Mon.Not.R.Astron.Soc.000,1–7(2012) Printed11December 2013 (MNLATEXstylefilev2.2) The mass of the black hole in GRS 1915+105: new constraints from IR spectroscopy 3 D. J. Hurley1⋆, P. J. Callanan1, P. Elebert1 and M. T. Reynolds2 1 1Department of Physics, University College Cork, Cork, Ireland. 0 2Department of Astronomy, Universityof Michigan, 500 Church St., Ann Arbor, MI 48109 2 n a J Accepted 2012December 31.Received2012December25;inoriginalform2012March15 2 ] ABSTRACT R GRS 1915+105 has the largest mass function of any Galactic black hole system, al- S thoughtheerrorisrelativelylarge.Herewepresentspectroscopicanalysisofmedium- h. resolutionIRVLTarchivaldataofGRS1915+105intheK-band.Wefindanupdated p ephemeris, and report on attempts to improve the mass function by a refinement of - the radial velocity estimate. We show that the spectra are significantly affected by o the presence of phase-dependent CO bandhead emission, possibly originating from r t the accretion disc: we discuss the impact this has on efforts to better constrain the s black hole mass. We report on a possible way to measure the radial velocity utilising a [ apparent H-band atomic absorption features and also discuss the general uncertainty of the system parameters of this well-studied object. 1 v Key words: stars: binaries – infrared: stars – stars: individual: GRS 1915+105. 4 7 2 0 . 1 INTRODUCTION andV404Cyg,whichuntilrecentlywasthoughttoharbour 1 a 12 ± 2 M⊙ (Shahbazet al. 1994) but has now been 30 SGiRncSe 19it1s5+1d0is5covreemryainbsyonCeasotfro-tTheiradmoosett ailn.ten(s1i9v9e4ly) reduced to 9.0+−00..26 M⊙ (Khargharia et al. 2010). 1 studied of all the Galactic X-ray sources, and is the : Following the estimate by Greiner et al. (2001b), v prototypical Galactic ‘microquasar’. This system is of Casares (2007) noted that this mass falls outside the i particular importance, not only for our understanding of X theoretical distribution range for our Galaxy as calculated the formation of jets near black holes, but also because by Fryer& Kalogera (2001). However, since the error on r it is regarded as a Galactic analogue to Active Galactic a the black hole mass is so large (±4.0 M⊙), it is clear that Nuclei (Mirabel & Rodr´ıguez 1994; Castro-Tirado et al. a more accurate estimate is required before we can make a 1996; Greiner et al. 2001a,b). Due to its location only 0.2 meaningfulcomparisontothetheoreticalmassdistribution. arcsecfrom theGalactic plane,GRS1915+105 suffersfrom In addition, McClintock et al. (2006) derive a lower limit isstr1o9n.g6±int1e.7rstmelalgar(Cabhsaoprupitsio&n:CtohrebeVl-2b0a0n4d).eFxotrintchtiisonre(aAsovn) for the dimensionless spin parameter of a∗ > 0.98 making GRS1915+105 a nearmaximally spinningKerrblack hole. observations in the K-band have been particularly critical, This conclusion was drawn from an analysis of RXTE leadingtotheidentificationofthesecondary(Greiner et al. and ASCA data for the thermal state of GRS 1915+105, 2001a), a radial velocity estimate for the secondary and a model of the X-ray continuum of a fully relativistic (Greiner et al. 2001b) and the possible determination of a accretion disc. As discussed by McClintock et al. (2006), photometric period (Neil et al. 2007). The radial velocity determining the mass of, and distance to, GRS 1915+105 studyyielded amassfunction of9.5±3.0 M⊙ which,when is of fundamental importance for verifying this method of combined with inclination estimates (Fenderet al. 1999; spin determination. Greiner et al.2001b)indicatedaprimarymassof14.0±4.0 M⊙. This not only confirmed the presence of a black hole in the system but indicated it to be one the most massive HerewepresentananalysisofarchivalVeryLargeTele- black holes discovered to date in a Galactic X-ray binary. scope(VLT)dataandare-analysisofthedatapresentedby Other examples of massive primaries include Cygnus X-1 Greiner et al. 2001a,b. In particular, we show in what fol- with a primary mass of 14.8±1.0 M⊙ (Orosz et al. 2011) lows that, near orbital phase 0.25, the absorption lines are corrupted by emission, which will have a significant effect ontheamplitudeoftheradialvelocityvariation.Wediscuss ⋆ E-mail:[email protected] possibleoriginsforthisemission,andtheeffectitmighthave 2 D. J. Hurley et al.: The mass of the black hole in GRS 1915+105 Table 1.Completelistofarchivaldatausedincludingtotal exposuretimesforeachnightfrom1999-2002. Date No. of Exp Total Phase Date No. of Exp Total Phase (UT) (s) (UT) (s) 1999 Jul 23 10×300s 3000 0.46271 2002 Jul 30 10×300s 3000 0.23063 2000 Apr 23 8×240s 1920 0.74271 2002 Aug 06 20×300s 6000 0.43595 2000 May 11 8×240s 1920 0.27410 2002 Aug 08 10×300s 3000 0.49377 2000 May 23 8×240s 1920 0.63179 2002 Aug 10 10×300s 3000 0.55101 2000 Jun 10 8×250s 2000 0.16171 2002 Aug 12 10×300s 3000 0.63898 2000 Jun 13 8×250s 2000 0.25325 2002 Aug 13 10×300s 3000 0.64005 2000 Jun 18 8×250s 2000 0.39898 2002 Aug 14 20×300s 6000 0.65954 2000 Jul 03 8×250s 2000 0.83817 2002 Aug 17 20×300s 6000 0.75728 2000 Jul 09 8×250s 2000 0.01589 2002 Aug 19 20×300s 6000 0.81673 2000 Jul 12 8×250s 2000 0.10582 2002 Aug 29 20×300s 6000 0.11109 2000 Jul 14 8×250s 2000 0.16342 2002 Aug 31 20×300s 6000 0.17236 2000 Jul 17 8×250s 2000 0.25120 2002 Sep 02 10×300s 3000 0.22992 2000 Jul 23 8×250s 2000 0.46333 2002 Sep 05 10×300s 3000 0.28863 2000 Jul 27 8×250s 2000 0.54704 2000 Aug 01 8×250s 2000 0.69426 2000 Aug 21 8×250s 2000 0.28594 onthemassfunctionandprimarymass.Wealsodiscussthe feasibility of using H–band spectroscopy tomeasure thera- 200 dialvelocityofthesecondaryfromnarroweratomicfeatures using new Gemini observations. 150 2 OBSERVATIONS 100 2.1 ESO VLT archival data s) TusheedEhuerreopweaerneStoaukethnewrnitOhbtsheervIantforaryred(ESSpOe)ctarrocmheivtaerl dAantda city (km/ 50 Array Camera (ISAAC) on the 8-m VLT Antu telescope elo V at Cerro Paranal (Chile). The short wavelength arm al 0 of ISAAC equipped with a 1024 × 1024 pixel Rockwell adi R HgCdTe array with an image scale of 0.147 arcsec pixel−1. Using the medium resolution grating (1.2 ˚A pixel−1 in -50 the K-band) yields a spectral resolution of ∼3000 with a 1 arcsec slit. Science data consisted of several 250–300 s -100 individual exposures each night (see Table 1 for summary of the data used). We present a re-analysis of 16 nights -150 of observations from 2000 May to August, which were the 0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 focusofGreiner et al.2001a,b,and13morefrom2002July Orbital Phase to Septemberwhich are presented herefor thefirst time. Figure 1. Radial velocity curve of the previously unpublished The bias subtraction, flat-fielding and sky subtraction 2002 dataset folded using the best-fit period of Greineretal. were performed using the iraf1 ccdproc package. The (2001b) Porb = 33.5 ± 1.5 days (marked with open circles). spectra were extracted using the iraf kpnoslit package. The fit overlaid has a reduced χ2 = 1. The peak occurs at Initial wavelength calibration was carried out using arc Kd=131.5±2.5kms−1withasystemicvelocityof33.0±2.1km spectra obtained before or after each science run. How- s−1 . Ascan beclearlyseen the phasing ofthe dataset withthe ever, cross-correlation of these spectra with the OH sky ephemerisofGreineretal.(2001a)leadstoanoffsetinthepeak lines indicated a residual wavelength offset of ∼3–5 ˚A, from the expected 0.25 to ∼0.5. Also plotted is the same data presumably due to flexure in the telescope/spectrograph folded on our refined ephemeris (marked with crosses) from the combineddataset(seeFigure3).Notethattherefinedephemeris between the times when science exposures and calibration exhibitsnosignificantoffsetfromtheexpected peakof0.25. arc frames were taken. This wavelength shift was corrected for before thetelluric features were removed. Tocorrect for 1 IRAFisdistributedbytheNationalOpticalAstronomyObser- telluric absorption the A0 V star HD 179913 was observed vatories,whichareoperatedbytheAssociationofUniversitiesfor either before or after the science exposures each night Research in Astronomy, Inc., under cooperative agreement with and the method as outlined by Vacca, Cushing & Rayner theNational ScienceFoundation (2003) utilized. The data were exported to molly2 where The mass of the black hole in GRS 1915+105 3 250000 6000 y200000 5000 nsit 4000 de150000 DF3000 wer100000 P2000 o P 50000 1000 0 0 25 30 32.5 35 37.5 40 45 25. 27.5 30. 32.5 35. 37.5 40. 42.5 45. Period HdL Period HdL 6000 0 5000 -2 F 4000 D -4 P F g10 -6 PD3000 Lo -8 2000 -10 1000 -12 0 25. 27.5 30. 32.5 35. 37.5 40. 42.5 45. 33 33.5 34 34.5 35 35.5 36 Period HdL Period HdL Figure 2.(topleft)Lomb-Scargleperiodogramofthecombined2000and2002datasetwithaT0=2451666.5.(topright)Bretthorst’s Bayesianperiodogram(Bretthorst2001)whichhasbeenmarginalizedtoexcludenoiseresultingfromanon-sinusoidalsource.(bottomleft) NormalizedLogplotofthePDFtoemphasiseanylowerpeaks.(bottomright)Closerlookatthetworesultingmajorpeakswhichoccur at33.8and35.38daysrespectively,witheachhavingaFWHMof0.1dayswhichwetakeastheerror.Oftheseonlythe33.8dayperiod successfullyphases bothdatasets correctly. they were re-binned onto a common velocity scale. Three yieldsaspectralresolutionof∼5000.Sciencedataconsisted techniques were then used to measure the radial velocity of 5×180s target exposures each night resulting in a S/N curve:eachspectrumofGRS1915+105wascrosscorrelated ∼24. The data were reduced using the iraf gemini nifs against (i) a template spectrum of the KIII standard star package and telluric absorption corrected for by observing HR 8117, (ii) the average of the (remaining) spectra of theA0VstandardHD182761beforeandafterscienceruns. GRS1915+105and(iii)theaverageofthespectraoccurring The data were again exported to molly where they were aroundphase0.75(seebelow).Allthreetechniquesresulted re-binned onto a common velocity scale. The spectra were in similar velocities within theerrors. then cross correlated against the KI III template star HD 83240, taken from the catalogue of high resolution VLT Asafinalcheck,thespectrawerealsore-extractedusing IR spectra of Lebzelter et al. (2012). A stringent mask of theVLTISAACpipeline,andtheresultingradialvelocities the spectra was used in each case to utilise only the lines were again found to be consistent with those determined identified in Figure 7 for thecross correlation. above once the shift due to instrument flexure had been removed. 2.2 Gemini-North NIFS data 3 RESULTS In Greiner et al. (2001a) the presence of atomic absorption 3.1 ESO VLT archival data featuresareidentifiedintheH-band.Motivatedbythehope that a radial velocity measurement based on these inher- In Figure 1 we plot the previously unpublished data from ently narrower features would lead to a more accurate and 2002 only, folded using the ephemeris of Greiner et al. reliablemassfunctionestimate,atotalofthree’pathfinder’ (2001b). The maximum velocity shift is seen to occur spectra were obtained using Gemini-North’s Near-Infrared at phase 0.5, whereas it should occur at phase 0.25 by Integral Field Spectrometer on the nights of 2010 June definition, suggesting that a refinement in the ephemeris of 29th, July 6th and July 13th. The observations were Greiner et al. (2001b) is needed. Both the 2000 and 2002 scheduled to sample the radial velocity variation at three datasetsyieldradialvelocityamplitudesconsistentwiththe orbital phases. NIFS provides 3D imaging spectroscopy result of 140±15 km s−1 found by Greiner et al. (2001b), witha3.0′′×3.0′′ fieldofview,equippedwitha2048×2048 although the systemic velocity (γ) differs somewhat to Rockwell HAWAII-2RGHgCdTe array with an image scale the previously published value of −3 ± 10 km s−1. This of0.103′′×0.043′′perpixelacross andalongtheslit respec- difference is most likely due to the skyline correction tively. Using the H-G5604 grating and the HK-60603 filter outlined in Section 2.1 as prior to this correction we find a systemic velocity that is consistent with that of Greiner et al. (2001b) for the same dataset. 2 http://www.warwick.ac.uk/go/trmarsh/software 4 D. J. Hurley et al.: The mass of the black hole in GRS 1915+105 interval), which is slightly lower than the 9.5 ± 3.0 M⊙ 200 determinedbyGreiner et al. (2001b),butconsistent within their 1σ error. The mass of the primary is found to be 12±1.4 M⊙. 150 However,inFigure4weplotthetrailedspectraforthe 100 combineddatasetphasedonourrefinedephemeris:emission s) can be seen in the first half of the orbital range, indicated m/ by the lighter regions, blueward of all the CO bandhead y (k 50 absorption features (this is more apparent in the electronic cit elo colour version). The emission rises and falls over the first al V 0 half of the orbit and is predominantly strongest near phase adi 0.25. To demonstrate this emission more clearly, in Figures R 5 & 6 we plot the summed spectra, phased on the revised -50 ephemeris,forboththe2000and2002datasets.Ineachcase thedataisfurtherseparated intotwodistinct orbitalphase -100 windows(0.15-0.35 and0.65-0.85) toillustrate thepresence andabsenceoftheemission near0.25and0.75respectively. As can be clearly seen, there is significant CO bandhead -150 emission present in the first half of theorbital phaserange, 0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 Orbital Phase notinthelatter.Thisemission ispresentinthetwoepochs ofobservation(mostnoticeablynearthe12CO(2-0)and(3- 1)transitions) indicatingthat itisnot just atransientphe- Figure 3. Radial velocity curve of the combined 2000 (marked nomenon. Whereas variability in the bandhead absorption with open squares) and 2002 (marked with crosses) datasets foldedoverthebest-fitperiodofPorb=33.8±0.1daysobtained hadbeenreportedbypreviousauthors(Harlaftis & Greiner from the Lomb-Scargle periodogram. The overlaid fit has a re- 2004), the fact that it occurs predominantly near orbital ducedχ2=1.ThepeakoccursatKd=131.9±3.3kms−1 with phase 0.25 has important implications for the radial veloc- asystemicvelocityof34.2±2.5kms−1.Thescalingoferrorbars ity curve,which we discuss in Section 4. to produce a reduced χ2 = 1 makes the 2002 error bars larger than those in Figure 1. This is due to the larger scatter of the pointsinthecombineddataset. 3.2 Gemini-North NIFS data From our NIFS spectra (see Figure 7) we confirm the pres- We next combined both the 2000 and 2002 datasets in enceof many of theatomic features noted byGreiner et al. order to derive an improved ephemeris byperforming a pe- (2001a): namely the Al triplet ( λ16723.5 ˚A, λ16755.2 ˚A, riod analysis via a Lomb-Scargle periodogram and then re- λ16767.9˚A),SiI(λ15964.4˚A,λ16685.3˚A),MgI(λ15029.1 fining the results using Bretthorst’s bayesian periodogram ˚A,λ15044.3 ˚A,λ15770.1 ˚A,λ17113.3 ˚A),FeI(λ15297.3 ˚A, (Bretthorst 2001) to clean out any false signals, the re- λ15969.1 ˚A,λ16047.1 ˚A)and OH(λ16753.8 ˚A).Equivalent sults of which are shown in Figure 2 (calculated using widths of some of the major lines can be found in Table 2. T0 =2451666.5fromGreiner et al.(2001b)).Thepeakthat We do not detect Br series (λ or η), CO or He I. We first corresponds toaperiod of 33.8±0.1 is theonly onetosuc- used these spectra to independently estimate the effective cessfullyphaseallthedataconsistentlyandsoweadoptthis temperature (Teff) of the secondary, from Equation 3 of asourrevisedorbitalperiodforGRS1915+105.Thisperiod Leet al. (2011), using the EW of the Mg I (λ17113.3 ˚A) is consistent with that of Greiner et al. (2001b) (33.5±1.5 line. From this we estimate an Teff of 4540+−550055 K which is days) to within 1–σ, but does not agree with the low am- reassuringlyconsistentwiththeK2IIIclassificationalready plitude K–band modulation discussed by Neil et al. (2007). ascribed by Greiner et al. (2001a). Folding all the data on this revised period yields the radial velocitycurveshowninFigure3.Asinewavefittothisplot Theresultingvelocitiesfromourcrosscorrelation anal- yields a secondary radial velocity semi-amplitude (Kd) of ysis(seeSection2.2fordetails)areshowninTable2.These 131.9±3.3kms−1 andasystemicvelocity (γ)of34.2±2.5 were then searched for the expected variation due to the km s−1. To estimate the uncertainties on derived quanti- motionofthesecondary,basedontheresultsoftheK–band ties, we used a Monte Carlo analysis. This involved select- observationsoutlinedinSection3.1.However,evenwiththe ing100000randomsamplesfromdistributionsofeachinput improvementintheerroroftheephemeris(Section3.1),over parameter i, Kd, Porb and MD (based on their means and theintervening110 cyclesbetweentheK &H–bandobser- standard deviations), and calculating the results based on vationstheuncertaintyinphaseinformationaccumulatesto theoutputdistributions.Combiningourinitialestimatesin ∼±0.33.Hence,inconstrainingKd,theabsolutephasewas Porb andKd withthesuggesteddonormass(MD)of0.8±0.5 leftasafreeparameter.WeusedtheγvelocityfoundinSec- M⊙ (Harlaftis & Greiner 2004) and the binary inclination tion 3.1. To our surprise, we do not find the radial velocity (i)of66±2degrees(Fenderet al.1999)themass function: variation expected, based on the CO band-head measure- ments discussed in Section 3.1 and Greiner et al. (2001b). f(M)= (Mcsini)3 = PorbKd3 Our measurements are consistent only with an upper limit (Mc+MD)2 2πG of∼50kms−1onKd,althoughthisdependsstronglyonthe is found to be8.0±0.6 M⊙ ( at the68 per cent confidence assumed valueof γ. The mass of the black hole in GRS 1915+105 5 Figure 4. Trailedspectrum of all data, repeated twice for clarity. Lighter colours indicate emissionand darker absorption. Notice the emission is visible only in the the first half of the orbital cycle, blueward of each bandhead, and is strongest near phase ∼0.25 (see electronicversionalso).Intermittent emissionatλ23160˚Aisaresidualtelluricfeature. Spectrum Mg (15770.9) Si (1668.53) Mg (17113.08) Al (16723.5) Velocity (˚A) (˚A) (˚A) (˚A) (km s−1) 2010 Jun 29 2.02±0.16 0.59±0.10 1.07±0.10 1.25±0.22 9.1±20.5 2010 Jul 06 2.47±0.15 0.44±0.10 0.92±0.10 1.31±0.15 54.4±21.3 2010 Jul 13 2.55±0.15 0.49±0.10 1.05±0.11 1.07±0.13 62.5±21.9 HD 83240 2.23±0.20 0.47±0.10 0.99±0.09 0.567±0.07 Table 2. Equivalent widths of H–band spectra. Only the deepest lines are listed. Also listed are the corresponding radial velocity measurements,andtheirassociatederrors,fromthecrosscorrelationanalysisinMOLLY. 4 DISCUSSION the observed emission. Higher order lines (> P 33) blend to form a continuum at shorter wavelengths (see Figure 5) H-Band Atomic Absorption and do not exhibit the discrete band structure blueward Itisclearthatsomecautionshouldbeusedwheninterpret- of each CO bandhead that is observed. Also, the intensity ing the radial velocity variation (or lack thereof) in the H- of Pfund emission lines should increase from shorter to band,aswehaveonlythreemeasurementsatourdisposal.If longer wavelengths, whereas the emission that we observe thelackofvariation(comparedtowhatisexpectedfromthe is approximately uniform at each site. Hence we believe K–band)isreal,thenwemustlookforalternativesourcesof that CO bandhead emission is the best explanation for the theH-bandemission. Forexample, thework of Muno et al. observed emission features. (1999) and Lee et al. (2002) suggests that GRS 1915+105 With a dissociation energy of 11.1 eV, the site of this may possess a circumbinary disc of material: if we assume emission must reside in a relatively cool, outer part of the such a discis thesource of theH-bandabsorption features, disc.SucharegioncannotbedirectlyilluminatedbyX-rays then we can use the FWHM of the latter to constrain the from the inner disc, and must instead be shielded from it. location of the emission. The observed FWHM of the ab- Thiscouldbeprovided,for example,byawarped accretion sorption features is ∼60 km s−1, and assuming a Keplerian disc, which in turn has been invoked to explain the long disc orbiting a binary of total mass 12.8 M⊙, we estimate termperiodicitiesintheX-raylightcurveofGRS1915+105 a radius of 530 R⊙, well outside the binary radius of 100 (Rau et al. 2003). R⊙. However, it remains suspicious that the H-band spec- trum,despitethelackofradialvelocityvariation,soclosely However, it is the fact that the emission occurs prefer- matches the expected spectral type of the secondary. It is entially near orbital phase 0.25 that most interests us here. clearthatmoresystematicobservationsofthesespectralfea- Such emission may become more visible near this phase tures are warranted. because the absorption lines of the secondary are at their maximum redshift at this phase, making emission from the CO Bandhead Emission accretion disc easier to observe. However, the same should be true at maximum blue shift (phase 0.75), which is not The CO emission is a rare phenomenon in accreting thecase for our data. binary systems: to our knowledge, it has only previously been observed in the well known cataclysmic variable WZ Sagittae (Howell, Harrison & Szkody 2004). As with Alternatively, it may simply be that the shielded re- Howell, Harrison & Szkody (2004), we also believe that H gion of the disc is best visible near binary phase 0.25 - Pfund emission is unlikely to be a major contributor to although, in the context of a warped and precessing disc 6 D. J. Hurley et al.: The mass of the black hole in GRS 1915+105 knowndistance.However,thedistancetoGRS1915+105 is still a matter of some debate; the conservative estimate of 9.0±3.0kpcasproposedbyChapuis & Corbel(2004)would suggest a binary inclination of 58 ± 11◦ (using Equation 4 from Fenderet al. 1999). Combining this with a donor mass of 0.8±0.5 M⊙ (Harlaftis & Greiner 2004), yields a considerablymoreuncertainprimarymassof16.7±7.4M⊙. To estimate the distance to GRS 1915+105, Greiner et al. (2001b) combined their estimate of γ with the Galactic rotation curve of Fich, Blitz, & Stark (1989) to yield D=12.1±0.8 kpc. Using the same technique, our estimate of γ (see Section 3.1) yields D = 9.4±0.2 kpc, where the low error is due to the fact that we assume a flat rotation curve and thereby only the uncertainty on γ contributes. This in turn would suggest an inclination of Figure 5. Bottom: Average of 2000 spectra between 0.15 and 62±2◦ (again from Fenderet al. (1999)), and a primary 0.35 in phase. Top: Average of 2000 spectra taken between 0.65 and 0.85. Notice the emission present in the former compared mass of 12.9±2.4 M⊙. to the latter. The positions of the Pfund lines are indicated to demonstrate that they cannot account for the structure of the Finally, it is also clear that the black hole mass esti- observedemission. mate will also be affected by the presence of the bandhead emission. Specifically, near phase 0.25, the absorption line measurementswillbeskewedtohighervelocities,artificially increasingtheamplitudeoftheradialvelocity curve.Inad- dition,thepresenceofthebandheademissionwillalsoaffect the measurement of γ in the same manner. Hence, in this sense,themassfunctionwehavederivedmustbeconsidered an upper limit only. Further attempts to refine the mass function and γ of GRS 1915+105 utilizing K-band spec- troscopy will depend on the feasibility of decontaminating theabsorption features from this emission contribution, es- pecially near phase 0.25. This will requirehigher resolution and S/N spectra than havebeen obtained thusfar. 5 CONCLUSIONS We have confirmed the presence of narrow atomic features in the H-band as noted by Greiner et al. (2001a), but our analysis does not provide evidence for the expected radial Figure 6. Bottom: Average of 2002 spectra between 0.15 and velocity variation. This may suggest that the origin of the 0.35 in phase. Top: Average of 2002 spectra taken between 0.65 and0.85.Again,aswiththe2000spectra,theemissionispresent absorptionisdistinctfromthesecondaryandoutsideofthe inthesameorbitalranges.Again,thepositionsofthePfundlines binary but more observations are clearly required to con- areindicatedforreference. firmthis.Alternatively,theCObandheademissionobserved specifically at phase0.25 bringsintoquestionthereliability of the radial velocity amplitude of the secondary based on (Lee et al. 2002),it isnot clearwhy thisshouldbethecase the K-band observations. Higher resolution K-band obser- for both epochs of observations. Another possible cause for vations are required to disentangle the emission from the the appearance of the blueshifted emission phenomenon is bandhead absorption so that a reliable radial velocity esti- P-Cygni type emission originating from a wind driven out- mate can bemade. flow of material, such as proposed by Lee et al. (2002) for GRS1915+105. This would imply a shell ofmaterial some- ACKNOWLEDGEMENTS wherebetween200-400R⊙ fromtheprimary(Clark et al. We gratefully acknowledge the use of the MOLLY software 1999, Equation 1). The dominant source of uncertainty for package developed by T. Marsh. We would also like to this estimate, however, is dueto theuncertainty in the dis- thank the anonymous referee for useful suggestions leading tance to thebinary. to an improved paper. The uncertaintyin thedistance toGRS 1915+105 also plays a significant role in any attempt to tightly constrain REFERENCES the black hole mass. This is because the estimate for the primarymassstatedinSection3.1reliesheavilyontheincli- Bretthorst G. L., Bayesian Inference And Maximum En- nationofthesystem.Whilecitedbymanyauthorsas66±2◦ tropyMethodsInScienceAndEngineering:20thInterna- (McClintock et al. 2006;van Oers et al. 2010;Rahoui et al. tional Workshop. AIP Conference Proceedings, Volume 2010), this estimate is based in turn on an accurately 568, pp.241-245, 2001 The mass of the black hole in GRS 1915+105 7 Figure7.H-Bandspectrumconfirmingthepresenceofatomicfeatures.Top:Aspectrumfromasinglenights’observationwithGemini. All features indicated are found in each of the three spectra. Equivalent widths of selected lines can be found in Table 2. Bottom: TemplatestarHD83240fromthecatalogueofLebzelteretal.(2012) Casares, J., Black Holes from Stars to Galaxies – Across A.;Davis, S. W.; Li, L. X., 2006, ApJ,652, 518 the Range of Masses. Edited by V. Karas and G. Matt. Mirabel I. F., Rodr´ıguez L. F., 1994, Nature, 371, 46 Proceedings of IAU Symposium,238, pg3-12, 2007 Muno, M. P., Morgan, E. H., Remillard, R. A. 1999, ApJ, Castro-Tirado A.J., Brandt S., Lund N., Lapshov I., Sun- 527, 321 yaev R. A., Shlyapnikov A.A., Guziy S., Pavlenko E.P., Neil, E., T., Bailyn, D., Cobb, B., E., 2007, ApJ, 647, 409 1994, ApJS, 92, 469 Orosz, J. A., McClintock, J. E., Aufdenberg, J.P., et al., Castro-Tirado, A. J., Geballe, T. R., Lund,N. 1996, ApJ, 2011, ApJ, 742, 84 461, L99 Rahoui,F.,Chaty,S.,Rodriguez,J.,etal.,2010,ApJ,715, Chapuis C., Corbel S., 2004, A&A,414, 659 1191 Clark,J.S.,Steele,I.A.,Fender,R.P.,&Coe,M.J.1999, Rau, A.; Greiner, J., McCollough, M. L., 2003, ApJ, 590, A&A,348, 888 L37 Fender,R.P.,Garrington,S.T.,McKay,D.J.,Muxlow, Shahbaz, T., Ringwald, F. A., Bunn, J. C., et al., 1994, T. W. B. , Pooley, G. G., et al., 1999, MNRAS,304, 865 MNRAS,271, L10 Fich M., Blitz L., Stark A.A., 1989, ApJ,342, 272 Vacca, W. D., Cushing, M. C., Rayner, J. T. 2003, PASP, Fryer, C. L., Kalogera, V. 2001, ApJ, 554, 548 115, 389 vanOers,P.,Markoff,S.,Rahoui,F.,etal.;2010,MNRAS, Greiner, J., Cuby, J., McCaughrean, M. 2001b, Nature, 409, 763 414, 522 Greiner,J.,Cuby,J.,McCaughrean,M.,Castro-Tirado,A., Mennickent,R.2001a, A&A,373, L37 Harlaftis, E. T.; Greiner, J., 2004, A&A,414, L13 Howell,S.B.,Harrison,T.E.,Szkody,2004,ApJ,602,L49 Khargharia, J., Froning, C. S., & Robinson, E. L., 2010, ApJ,716, 1105 Lebzelter T., et al., 2012, A&A,539, A109 Le, H. A. N., Kang, W., Pak, S., Im M., Lee, J.E., Ho, L. C., Pyo, T. S.,Jaffe, D.T., 2011, arXiv, arXiv:1108.1499 Lee, J. C., Reynolds, C. S., Remillard, R., Schulz, N. S., Blackman, E. G., Fabian, A. C. 2002, Apj, 567, 1102 McClintock, J. E.; Shafee, R.; Narayan, R.; Remillard, R.