SS433's circumbinary ring and accretion disc viewed through its attenuating disc wind
aa r X i v : . [ a s t r o - ph . H E ] M a r Mon. Not. R. Astron. Soc. , 1–7 (2009) Printed 29 October 2018 (MN L A TEX style file v2.2)
SS 433’s circumbinary ring and accretion disc viewed through itsattenuating disc wind
Sebastian Perez ⋆ and Katherine M. Blundell University of Oxford, Department of Physics, Keble Road, Oxford, OX1 3RH, U.K.
ABSTRACT
We present optical spectroscopy of the microquasar SS 433 covering a significant fraction of aprecessional cycle of its jet axis. The components of the prominent stationary H α and H β linesare mainly identified as arising from three emitting regions: (i) a super-Eddington accretiondisc wind, in the form of a broad component accounting for most of the mass loss from thesystem, (ii) a circumbinary disc of material that we presume is being excreted through thebinary’s L2 point, and (iii) the accretion disc itself as two remarkably persistent components.The accretion disc components move with a Keplerian velocity of > ∼
600 km s − in the outerregion of the disc. A direct result of this decomposition is the determination of the accretiondisc size, whose outer radius attains ∼ R ⊙ in the case of Keplerian orbits around a blackhole mass of 10 M ⊙ . We determine an upper limit for the accretion disc inner to outer radiusratio in SS 433, R in /R out ∼ . , independent of the mass of the compact object. The Balmerdecrements, H α/ H β , are extracted from the appropriate stationary emission lines for eachcomponent of the system. The physical parameters of the gaseous components are derived.The circumbinary ring decrement seems to be quite constant throughout precessional phase,implying a constant electron density of log N e (cm − ) ≃ . for the circumbinary disc. Theaccretion disc wind shows a larger change in its decrements exhibiting a clear dependence onprecessional phase, implying a sinusoid variation in its electron density log N e (cm − ) alongour line-of-sight between 10 and 13. This dependence of density on direction suggests thatthe accretion disc wind is polloidal in nature. Key words: stars: individual: SS 433 – stars: winds, outflows – binaries: spectroscopic
Microquasars are X-ray binaries which undergo a wide rangeof physical processes including accretion onto a compact object(black hole or neutron star) and the launch of relativistic jets.SS 433, one of the most studied microquasars, became famous asthe first known source of relativistic jets in the Galaxy, and it isthe only system, X-ray binary or active galactic nucleus (AGN),for which atomic emission lines have so far been associated withthe jets, hence implying a baryonic content (i.e., e − p + , Milgrom1979; Crampton & Hutchings 1981; Fender et al. 2000).The optical spectrum of SS 433 is characterised by the pres-ence of numerous broad emission lines with complex profiles, ontop of a bright continuum. As at near-infrared and X-ray wave-lengths, these emission lines can be divided into two groups: linesthat are referred to as stationary , albeit highly variable in strengthand profile, and lines that are moving . The latter are thought tooriginate in the two oppositely-directed relativistic jets movingwith a speed v ∼ . c . The system SS 433 shows four mainperiodicities: the binary’s orbital motion, with a period of about ⋆ E-mail: [email protected] kinematical model .The stationary optical and near-infrared spectrum of SS 433is dominated by hydrogen and He I emission lines (Margon 1984)and at least 15 per-cent of the flux in the hydrogen lines is con-tributed by the accretion disc itself (Perez M. & Blundell 2009).The emission-line spectra of accretion discs, and their evolutionwith orbital and precessional phases, comprise most of the infor-mation we can obtain about the temperature and density variationsas well as velocity gradients within the disc (Skidmore et al. 2000).The Balmer decrements of the stationary lines in SS 433 havenot previously been studied due to the high interstellar extinctiontowards the object, which makes the detection of the H β line ratherdifficult (Panferov & Fabrika 1997). The decrements are highly de-pendent on physical parameters of the gas such as its tempera- c (cid:13) S. Perez and K. M. Blundell ture, optical depth and also the nature of the source of radiation(Drake & Ulrich 1980).The hard radiation field around compact objects and the ex-pected high electron densities for SS 433 ( N e > cm − ,Panferov & Fabrika 1997) require that we use an adequate treat-ment for such environments in order to study the emission-line gas.At high densities, and in the presence of heavy elements, excitationby collisional processes become a relevant factor (Ferland & Rees1988). Drake & Ulrich (1980) performed theoretical calculationsof the emission-line spectrum from a slab of hydrogen at moder-ate to high densities ( < N e < cm − ) over a wide rangeof physical parameters, including values close to those observedin SS 433’s gas (Panferov & Fabrika 1997). We compare our esti-mates with Drake & Ulrich’s findings in Section 4.The existence of dust mixed with the emitting gas wouldhave effects on the Balmer decrement that are by no means neg-ligible (Osterbrock & Ferland 2006). SS 433’s optical spectrum re-veals evidence of severe dust extinction, such as the remarkablyred continuum and the presence of prominent interstellar absorp-tion lines in the form of diffuse interstellar bands (Margon 1984). Inour optical data described in subsequent sections, and in agreementwith previous observations (e.g., Murdin et al. 1980; Margon 1984;Gies et al. 2002, among others), we have detected the diffuse inter-stellar bands at 4430, 5778 and 5780 ˚A and also the interstellar linesCa H λ , Ca K λ as well as Na D λ . Murdin et al.(1980) reported an interstellar absorption (i.e., Galactic extinction)of A V ∼ mag toward SS 433, obtained from infrared measure-ments. This result has been corroborated by Gies et al. (2002) byfitting the spectral energy distribution, and the currently most ac-cepted value is A V = 7 . mag.Fig. 1 shows the map of Galactic extinction in the directionof the SS 433/W50 complex. We constructed this extinction im-age by converting the colour excess map of Schlegel et al. (1998)into an estimate of the reddening, assuming a selective extinction, R V ≡ A V /E ( B − V ) , equal to the average value for the Galaxy(i.e., R V = 3 . , Cardelli et al. 1989), where E ( B − V ) is thecolour excess. It is clear from the image that the ∼ A V = 7 . mag. Although the value for E ( B − V ) is quiteaccurate, the choice of R V is not. The intrinsic error introduced by R V implies that the accuracy of the extinction estimate is not betterthan to a tenth of a magnitude. Therefore, we corrected our spec-tra (whose reduction is described in Section 2) for A V = 7 . magof interstellar absorption using the reddening law of Cardelli et al.(1989), before calculating the Balmer decrements.In this paper we decompose the profiles of the stationary emis-sion lines, H α and H β , with a number of Gaussian components(Section 3). Each model component is identified with its corre-sponding emitting origin. In Section 4 we analyse the physical con-ditions of each component present in SS 433 and its behaviour as afunction of orbital and precessional phases, via the Balmer decre-ment. In Section 5 we present our calculations for the accretiondisc radii and compare our findings with discs from other objectsthat share some common features with SS 433. We present optical spectroscopy data covering most of a preces-sional cycle of SS 433’s jet-axis (and presumably therefore accre-tion disc). The data were acquired with the Supernova Integral Field
Figure 1.
Colour-scale image shows the Galactic extinction ( A V ) to-wards the W50/SS 433 system from the IRAS/COBE all-sky survey(Schlegel et al. 1998). Colour scale corresponds to visual extinction in mag-nitudes and it is centred on SS 433’s position (given by the red circle). Greycontours on the colour image are 3, 3.6, 5.4, 8.4, 12.6 and 18 mag. Theradio continuum emission at 1465 MHz is shown as black contours (datafrom Dubner et al. 1998). Radio contours are 12, 14 and 20 mJy beam − . Spectrograph (SNIFS) at the University of Hawaii 2.2-m telescopeby the Nearby Supernova Factory (Aldering et al. 2002). The spec-trograph is composed of two modules, one for blue wavelengthscovering the region between 320 nm to 560 nm with a resolutionof ∼ ∼ ψ pre ∈ [0 . , . ), and an-other one between 2006 October 1 and November 10 (precessionalphases ψ pre ∈ [0 . , . ). We use the convention in which or-bital phase ( φ orb ) zero is when the donor star is eclipsing out thecompact object (Goranskii et al. 1998). Precessional phases are cal-culated based on the ephemeris reported in Fabrika (2004), whereprecession phase zero is when the moving jet lines are maximallyseparated hence the inclination of the jet axis with our line-of-sightattains a minimum, i.e., it corresponds to maximum exposure of theaccretion disc to the observer.The data were reduced by the Nearby Supernova Factorygroup through their standard pipeline. Frequent observations ofstandard stars and the arc lamp were performed during eachnight to perform flux calibration and provide an accurate esti-mate of the wavelength axis. Wavelength calibration was carriedout by fitting a high-order polynomial to the lamp spectra. Stan-dard star division was applied to each frame in order to flux cal-ibrate the data. Telluric features were not removed. All the sub-sequent analyses were carried out using the Perl Data Language( http://pdl.perl.org, Glazebrook et al. 1997).
Examples of the reduced spectra in the vicinity of the H α and H β lines are shown in Fig. 2. This figure also shows the best combi-nation of Gaussian components fitted to those spectra. The blue c (cid:13) , 1–7 almer decrements in SS 433 Figure 2.
Examples of H β (upper panel) and H α (lower panel) so-calledstationary emission lines. Both spectra were acquired on MJD 53009.23,which corresponds to ψ pre = 0 . and φ orb = 0 . (for a description ofthe convention used please see text in Section 2). The different componentsare explained in the text. The narrow blue- (dot-dashed line) and red-shifted(dashed line) Gaussian components correspond to the accretion disc’s flow.The top x -axis corresponds to rest velocity in units of km s − and both y -axes are in flux density units of W m − µ m − . The broad wind compo-nent has a FWHM of ∼ km s − and ∼ km s − for H β and H α ,respectively. The error bars correspond to 3 σ , which are about half the sym-bol size in the case of the H α line. The shaded area represents that whichwe consider to be the wings of the accretion disc profile (see Section 5). spectra were rebinned to the spectral resolution of the red spectra(130 km s − pixel − ). The underlying continuum was subtractedby fitting a low-order polynomial (linear in most cases and only insome cases parabolic). The best-fitting parameters were determinedby χ minimisation in the pixel space. We explain the fitting modeland the deconstruction of these profiles in the next section. The stationary emission lines in SS 433 possess complex andvariable profiles (Blundell et al. 2008; Perez M. & Blundell 2009).Falomo et al. (1987) fitted the stationary H α line profile of theirlow-resolution spectra using three to four Gaussian components.Their best fit consisted of one broad component plus three nar-rower components. Blundell et al. (2008) carried out decomposi-tion of the H α line using higher-resolution data. They found onebroad component (FWHM ∼
700 km s − ) whose width was ob-served to decrease with precessional phase (i.e., as the jets becomemore in the plane of the sky) identified as the accretion disc wind, and two narrower, red- and blue-shifted, components which werestationary in wavelength, being radiated from a glowing circumbi-nary ring. In the data presented in that paper, there was no evidenceof the accretion disc lines being apparent, in contrast with their per-sistent appearance in the data presented here.Recently, Perez M. & Blundell (2009) were able to model thecomplex Brackett- γ (Br γ ) stationary line with usually 5 Gaussiancomponents: a broad wind component (FWHM ∼ − ),two Gaussian components accounting for the circumbinary ringand they also found the presence of two components being radi-ated by matter spiralling in the accretion disc. The presence ofSS 433’s accretion disc in the near-infrared line seems to be per-sistent and their separation implies a circular velocity between 500and 700 km s − . We modelled our optical data following a similarapproach.Following the analysis done by Perez M. & Blundell (2009),the H α and H β stationary lines were fitted with four Gaussiancomponents in order to account for the complexity of the profiles.Although two narrow components have been used in the past torepresent the split lines from the circumbinary ring (Blundell et al.2008; Perez M. & Blundell 2009), because of the low resolution ofthe spectra we present in this paper ( ∼ km s − ) just one centralnarrow component was needed to represent the contribution fromthe circumbinary. Fig. 2 shows examples of fitted profiles.The presence of a P Cygni feature in the stationary linesat certain precessional phases has been noted by several authors(Crampton & Hutchings 1981; Filippenko et al. 1988; Gies et al.2002). An absorption feature in the blue wing of the stationaryline profile would indeed complicate the analysis and it would haveto be taken into account using models of outflowing winds (e.g.,Castor & Lamers 1979). We detected the P Cygni absorption sig-nature at the epochs ψ pre ∈ [0 . , . and we have excluded thesedata from our analysis.The reason why we identify the accretion disc with two sep-arate components representing each peak of the disc’s profile isbased on the model of emission line formation in accretion discspresented by Horne & Marsh (1986), in which the velocity of theouter regions of the disc is represented by half the velocity separa-tion of the line peaks (Horne & Marsh 1986). This is caused by theDoppler effect and the rotational motion of the matter within thedisc. Therefore, the speed with which the radiating material spiralswithin the accretion disc corresponds to half the difference betweenthe speed at which those lines are moving (given by the relative po-sition of the peaks), under the assumption that the fitted centroidscorrespond to the tangent speed. This reveals material that is orbit-ing in the potential well at speeds of about > ∼ km s − (see blueand red tracks in Fig. 3).Fig. 3 does not reveal a clear signature of the expected motionfor the accretion disc centroid and certainly nothing correspondingto the Fabrika & Bychkova (1990) sinusoidal plot that assumes acircular orbit. It is not possible to identify what would be the correctphasing of the motion of the disc due to lack of knowledge aboutthe circularity or eccentricity of the orbit. For certain eccentrici-ties, there are not necessarily two well-defined peaks, for example.A detailed analysis of this problem requires much more long-termmonitoring and sophisticated modelling.Fig. 3 (left and centre panels) clearly shows that the accre-tion disc lines appear slower at orbital phase zero. This is whenthe donor star obscures the inner (faster) region of the accretiondisc. This is seen most clearly in H α . Additionally, there appearsto be an episode around MJD 53930, which corresponds to orbitalphase 0.5, where the accretion disc lines seem to be rotating faster c (cid:13) , 1–7 S. Perez and K. M. Blundell
Figure 3.
Tracks of the fitted centroids for each component of the stationary H β ( left ) and H α ( middle ) lines as a function of modified Julian day ( y -axis,increases vertically). The tick mark heights are proportional to the FWHM of each component. Red and blue ticks correspond to the accretion disc lines, thelarger black ticks represent the disc wind while the light blue ones represent the circumbinary ring. The top x -axis corresponds to rest velocity in units of km s − for an assumed systemic velocity of zero. The dot-dashed sinusoidal depicts the orbital motion of the binary. The dotted horizontal lines correspondto epochs when the orbital phase is zero. The plot on the right hand side corresponds to the evolution of the FWHM of the wind component of the stationaryH β (gold triangles) and H α (blue circles) lines. than usual and may simply correspond to a time when we have anunimpeded view into the inner, hence faster, part of the accretiondisc. The right panel in Fig. 3 shows the evolution of the FWHMof the wind components of the stationary H α and H β lines. Thispanel also shows that the wind lines appear to get broader at longerwavelengths, which is consistent with previous near-infrared ob-servations of the Br γ line, where the wind component has beenreported to be broader than in H α (Perez M. & Blundell 2009). Itis remarkable that near the end of the data set we present in thispaper, the width of the accretion disc wind attains a velocity of al-most 2500 km s − , which translates into a factor 1.7 increase in the10 − M ⊙ yr − mass-loss rate estimated from recombination linefluxes by Perez M. & Blundell (2009). Modelling of the stationary Balmer emission lines, H α (thiswork and Blundell et al. 2008), H β (this work) and Br γ (Perez M. & Blundell 2009), yields a clear and consistent interpre-tation of the radiating components present in the SS 433 system.The main components of the system, as inferred from the decom- position of the stationary line complexes, are: a rotating accretiondisc, a fast outflowing wind that accounts for most of the mass lossin the system and a circumbinary ring probably being fed from theL2 point. The moving emission lines from the jets sometimes blendwith the stationary ones at around ψ pre ∼ . or 0.4, contributingsignificantly to the intensity variation and profile shape. Moreover,care must be taken since the intensities and profiles of the stationarylines vary strongly during flares (Blundell et al. in prep., Margon1984). The hydrogen Balmer lines show very variable Balmer decrementsin SS 433. Fig. 4 shows that the computed Balmer decrements forthe stationary emission-lines scatter from < ∼ H α/ H β < ∼ . In theinterpretation of these results we use the calculations carried outby Drake & Ulrich (1980) who modelled the emission-line spec-trum from a slab of hydrogen at high densities, by taking into ac-count important mechanisms such as collisional excitation and de-excitation, as well as self-absorption processes. It is important tochose reasonable temperature and optical depth ( τ L α ) parametervalues in order to apply their calculations to SS 433. We mainlyutilise the decrements computed for a model with T e > ∼ K. For a c (cid:13) , 1–7 almer decrements in SS 433 Figure 4.
The Balmer decrement, H α /H β , as a function of precessionalphase (top x -axis) or modified Julian day (bottom x -axis), for the accretiondisc wind (top), circumbinary ring (middle panel) and accretion disc lines(bottom). The dot-dashed line corresponds to the best fitted sinusoidal tothe disc wind decrements, with a fixed period of 162 days. In the bottomplot the red- and blue-shifted accretion disc lines are represented by solidtriangles and circles, respectively. The vertical dotted lines represent timesat which the orbital phase of the binary is zero. theoretically-modelled accretion flow, Kramer’s approximation forthe optical depth yields τ L α = 4 × (Drake & Ulrich 1980).Since decrements for this exact value are not reported in their paper,we use τ L α ∼ instead. According to Drake & Ulrich (1980)these ratios ( < ∼ H α/ H β < ∼ ) then imply up to four orders of mag-nitude variation in density throughout the precession period. Thislarge inferred variation is not a problem for plausible physical mod-els. We will analyse and discuss the Balmer decrements calculatedfor each component of SS 433 in the following sub-sections. The Balmer line ratios of the components corresponding to the cir-cumbinary excretion disc observed by Blundell et al. (2008) areshown in the middle panel of Fig. 4. They are roughly constantthroughout precessional phase with an average value of H α /H β =2 . ± . (1 σ scatter). This implies a moderate electron densityof log N e ≃ . , approximately constant throughout the preces-sional cycle and given the similarity of this Balmer decrement to theCase B canonical value of ∼ circumbi-nary nature of this emitting region (Doolin & Blundell 2009). Fig. 4 shows that in the case of the accretion disc wind, theBalmer decrements have a very clear precessional phase depen-dence, showing a clear tendency towards lower electron density (steeper decrements) as the jet axis becomes more in the plane ofthe sky. This dependence of density on direction is evidence forthe polloidal nature of the wind. The decrement values for the discwind vary between quite flat H α /H β = 1 . ± . between pre-cessional phases 0.95–0.05 and steep H α /H β = 5 . ± . forphases 0.65–0.84. These values correspond to variations in electrondensity of log N e ≃ and log N e ≃ , respectively. The accretion disc lines are the ones that show the most dra-matic and scattered fluctuations in decrement. Both lines, H α andH β , show a similar level of scatter. The average decrement forthe red and blue components are H α /H β = 2 . ± . andH α /H β = 3 . ± . , respectively. Inverted Balmer decrementscan be seen at some epochs, implying very high densities and pos-sibly an enhanced inflow of matter from the companion onto theaccretion disc. The decrement of the accretion disc itself appearsto be dominated by fluctuations in the instantaneous wind densityrather than being dominated by any precessional phase dependence. A direct consequence of Horne & Marsh’s 1986 model of the emis-sion line profile formed in an accretion disc is that the emissionclose to the inner radius of the disc contributes to the outer wings ofthe double-peaked profile, while the regions nearer the outer radiusprovide the emission for the peaks (see fig. 1 in Horne & Marsh1986). By considering this along with the assumption that the ve-locity of the matter spiralling in the accretion disc is Keplerian, wecan estimate the size of the disc (Mason et al. 2000).For a test particle in Keplerian motion at a distance r from acentral object of mass M bh , the Keplerian speed projected onto theplane of the sky is V Kep = r GM bh r sin i, (1)where i is the angle between the normal to the disc and our line-of-sight. The disc is edge-on when this inclination angle is ◦ andpole-on when i = 0 ◦ . In our decomposition of the Balmer lines’profiles, the Keplerian velocity of the gas at the outer radius R out of the disc is given by half the separation between the peaks of thetwo Gaussian components representing the accretion disc, whilethe Keplerian velocity of the gas closer to the disc inner radius R in corresponds to half the separation between the red wing of the red-shifted component and the blue wing of the blue-shifted component(see hatched area in Fig. 2).The most suitable epoch to measure the inner and outer radiiof the accretion disc is when the disc is in between the companionstar and the observer (orbital phase 0.5) and when the disc is nottotally edge-on (i.e., sin i ∼ ) at precessional phase ψ prec closeto 0.33 or 0.66. The closest spectrum in our dataset to those epochswas taken on MJD 54009 which corresponds to ψ prec = 0 . and φ orb = 0 . . This spectrum is depicted in Fig. 2. Both the halfwing separation and the half peak separation for the accretion disclines were measured from this spectrum.Half the separation between the peaks of the fitted Gaussiancomponents of the disc yields a Keplerian velocity for the outerpart of the disc of ∼ km s − , measured from both the H α andH β emission. The Keplerian velocity of the inner part of the discwas determined by measuring the positions in velocity of the line c (cid:13) , 1–7 S. Perez and K. M. Blundell
Figure 5.
Variation of the outer radius of the accretion disc as a function ofthe Keplerian velocity of the outer edge of the disc (half the separation ofthe line peaks, in km s − ), for different values of the mass of the compactobject. The solid area represents the range of parameters that we think arerelevant to SS 433. The thick orange line corresponds to M bh = 10 M ⊙ . wings (centre of the hatched areas in Fig. 2) by inspection. BothH α and H β present wing velocities of ∼ km s − .The central compact object in SS 433 is most likely to be astellar mass black-hole, rather than a neutron star (Gies et al. 2002;Fabrika 2004; Lopez et al. 2006; Blundell et al. 2008). It is evidentfrom Equation 1 that large black-hole masses imply larger outerradii for the accretion disc. The total mass of the accretion disc plusthe compact object has been calculated by Blundell et al. (2008)and it attains 16 M ⊙ . However, we point out that there has been alarge dispersion reported for the component masses of SS 433 (e.g.,Lopez et al. 2006). We will assume that the mass of the black holeitself is of the order of 10 M ⊙ .The Keplerian velocity for the inner region of the disc of1200 km s − implies that the accretion disc’s inner radius is R in ∼ . M R ⊙ , where M = M bh /
10 M ⊙ . The estimatedvalue for R in corresponds to an upper limit to the inner radius ofthe accretion disc at the time of observation. This is a direct conse-quence of the high obscuration presented in the system due to theaccretion disc wind, which hinders the emission propagating fromthe inner parts of the disc. Additionally, the determination of thefull-width at zero intensity (FWZI), or distance between the wings,could be imprecise sometimes and it could add certain degree ofuncertainty to the R in value. Also, it is important to note that theinner radius of an accretion disc is known to recede as a conse-quence of changes in the accretion rate onto the compact object,for a range of masses of the compact object.On the other hand, the distance between the peaks of the ac-cretion disc components depicted in Fig. 2 yields a velocity for thedisc outer region of about 620 km s − , which if Keplerian yields R out ∼ M R ⊙ . Fig. 5 shows how R out varies as a functionof the Keplerian velocity of the outer edge. This is fully consis-tent with a previous estimate of the overall size of the system,whose semi-major axis of the orbit has been reported to be ∼ R ⊙ (Perez M. & Blundell 2009).The Keplerian velocity of the outer region of the accretiondisc has been reported to be lower in the near-infrared Br γ line(Perez M. & Blundell 2009) than in the optical lines analysed inthis paper. This is consistent with observations of other accretiondiscs such as the one in WZ Sge. By analysing Doppler tomographymaps, Skidmore et al. (2000) found that the Br γ emission is being emitted from a region further away from the central object, imply-ing that this emission is more representative of the outer regions ofthe disc. This is probably the case in SS 433 since the peak sepa-ration in Br γ is smaller than in the Balmer lines. The reported halfseparation of the peaks of ∼ km s − by Perez M. & Blundell(2009) implies an outer radius R out = 8 M R ⊙ .The black-hole mass does not affect the inner to outer radiusratio, since R in /R out = ( V out /V in ) . In the case of SS 433’s disc, R in /R out ≈ . (for an outer radius R out ∼ M R ⊙ ).The size of SS 433’s accretion disc is quite large comparedwith other close binary systems. Mason et al. (2000) determinedthe size of WZ Sge to be at most 0.43 R ⊙ for a white dwarf with theChandrasekhar mass. This is about 20 times smaller than SS 433’sdisc. The R in : R out ratio is approximately the same in both thesesystems. We have presented optical spectroscopy of the microquasar SS 433covering a significant fraction of the precessional cycle of itsaccretion disc. The components of the prominent stationary H α and H β lines have been identified as arising from three emit-ting regions: an accretion disc wind which is super-Eddington(Perez M. & Blundell 2009), in the form of a broad component ac-counting for most of the mass loss in the system, a circumbinarydisc of material probably excreted through the binary’s L2 point,and the accretion disc itself as two persistent components, havingan outer region Keplerian velocity of > ∼
600 km s − .A direct result of this decomposition using our UKIRT datapublished in Perez M. & Blundell (2009) was the determination ofthe accretion disc size, whose outer radius attains 8 R ⊙ , for anassumed black hole mass of 10 M ⊙ . With the data presented in thispaper we determined the accretion disc inner to outer radius ratioin SS 433, R in /R out to be ∼ α/ H β , were extracted from thestationary emission lines for each component of the system. Thedecrement of the circumbinary ring seems to be quite constantthroughout precessional phase, implying a fairly constant electrondensity of log N e ≃ . for the circumbinary disc. The accretiondisc wind shows larger changes in its decrements as a function ofprecessional phase, implying variations in its density, N e , between and cm − . Thus, the physical parameters of the gaseouscomponents imply rather dense environments emitting the Balmerlines. ACKNOWLEDGMENTS
We thank the referee for a careful reading of the paper. We are verygrateful to the Sciences and Technology Facilities Council (
STFC )and Conicyt for the award of a STFC-Gemini studentship. We arespecially grateful to Greg Aldering, the Nearby Supernova Fac-tory collaboration and the University of Hawaii for their generosityin allowing us to obtain some spectra on SS 433 during intersti-tial time within their main supernovae follow up campaign. K. B.thanks the Royal Society for a University Research Fellowship.
REFERENCES
Aldering, G., et al. 2002, SPIE, 4836, 61 c (cid:13) , 1–7 almer decrements in SS 433 Blundell, K. M., Bowler, M. G., & Schmidtobreick, L. 2008, ApJ,678, L47Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345,245Castor, J. I., & Lamers, H. J. G. L. M. 1979, ApJS, 39, 481Crampton, D., & Hutchings, J. B. 1981, ApJ, 251, 604Doolin, S., & Blundell, K. M. 2009, ApJL, 698, L23Drake, S. A., & Ulrich, R. K. 1980, ApJS, 42, 351Dubner, G. M., Holdaway, M., Goss, W. M., & Mirabel, I. F. 1998,AJ, 116, 1842Fabian, A. C., & Rees, M. J. 1979, MNRAS, 187, 13PFabrika, S. N., & Bychkova, L. V. 1990, A&A, 240, L5Fabrika, S. 2004, Astrophysics and Space Physics Reviews, 12, 1Falomo, R., Boksenberg, A., Tanzi, E. G., Tarenghi, M., & Treves,A. 1987, MNRAS, 224, 323Fender, R., Rayner, D., Norris, R., Sault, R. J., & Pooley, G. 2000,ApJL, 530, L29Ferland, G. J., & Rees, M. J. 1988, ApJ, 332, 141Filippenko, A. V., Romani, R. W., Sargent, W. L. W., & Blandford,R. D. 1988, AJ, 96, 242Gies, D. R., Huang, W., & McSwain, M. V. 2002, ApJ, 578, L67The Perl Data Language, Glazebrook K., Economou F., The PerlJournal, 1997, Issue 5, 5 - also reprinted in Dr Dobbs Journal,Fall 1997 issueGoranskii, V. P., Esipov, V. F., & Cherepashchuk, A. M. 1998,Astronomy Reports, 42, 209Horne, K., & Marsh, T. R. 1986, MNRAS, 218, 761Katz, J. I., Anderson, S. F., Grandi, S. A., & Margon, B. 1982,ApJ, 260, 780Lopez, L. A., Marshall, H. L., Canizares, C. R., Schulz, N. S., &Kane, J. F. 2006, ApJ, 650, 338Margon, B. 1984, ARA&A, 22, 507Mason, E., Skidmore, W., Howell, S. B., Ciardi, D. R., Littlefair,S., & Dhillon, V. S. 2000, MNRAS, 318, 440Milgrom, M. 1979, A&A, 76, L3Murdin, P., Clark, D. H., & Martin, P. G. 1980, MNRAS, 193, 135Osterbrock, D. E., & Ferland, G. J. 2006, Astrophysics of gaseousnebulae and active galactic nuclei, 2nd. ed. by D.E. Osterbrockand G.J. Ferland. Sausalito, CA: University Science Books,2006,Panferov, A. A., & Fabrika, S. N. 1997, Astronomy Reports, 41,506Perez M., S., & Blundell, K. M. 2009, MNRAS, 397, 849Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500,525Skidmore, W., Mason, E., Howell, S. B., Ciardi, D. R., Littlefair,S., & Dhillon, V. S. 2000, MNRAS, 318, 429This paper has been typeset from a TEX/ L A TEX file prepared by theauthor. c (cid:13)000