Finkelstein S.L.,Texas A&M University |
Hill G.J.,McDonald Observatory |
Hill G.J.,University of Texas at Austin |
Gebhardt K.,University of Texas at Austin |
And 9 more authors.
Astrophysical Journal | Year: 2011
We present the results of Keck/NIRSPEC spectroscopic observations of three Lyα emitting galaxies (LAEs) at z 2.3 discovered with the HETDEX pilot survey. We detect Hα, [O III], and Hβ emission from two galaxies at z= 2.29 and 2.49, designated HPS194 and HPS256, respectively, representing the first detection of multiple rest-frame optical emission lines in galaxies at high redshift selected on the basis of their Lyα emission. We find that the redshifts of the Lyα emission from these galaxies are offset redward of the systemic redshifts (derived from the Hα and [O III] emission) by Δv = 162 37 (photometric) 42 (systematic) km s-1 for HPS194 and Δv = 36 35 18 km s-1 for HPS256. An interpretation for HPS194 is that a large-scale outflow may be occurring in its interstellar medium. This outflow is likely powered by star-formation activity, as examining emission line ratios implies that neither LAE hosts an active galactic nucleus. Using the upper limits on the [N II] emission, we place meaningful constraints on the gas-phase metallicities in these two LAEs of Z< 0.17 and < 0.28 Z (1σ). Measuring the stellar masses of these objects via spectral energy distribution (SED) fitting (1010 and 6 × 108 M, respectively), we study the nature of LAEs in a mass-metallicity plane. At least one of these two LAEs appears to be more metal poor than continuum-selected star-forming galaxies at the same redshift and stellar mass, implying that objects exhibiting Lyα emission may be systematically less chemically enriched than the general galaxy population. We use the SEDs of these two galaxies to show that neglecting the contribution of the measured emission line fluxes when fitting stellar population models to the observed photometry can result in overestimates of the population age by orders of magnitude and the stellar mass by a factor of 2. This effect is particularly important at z≳ 7, where similarly strong emission lines may masquerade in the photometry as a 4000 break. © 2011. The American Astronomical Society. All rights reserved. Source
This is especially true of Miranda, the smallest and innermost of Uranus' large moons – and some would say, the oddest-looking! Like the other major Uranian moons, its orbits close to its planet's equator, is perpendicular to the Solar System's ecliptic, and therefore has an extreme seasonal cycle. Combined with one of the most extreme and varied topographies in the Solar System, this makes Miranda an understandable source of interest! Miranda was discovered on February 16th, 1948, by Gerard Kuiper using the McDonald Observatory's Otto Struve Telescope at the University of Texas in Austin. Its motion around Uranus was confirmed on March 1st of the same year, making it the first satellite of Uranus to be discovered in almost a century (the previous ones being Ariel and Umbriel, which were both discovered in 1851 by William Lassell). Consistent with the names of the other moons, Kuiper decided to the name the object "Miranda" after the character in Shakespeare's The Tempest. This continued the tradition set down by John Herschel, who suggested that all the large moons of Saturn – Ariel, Umbriel, Titania and Oberon – be named after characters from either The Tempest or Alexander Pope's The Rape of the Lock. With a mean radius of 235.8 ± 0.7 km and a mass of 6.59 ± 0.75 ×1019 kg, Miranda is 0.03697 Earths times the size of Earth and roughly 0.000011 as massive. Its modest size also makes it one of the smallest object in the Solar System to have achieved hydrostatic equilibrium, with only Saturn's moon of Mimas being smaller. Of Uranus' five larger moons, Miranda is the closest, orbiting at an average distance (semi-major axis) of 129,390 km. It has a very minor eccentricity of 0.0013 and an inclination of 4.232° to Uranus' equator. This is unusually high for a body so close to its parent planet – roughly ten times that of the other Uranian satellites. Since there are no mean-motion resonances to explain this, it has been hypothesized that the moons occasionally pass through secondary resonances. At some point, this would have led Miranda into being locked in a temporary 3:1 resonance with Umbriel, and perhaps a 5:3 resonance with Ariel as well. This resonance would have altered the moon's inclination, and also led to tidal heating in its interior (see below). With an average orbital speed of 6.66 km/s, Miranda takes 1.4 days to complete a single orbit of Uranus. Its orbital period (also 34 hours) is synchronous with its rotational period, meaning that it is tidally-locked with Uranus and maintains one face towards it at all times. Given that it orbits around Uranus' equator, which means its orbit is perpendicular to the Sun's ecliptic, Uranus goes through an extreme seasonal cycle where the northern and southern hemispheres experience 42 years of lightness and darkness at a time. Miranda's mean density (1.2 g/cm3) makes it the least dense of the Uranian moons. It also suggests that Miranda is largely composed of water ice (at least 60%), with the remainder likely consisting of silicate rock and organic compounds in the interior. The surface of Miranda is also the most diverse and extreme of all moons in the Solar System, with features that appear to be jumbled together in a haphazard fashion. This consists of huge fault canyons as deep as 20 km (12 mi), terraced layers, and the juxtaposition of old and young surfaces seemingly at random. This patchwork of broken terrain indicates that intense geological activity took place in Miranda's past, which is believed to have been driven by tidal heating during the time when it was in orbital resonance with Umbriel (and perhaps Ariel). This resonance would have increased orbital eccentricity, and along with varying tidal forces from Uranus, would have caused warming in Miranda's interior and led to resurfacing. In addition, the incomplete differentiation of the moon, whereby rock and ice were distributed more uniformly, could have led to an upwelling of lighter material in some areas, thus leading to young and older regions existing side by side. Another theory is that Miranda was shattered by a massive impact, the fragments of which reassembled to produce a fractured core. In this scenario – which some scientists believe could have happened as many as five times – the denser fragments would have sunk deep into the interior, with water ice and volatiles setting on top of them and mirroring their fractured shape. Overall, scientists recognize five types of geological features on Miranda, which includes craters, coronae (large grooved features), regiones (geological regions), rupes (scarps or canyons) and sulci (parallel grooves). Miranda's cratered regions are differentiated between younger, lightly-cratered regions and older, more-heavily cratered ones. The lightly cratered regions include ridges and valleys, which are separated from the more heavily-cratered areas by sharp boundaries of mismatched features. The largest known craters are about 30 km (20 mi) in diameter, with others lying in the range of 5 to 10 km (3 to 6 mi). Miranda has the largest known cliff in the Solar System, which is known as Verona Rupes (named after the setting of Shakespeare's Romeo and Juliet). This rupes has a drop-off of over 5 km (3.1 mi) – making it 12 times as deep as the Grand Canyon. Scientists suspect that Miranda's ridges and canyons represent extensional tilt blocks – a tectonic event where tectonic plates stretch apart, forming patterns of jagged terrain with steep drops. The most well known coronae exist in the southern hemisphere, with three giant 'racetrack'-like grooved structures that measure at least 200 km (120 mi) wide and up to 20 km (12 mi) deep. These features, named Arden, Elsinore and Inverness – all locations in Shakespeare's plays – may have formed via extensional processes at the tops of diapirs (aka. upwellings of warm ice). Other features may be due to cryovolcanic eruptions of icy magma, which would have been driven by tidal flexing and heating in the past. With an albedo of 0.32, Miranda's surface is nearly as bright as that of Ariel, the brightest of the larger Uranian moons. It's slightly darker appearance is likely due to the presence of carbonaceous material within its surface ice. Miranda's apparent magnitude makes it invisible to many amateur telescopes. As a result, virtually all known information regarding its geology and geography was obtained during the only flyby of the Uranian system, which was made by Voyager 2 in 1986. During the flyby, Miranda's southern hemisphere pointed towards the Sun (while the northern was shrouded in darkness), so only the southern hemisphere could be studied. At this time, no future missions have been planned or are under consideration. But given Miranda's "Frankenstein"-like appearance and the mysteries that still surround its history and geology, any future missions to study Uranus and its system of moons would be well-advised. Explore further: Miranda: An icy moon deformed by tidal heating
In the first section below, we describe spectroscopic data for NGC 1600 and our procedures for measuring stellar kinematics. In the second section, we describe photometric data and the surface brightness profile of NGC 1600. In the third section, we describe our stellar orbit modelling procedures. In the fourth section we compare the stellar mass-to-light ratio obtained from the dynamical models with independent constraints from a stellar population analysis. The implied distribution of stellar orbits in NGC 1600 is described in the fifth section. Finally, in the last section we present the black-hole scaling relations for core radii, determined from an alternative fit to the light profiles. We obtained high-spatial-resolution spectroscopic data from GMOS-N, an IFS on the 8-metre Gemini North Telescope. We observed the central region of NGC 1600 with GMOS-N, which provides continuous two-dimensional coverage of a 5 arcsec × 7 arcsec science field and simultaneously covers a sky field offset by 60 arcsec. Our spectra were centred on the triplet of calcium absorption lines from 8,480 Å to 8,680 Å, a well studied region frequently used for stellar kinematic measurements31, 32. We obtained nine 1,230-second exposures of NGC 1600 over three nights of queue-mode observations in November 2014. GMOS-N is seeing-limited, with spatial sampling of 0.2 arcsec on the IFS. We estimated the point-spread function (PSF) on each observing night by measuring the width of foreground stars in the acquisition images of NGC 1600. Our average PSF for GMOS-N has a full width at half-maximum (FWHM) of 0.6 arcsec. A Gaussian model of the PSF is included in our orbit superposition models. We used the image reduction and analysis facility (IRAF) software package supplied by the Gemini Observatory to flat-field and wavelength-calibrate the GMOS-N data, and to extract a one-dimensional spectrum for each IFS lenslet. We developed custom routines to construct collapsed images of the galaxy and record the position of each one-dimensional spectrum with respect to the galaxy centre. The spectra were then resampled on a two-dimensional grid and binned to a consistent signal-to-noise ratio (of about 100 per pixel) using Voronoi tessellation33. Our binning implementation imposed symmetry over four projected quadrants of the galaxy, so that the kinematic measurements could be folded into a single quadrant before orbit modelling. We obtained wide-field spectroscopic data from the Mitchell Spectrograph34 on the 2.7-metre Harlan J. Smith Telescope at the McDonald Observatory. The Mitchell Spectrograph is an optical IFS with a 107 arcsec × 107 arcsec field of view and 246 fibres, each of 4.1-arcsec diameter. The low-resolution blue setting (R ≈ 850) was used, providing wavelength coverage from 3,650 Å to 5,850 Å, including the Ca H+K region, the G-band region, Hβ, the Mgb region, and several Fe absorption features. The spectral resolution varied spatially and with wavelength, with an average of 5 Å FWHM, corresponding to a dispersion of ~1.1 Å pixel−1 and σ ≈ 100 km s−1 in the red and σ ≈ 150 km s−1 in the blue part of the spectrum (where σ is the instrumental resolution). Data reduction was performed using the Vaccine package35. We fit Mitchell spectra with the MILES library of 985 stellar spectra36 and determined the best-fit line-of-sight velocity distribution (LOSVD) for each of the 58 spatial bins6, 37. Our GMOS and Mitchell data are the first IFS observations of NGC 1600 and reveal a stellar velocity distribution that is well aligned with the galaxy’s light distribution, indicating that NGC 1600 is axisymmetric. We used a high-resolution image taken with the near-infrared camera and multi-object spectrometer (NICMOS) instrument on the HST to measure the central light distribution of NGC 1600. The observation (from General Observer Program number 7886) consisted of four dithered exposures of NGC 1600 taken with NICMOS camera 2 in the F160W bandpass. We downloaded the calibrated, combined image from the Hubble Legacy Archive. The image had a pixel scale of 0.05 arcsec per pixel and total exposure time of 460 seconds. We combined the HST observations with ground-based photometric data at large radii taken from the literature7. The NICMOS data were calibrated to the R-band of the ground-based data by minimizing the squared magnitude differences between the two surface brightness profiles in the radial region where both data sets overlap and PSF effects are negligible (r = 2–10 arcsec). Single one-dimensional profiles of the surface brightness, the ellipticity and the isophotal shape parameters38 a and a were then constructed by using the NICMOS data at r < 8 arcsec and the ground-based data at r ≥ 8 arcsec. The position angle (PA) of the isophotes is constant with radius7 (ΔPA < 2°), consistent with an axisymmetric stellar distribution. The resulting circularized surface brightness distribution of NGC 1600 is well described by a core-Sérsic function with a core radius of r = 2.15 arcsec (Extended Data Fig. 2). Over several orders of magnitude in radius, the light profiles of lower-luminosity elliptical galaxies follow a single Sérsic function characterized by the Sérsic index n, the half-light radius r and a surface brightness scale μ = μ(r ). The core-Sérsic function combines a Sérsic profile at r > r and a power-law distribution with slope at r < r . The transition is controlled by a smoothness parameter α and the surface brightness scale is μ = μ(r ). Inside r < 5 arcsec, the inward extrapolation of the outer Sérsic component overpredicts the central surface brightness of NGC 1600 by about three magnitudes. From the difference between the integrated light of the Sérsic component and the actual core-Sérsic fit, we derive a ‘light deficit’ of ΔL = 9.47 × 109L in the centre of NGC 1600. We use the isophotal model of NGC 1600 to compute the galaxy’s intrinsic luminosity density distribution. The deprojection is nonparametric39 and accounts for the observed ellipticity profile and boxy shape of NGC 1600’s isophotes7. The same technique has been used for the dynamical modelling of other galaxies12, 40. We generate dynamical models of NGC 1600 using Schwarzschild’s orbit superposition method41. Because the two-body relaxation time of stars in massive elliptical galaxies exceeds the age of the Universe, their dynamics is governed by the collisionless Boltzmann equation. Any steady-state equilibrium solution of the Boltzmann equation can be written as a sum over single-orbit distribution functions, where the phase-space density of stars along each trajectory is constant. The total number of stars on each orbit—that is, the orbital luminosity or orbital occupation number—can take arbitrary (positive) values. For the modelling, we assume that NGC 1600 is axisymmetric and that the stellar mass profile follows the observed light distribution with a constant stellar mass-to-light ratio, M /L. The models also assume a central black hole of mass M and a cored isothermal dark-matter halo with core radius r and asymptotic circular velocity v . The four parameters of the mass model are constrained by the photometrically derived luminosity density and by 1,978 LOSVD data points measured from the galaxy spectra between 0.4 arcsec and ~45 arcsec. Given some specific values for M /L, M , r and v , the Poisson equation is solved for the gravitational potential generated by the respective mass model and the luminosity densities, and the LOSVDs of ~29,000 representative stellar orbits are computed9. The orbits are convolved with the PSF of the observations and integrated over the respective areas on the sky. We use a maximum entropy technique42 to determine the orbital occupation numbers that minimize the χ2 difference among the observational data and the orbit superposition model. Thousands of different mass distributions are compared to the data by systematically varying M /L, M , r and v . We repeat the computation of the stellar orbits and the phase-space optimization independently for each model. The best-fit values and confidence intervals in M and M /L are determined by evaluating the relative likelihood43 for all models with different assumed values of M , M /L and dark halo parameters (Extended Data Fig. 3). The same modelling technique to determine the mass of the stars, the central black hole and the dark matter halo was first applied to the central galaxy of the Virgo Cluster, M87 (ref. 44). Previous models of NGC 1600, based on stellar velocity data only along the major and minor axes of the galaxy and with a lower spatial resolution, did not include all three mass components45, 46, 47, 48. We have tested that models without a dark-matter halo and/or without a central black hole cannot reproduce the full set of our new observations. Extended Data Fig. 4 shows our best-fit orbit model together with the velocity data for NGC 1600. The stellar mass-to-light ratio derived from our dynamical modelling is M /L = (4.0 ± 0.15)M /L . The stellar mass deficit in the core of NGC 1600 is thus ΔM = 3.8 × 1010M . In other core galaxies, mass deficits have been reported to range from one to ten times the mass of the central black hole15, 17, 49. For NGC 1600 we find a mass deficit, ΔM , of 2.2 × M . Results from numerical simulations of mergers of galaxies with central black holes50 suggest mass deficits of ~N × 0.5M , where N is the number of mergers. We measure a stellar population age of τ ≈ 10 Gyr, a total metallicity of [Z/H] = 0.03 and an iron abundance of [Fe/H] = −0.15 for NGC 1600, from the absorption-line strengths of hydrogen, iron and other metallicity-dependent stellar lines in the 1-kpc galaxy core. It has been reported that the fraction of low-luminosity dwarf stars in elliptical galaxies less luminous than NGC 1600 is larger than that in the Milky Way51, 52, 53. Extrapolating correlations with the galaxy velocity dispersion obtained for these smaller elliptical galaxies would yield a higher stellar mass of M /L ≈ 6.0 for NGC 1600. We find the galaxy-wide contribution of dwarf stars to the stellar mass in NGC 1600 to be consistent with observations of the Milky Way. Dynamical constraints on the fractional mass of dwarf stars depend on the assumed dark-matter halo profile54 and on the assumed shape of the stellar initial mass function55. Some nearby massive galaxies have, like NGC 1600, dwarf-star fractions consistent with that of the Milky Way56. The spectroscopic analysis of the age and chemical composition of the stars in NGC 1600 does not provide evidence for a notable change in the stellar population with radius. In the dynamical modelling we assume a constant M /L ratio throughout the galaxy. Individual massive galaxies have been reported to host extreme populations of dwarf stars at their centres57 that would not be detectable in our optical spectra. The respective dwarf stars can increase the central M /L by up to a factor of three. If the assumed constant M /L does not account for all of the stellar mass at the centre of NGC 1600, then the dynamical models may compensate for the missing stellar mass by overestimating M . Extended Data Fig. 5 shows the enclosed mass distribution of NGC 1600 over the region for which we have obtained stellar velocity data. For a constant M /L, we find the enclosed stellar mass at the smallest observed radius (r ≈ 0.2 kpc) to be 100 times smaller than M . A central increase of M /L by a factor as extreme as ten would imply an unaccounted-for extra stellar mass of 10% of M (dotted lines in Extended Data Fig. 5). Even in this unrealistic case, we would overestimate M only by its one sigma measurement error. Because there is so little stellar light in the core of NGC 1600, uncertainties in the central stellar population have a negligible effect on our black-hole mass measurement. The stars in a galaxy are collisionless and their velocity distribution can be anisotropic. We compute the intrinsic velocity dispersions of the stars along the radial and the two angular directions of a polar coordinate system—σ , σ , and σ , respectively—from the orbital occupation numbers of our best-fit dynamical model in 20 spherical shells centred on NGC 1600’s black hole. The classic measure for the anisotropy of the stellar velocities is , where the tangential velocity dispersion is the average of the motions in the two angular directions. Stellar orbits in core galaxies have been reported to be very uniform18. In NGC 1600 and similar galaxies with a flat central surface brightness, most of the stars inside the diffuse core region (r < r ) are moving along tangential directions. With increasing distance from the centre, more and more stars are found on radially elongated orbits (Extended Data Fig. 6). In the black-hole-binary model, the observed stellar motions are naturally explained as the leftover of the core scouring process (shaded regions in Extended Data Fig. 6). Central stars originally on radial trajectories are subject to interactions with the black-hole binary as they frequently pass the galaxy centre. Eventually, these stars get ejected to larger radii via gravitational slingshot. The stars that we observe today in the centres of core galaxies remained there because they moved (and still move) on tangential orbits that avoid the centre58. It has not yet been tested whether other black-hole activities—such as their feedback processes on ambient accreting gas—can produce the tight relations between black-hole mass, core radius, sphere of influence and mass deficit together with the observed orbital structure. While the core-Sérsic function describes galaxy light profiles from the core region out to large radii30, 59, the Nuker function60 has been widely used to fit the central light profiles of galaxies observed with HST. Fifteen out of the 21 core galaxies discussed in the text also have core radii measured from Nuker fits13. For the Nuker r , we obtain log (r /kpc) = (−0.18 ± 0.21) + (1.00 ± 0.09)log (r /kpc) with an intrinsic scatter of ϵ = 0.16, and log (M /M ) = (10.06 ± 0.45) + (1.25 ± 0.17)log (r /kpc) with an intrinsic scatter of ϵ = 0.31. The Nuker r values were measured along the major axis of the galaxies, while the core-Sérsic r values discussed in the text come from the galaxies’ circularized light profiles15. Galaxy core sizes have also been quantified by the cusp radius61, r —that is, the radius at which the negative logarithmic slope of the surface brightness profile equals 1/2. We obtained the cusp radii of the galaxies that are shown in Figs 3 and 4 from their core-Sérsic models. Using r as a measure of the core size, we find log (r /kpc) = (0.06 ± 0.28) + (0.94 ± 0.09)log (r /kpc) (intrinsic scatter ϵ = 0.16) and log (M /M ) = (10.37 ± 0.60) + (1.20 ± 0.19)log (r /kpc), ϵ = 0.33. Note that the core of NGC 1550 has a slope15 of γ = 0.52 ± 0.05. The scaling relations with r have been computed without including NGC 1550. The slope and the scatter in the above correlations are consistent with the results for the core-Sérsic r .
An international team of astronomers, led by Marshall C. Johnson of the University of Texas at Austin, has used the data from K2's Campaign 4, which lasted from February 7 to April 23, 2015, to search for possible transiting planets. They found two periodic transit-like signals associated with two targets designated EPIC 211089792b (K2-29b) and EPIC 210957318b (K2-30b). While K2-30b was confirmed as a "hot Jupiter" exoplanet during previous observations, K2-29b is a new addition to the long list of Kepler's confirmed extrasolar worlds. The astronomers also used three different ground-based spectrographs to conduct high-resolution spectroscopic observations of K2-29b, in order to definitely verify it as a "hot Jupiter." The Robert G. Tull Coudé spectrograph, mounted on the 2.7m Harlan J. Smith Telescope at the McDonald Observatory, Texas, allowed the scientists to obtain both reconnaissance spectroscopy and radial velocity measurements. Similar observations were conducted using the Fiber-fed Échelle spectrograph (FIES) on the 2.56m Nordic Optical Telescope at the Observatorio del Roque de los Muchachos, La Palma (Spain) and the HARPS-N spectrograph on the 3.58m Telescopio Nazionale Galileo, also at La Palma. "Here, we present K2 photometry for two late-type dwarf stars, EPIC 211089792 (K2-29) and EPIC 210957318 (K2-30), for which we identified periodic transit signals, and our follow-up spectroscopic observations. These have allowed us to confirm both transiting objects as bona fide hot Jupiters, and to measure the stellar and planetary parameters," Johnson and his colleagues wrote in a paper. Hot Jupiters are gas giant planets, similar in characteristics to the solar system's biggest planet, with orbital periods of less than 10 days. They have high surface temperatures as they orbit their parent stars very closely—between 0.015 and 0.5 AU. While the newly discovered K2-29b exoplanet has a radius that is about the same as Jupiter's, it's less massive (0.6 Jupiter masses) than our solar system's biggest planet. It has an orbital period of 3.26 days and an equilibrium temperature of approximately 800 degrees Celsius, making it a textbook example of a hot Jupiter. The planet's parent star K2-29 is slightly smaller than our sun, with 0.75 solar radii and 0.86 solar masses. The star is about 2.6 billion years old and is located some 545 light years from the Earth. The researchers also found that the orbit of K2-29b is slightly eccentric. This suggests that either the planet migrated to its current location via high-eccentricity migration, or that there is an additional planet in the system exciting the eccentricity. "In general, eccentric orbits of hot Jupiters might be generated in two different manners: Either the eccentricity is primordial, a relic of high-eccentricity migration that emplaced the planet on a short-period orbit, or the eccentricity is being excited by an external perturber," the paper reads. However, to investigate these possibilities, future observations using long-term radial velocity and transit timing variation methods are required. "These possibilities could be distinguished using long-term radial velocity and transit timing variation monitoring to detect an additional companion," the team concluded. More information: Two Hot Jupiters from K2 Campaign 4, arXiv:1601.07844 [astro-ph.EP] arxiv.org/abs/1601.07844 Abstract We confirm the planetary nature of two transiting hot Jupiters discovered by the Kepler spacecraft's K2 extended mission in its Campaign 4, using precise radial velocity measurements from FIES@NOT, HARPS-N@TNG, and the coud'e spectrograph on the McDonald Observatory 2.7 m telescope. K2-29 b (EPIC 211089792 b) transits a K1V star with a period of 3.2589263±0.0000015 days; its orbit is slightly eccentric (e=0.084+0.032−0.023). It has a radius of RP=1.000+0.071−0.067 RJ and a mass of MP=0.613+0.027−0.026 MJ. Its host star exhibits significant rotational variability, and we measure a rotation period of Prot=10.777±0.031 days. K2-30 b (EPIC 210957318 b) transits a G6V star with a period of 4.098503±0.000011 days. It has a radius of RP=1.039+0.050−0.051 RJ and a mass of MP=0.579+0.028−0.027 MJ. The star has a low metallicity for a hot Jupiter host, [Fe/H]=−0.15±0.05.
Though blue stragglers were first identified 62 years ago, astronomers have yet to converge on a solution for their odd appearance. The most popular explanation among several competing theories is that an aging star spills material onto a smaller companion star. The small star bulks up on mass to become hotter and bluer while the aging companion burns out and collapses to a white dwarf – a burned out cinder. To test this theory Gosnell's team conducted a survey of the open star cluster NGC 188 that has 21 blue stragglers. Of those, she found that seven had white dwarf companions, by identifying their ultraviolet glow that is detectable by Hubble. Of the remaining 14 of the 21 blue stragglers, a further seven show evidence of so-called mass transfer between stars in other ways. Gosnell said she believes these are older white dwarf-blue straggler binaries, and indicate two-thirds of blue stragglers form through mass transfer. "This was really great," Gosnell says. "Until now there was no concrete observational proof, only suggestive results," Gosnell said. "It's the first time we can place limits on the fraction of blue stragglers formed through mass transfer." This discovery sheds light on the physical processes responsible for changing the appearance of 25 percent of evolved stars. Gosnell's work, which closes gaps in our understanding of how stars age, is published in the current issue of the Astrophysical Journal. The problem came to light because in recent years, astronomers have been able to make a complete and accurate census of stars in a number of open star clusters, Gosnell said. "Open clusters really are the best laboratory for the study of stellar evolution," Gosnell said. "They have a simple stellar population." The stars in a cluster form at the same time and from the same materials, she explained. The cluster population studies revealed that up to a quarter of the oldest stars "are not evolving like we think they're supposed to," Gosnell said. Stars that astronomers expected to become red giants (like Aldebaran, the eye of Taurus, the bull) instead became "blue stragglers," unexpectedly bright, blue stars with a host of strange characteristics. Gosnell wanted to find out what happened to them. So she, along with Bob Mathieu at the University of Wisconsin-Madison and their collaborators, designed a study using Hubble Space Telescope's Advanced Camera for Surveys to try to differentiate between three theories of how these stars became blue stragglers. The theories included: collisions between stars in the cluster (with debris coalescing to form a blue straggler), the merger of two of the stars in a triple star system, or mass transfer between two stars in a binary pair. In a binary pair of stars, the larger star will evolve faster, Gosnell said. That star becomes a red giant. A red giant is so bloated that the outermost layers of gas on its surface are only tenuously held by the star's gravity. They can be pulled off by the gravity of the companion star. This is mass transfer. As the gas is siphoned off by the partner, the red giant is left with only its core, making it into a white dwarf. The partner—initially the less massive of the pair, but now the heavier one—becomes a blue straggler. Gosnell's method is limited by the fact that it will not detect white dwarfs that have cooled down enough so that they don't glow in UV light detectable by Hubble, she said. That means that only those white dwarfs formed in the last 250 million years (youngsters, astronomically speaking) are detectable. Knowing more about how these stars form is important because astronomers use their assumptions to model the stellar populations of distant galaxies (where the light from all the stars blends together). "You don't want to be ignoring 25 percent of the evolved stars" in those galaxies, Gosnell said. Such models are important because distant galaxies figure into many different types of cosmological studies. Right now, Gosnell said, "the models have a lot of room for improvement." "If we tweak the way models treat mass transfer, that would bring the observations and theory together," Gosnell said. "They would agree. And we can use this to inform our understanding of unresolved stellar populations"—that is, those stars in galaxies so far away that all their light is blended together. Gosnell plans to continue studying these stars using the 2.7-meter Harlan J. Smith Telescope at McDonald Observatory and its IGRINS spectrograph to constrain the number of blue stragglers that could form through mergers in triple systems. Explore further: Even stars get fat -- And 'stellar cannibalism' is the reason More information: "Implications for the Formation of Blue Straggler Stars from HST Ultraviolet Observations of NGC 188," Natalie M. Gosnell et al., 2015 Dec. 1, Astrophysical Journal, Vol. 814, No. 2 iopscience.iop.org/article/10.1088/0004-637X/814/2/163 , Arxiv: arxiv.org/abs/1510.04290, PDF: hubblesite.org/pubinfo/pdf/2015/43/pdf.pdf