Mon.Not.R.Astron.Soc.000,1–5() Printed1March2016 (MNLATEXstylefilev2.2) ˙ Large proper motion of the Thorne-Zytkow object candidate HV 2112 reveals its likely nature as foreground Galactic S-star 6 Thomas J. Maccarone 1 Department of Physics, Box 41051, Science Building, Texas Tech University, Lubbock TX, 79409-1051, USA 0 email:[email protected] 2 Selma E. de Mink b e Astronomical Institute “Anton Pannekoek”, University of Amsterdam, Postbus 94249, 1090 GE Amsterdam, The Netherlands F email:[email protected] 9 2 ] R ABSTRACT S UsingtheSouthernProperMotion(SPM)catalog,weshowthatthecandidateThorne- . Z˙ytkow object HV 2112 has a proper motion implying a space velocity of about h p 3000kms−1if the object is located at the distance of the Small Magellanic Cloud. - The proper motion is statistically different from that of the SMC at approximately o 4σ in SPM, although the result can drop to about 3σ significance by including the r UCAC4 data and considering systematic uncertainties in addition to the statistical t s ones. Assuming the measurement is robust, this proper motion is sufficient to exclude a itsproposedmembershipoftheSmallMagellanicCloudandtoargueinsteadthatitis [ likely to be a foreground star in the Milky Way halo. The smaller distance and there- 2 forelowerbrightnessargueagainstitsproposednatureasaThorne-Z˙ytkowobject(the v hypothesizedstar-likeobjectformedwhenanormalstarandaneutronstarmerge)or 5 a super-Asymptotic Giant Branch (AGB) star. Instead we propose a binary scenario 5 where this star is the companion of a former massive AGB star, which polluted the 4 object with via its stellar wind, i.e. a special case of an extrinsic S star. Our new 5 0 scenario solves two additional problems with the two existing scenarios for its nature . as Thorne-Z˙ytkow object or present-day super-AGB star. The puzzling high ratio of 1 the strength of calcium to iron absorption lines is unexpected for SMC supergiants, 0 but is fully consistent with the expectations for halo abundances. Secondly, its strong 6 1 variabilitycannowbeexplainednaturallyasamanifestationoftheMiraphenomenon. : We discuss further observational tests that could distinguish between the foreground v and SMC scenarios in advance of the improved proper motion measurements likely to i X come from Gaia. r Key words: astrometry – stars:abundances – stars: chemically peculiar – a stars:individual:HV 2112 1 INTRODUCTION Object” (TZ˙O), the hypothesized red-giant like stellar ob- jects with neutron stars inside their cores. The nature of the star HV 2112, located in the same part TZ˙Os,firstmodeledbyThorne&Z˙ytkow(1975,1977), of the sky as the Small Magellanic Cloud (SMC), has re- may form as a result of unstable mass transfer in a massive cently become the center of a vivid debate. Levesque et al. X-ray binary after the neutron star is engulfed in the enve- (2014) showed that it exhibits anomalous spectral features lopeofitscompanionstar(Taametal.1978).Alternatively, suggestingenhancedabundancesofmolybdenum,rubidium, these objects may form when the supernova kick of newly lithium, calcium and possibly also potassium when com- formed neutron star accidentally sends it towards and into paredtootherSMCsupergiants.Thehighinferredbolomet- itscompanion(Leonardetal.1994).WhetherTZ˙Osactually ricluminosityandtheenhancedabundancesofthefirstthree formisamatterofdebate.Fryeretal(1996)andChevallier elementsleadtotheintriguingsuggestionbyLevesqueetal. (1993) argued that the neutron star would accrete enough (2014) that HV 2112 could be a massive “Thorne-Z˙ytkow material to collapse to a black hole before forming a stable (cid:13)c RAS 2 Maccarone & De Mink TZ˙O. Biehle (1991, 1994) and Cannon (1993) predict that 2 PROPER MOTION MEASUREMENTS AND massive TZOs, if stable, are powered by nuclear burning THEIR IMPLICATIONS through the exotic interrupted rapid proton (irp)-process. The star HV 2112 has had its proper motion measured as Therefore they may be identified by enhanced abundances part of the Southern Proper Motion survey. This is a sys- of rare proton-rich elements, including Mo and Rb. Also tematic astrometric study with a baseline of approximately enhancements of Li have been predicted (Podsiadlowski et fourdecadescoveringnearlytheentiresouthernskydownto al. 1995). The predictions triggered several (proposed) ob- approximately V =17.5 (Girard et al. 2011). For HV 2112, serving campaigns hoping to identify massive TZOs (van thefirstepochwasobtainedin1972andthesecondin2007. Paradijs et al. 1995, Kuchner et al. 2002, Levesque et al. Thestarshowsapropermotionof2.8±2.3mas/yearinright 2014).SofarHV2112isthemostpromisingcandidate,even ascension and -9.8±2.3 mas/year in declination. thoughtheabundancesofRbandMoarenotashighasex- pectedandtheenhancementofCaandKwasnotpredicted. While the star is strongly variable, there are two rea- sons to believe that the proper motion measurements from Tout et al. (2014) discuss an alternative explanation SPM4.0arereliable.First,inbothepochs,itwasamongthe whereHV2112isa“superasymptoticgiantbranch(AGB) brightest stars in its general vicinity – within two arcmin- star”. This is the late evolutionary phase of stars with ini- utes, only one star is brighter in the second epoch V band tialmassaround8M ,massiveenoughtoignitecarbon,but CCDdata,andonlytwostarsarebrighterinthefirstepoch (cid:12) still low enough to experience thermal pulses and dredge B band photometry. Additionally, within 2’, there are only up episodes typical for AGB stars. Super AGB stars have 8starsbrighterthanB =16,sotheprobabilityoffindinga higher core masses than typical red giants, giving them lu- star within 10” which is even 10% as bright as HV 2112 is minositiesashighasredsupergiantsandTZ˙Osintheirearly about10−4.Second,withthe35yearbaseline,thetotalas- phases(Siess2010).SimilartoAGBstarstheycan,inprin- trometricshiftmeasuredisabout0.35arcseconds.Thisshift ciple, produce Mo and Rb through the slow neutron cap- issufficientlylargethatitwouldrequireextremefinetuning ture process (Karakas & Lattanzio 2014) and Li through to produce an astrometric shift due to the combination of the Cameron & Fowler (1971) mechanism. Enhancement of blending and source variability without the second object Ca or K is, however, not predicted. beingapparentintheimages,andasshown,theprobability of such a star being that nearby is of order 10−4. InbothscenariosHV2112isassumedtobememberof The UCAC4 catalog (Zacharias et al. 2013) often gives the SMC. This view is supported by its projected location anadditionalusefulsecondepochforpropermotionstothat intheskyandthekinematicproperties:themeasuredradial fromtheSPM,althoughUCAC4andSPMmakeuseofthe velocity (RV) is similar to that of the SMC (Neugent et al. same first epoch data. For this object, however, the second 2010). However, as Levesque et al. (2014) state, the possi- epoch from UCAC caught this variable star in a relatively bility that it may be a halo giant with a similar RV cannot faintstate,sothatthepositionaluncertaintyisseveraltimes be ruled out, but it would require a novel explanation for aslargeasintheSPMdata.TheUCAC4propermotiones- the peculiar spectral features. timateis1.8±2.9mas/yrinrightascensionand−3.3±2.7 mas/yrindeclination.Ifonetakesaweightedaverageofthe In this paper, we report the detection of a very large twomeasurements,thenpropermotionis2.4±1.8mas/year propermotionbasedondatafromtheSouthernProperMo- in right ascension and −6.8±1.8 mas/year in declination. tion survey (Girard et al. 2011). If confirmed, the proper This differs from zero at roughly 3.8σ, and from the -1.1 motionexcludesmembershipintheSMC,makingHV2112 mas/yr in declination of the SMC (Kallivayalil et al. 2013) aforegroundMilkywaystar(likelyagiantduetoitspulsa- by3.2σ.ThedeviationfromthepropermotionoftheSMC tional properties). A location in the Milky Way would rule in right ascension, which is -0.8 mas/yr (Kallivayalil et al. out both proposed scenarios, the TZ˙O-nature and sAGB- 2013) is statistically insignificant, being about 1.7σ. Av- nature of HV 2112. eraging SPM and UCAC using weights determined solely bytheirstatisticaluncertaintiesisaconservativeapproach, Weproposeanewscenariowhichinvolvespollutionby since any systematic uncertainties should be far more seri- the wind of a former massive AGB companion star. In this ous for the fainter UCAC4 epoch than for the SPM data, manner, the enhancement of molybdenum, rubidium, and andtheSPMmeasurementindicatesalargerpropermotion possiblylithiumcomesfromtheprocessesthatwouldenrich than does the UCAC4 measurement. amassiveAGBstar.Bynowtheformercompanionhaslost Wehavealsocheckedforsystematicsintheastrometry itsentireenvelope.Itsremnant,mostlikelyamassivewhite of the field by taking the Hipparcos proper motion mea- dwarf,isexpectedtobestillaroundinawideorbit.HV2112 surements for stars within 2 degrees of HV 2112 and cross- in this scenario is the initially less massive secondary star. correlating them against the SPM 4.0 measurements. We This scenario is similar to the scenario proposed by Iben have found that there is no systematic offset at a level of & Renzini (1983) to explain extrinsic S stars as well as Ba about 0.5 mas/year, and that the differences between the and CH stars (McClure 1984). One of the very attractive SPMandHipparcosmeasurementsarecomparabletothose features of this scenario (and any scenario were it is placed expected from the stated statistical errors. intheGalactichalo)isthenaturalexplanationprovidedfor We have test whether the internal errors for the SPM the Ca (and K) features. These elements appear enhanced data are comparable to what is stated in the SPM cata- whencomparedtoSMCgiants,buttheyarenormalforthe logue. Using the TOPCAT package, we have taken all the Galactic halo where [Ca/Fe] enhancements are typical. We SPM stars within 1.5 degrees of HV 2112. We have then discuss various testable predictions of this scenario. filtered to include only stars whose 2MASS colours (Cutri (cid:13)c RAS,MNRAS000,1–5 HV 2112: an extrinsic S star in our Galaxy 3 1. 2. v1014rgb -15 Rb -10 Mo -5 main sequence main sequence wind accretion primary secondary massive AGB onto secondary as/yr 0 3. 4. m E / 5 pmD 10 RbMo LiMRbo 15 white dwarf enriched secondary white dwarf TZO-impostor 20 25 Figure 2. Cartoon of the alternative evolutionary scenario pro- posedinvolvingawidebinarysystemlocatedintheGalactichalo 30 (1).TheprimarystarevolvestobecomeamassiveAGBstarand -30 -20 -10 0 10 20 30 40 50 pollutesitslowmasscompanionwithitsenrichedwind(2),leav- pmRA / mas/yr ing behind a white dwarf after shedding its entire envelope (3). Figure 1. The proper motions of stars within 2 degrees of Eventually, the enriched secondary becomes an AGB star itself HV 2112, which have magnitudes from V=10 to V = 14 and showing the, now diluted, abundance patterns of enrichment in J −K colors consistent with the giant branch for metal poor additiontoLimadeinsitu.Thecartoonsuggeststhatthewhite stars (from J −K=0.78 to 1.35). HV 2112 has the black circle dwarf remnant of the primary is now accreting the wind of the around it in the figure. It can be seen to be an outlier in the “TZO-impostor”.Thisisapossibility,butatpresentwehaveno propermotiondiagram,withthemainlocusofpointstowardthe strongevidenceprovingordisprovingthishypothesis. center being the SMC stars, and the outliers being foreground stars. 3 A FORMATION SCENARIO FOR HV 2112 etal.2003)placethemononeofthegiantbranchesforthe At a distance of a few kpc, the luminosity of this object is SMCandwhichhaveV between10and14(i.e.amagnitude likely to be ∼103L , rather than the 105L it would have (cid:12) (cid:12) rangeforwhichweexpecttheerrorstobesimilartothosefor in the SMC. Furthermore, is has been classified as a Mira- HV 2112), and then examined only those stars with proper like variable from photometry (Samus et al. 2012), and its motions of 5 masec/year or less. We find that for these 199 emission lines in its bright phases are also consistent with stars,thestandarddeviationinthevaluesofthepropermo- Mira phenomenology (Levesque et al. 2014). The hypothe- tions in both dimensions is about 90% of the mean values sis of a Mira star also fits well with the inferred luminosity of the uncertainties on the proper motions from SPM 4.0. for the closer distance. These points were already noted in Theexpectation,ifallthestarshadmeasurementuncertain- Levesqueetal.(2014),butrejectedonthebasisofthethen- ties of the mean uncertainty, and there were no interloper likely association with the SMC. Levesque et al. (2014) did stars,wouldbethatthestandarddeviationwouldbeabout briefly,butpresciently,notethatfuturekinematicsobserva- 75% of the 1σ uncertainty (the expectation is not exactly tions could potentially place the object in the Milky Way, 1σ because the more extreme outliers will be eliminated by andthatiftheydid,thiswouldleadtoadifferentinterpreta- the 5 mas/year cut). It may then be that the measurement tionthantheThorne-Z˙ytkowobjectinterpretationpursued uncertainties are underestimated by about 20%, and that intheirpaper.Likewise,thesuperasymptoticgiantbranch the significance of the proper motion measurement relative (sAGB)starinterpretationconsideredbyToutetal.(2014) to the SMC proper motion is at only the 3.0σ level, but alsocannotexplaintheobjectifitislocatedintheGalactic this gives us additional confidence that there are no major foreground due to brightness considerations. systematic uncertainties in SPM 4.0 that bias our results. On the other hand, the abundances may be explained Any statistically significant measurement of a proper by a scenario in which matter is transferred from a mas- motion in data with the precision of SPM 4.0 immediately sive AGB primary star to a binary companion. Eventually, providesastrongargumentagainstalocationofthestarin thesecondarywillevolveoffthemainsequenceandbecome the Small Magellanic Cloud, because the statistical uncer- anAGBstaritself,albeitofsignificantlylowermass,allow- taintiesonthepropermotionsarelargerthantheescapeve- ing such an object still to exist in the old population in the locityfromtheSMC.Thepropermotioninδof10mas/year halo.ThesestarsaregenerallyreferredtoasextrinsicSstars atadistanceof62.1kpc(Graczyketal2014)yieldsaspace (e.g. Iben & Renzini 1983), following an initial purely phe- velocity of 3100 km/sec, well above the escape velocity of nomenologicalclassificationof Sstars (Merrill1922),which the SMC. If one instead takes this to be the velocity of a were later found, in a mnemonically fortunate coincidence halostarintheMilkyWayhalo,andassumesa150km/sec to be rich in s-process elements (Smith & Lambert 1986). speed (a reasonable value given the star’s radial velocity of The primary must be massive enough to produce substan- 157 km/sec – Levesque et al. 2014), then it is likely to be tial enhancements of molybdenum and rubidium. This cer- locatedatadistanceofabout3kpc,althoughthestatistical tainly is expected for super-AGB stars (Lau et al. 2011; errors on the proper motion measurement allow for a wide Tout et al. 2014), but is likely to happen down to even 4 rangeofpossibledistances.Boththespacevelocity,andthe M (Lugaroetal.2003),andisclearlyseeninsomeSstars (cid:12) height, probably a few kpc below the Galactic Plane, given (e.g.Lambertetal.1995showthatRbabundancesincrease the Galactic latitude of 45 degrees, suggest that this is a withincreasings-processelementalabundances,whileAllen halo star rather than a disk star in the Milky Way. & Barbuy 2006 show an excess of molbydenum in most of (cid:13)c RAS,MNRAS000,1–5 4 Maccarone & De Mink the members of a large sample of barium stars1.). The ex- for such a star to have a pulsational period of about 600 trinsic S stars form through the binary evolution channel dayscausedbythesamepulsationmechanismastheclassi- above, while the intrinsic S stars are formed by dredge-up calMiras,eventhoughitsradiuswouldbemuchlargerthan ofs-processelements,andshowenhancementofTc.TheTc a classical Mira. signature indicates that the material has been placed into Ontheotherhand,theradialvelocityamplitudeofthe the stellar atmosphere on a timescale of order or less than pulsationsshouldbequiteabitlargeriftheobjectisatSMC the radioactive decay time of Tc. distancesthanifitisaforegroundGalacticMiravariable.In The placement of the star in the Milky Way halo also essence,onecanapplytheBaade-Wesselinkmethod(Baade helpstoexplainthecalciumabundanceanomalythatcould 1926;Wesselink1946)toestimatetheradiusandhencedis- not be explained by the super-AGB star scenario discussed tance to the object. At SMC distances, the radius of the inToutetal.(2014)norbytheTZOscenarioofLevesqueet star should be about 1500 R , so a variation of 40% about (cid:12) al. (2014). Levesque et al. (2014) did not have a calibrated themeanradiusisavariationof600R .Thepulsationsof (cid:12) estimateoftheCaabundanceofthestar,butdidshowthat Miras in general, and HV 2112 in particular are such that the ratio of strengths of Ca lines to Fe lines was a factor of most of the variation takes place in a relatively small range about 3 stronger than that for red supergiants in the SMC. ofpulsephase.Ifweassumethatsuchasizevariationtakes Thisisindicativeofa[Ca/Fe]ratioabout3.Kobayashietal. placein0.2timesthepulseperiod,or120days,thenweex- (2006)showsthathalostarsinthesolarneighborhoodwith pectavelocityofexpansionofabout50km/sec.Thisspeed [Fe/H]<−1 tend to have [Ca/Fe] of about 0.4, which indi- is not large enough to be inherently problematic for either catesacalciumtoironabundanceratioabout2.5.Themost scenario discussed here, but does provide a means for esti- metal rich (and presumably youngest) SMC giants have ra- matingthestellarradius,andhencethedistancetothestar tiosofαelementabundancestoironabundancesquitesimi- if a radial velocity monitoring campaign is undertaken. lartosolar(Mucciarelli2014).Relativelylittlediscussionex- istsaboutthepotassiumexcessinHV2112,butitisofsim- ilarsizetotheCaexcessjudgingfromtheplotsinLevesque 4.1 Population estimate et al. (2014), and this, also, can be explained well in terms We can make a crude estimate of the number of extrinsic of empirical halo abundances (Kobayashi et al. 2006). S stars in the Galaxy with a massive AGB progenitors as Lithium is a fragile element and is easily destroyed at partoftestingtheplausibilityofthescenario.Sstarsmake temperatures in excess of about 2.5x106K. Even a modest about4%ofgiantsinthetemperaturerangewhereMstars depth of mixing can lead to efficient depletion (Stancliffe are normally found, and extrinsic S stars make about 1/3 2009). In principle, Lithium can be in the AGB phase of of S stars (van Eck & Jorissen 2000). Extrinsic S stars are the primary star (Cameron & Fowler 1971) and accreted thus about 1% of the number of stars that appear to be M onto the secondary. However, all Lithium originating from giantsincolorspace.TheSMCspansabout3×10−4 ofthe the primary has most likely been destroyed when HV 2112 solid angle of the sky. The Milky Way has ∼ 1011 stars, of evolved to become an AGB star and developed a deep con- which about 109 are in the halo, and about 107 are halo vectiveenvelope.Thepresentdaylithiumabundanceismore M giants. There should then be approximately 30 extrinsic likelytheresultofproductioninsitu,inHV2112itself.Low S stars in the Milky Way halo projected against the SMC. mass AGB stars can produce lithium as a result of thermal If we further assume that only the extrinsic S stars whose halinemixing(Stancliffe2010).Analternativepossibilityis progenitors were more massive than 4 M can have the re- (cid:12) thatthewhitedwarfleftbehindaccretesenoughmaterialto quired abundances of rubidium and molybdenum, we must produce a substantial nova rate. Evidence for lithium pro- correctforthefactthatonly11%ofthestarsbetween1and duction has been found in the classical nova V339 (Tajitsu 10 M (i.e. the ones that produce AGB stars without un- (cid:12) etal.2015).ThiscouldpollutetheenvelopeofHV2112.The dergoingsupernovae)areabove4M (andhenceareheavy (cid:12) model predictions remain uncertain, but lithium is not un- enough for production of these elements). We then expect common in symbiotic Mira’s. For example V407 Cyg shows a few extrinsic S stars with massive enough progenitors in a strong lithium excess (Tatarnikova et al. 2003). the SMC field. Since we additionally require that they be in an AGB state themselves to explain the Mira variations andtheLiproduction,theexpectednumberfallsbyafactor of about 30 (Girardi et al. 2010), so our expected number 4 THE VARIABILITY OF THE SYSTEM AND of objects is about 0.1. The scenario we propose thus quite ITS IMPLICATIONS plausiblyplacesasourceofthetypeweinvokeinalocation Radialpulsationshaveperiodswhichscalewiththedynam- projected against the SMC, while also not placing so many ical timescale of the star – that is, P ∝ ρ−1/2, where P is suchobjectsprojectedagainsttheSMCthatwewouldhave the period of oscillation, and ρ is the density of the star. expected to have found many of them already. Miravariableshavepulsationperiodofabout140-700days. Their masses tend to be ≈1M and their radii tend to be (cid:12) about 300 R . The typical variations in radius are a factor (cid:12) 5 ADDITIONAL TESTABLE PREDICTIONS of about 1.5 (e.g. Mahler et al. 1997). Because the mass of OF THE SCENARIO the Thorne-Z˙ytkow object could be quite large, it is viable The scenario makes several additional testable predictions. First and foremost, the object should be in a wide binary 1 Thebariumstarsarethoughttobeanearlierevolutionarystate system. Orbital periods of extrinsic S stars are generally at oftheextrinsicSstars least 600 days (Jorissen & Mayor 1992). The white dwarf (cid:13)c RAS,MNRAS000,1–5 HV 2112: an extrinsic S star in our Galaxy 5 in this system should likely be a ONeMg white dwarf, and REFERENCES henceshouldbeofconsiderablyhighermassthattheSstar. AllenD.M.,BarbuyB.,2006A&A,454,895 Nonetheless,themotionsshouldstillbeofrelativelylowor- Anguloetal.1999,NuclearPhysicsA,656,3 bitalvelocity.Assumingatotalsystemmassofabout2M (cid:12) BaadeW.,1926,AN,228,359 and an orbital period of about 2 years, the orbital velocity BiehleG.T.,1991,ApJ,380,167 oftheopticallybrightstarshouldbeabout20km/sec,while BiehleG.T.,1994,ApJ,420,364 the orbital velocity of the white dwarf should be about 10 BondiH.,HoyleF.,1944,MNRAS,104,273 km/sec. BrittC.T.,etal.,2015,MNRAS,448,3455 Additionally, the system may show some accretion by CameronA.G.W.,FowlerW.A.,1971,ApJ,161,111 the white dwarf in the system. In Mira itself, for example, CannonR.C.,1993,MNRAS,263,817 ChevalierR.,1993,ApJL,411,33 the white dwarf companion of the optically bright star is seen to be accreting at about 10−10M per year (Sokoloski CutriR.M.,etal.,2003,TheIRSA2MASSAll-SkyPointSource (cid:12) Catalog,NASA/IPACInfraredScienceArchive & Bildsten 2010), but the orbital period of Mira is about FryerC.L.,BenzW.,HerantM.,1996,ApJ,460,801 500 years. If, instead, this system has an orbital period of GirardT.M.,etal.,2011,AJ,142,15 only a few years, the accretion rate can be expected to be GirardiL.,etal.,2010,ApJ,724,1030 much higher. To first order, the mass accretion rate from a GraczykD.,etal.,2014,ApJ,780,59 stellar wind can be estimated to be: IbenI.,RenziniA.,1983,ARA&A,21,271 (cid:18)m˙ (cid:19) (cid:18)v (cid:19)4(cid:18) M (cid:19)2 JorissenA.,MayorM.,1992,A&A,260,115 acc = acc acc . (1) KallivayalilN.,vanderMarelR.P.,BeslaG.,AndersonJ.,Alcock m˙w vw Macc+Mdon C.,2013,ApJ,764,161 KarakasA.I.,LattanzioJ.C.,2014,PASA,31,30 Since the orbital speed of the donor star and the wind KobayashiC.,UmedaH.,NomotoK.,TominagaN.,OkhuboT., speedsofAGBstarsareofroughlythesameorder,andthe 2006,ApJ,653,1145 accretor is heavier than the donor, a substantial fraction of Lambert D.L., Smith V.V., Busso M., Gallino R., Straniero O., thetotalwindmaybeaccreted(Bondi&Hoyle1944;Struck 1995,ApJ,450,302 et al. 2004; Mohamed & Podsiadlowski 2012). Lau H.H.B., Doherty C.L., Gil-Pons P., Lattanzio J.C., 2011, Given that the wind speed and orbital speed are not ASPC,445,45 known in this system, and the accretion rate depends on KuchnerM.J.,VakilD.,SmithV.V.,LambertD.L.,PlezB.,Phin- the ratio of these two speeds to the fourth power, it is diffi- neyE.S.,2002,ASPC,263,131 culttomakeaspecificpredictionhere.If∼1%ofthewind LeonardP.J.T.,HillsJ.G.,DeweyR.J.,1994,ApJ,423,L19 of a typical AGB star (i.e 1% of 10−6M /yr) is accreted, Levesque E.M., Massey P., Z˙ytkow A.N., Morrell N., 2014, MN- (cid:12) RAS,443L,94 however, then the system would be expected to be bright Lugaro M., Davis A.M., Gallino R., Pellin M.J., Straniero O., intheultravioletduetostrongaccretion.Thesystemisnot K¨appelerF.,2003,ApJ,593,486 detectedinGALEXdata,butdoesshowablueexcessinthe MahlerT.A.,WasatonicR.,GuinanE.F.,1997,IBVS,4500,1 spectrum of Levesque et al. (2014). More targeted searches McClureR.D.,1984,ApJ,208L,31 for an accretion signature in this object are warranted, and MerrillP.W.,1922,ApJ,56,457 should be spread over the ∼600 day period of the optically MohamedS.,PodsiadlowskiP.,2012,BalticAstronomy,21,88 bright star, given that the accretion rate is likely to change MucciarelliA.,2014,AN,335,79 with the pulsation phase. The system does not show X-ray NeugentK.F.,MasseyP.,SkiffB.,DroutM.R.,MeynetG.,Olsen emission,witha90%upperlimitonitsluminosityofabout K.A.G.,2010,ApJ,719,1784 1031(d/6kpc)2 in data taken on March 25, 2013 (V. An- Samus, N.N., et al., 2012, General Catalog of Variable Stars, VizieROnlineDataCatalog:B/gcvs toniou, private communication). Relatively few cataclysmic SmithV.V.,LambertD.L.,1986,ApJ,311,843 variableshaveX-rayluminositieshigherthanthisvalue,and SokoloskiJ.,BildstenL.,2010,ApJ,723,1188 those which do are frequent dwarf nova outbursters (e.g. StancliffeR.J.,2009,MNRAS,394,1051 Britt et al. 2015). This limit, again, is not particularly con- Stancliffe,R.J.,2010,MNRAS,403,505 straining, as many symbiotic stars have X-ray luminosities StruckC.,CohanimB.E.,WillsonL.A.,2004,MNRAS,347,173 below 1031 erg/sec. TaamR.E.,BodenheimerP.,OstrikerJ.P.,1978,ApJ,222,269 Tajitsu, A., Sadakane, K., Naito, H., Arai, A., Aoki, W., 2015, Nature,518,381 TatarnikovaA.A.,MarreseP.M.,MunariU.,TomovT.,White- 6 ACKNOWLEDGMENTS lockP.A.,&YudinB.F.,2003,MNRAS,344,1233 ThorneK.S.,Z˙ytkowZ.N.,1975,ApJ,199L,19 We thank the Lorentz Center for having hosted the work- ThorneK.S.,Z˙ytkowA.N.,1977,ApJ,212,832 shop on “The Impact of Massive Binaries throughout the Tout C.A., Z˙ytkow A.N., Church R.P., Lau H.B., Doherty C.L., Universe”, where the ideas in this paper were first dis- IzzardR.G.,2014,MNRAS,445L,36 cussed. We thank Stephen Justham, Ben Davies, Lennart VanEckS.,JorissenA.,2000,A&A,360,196 van Haaften and Onno Pols for useful discussions, Vallia van Paradijs J., Spruit H.C., van Langevelde H.J., Waters Antoniou for sharing the X-ray upper limit in advance of L.B.F.M.,1995,A&A,303L,25 publication, and Terry Girard for helpful and detailed dis- WesselinkA.J.,1946,Bull.Ast.Int.Neth.,10,91 cussionsofpossiblesystematicerrorsintheSouthernProper ZachariasN.,FinchC.T.,GirardT.M.,HendenA.,BatlettJ.L., MonetD.G.,ZachariasM.L.,2013,AJ,145,44 Motion survey data. SdM acknowledges support by the EU Horizon 2020 program through a Marie Sklodowska-Curie ReintegrationFellowship,H2020-MSCA-IF-2014,projectid 661502. (cid:13)c RAS,MNRAS000,1–5