, G., Rood, R.
T., Mapelli, M., & Sigurdsson, S. 2007, ApJ, 663, 267
Leigh, N., Sills, A., & Knigge, C. 2007, ApJ, 661, 210
Luther, E. 1860, Astron. Nachr., 53, 293
Mapelli, M., Sigurdsson, S., Ferraro, F. R., Colpi, M., Possenti, A., & Lanzoni,
B. 2006, MNRAS, 373, 361
Momany, Y., et al. 2004, A&A, 420, 605
Oosterhoff, P. T. 1939, The Observatory, 62, 104
Piotto, G., et al. 2002, A&A, 391, 945
Piotto, G., et al. 2004, ApJ, 604, L109
Pogson, N. R. 1860, MNRAS, 21, 32
Servillat, M., et al. 2008, A&A, 490, 641
Shara, M. M., & Drissen, L. 1995, ApJ, 448, 203
Shara, M. M., Hinkley, S., & Zurek, D. 2005, ApJ, 634, 1272
Shara, M. M., & Hurley, J. R. 2006, ApJ, 646, 464
Shawl, S. J., & White, R. E. 1986, AJ, 91, 312
Sills, A., Bailyn, C. D., Edmonds, P. D., & Gilliland, R. L. 2000, ApJ, 535,
298
Sirianni, M., et al. 2005, PASP, 117, 1049
Stetson, P. B. 1991, in 3rd ESO/ST-ECF Garching, Data Analysis Workshop,
ed. P. J. Grosbøl & R. H. Warmels (Garching: ESO), 187
Townsley, D. M., & Bildsten, L. 2003, ApJ, 596, L227
Wehlau, A., Butterworth, S., & Hogg, H. S. 1990, AJ, 99, 1159
White, N. E., & Angelini, L. 2001,
ApJ, 561, L101
Zacharias, N., Urban, S. E., Zacharias, M. I., Wycoff, G. L., Hall, D. M., Monet,
D. G., & Rafferty, T. J. 2004, AJ, 127, 3043
Zurek, D. R., Knigge, C., Maccarone, T. J., Dieball, A., & Long, K. S. 2009, ApJ,
699, 1113
