WATCHDOG: A COMPREHENSIVE ALL-SKY DATABASE OF GALACTIC BLACK HOLE X-RAY BINARIES

Here we outline four different methods (discussed in detail in Jonker & Nelemans 2004) to estimate distance to a XRB system. First, a direct, model-independent distance can be obtained using trigonometric parallax. While this method may provide the most accurate estimates of distance, for many sources this method is not feasible. XRBs may be (i) located several kpc away, and therefore require sub-milliarcsecond astrometry to measure their parallaxes² , (ii) too faint to be detected at radio wavelengths, or (iii) located in the Galactic plane, where high-precision astrometry is next to impossible due to scatter broadening along the line of sight (Miller-Jones et al. 2009). Currently, there are only three trigonometric parallax distances to BH systems (Miller-Jones et al. 2009; Reid et al. 2011, 2014).

Second, distance can be estimated by comparing the absolute magnitude to the (dereddened) apparent magnitude of the counterpart star, taking into account a possible contribution from residual accretion (Jonker & Nelemans 2004). In this case, the absolute magnitude can be determined by either: (i) determining the spectral type by fitting the data (e.g., Marsh et al. 1994), and assuming the absolute magnitude is that of a main-sequence star of this spectral type; or (ii) determining radius, spectral type and luminosity class from the data, and then using the surface brightness of the determined spectral type (e.g., from Barnes & Evans 1976; Popper 1980), or a combination of the determined radius and effective temperature along with an appropriate bolometric correction, to estimate absolute magnitude. See, for example, GS 2000+251 (Barret et al. 1996b) or 1A 0620−00 (Gelino et al. 2001b).

Third, limits can be placed on the distance to a source using the observed proper motions of approaching and receding jet ejections. See Mirabel & Rodríguez (1999) for a detailed description of this method and Hjellming & Rupen (1995) for an example of this method being employed for GRO J1655−40.

Finally, the interstellar absorption properties of a source may also be used to determine distance. In this case, distance can be estimated by either: (i) using the correlation between the observed equivalent widths of interstellar absorption lines and diffuse interstellar bands and color excess (Herbig 1995) and then converting the color excess to distance (e.g., see Beals & Oke 1953); or (ii) tracing the movement of individual gas clouds in velocity space through high-resolution spectroscopic observations of interstellar absorption lines and associating them with distance by assuming the velocity is due to Galactic rotation. See for example GX 339−4 (Hynes et al. 2004).

During quiescence, optical/infrared (OIR) observations of BHXBs permit detailed studies of the binary counterpart, allowing for the determination of key orbital parameters involved in the mass measurement of the compact object. Once the counterpart has been identified, the orbital period (${P}_{{\rm{orb}}}$) can be measured from periodicity in either photometry (X-ray or optical/infrared), or radial velocity variations. Once the orbital period is known, radial velocity measurements of the secondary star can give the mass function,

$$\begin{eqnarray}&&f(M)=\displaystyle \frac{{P}_{{\rm{orb}}}}{2\pi G}{({K}_{2})}^{3}=\displaystyle \frac{{M}_{1}{\mathrm{sin}}^{3}i}{{({M}_{1}+{M}_{2})}^{2}},\end{eqnarray} \tag{ 1 }$$

where ${K}_{2}={v}_{2}\mathrm{sin}i$, the semi-amplitude of the radial velocity curve. The mass function gives a minimum mass of the compact object (M₁). A mass function greater then 3M_⊙ (Rhoades & Ruffini 1974; Kalogera & Baym 1996) proves that the compact object is a BH. Next, either measuring ${v}_{2}\mathrm{sin}i$ (the radial velocity of the compact object, by tracing the motion of the accretion disk), or independently measuring the mass of the counterpart star (for instance, by spectral typing and assuming it is a main-sequence star), allows for calculation of the mass ratio ($q={M}_{2}/{M}_{1}$). Finally, analysis of ellipsoidal variability in the photometric light curve provides a method to determine the inclination i. The ellipsoidal variability occurs as a result of gravitational distortion of the counterpart star, causing the projected area and average temperature seen by the observer to vary differently with orbital phase depending on the inclination. See for example Charles & Coe (2006) for a detailed discussion of the entire process.

Recently, however, the accuracy of some existing mass measurements for individual BHXBs has been brought into question (Cantrell et al. 2010; Kreidberg et al. 2012). It has been suggested that the most commonly used model to analyze ellipsoidal variability (i.e., the "star-only" model), which works under the assumption that sources of light not due to the star are negligible, may result in mass estimates that are inaccurate by substantial amounts (Kreidberg et al. 2012). This argument stems from the numerous observations of a non-stellar flux component contributing a significant fraction of the total flux of the system at both optical (Zurita et al. 2002; Orosz et al. 2004; Cantrell et al. 2010) and IR (Hynes et al. 2005; Gelino et al. 2010) wavelengths, implying that OIR spectra of quiescent BHs are likely shaped by a number of competing emission mechanisms.

Both direct thermal emission from the outer disc (Cantrell et al. 2010; Gelino et al. 2010) and synchrotron emission from a relativistic jet (Russell et al. 2006; Gallo et al. 2007) have been suggested as culprits for this flux contamination. This non-stellar contribution is not modulated (as the counterpart star is) leading to flattening of the light curve (reducing the size of the modulation). This flattening tends to bias the inferred inclinations toward lower values, and can lead to differences of up to a factor of 2 in the derived mass of a compact object (Gelino et al. 2001b). As such, orbital inclination is by far the largest source of systematic error involved in estimating the mass of the BH. In addition, as an accurate measurement of inclination relies on the ability to quantify the ratio of non-stellar flux to total flux, distinguish between the non-stellar emission mechanisms and characterize the degree to which these individual sources of emission contribute to the OIR light, it is also the most difficult orbital parameter to measure.

Fortunately, Kreidberg et al. (2012) recently published a study in which they characterize the systematic error caused by the effects discussed above. In particular, they build on the study of 1A 0620−00 quiescent light curves by Cantrell et al. (2008), who define two separate states existent in the quiescent optical light curves, (i) passive: displaying minimum aperiodic variability, resulting in a stable light curve shape over short timescales; and (ii) active: brighter, bluer, and more variable than the passive state, possibly as a result of increased accretion activity. Doing so has allowed them to use inclination estimates and light curves in the literature to assess the most probable value of inclination for 16 BHXB systems. We have made extensive use of this study in determining the orbital parameters used in our analysis.

XTE J0421+560 was discovered by the All-Sky Monitor (ASM) on board RXTE, when it underwent its first and only outburst in 1998 (Smith et al. 1998a). The outburst peaked first in X-rays followed by optical and radio wavelengths, a behavior commonly associated with XRN. Several soft X-ray flares were observed after the initial flare (Frontera et al. 1998). The optical counterpart, CI Cam, is a B0-2 supergiant Be star (assuming a BH or NS accretor; Robinson et al. 2002), so the system is a HMXB.

The distance is quite uncertain, with distance estimates ranging from 1 kpc (Barsukova et al. 2006) to $\gt 10\;{\rm{kpc}}$ (Robinson et al. 2002), and thus the X-ray luminosity is poorly constrained. This large uncertainty in luminosity has impacted accurate determination of the nature of both the mass donor and the accretor (Bartlett et al. 2013). If the distance is >2 kpc, the resulting X-ray luminosity would indicate the presence of a NS or BH (Belloni et al. 1999a), while a closer distance would suggest a white dwarf accretor (Orlandini et al. 2000; Ishida et al. 2004) and a ${\rm{B}}{4}_{\mathrm{III}-\mathrm{IV}}$ counterpart (Barsukova et al. 2006). Despite the uncertainty in its nature, we include this system in our BHC sample because its X-ray properties are similar to known BHXBs.

GROJ0422+32 is an LMXB discovered in 1992 by BATSE on the Compton Gamma-ray Observatory (CGRO) when it underwent a FRED type outburst (Paciesas et al. 1992). The spectrum was well described by a hard power-law (Sunyaev et al. 1994) and timing analysis revealed properties commonly associated with the HCS (van der Hooft et al. 1999a), indicating that this was a "hard-only" outburst. The X-ray and optical light curves of this outburst were similar to 1A 0620−00, except that they showed two unusual optical mini-outbursts following the main burst in 1993, which were not detected in hard X-rays (Shrader et al. 1997; Callanan et al. 1995). A series of mini-outbursts, with a recurrence period of ∼120 days, were also detected in 1997 (Iyudin & Haberl 1997).

Rigorous determination of binary parameters showed this system to contain a BH primary (Casares et al. 1995b; Filippenko et al. 1995) and an M1–4 V optical counterpart (Harlaftis et al. 1999; Webb et al. 2000). The detection of a radio counterpart and evolution of its spectrum is discussed by Shrader et al. (1994). We adopt the distance derived by Gelino & Harrison (2003) of 2.49 ± 0.30 kpc.

There have been numerous discussions of the inclination of GRO J0422+32. Orosz & Bailyn (1995), Gelino & Harrison (2003), Reynolds et al. (2007), Filippenko et al. (1995) all estimate $i\gt 45^\circ$. However, (i) the light curve of Orosz & Bailyn (1995) exhibits shape changes between two different nights and a difference in mean I magnitude of 0.05 suggesting it is probably in the active state (i.e., variable and poorly defined light curve shape as a result of contributions from the accretion disk and/or jet; see Cantrell et al. 2010), (ii) Gelino & Harrison (2003) and Reynolds et al. (2007) find similar mean magnitudes in the H and K light curves, but because there is evidence for IR disk contamination in the former and no detection of ellipsoidal variability in the light curves of the latter, the conflicting results suggest that both authors' light curves are in the active state, and, (iii) while Filippenko et al. (1995) find an $i=48^\circ \pm 3^\circ$, which is consistent with the above three results, they assume a normal mass M2V secondary.

As discussed in Kreidberg et al. (2012), only lower limits on the inclination may be derived if the light curves are in the active state, or if the secondary star is assumed to have a mass equal to or less than that of a main-sequence star with the same radius. The latter limitation stems from the fact that the relationship between mass, radius, and surface temperature for a star which fills its Roche-lobe may be very different from that of a spherical star. Thus, if q is known, assuming a standard mass for the secondary star's spectral type (i.e., an upper limit) will result in overestimated M₂, and therefore an overestimated ${M}_{{\rm{BH}}}$. If ${M}_{{\rm{BH}}}$ is overestimated, then i will be underestimated (for a fixed q and $f(M)$).

Several authors measure $i\lt 45^\circ$, including Casares et al. (1995a), Callanan et al. (1996), Beekman et al. (1997). However, all three make use of binned light curves. The act of binning light curves may flatten their shape, therefore implying a lower inclination. We therefore adopt the value of $i=63\buildrel{\circ}\over{.} 7$ calculated by Kreidberg et al. (2012) by adjusting the $i=45^\circ$ value assuming the source is active. We adopt a ${M}_{{\rm{BH}}}$ from this corrected inclination.

Commonly known as LMC X-3, it was discovered by the UHURU satellite in 1971 (Leong et al. 1971). LMC X-3 is one of three known persistently accreting, dynamically confirmed BHs in the Milky Way and Magellanic Clouds (McClintock & Remillard 2006). From the large mass function, the absence of X-ray eclipses, and an estimated mass of the B3V optical counterpart, Cowley et al. (1983) calculated a lower limit on the mass, confirming the presence of a BH (Kuiper et al. 1988; Orosz et al. 2014).

LMC X-3 almost continuously maintains itself in a bright state like persistent HMXB wind-fed systems, but the BH is actually fed by Roche lobe overflow, similar to transient systems (Steiner et al. 2014). The X-ray spectrum is nearly constantly thermal and disk-dominated, spending most of its time in the SDS (Treves et al. 1988; Ebisawa et al. 1993; Nowak et al. 2001) with occasional transitions to the HCS (Wilms et al. 2001; Smale & Boyd 2012). Due to its location in the Large Magellanic Cloud (LMC), the distance is well constrained (Cowley et al. 1983).

Recently, Orosz et al. (2014) established a new dynamical model for the system. More high quality radial velocity data (yielding an improved determination of K-velocity) and an accurate measurement of the projected rotational velocity for the companion star, along with a much larger array of ellipsoidal light curves, have yielded a more precise ${M}_{{\rm{BH}}},q$ and i than previous studies.

Commonly known as LMC X-1, it is among the three known persistently accreting, dynamically confirmed BHs in the Milky Way and Magellanic Clouds (McClintock & Remillard 2006). Spectroscopic studies of the optical counterpart (Hutchings et al. 1983, 1987) established it as a firm BHC. However, the presence of a BH in this system was not confirmed until an accurate measurement of the mass function was made (Haardt et al. 2001).

Similar to LMC X-3, LMC X-1 has been observed to spend most of its time in the SDS (Ebisawa et al. 1993; Nowak et al. 2001). Given its location in the LMC (Cowley et al. 1983), the distance to the system is well determined (Freedman et al. 2001).

Orosz et al. (2009) have improved the dynamical model of Hutchings et al. (1987). High and medium resolution spectra have reduced the uncertainty in radial velocity amplitude K by a factor of 6 and produced a secure value of the rotational line broadening. Additionally, they present the first optical light curves and infrared magnitudes and colors of LMC X-1, allowing strong constraints on the inclination of the system, as well as the temperature and radius of the companion star.

The X-ray transient 1A 0620−00 was discovered by the Ariel V Satellite during an outburst in 1975 (Elvis et al. 1975). A radio counterpart was detected almost two weeks into the outburst, remaining visible for approximately one week (Owen et al. 1976). Kuulkers et al. (1999) collected data from the 1975 outburst and found multiple (jet) ejections with expansion velocities in excess of $0.5c$, a behavior associated with BHXB systems (McClintock & Remillard 2006). A previous outburst, occurring in 1917, was discovered retrospectively on photographic plates at the Harvard College Observatory (Eachus et al. 1976).

Oke (1977) identified the optical counterpart as a K5 dwarf. A radial velocity study of the bright, low-mass counterpart led to the measurement of the mass function, solidifying 1A 0620−00 as a strong BHC (McClintock & Remillard 1986).

Many attempts have been made to measure inclination of the system: Haswell et al. (1993) found $i\gt 62^\circ$; Shahbaz et al. (1994a) found $i=36\buildrel{\circ}\over{.} 7;\;$Marsh et al. (1994) found $i=37^\circ$; Gelino et al. (2001c) found $i=40\buildrel{\circ}\over{.} 8;$ and Froning & Robinson (2001) found a wide range of inclinations ($38^\circ \lt i\lt 75^\circ$) corresponding to different epochs of data. Overall inclination estimates remained inconsistent at best. The inconsistency stems from the highly asymmetric ellipsoidal variations in the light curve (Leibowitz et al. 1998) and contamination of the K-star flux by light from the accretion disk (Neilsen et al. 2008).

Cantrell et al. (2010) rectify these issues, making use of an extensive data set spanning a decade. They restrict their sample to light curves that are in the passive state (i.e., minimal aperiodic variability from accretion disk contributions) and those where the non-stellar luminosity (NSL) fraction and magnitude calibration are well constrained. They also require each light curve to maintain the same shape over its duration. As a result of fitting an 11 parameter model to the 8 remaining light curves, they estimate a weighted average of $i=51^\circ \pm 0\buildrel{\circ}\over{.} 9$. We assume this value to be unbiased by systematic error (Kreidberg et al. 2012), and use this inclination along with ${M}_{{\rm{BH}}}$ inferred from it in our calculations. We also adopt the distance estimated by Cantrell et al. (2010), using their dynamical model, in our analysis.

GRS 1009−45 was discovered by the GRANAT/WATCH ASM (Lapshov et al. 1993) and by BATSE on board CGRO (Harmon et al. 1993a). It was shown to have an ultra-soft spectrum typical of BHXB systems (Kaniovsky et al. 1993). della Valle & Benetti (1993) discovered a blue optical counterpart that showed a spectral type of G5–K7 (della Valle et al. 1997; Filippenko et al. 1999). Further optical photometry revealed a secondary outburst 6 months later, followed by a series of "mini-outbursts," reminiscent of BH XRN systems like GRO J0422+32 (Bailyn & Orosz 1995).

While Filippenko et al. (1999) were able to obtain a mass function and mass ratio, their estimate of inclination assumed that the secondary is a K7–K8 star that is not under massive. Shahbaz et al. (1996b) performed an ellipsoidal variability analysis; however, because the light curve shows clear evidence of a significant non-stellar contribution and the star-only model fit yields a large ${\chi }^{2}$, the light curve is likely active.

The inclination found by Shahbaz et al. (1996b) conflicts with the estimate by Filippenko et al. (1999), and the spectral type of the secondary is not clear. Assuming it is a G5V star, as suggested by della Valle et al. (1998), Kreidberg et al. (2012) find a lower limit on inclination of $i=62^\circ$, consistent with their corrected estimate. We therefore adopt this inclination in our calculation of ${M}_{{\rm{BH}}}$. Hynes (2005) discusses their preference for the distance estimate of 3.82 kpc by Gelino & Harrison (2002), which we also adopt.

XTE J1118+480 was discovered by RXTE/ASM in 2000 as a weak, slowly rising X-ray source (Remillard et al. 2000). Both optical (Uemura et al. 2000a, 2000b) and radio (Pooley & Waldram 2000) counterparts were quickly found. Strong low-frequency variability and a spectrum dominated by a hard power-law component extending past 100 keV (Revnivtsev et al. 2000b), accompanied by a persistently inverted radio spectrum (Hynes et al. 2000b), suggested it was a BHC in the HCS (Fender et al. 2001).

This outburst lasted ∼7 months and exhibited some complex behavior. After the first peak in 2000 January (Remillard et al. 2000; Wren & McKay 2000), the source decayed only to re-brighten to a plateau state (of similar brightness to the first peak; Uemura et al. 2000a) for ∼5 months (Brocksopp et al. 2010b). XTE J1118+480 remained in the HCS for the duration of the outburst, making it a "hard-only" outburst source (Brocksopp et al. 2004; Zurita et al. 2006).

Its second outburst was discovered in the optical (Zurita et al. 2005) and confirmed by X-ray and radio observations (Pooley 2005; Remillard et al. 2005) in 2005. RXTE would confirm that the source again remained in the HCS for the duration of the outburst (Swank & Markwardt 2005; Zurita et al. 2006). However, the 2005 event exhibited behavior more typical of a soft X-ray transient (SXT), with short lived jet ejections, a more dominant disk component to the spectrum, and a FRED type light curve (Brocksopp et al. 2010b).

Dynamical measurements establishing a very large mass function ($\gt 6\;{M}_{\odot }$) confirmed a BH primary in this system (McClintock et al. 2001a; Wagner et al. 2001). Observations in quiescence have allowed refinement of the system parameters (Gelino et al. 2006; Gonzalez Hernandez et al. 2008); however, estimates of inclination have proven more challenging. While the consensus is that XTE J1118+480 has a high inclination, accurate measurements are challenging due to strong superhump modulation (for a discussion of superhumps in LMXBs, see Haswell et al. 2001) in addition to ellipsoidal variability (Zurita et al. 2002), and a large and variable NSL fraction (Wagner et al. 2001). The inclination measurements by Wagner et al. (2001), McClintock et al. (2001b), Zurita et al. (2002) and Gelino et al. (2006) all lie in the range $68^\circ \lt i\lt 82^\circ$. While these estimates are all consistent with each other, none are free of significant systematic errors (Kreidberg et al. 2012). Therefore, we adopt the full range of inclinations in our calculation of ${M}_{{\rm{BH}}}$. We adopt the distance derived by Gelino et al. (2006) using their established dynamical model.

The X-ray nova GS 1124−684 was discovered by the GINGA ASM (Lundt et al. 1991) and by the GRANAT/WATCH ASM (Lundt et al. 1991) in 1991. It was observed extensively from radio to hard X-rays (Kitamoto et al. 1992). Its X-ray spectra and decay timescales similar to 1A 0620−00 (della Valle et al. 1991; Kitamoto et al. 1992), a known BH (McClintock & Remillard 1986), established it as a firm BHC. The orbital period and radial velocity curve of the secondary by Remillard et al. (1992) established it as a dynamically confirmed BH. Orosz et al. (1996) refined the determination of the mass function, spectral type of the secondary, and inclination.

This estimate of inclination is both higher and less precise than the estimate in Gelino et al. (2001a). However, as there is clear evidence for non-stellar flux in the IR (Gelino et al. 2010) when the source was active, the difference may be due to the non-stellar contributions. Even though the source may have been passive during the Gelino et al. (2001a) observations, this does not guarantee a negligible NSL fraction. The Orosz et al. (1996) estimate is also higher than that of Shahbaz et al. (1994a). However, the Shahbaz et al. (1994a) best-fit inclination has a large ${\chi }^{2}$, suggested to be the result of incorrect sky subtraction. Following Kreidberg et al. (2012), we adopt the inclination range given in Orosz et al. (1996) of 54°–65° in our calculation of ${M}_{{\rm{BH}}}$.

Using infrared photometry, corrected for reddening, Gelino et al. (2001a) find a distance of ∼5.1 kpc. Gelino (2001) refined their estimate with simulations to include error bars. Hynes (2005) prefer this estimate, and we adopt it in our calculations.

The transient hard X-ray source IGR J11321−5311 was discovered by ISGRI on board INTEGRAL in 2005 during a short flare lasting ∼3.5 hr (Krivonos et al. 2005). During this time the source exhibited a very hard spectrum (${\rm{\Gamma }}\sim 0.55$), with no evidence for a break up to 300 keV leading Sguera et al. (2007) to suggest that it was probably a magnetar. However, Krivonos et al. (2005) suggest that the spectrum is reminiscent of a BHXB. Therefore, we include this source in our sample as a possible BHC.

The X-ray transient MAXI J1305−704 was discovered by the Gas Split Camera (GSC) on board MAXI in 2012 (Sato et al. 2012). This source was proposed to be a BHXB based on X-ray and optical spectra as well as light curve and hardness variations over time (Greiner et al. 2012; Kennea et al. 2012b; Suwa et al. 2012; Morihana et al. 2013). Kennea et al. (2012a, 2012d) observed dip-like features and Miller et al. (2012a, 2012b) observed possible absorption line features. These dips and absorption profiles provide a strong indication that this source has a large inclination (Shidatsu et al. 2013).

Shidatsu et al. (2013) identified a 9.74 hr orbital period from the recurrence interval between absorption dips and inferred an inclination between $60^\circ \lt i\lt 75^\circ$, most likely $\sim 75^\circ$ as the source shows dips but no eclipses. A precise value of inclination has not yet been determined.

Swift J1357.2−0933, a new Galactic BHC (Casares et al. 2011), was discovered by the Swift/BAT in January of 2011 when it went into outburst (Krimm et al. 2011c). The source remained in the HCS for the duration of the outburst, classifying it as a "hard-only" outburst source (Armas Padilla et al. 2013a). Both the X-ray spectrum (Krimm et al. 2011a) and the magnitude difference (between quiescence and outburst) of the detected optical counterpart (Rau et al. 2011b) pointed to an LMXB nature.

Photometry from the Sloan Digital Sky Survey (SDSS) suggested an M4 counterpart (Rau et al. 2011b; Shahbaz et al. 2013). The distance is debated; Corral-Santana et al. (2013) suggest 1.5 kpc, but Shahbaz et al. (2013) compare the estimated magnitude of the companion (not detected) with the expected magnitude of an M4.5 V star to infer a possible distance range from 0.5 to 6 kpc. Given this information, we choose to use 1.5 kpc as a lower limit and 6 kpc as an upper limit on the distance to this system.

Armas Padilla et al. (2013a) calculate the peak luminosity for the outburst (for a distance of 1.5 kpc) to be ${L}_{{\rm{X}}}=1.1\times {10}^{35}\;{\rm{erg}}$ ${{\rm{s}}}^{-1}$, making it the only confirmed BH Very Faint X-ray Transient⁵ (VFXT; Armas Padilla et al. 2013a; Corral-Santana et al. 2013).

Corral-Santana et al. (2013) established ${M}_{{\rm{BH}}}\gt 3.6\;{M}_{\odot }$, dynamically confirming this as a BH. Corral-Santana et al. (2013) also find an orbital period of 2.8 hr and presented an observation of intense dips in the optical light curve, which they explained as toroidal structure in the inner region of the disk, seen at high inclinations ($i\geqslant 70^\circ$), moving outward as the outburst progressed. This implies we may be observing the system close to edge on. In addition, Shahbaz et al. (2013) find evidence for quiescent optically thin synchrotron emission, which they discuss could possibly arise from a jet in the system.

In 1987 the GINGA ASM discovered GS 1354−64 in outburst (Makino 1987). The X-ray spectrum was well fit with soft disk blackbody and hard power-law components (Kitamoto et al. 1990b) typical of X-ray transient outbursts, and suggestive of a BH nature (Brocksopp et al. 2001). The position of GS 1354−64 is consistent with two transient sources, Cen X-2 (Francey 1971) and MX 1353−64 (Markert et al. 1979), which were observed in outburst in 1967 and 1972, respectively. Both sources show different X-ray spectral properties than the 1987 outburst of GS 1354−64. If all three were in fact the same source, then GS 1354−64 must show at least four different spectral states (Kitamoto et al. 1990b). Brocksopp et al. (2001) argue that this is not unfeasible as sources such as GX 339−4 routinely show multiple state behaviors during outburst (McClintock & Remillard 2006). We therefore assume all three are in fact the same source.

GS 1354−64 was again observed in outburst in 1997 (Revnivtsev et al. 2000a; Brocksopp et al. 2001). During this time, optical (Castro-Tirado et al. 1997), IR (Soria et al. 1997), and radio (Fender et al. 1997c) counterparts were detected. The radio source was too faint to detect extended structure. However, analysis of the radio spectrum showed a weak flat synchrotron spectrum, which suggested possible mass ejections in the form of a jet (Brocksopp et al. 2001). This source has recently (2015 June) gone into outburst for the fifth time, where it has been detected across X-ray (Miller et al. 2015a, 2015b), optical (Russell & Lewis 2015), and radio (Coriat et al. 2015) wavelengths.

While the first and third outbursts of GS 1354−64 show very soft X-ray spectra (Brocksopp et al. 2001, 2004), the second and fourth events display spectra dominated by a hard power-law, typical of XRBs in the hard state, indicating the source did not reach the softer states in these cases (i.e., "hard-only" outbursts; Revnivtsev et al. 2000a; Brocksopp et al. 2001).

Casares et al. (2004) obtained the first radial velocity curve of the optical counterpart, BW Cir, identified its spectral type, mass ratio, period, and in turn a mass function of $f(M)=5.75\pm 0.30\;{M}_{\odot }$, confirming a BH primary. Casares et al. (2009) infer a lower limit on the distance of 25 kpc (from the companion's luminosity) and estimate an upper limit of 61 kpc (assuming a $10\;{M}_{\odot }$ BH) from calculations by Kitamoto et al. (1990b). We take the distance to be characterized by a uniform distribution between 25 and 61 kpc for the purpose of our analysis.

While Casares et al. (2009) have multi-wavelength photometry and spectroscopy between 1995 and 2003 of the source, the data is characterized by strong aperiodic variability without discernible ellipsoidal modulation. Therefore, no lower limit on inclination can be found. From the spectral type and eclipse limits Kreidberg et al. (2012) find $27\buildrel{\circ}\over{.} 2\lt i\lt 80\buildrel{\circ}\over{.} 8$ We take $80\buildrel{\circ}\over{.} 8$ as the upper limit on the inclination of GS 1354−64 to calculate a lower limit on the mass of the system.

The X-ray transient 1A 1524−62 was discovered by Ariel V (Pounds 1974) and has been observed in outburst twice. It is considered to be a BHC due to its ultra-soft spectrum, bi-modal spectral behavior, and absence of type I X-ray bursts.

The 1990 outburst was observed in both the hard (Barret et al. 1992) and soft (ROSAT All-Sky Survey; RASS) X-rays. The soft X-ray spectrum could be fit equally well by a cool blackbody or a power law. The presence of an ultra-soft component to the spectrum could not be ruled out (Barret et al. 1995). However, the outburst was insufficient to trigger the soft X-ray ASMs (WATCH and Ginga) and Ginga provided an upper limit consistent with the ROSAT detection (Barret et al. 1995; Brocksopp et al. 2004). For these reasons we include this outburst as a possible "hard-only" outburst (Brocksopp et al. 2004).

Murdin et al. (1977) identify a possible optical counterpart. From similarities with 1A 0620−00, Murdin et al. (1977) estimate a distance of >3 kpc, while van Paradijs & Verbunt (1984) propose 4.4 kpc assuming ${M}_{v}=1.0$ and E(B − V) = 0.7. We adopt 3 kpc as a lower limit, set the upper limit to 8 kpc and take 4.4 kpc as the most likely value.

Swift J1539.2−6227 was discovered by Swift/BAT in 2008 (Krimm et al. 2008b). Krimm et al. (2011b) present the complete evolution of spectral and timing properties during the outburst, including the rise of the disk component in the SDS and power density spectra signatures of transitions between the HCS and SDS. These features, coupled with a lack of observed pulsations, establish the source as a possible BHC (Krimm et al. 2013c).

Torres et al. (2009a) performed optical spectroscopy and found a possible optical counterpart with a blue continuum. No Balmer lines, He ii 4686 Å or Bowen blend emission were detected. Krimm et al. (2011b) suggest that the lack of emission features in the outburst spectrum paired with the faintness of the source in quiescence points to a low-mass, main-sequence or degenerate donor star companion to the compact accretor.

MAXI J1543−564 was discovered by the GSC on board MAXI in 2011 (Negoro et al. 2011a). Type-C QPOs, and an observed decrease in fractional rms and hardness ratios and steepening of the photon index during the outburst led Munoz-Darias et al. (2011a) to classify the source as a BHC. Stiele et al. (2012) present a full spectral and timing analysis. Miller-Jones et al. (2011a) detect a radio counterpart with an optically thin spectrum.

Three possible optical/IR counterparts have positions consistent with the XRT, ATCA, and Chandra error circles (Chakrabarty et al. 2011; Rau et al. 2011a; Rojas et al. 2011; Russell et al. 2011a). None of these three candidates show any variability (Rau et al. 2011a; Rojas et al. 2011; Russell et al. 2011a), leaving the optical counterpart unknown.

4U 1543−475 is a recurrent X-ray transient discovered in 1971 (Matilsky et al. 1972), and also observed in outburst in 1983 (Kitamoto et al. 1984), 1992 (Harmon et al. 1992), and 2002 (Park et al. 2004; Kalemci et al. 2005). 4U 1543−475 displayed classical SXT behavior during the first, second, and fourth outbursts (Matilsky et al. 1972; Kitamoto et al. 1984; Park et al. 2004). However, hard X-ray observations during the third event reveal a power-law spectrum (Harmon et al. 1992). Unfortunately, there are no soft X-ray observations of this source during the 1992 outburst. Following Brocksopp et al. (2004) we include this outburst as a possible "hard-only" outburst. Overall, the observation of a wide array of spectral features make 4U 1543−475 a strong BHC (Orosz et al. 1998b).

4U 1543−475 is one of the few sources that has near-simultaneous photometry and spectroscopy (Kreidberg et al. 2012). The optical counterpart, IL Lupi, was discovered by Pederson (1983) and classified as spectral type A2V (Chevalier & Ilovaisky 1992). A radio counterpart was detected by Hunstead & Webb (2002).

4U 1543−475 has been subject to many detailed dynamical studies (Orosz et al. 1998b, 2002; Orosz 2003), confirming its BH primary. However, when estimating inclination, Orosz et al. (1998b) include the mass ratio as a free parameter, which results in a large source of error (Kreidberg et al. 2012).

A more precise inclination measurement is in a conference proceeding (Orosz et al. 2002), however, there is no formal published record of the light curve. We therefore follow Kreidberg et al. (2012) in taking the inclination estimates from Orosz et al. (1998b) as the boundaries for a uniform distribution. A precise measurement of distance is found in Ozel et al. (2010).

The Galactic microquasar⁶ XTE J1550−564 was discovered by the RXTE ASM (Smith 1998). Shortly after discovery, optical (Orosz et al. 1998a), and radio (Campbell-Wilson et al. 1998) counterparts were detected along with a superluminal ejection observed in the radio (Hannikainen et al. 2001). Observations of rapid X-ray variability, hard spectrum, and the absence of X-ray bursts or pulsations suggested a BHC (Cui et al. 1999).

Later, Orosz et al. (2002) confirmed the BH nature of the primary. It has been seen in outburst five times: 1998/1999 (Sobczak et al. 2000; Remillard et al. 2002; Kubota & Makishima 2004), 2000 (Rodriguez et al. 2004; Kalemci et al. 2001; Miller et al. 2001b; Tomsick et al. 2001a), 2001 (Tomsick et al. 2001b), 2001/2002 (Belloni et al. 2002) and 2003 (Sturner & Shrader 2005; Aref'ev et al. 2004). Radio observations during the 2001/2002 outburst by Corbel et al. (2002) confirmed the presence of a hard state jet spectrum.

The first two outbursts showed the typical "turtlehead" pattern through BH states. Complete spectral and timing analysis for the 1998/1999 and 2000 outbursts can be found in Sobczak et al. (2000), Homan et al. (2001), Cui et al. (1999), Remillard et al. (1999a) and Tomsick et al. (2001a); Miller et al. (2001b); Kalemci et al. (2001); Belloni et al. (2002), respectively. The last three have been shown to be under-luminous "hard-only" outbursts (Tomsick et al. 2001b; Swank et al. 2002; Belloni et al. 2002; Corbel et al. 2002; Sturner & Shrader 2005).

Orosz et al. (2011b) provide an improved dynamical model including ${P}_{\mathrm{orb}}$, $f(M)$, q, and i. They have determined an inclination using photometry and spectroscopy over 7 years of data. However, they use NSL fractions determined at a different time than the photometry measurements, which can produce unreliable inclination measurements (Kreidberg et al. 2012). Orosz et al. (2011b) acknowledge the uncertainty and fit a model, which includes a disk and four free parameters, using eight different combinations of light curves and NSL fractions. They find a reasonably narrow possible range in inclination of $57\buildrel{\circ}\over{.} 7\lt i\lt 77\buildrel{\circ}\over{.} 1$. Following Kreidberg et al. (2012), we adopt an isotropic distribution over this range for our inclination, a mass ratio uniformly distributed over the range given in Orosz et al. (2011b), and use these values to calculate a ${M}_{{\rm{BH}}}$.

The recurrent X-ray transient 4U 1630−472 was discovered by the VELA 5B and UHURU satellites (Priedhorsky 1986; Jones et al. 1976). Over the last 46 years, 4U 1630−472 has undergone 23 outbursts occurring quasi-regularly (∼600–700 days; Kuulkers et al. 1997a) and exhibiting a wide range of complex outburst behavior (see Table 14 for a complete list of references for each outburst).

Highly polarized radio emission was observed in 1998, confirming the presence of jets (Hjellming et al. 1999d). The most recent studies of the radio jets in 4U 1630−472 discuss possible baryonic matter content within the jets (Díaz Trigo et al. 2013; Neilsen et al. 2014). An additional outflow, in the form of an accretion disk wind, has also been detected in this source (Ponti et al. 2012; Diaz Trigo et al. 2014).

No optical counterpart is known, most likely due to its high reddening and crowded field (Parmar et al. 1986) resulting in difficulty performing optical and infrared studies. While no compact object mass is known, McClintock & Remillard (2006) classify it as a very likely class "A" BHC. It is considered likely to contain a BH based on spectral properties (Parmar et al. 1986) and fast timing behavior (Kuulkers et al. 1997b).

The X-ray transient XTE J1637−498 was discovered by PCA on board RXTE during a regular scan of the Galactic bulge and ridge regions in 2008 (Markwardt et al. 2008c). During this time, the source was also detected as a weak source with Swift/BAT (H. Krimm 2015, private communication). Wijnands et al. (2008) obtained an X-ray spectrum described by an absorbed power law with a photon index of ∼1.5, which is consistent with the source being a LMXB, but due to the large errors on the spectral parameters they stress that other types of systems cannot be excluded. Curran et al. (2011a) identify a possible optical/IR counterpart. Given the uncertain nature of this object, we include XTE J1637−498 in our sample as a possible BHC.

XTE J1650−500 is a SXT that was discovered by the ASM on board RXTE in 2001 (Remillard 2001). The X-ray spectrum (Markwardt et al. 2001), power density spectra (Revnivtsev & Sunyaev 2001; Wijnands et al. 2001), evolution through the "turtlehead" pattern of accretion states (Rossi et al. 2004; Tomsick et al. 2004), and observation of QPOs (Homan & Wijnands 2003; Kalemci et al. 2003) during the outburst confirmed the source to be a BHC.

The optical counterpart was discovered by Castro-Tirado et al. (2001), confirmed by Groot et al. (2001), Augusteijn et al. (2001), and later classified as a star of spectral type G5–K4III (Orosz et al. 2004). The radio counterpart was discovered by Groot et al. (2001), and Corbel et al. (2004) observed the source at radio frequencies for the duration of the outburst, finding evidence for the existence of a steady compact jet. An additional outflow, in the form of an accretion disk wind, has also been detected in this source (Miller et al. 2004; Ponti et al. 2012).

Further optical observations of the source revealed the $f(M)$, ${P}_{{\rm{orb}}}$, and i for the system (Orosz et al. 2004). While Orosz et al. (2004) are able to determine a lower limit on the inclination of $i\gt 50^\circ$, by fitting a star only model with photometry obtained between 2003 May and August, Kreidberg et al. (2012) suggests that the source was active during this time, as there was more scatter in the light curve than one would expect from photometric errors alone. Therefore they use this lower limit to calculate their own corrected inclination. We adopt their estimate of $i=75\buildrel{\circ}\over{.} 2$.

The transient source XTE J1652−453 was discovered in 2009 during PCA monitoring of the Galactic region (Markwardt et al. 2009b). Further observations showed a quickly rising flux and an X-ray spectrum that evolved from a soft disk blackbody (Markwardt & Beardmore 2009; Markwardt & Swank 2009) to a hard power law (Coriat & Rodriguez 2009), which suggested a BHC (Han et al. 1999). For complete spectral and timing analysis of the outburst, see Han et al. (1999) and Hiemstra et al. (2011).

Near-IR observations show at least two possible counterparts (<18 from each other) within the XRT and ATCA error circles (Calvelo et al. 2009; Markwardt & Beardmore 2009; Reynolds et al. 2009). While the ATCA observation favors the fainter of the two as the probable counterpart, further near-IR observations have detected no significant variability in this source, suggesting that it may be an unrelated interloper star (Torres et al. 2009c). No conclusive argument has been made regarding the true counterpart.

A radio counterpart was detected by Calvelo et al. (2009), whose observations indicated emission from the decay of an optically thin synchrotron event associated with the activation of XTE J1652−453.

GRO J1655−40 was discovered in 1994 by BATSE on board CGRO (Harmon et al. 1995). Radio jets travelling with apparent superluminal motion were discovered (Hjellming & Rupen 1995; Tingay et al. 1995), allowing for a precise measurement of distance to the source⁷ (Hjellming & Rupen 1995).

GRO J1655−40 underwent additional outbursts in 1996/1997, exhibiting complex multi-peak behavior (Zhang et al. 1997; Remillard et al. 1999b; Sobczak et al. 1999), and 2005 (Brocksopp et al. 2006; Saito et al. 2006; Cabellero Garcia et al. 2007; Diaz Trigo et al. 2007; Migliari et al. 2007; Shaposhnikov et al. 2007; Joinet et al. 2008), in which variable radio emission was detected again (Rupen et al. 2005a, 2005b, 2005c; Brocksopp et al. 2006).

An additional outflow, in the form of an accretion disk wind, has also been detected in this source (Miller et al. 2006b, 2008). The magnetically driven wind in GRO J1655−40 (Kallman et al. 2009) has the largest known mass loss rate of all BH sources in which such winds have been detected (Ponti et al. 2012).

Studies of the orbital parameters in full quiescence published by Orosz & Bailyn (1997), Greene et al. (2001) and Beer & Podsiadlowski (2002), all of which are consistent, provided ${P}_{{\rm{orb}}}$ and ${M}_{{\rm{BH}}}$, and confirmed the BH nature of the primary.

However, Kreidberg et al. (2012) suggested that the distance of 3.2 kpc from Beer & Podsiadlowski (2002) is more accurate than that of Greene et al. (2001). We adopt the inclination and other orbital parameters from Beer & Podsiadlowski (2002), assuming it was passive during their observation.

MAXI J1659−152 was detected by Swift/BAT (Mangano et al. 2010b) and MAXI (Negoro et al. 2010) in 2010. Optical spectroscopy proved the Galactic origin of the source and its X-ray binary classification (De Ugarte Postigo et al. 2010; Kaur et al. 2012). It was established as a BHC by observations of fast timing behavior, similar to other BH transients (Kalamkar et al. 2011; Kennea et al. 2011b; Muñoz-Darias et al. 2011b; Yamaoka et al. 2012). Its orbital period of ∼2.4 hr (Kuulkers et al. 2010; Kennea et al. 2011b; Kuulkers et al. 2013) makes MAXI J1659−152 the shortest period BHXB source known (Kennea et al. 2010).

Kuulkers et al. (2013) constrain its inclination between $65^\circ \lt i\lt 80^\circ$, from the obscuration of ∼90% of the total emission on cyclical timescales, as well as the absence of eclipses. Mass estimates range from 2.2 to $20\;{M}_{\odot }$ (Kennea et al. 2011b; Shaposhnikov et al. 2012b; Yamaoka et al. 2012).

Distance estimates range from 1.6 to 4.2 kpc (Miller-Jones et al. 2011b) and 8.6 kpc (Yamaoka et al. 2012). We take a range from 1.6 to 8 kpc in our analysis. A possible optical counterpart (Kong et al. 2010a; Kong 2012) has a spectral type between M2 and M5 (Miller-Jones et al. 2011b; Kong 2012; Kuulkers et al. 2013).

The Galactic X-ray binary GX 339−4, discovered in 1972 by the MIT X-ray detector on board the Orbiting Solar Observatory (OSO) 7 satellite (Markert et al. 1973), is the most extensively studied transient Galactic BHXB system (Zdziarski et al. 2004). Over 43 years, GX 339−4 has undergone 20 outbursts in which the entire array of spectral accretion states have been observed (Belloni et al. 1999b). GX 339−4 has undergone numerous hard state outbursts, making it a "hard-only" outburst source (Rubin et al. 1998; Kong et al. 2002; Buxton et al. 2012; Belloni et al. 2013). For a complete list of references for each outburst see Table 14.

The secondary star is not clearly detected during quiescence, with most of the observed optical emission originating from the accretion disk. Its LMXB nature has been inferred from the upper limits on the luminosity of this companion star (Shahbaz et al. 2001). The system was classified as a BHC (Zdziarski et al. 1998; Sunyaev & Revnivtsev 2000) from its spectral and temporal characteristics. Fluorescence spectroscopy of N ii and He ii emission lines during outburst, formed on the donor star surface due to X-ray irradiation, allowed Hynes et al. (2003) to measure ${P}_{{\rm{orb}}}$ and put an upper limit on q and a lower limit on the mass function of $\gt 2\;{M}_{\odot }$.

The radio source associated with GX 339−4 was discovered by Sood & Cambell-Wilson (1994) in 1994. Wilms et al. (1999) argued that the radio emission could come from a compact self-absorbed jet. The picture of the radio jet existing only in the hard states, and being quenched in the soft states (Fender et al. 1999a, 2004), and disk-jet coupling implied from observed X-ray-Radio correlations (Hannikainen et al. 1998; Corbel et al. 2000, 2003; Markoff et al. 2003; Homan et al. 2005; Corbel et al. 2013), arose from numerous observations of GX 339−4.

An additional outflow, in the form of an accretion disk wind, has also been detected in this source (Miller et al. 2004; Ponti et al. 2012).

Hynes et al. (2004) (based on optical spectra) argue that GX 339−4 is located beyond the Galactic tangent point (implying a lower limit of ≥6 kpc) and favor $d\geqslant 15\;{\rm{kpc}}$. However, Zdziarski et al. (2004) prefer GX 339−4 to be in the Galactic bulge, and use optical/IR data to estimate a favored distance of 8 kpc, a result still consistent with the lower limit given by Hynes et al. (2004). We adopt the Zdziarski et al. (2004) estimate.

H 1705−250 was discovered by the ASMs on board the Ariel V (Griffiths et al. 1978) and HEAO 1 satellites (Kaluzienski & Holt 1977). The optical counterpart, discovered on plates taken at the Anglo-Australian Telescope and UK Schmidt Telescope (Griffiths et al. 1977, 1978), is a star of spectral type K3–7 V (Harlaftis et al. 1997).

The light curve behavior and observed soft (Griffiths et al. 1978) and hard (Wilson & Rothschild 1983) components to the spectrum resembled that of other SXTs (Martin et al. 1995). This evidence, along with a dynamical mass function measurement of $f(M)=4.86\pm 0.13\;{M}_{\odot }$ (Filippenko et al. 1997) led to the confirmation of a BH primary.

Martin et al. (1995) made the first inclination measurement of H 1705−250 to be $48^\circ \lt i\lt 51^\circ$. Remillard et al. (1996) obtained a conflicting result of $i\gt 60^\circ$. However, the former only show folded light curves and the latter analyzed a light curve which exhibits uneven maxima. Due to the uncertainty as to whether the source was active or passive during these observations, we agree with Kreidberg et al. (2012) and adopt the Martin et al. (1995) $i=48^\circ$ as a lower limit on the inclination, yielding an upper limit on ${M}_{{\rm{BH}}}$. As the source does not eclipse, is not a dipper or an accretion disk corona source, we estimate a lower limit on ${M}_{{\rm{BH}}}$ by taking $i=80^\circ$. The distance to the source is quoted in Barret et al. (1996b).

IGR J17091−3624 was discovered by INTEGRAL in 2003 (Kuulkers et al. 2003). Spectral analysis of the outburst revealed typical state transition behavior (i.e., "turtlehead"), where the X-ray emission softened as the outburst progressed (Lutovinov & Revnivtsev 2003; Capitanio et al. 2006a). IGR J17091−3624 has been classified as a probable BHC (Lutovinov & Revnivtsev 2003).

An archival search of TTM-KVANT and BeppoSAX Wide Field Camera (WFC) data revealed previous outbursts in 1994, 1996, and 2001 (in't Zand et al. 2003; Revnivtsev et al. 2003; Capitanio et al. 2006a). The radio counterpart showed a flux increase over two weeks and an inverted spectrum, characteristic of a compact jet (Capitanio et al. 2009b). An accretion disk wind has also been detected (King et al. 2012b).

Two more outbursts have been reported, in 2007, showing spectral behavior typical of a BHC (Capitanio et al. 2009b), and 2011–2013 (Capitanio et al. 2012). The most recent outburst has been well studied across the X-ray regime (Krimm & Kennea 2011; Capitanio et al. 2011, 2012; Del Santo et al. 2011; Rodriguez et al. 2011b). Radio observations again revealed evidence for a self-absorbed compact jet (Corbel et al. 2011; Rodriguez et al. 2011a; Torres et al. 2011) as well as discrete jet ejections (Rodriguez et al. 2011a). This outburst, unlike the others, showed peculiar pseudo periodic flare-like events ("heartbeats") (Altamirano et al. 2011c) closely resembling those observed in GRS 1915+105 (Altamirano et al. 2011a, 2011b; Belloni et al. 2000).

The transient IGR J17098−3628 was discovered by INTEGRAL in 2005 (Grebenev et al. 2005b), and remained detectable until 2007. The source was also briefly detected in 2009 (Prat et al. 2009). Its evolution in brightness and spectral shape (Grebenev et al. 2005a; Capitanio et al. 2009b) formed the basis for its BHC classification (Grebenev et al. 2007). Probable radio (Rupen et al. 2005e) and optical (Steeghs et al. 2005a) counterparts have been detected for this source.

The transient SAX J1711.6−3808 was discovered by the WFC on board BeppoSAX in 2001 (in't Zand et al. 2001). The Galactic latitude, flux, spectral, and timing properties confirmed the XRB nature of SAX J1711.6−3808, while the lack of observation of type I X-ray bursts and lack of coherent oscillations along with the appearance of a broad Fe–K emission feature suggested that the primary could be a BH (in't Zand et al. 2002b; Wijnands & Miller 2002). McClintock & Remillard (2006) classify the system with a grade "B" likelihood of harbouring a BH.

The spectrum, which was dominated by a Comptonized continuum, and timing properties of SAX J1711.6−3808 indicate that it never left the HCS during outburst, making it one of the "hard-only" outburst sources (in't Zand et al. 2002b).

The search for an optical counterpart has proven difficult due to large extinction (in't Zand et al. 2002b).

The transient Swift J1713.4−4219 was discovered by Swift/BAT in 2009 (Krimm et al. 2009b). Unfortunately, due to Sun constraints the source could not be observed by Swift/XRT or UVOT, and was only visible with PCA for 3 days. The PCA spectrum was well fit with a power law of photon index ${\rm{\Gamma }}=1.68$, and timing analysis revealed strong, aperiodic variability in the power spectrum, both of which point to a possible BH transient in the HCS (Krimm et al. 2013c). For this reason, we include Swift J1713.4−4219 in our sample as a possible BHC.

The hard X-ray transient XMMSL1 J171900.4-353217 was discovered in an XMM-Newton slew (Read et al. 2010a). Markwardt et al. (2010) argued for its association with XTE J1719−356, a faint transient seen by PCA on board RXTE in the same month (Markwardt et al. 2010). Decreases in flux (Armas Padilla et al. 2010a) and re-brightening events (Armas Padilla et al. 2010b) were observed in the source over the course of a few months, further indicating its transient nature.

As no definitive arguments have been made regarding the nature of the compact object, we include XMMSL1 J171900.4−353217 in our sample as a possible BHC.

XTE J1719−291 was discovered by the RXTE/PCA bulge scans in 2008 (Markwardt & Swank 2008). Over ∼46 days, it showed both flux decreases and rebrightening events (Degenaar et al. 2008a, 2008b; Markwardt & Swank 2008). Later, Degenaar & Wijnands (2008) (with Swift/XRT observations) observed large X-ray variability during an outburst that lasted almost two months.

Greiner et al. (2008) found a possible optical counterpart, deriving a spectral type K0V or later. Armas-Padilla et al. (2011) put constraints on the orbital period ($0.4\lt {P}_{{\rm{orb}}}\lt 12$ hr) and calculated a long-term mass transfer rate of $\sim 3.7\quad \times {10}^{-13}\;{M}_{\odot }\;{{\rm{yr}}}^{-1}$ if it does contain a BH.

Assuming a distance of 8 kpc, Armas-Padilla et al. (2011) estimate that XTE J1719−291 would have had a 2–10 keV peak luminosity of $7\times {10}^{35}\;{\rm{erg}}\;{{\rm{s}}}^{-1}$ during its 2008 outburst, therefore classifying the system as a VFXT.

To date, no conclusive evidence is available about the nature of the accretor. As such, we include XTE J1719−291 in our sample as a possible BHC.

X-ray Nova Ophiuchi (GRS 1716−249) was discovered in 1993 by SIGMA on board the GRANAT satellite (Ballet et al. 1993) and BATSE on board CGRO (Harmon et al. 1993b). Follow up observations discovered the optical and radio counterparts, derived the distance, and suggested that GRS 1716−249 was a LMXB system (della Valle 1994).

The X-ray spectrum was comparable to Cyg X-1 in the hard state (Revnivtsev et al. 1998b), and the power spectra showed a QPO which varied in frequency (van der Hooft et al. 1996). This confirmed that GRS 1716−249 never left the HCS during the 1993 outburst, and solidified it as a "hard-only" outburst source (Brocksopp et al. 2004).

In early 1995, Mir/Kvant detected five slow-rise, fast decay hard X-ray flares, "mini-outbursts" resembling those of the SXTs, GRS 1009−45 and GRO J0422+32 (Masetti et al. 1996). Hjellming et al. (1996b) found the relation between high energy X-ray and radio emission was very similar (i.e., radio emission follows the peak or onset of decay in X-ray flares seen between 20 and 200 keV) to that observed in GRO J1655−40 and GRS 1915+105 (Foster et al. 1996), implying the presence of a jet linked to state changes in the accretion disk.

Masetti et al. (1996) estimate a period of ∼14.7 hr and a lower limit on the mass of the primary from the super hump period⁸ to be $\gt 4.9\;{M}_{\odot }$.

XTE J1720−318 was discovered by the ASM on board RXTE undergoing an XRN-like outburst in 2003 (Remillard et al. 2003). Its X-ray spectrum included a 0.6 keV thermal component and a hard tail. Both the spectral characteristics and source luminosity were typical of a BH in the soft state (Markwardt 2003a). Cadolle Bel et al. (2004) perform a detailed spectral analysis of the outburst and conclude that XTE J1720−318 is a BHC.

The radio counterpart was discovered by Rupen et al. (2003d), O'Brien et al. (2003). Brocksopp et al. (2005) observed an unresolved radio source during the rise phase of the outburst, which reached a peak approximately coincident with the X-ray light curve. Through study of the spectral indices, they conclude that at least two ejection events occurred, similar to behavior observed in XTE J1859+226. Following a period in which the radio source was not detected, the source again switched on as XTE J1720−318 transitioned back into the hard state.

The X-ray transient XTE J1727−476 was discovered by RXTE (Levine et al. 2005a) and INTEGRAL (Turler et al. 2005a) in 2005. Observations yielded a soft spectrum, reminiscent of a BHXB in outburst (Kennea et al. 2005; Levine et al. 2005a). We therefore include XTE J1727−476 in our sample as a possible BHC. An optical counterpart was discovered by Maitra et al. (2005).

IGR J17285−2922 was discovered with INTEGRAL in 2004 (Walter et al. 2004). Based on the characteristic evolution of the spectrum from soft (Markwardt & Swank 2010) to hard (Barlow et al. 2005) observed during the ∼2 week outburst and its location in the Galactic bulge, IGR J17285−2922 was suggested to be an LMXB undergoing transient activity (Barlow et al. 2005). Given the lack of type I X-ray bursts and the relative hardness of the spectrum, Barlow et al. (2005) tentatively suggested the system may be harbouring a BH.

Assuming a distance of 8 kpc, the peak luminosity of this outburst was $\sim 8\times {10}^{35}\;{\rm{erg}}\;{{\rm{s}}}^{-1}$, classifying it as a VFXT (Sidoli et al. 2011).

Renewed activity in the transient identified as XTE J1728−295 was observed in 2010 (Markwardt & Swank 2010). During this time, the source was also detected by Swift/BAT (although never officially reported), reaching a peak of ∼16 mCrab on August 28–29, coincident with the RXTE detection (H. Krimm 2015, private communication). Turler et al. (2010) confirmed that IGR J17285−2922 and XTE J1728−295 were the same source. Using XMM and INTEGRAL, Sidoli et al. (2011) were able to obtain the first broadband spectrum, dominated by a power law with a slope consistent with the canonical range for BHXBs in the HCS (${\rm{\Gamma }}\sim 1.5-1.7;\;$Belloni 2010). IGR J17285−2922 remained in the HCS for the entire 2010 outburst, making it a "hard-only" outburst source and further suggesting a BH nature (Sidoli et al. 2011).

A possible optical counterpart is known (Kong et al. 2010b; Russell et al. 2010b; Torres et al. 2010).

The X-ray source GRS 1730−312 was discovered by SIGMA on board GRANAT (Churazov et al. 1994) and the TTM telescope on board Mir-Kvant (Borozdin et al. 1994) in 1994. Both Borozdin et al. (1995) and Vargas et al. (1996) observed that the source peak luminosity, contribution of the hard and soft spectral components, and hard to soft state transition during outburst, resemble the other BH sources GS 1124−684 (Ebisawa et al. 1994) and 1A 0620−00 (Ricketts et al. 1975), and therefore classified the source as a BHC.

GS 1732−273 was discovered by the GINGA satellite during scanning observations and was first designated GS 1734−275 (Makino 1988). The spectrum was fit well with a blackbody model similar to those of BHXBs (Yamauchi & Koyama 1990).

The source position was redetermined, now finding agreement with KS 1732−273 (van Paradijs & McClintock 1995; Liu et al. 2001; Yamauchi & Nakamura 2004), discovered by Mir-Kvant in 1989 (in't Zand et al. 1991) and 1RXS J173602.0-272541, discovered with the RASS in 1990 (Voges et al. 1999). Given the ultra-soft spectrum and transient behavior as evidence, Yamauchi & Nakamura (2004) classify the source as a BHC.

IGR J17379−3747, originally designated XTE J1737−376, was discovered by RXTE/PCA in 2004. The outburst lasted ∼9 days. Although not published formally, it appeared on the PCA bulge scan webpage.⁹ The source position coincided with IGR J17379−3747, a weak hard X-ray source reported in the 3rd INTEGRAL catalog (Bird et al. 2007). Given limits on its positional accuracy, Markwardt et al. (2008b) concluded that XTE J1737−376 and IGR J17379−3747 were most likely the same source.

This source was again detected by PCA, INTEGRAL, and BAT in 2008 (Markwardt et al. 2008b). Swift/XRT observations revealed a hard spectrum with photon index Γ ∼ 1.78, consistent with the RXTE/PCA spectrum (Krimm et al. 2008e). The observed spectrum suggests a BHC. In addition, Curran et al. (2011a) identify a possible optical/NIR counterpart.

GRS 1737−31 was discovered by SIGMA on board GRANAT in March of 1997 (Sunyaev et al. 1997). It was also found in RXTE and BeppoSAX data (Cui et al. 1997b; Heise 1997; Marshall & Smith 1997). Both the hard spectrum of the source (Sunyaev et al. 1997) and the observed chaotic variability (Cui et al. 1997a) were similar to properties observed by XRN and BH source Cyg X-1 (Sunyaev & Truemper 1979; Sunyaev et al. 1992; Vikhlinin et al. 1994; Trudolyubov et al. 1999).

Observations from Trudolyubov et al. (1999) and Cui et al. (1997a) revealed a hard power-law spectrum. Further observations using BeppoSAX and ASCA (Heise 1997; Ueda et al. 1997) confirmed this hard spectrum ∼2 weeks after the initial detection. Based on spectral and temporal properties, it was suggested that GRS 1737−31 was a distant XRN and BHC in the HCS (Cui et al. 1997a; Sunyaev et al. 1997; Trudolyubov et al. 1999). GRS 1737−31 appears to have never left the hard state for the duration of the 1997 outburst, making it a "hard-only" outburst source (Brocksopp et al. 2004).

No optical or radio observations have ever been published (Brocksopp et al. 2004).

The X-ray source GRS 1739−278 was discovered by SIGMA on board GRANAT in 1996 (Paul et al. 1996). The outburst was observed by Mir-Kvant (Borozdin et al. 1996), ROSAT (Greiner et al. 1997), GRANAT (Vargas et al. 1997), and RXTE (Takeshima et al. 1996).

A radio counterpart was discovered with Very Large Array (VLA) data (Durouchoux et al. 1996; Hjellming et al. 1996a) and an optical counterpart was discovered by Marti et al. (1997) to be either a luminous early/middle B-type, main-sequence star or a middle G/early K giant star.

Borozdin et al. (1998) and Vargas et al. (1997) find the light curve behavior, optical and radio observations, evolution of the spectrum and spectral characteristics correspond to the SDS and SPL states of BHCs, and Borozdin & Trudolyubov (2000) observe QPOs, present when the source was in the SPL and SDS, allowing GRS 1739−278 to reliably be classified as a BHC and soft XRN.)

In 2014 March, GRS 1739−278 was again detected by Swift (Krimm et al. 2014b) and INTEGRAL (Filippova et al. 2014) where, like the 1996 outburst, it completed the "turtlehead" BHXB pattern through the spectral states.

The micro-quasar 1E 1740.7−2942, located near the Galactic center, was discovered by Einstein in 1984 (Hertz & Grindlay 1984). Its hard X-ray emitting nature was first reported by Skinner et al. (1987). Given the spectral shape of its soft γ-ray emission and the similarities to Cyg X-1, Sunyaev et al. (1991a) classified it as a BHC and the strongest persistent source in the Galactic center region. Its micro-quasar classification came with the discovery of a double-sided radio emitting jet (Mirabel et al. 1992). Its radio emission has been found to be variable and correlated with the X-ray flux (Paul et al. 1991).

1E 1740.7−2942 has been suggested as a possible source of electron-positron annihilation due to an observed high energy spectral feature with GRANAT (Bouchet et al. 1991; Sunyaev et al. 1991b), hence the common name "The Great Annihilator." However, near simultaneous observations by CGRO (Jung et al. 1995) and BATSE (Smith et al. 1996) and high energy observations by INTEGRAL (Bouchet et al. 2009) could never confirm this feature.

1E 1740.7−2942 is one of only three BHCs that not only remain persistently near their maximum luminosity but also spend most of their time in the HCS (Churazov et al. 1993; Main et al. 1999; del Santo et al. 2004), with an X-ray spectrum usually described by an absorbed power law with photon index ${\rm{\Gamma }}\sim 1.4-1.5$ (Gallo & Fender 2002) and high energy cutoff (Sidoli et al. 1999; Natalucci et al. 2014). Occasionally the source has been observed to make the transition to the softer states (Sunyaev et al. 1991b; del Santo et al. 2004).

Due to a source environment characterized by a high concentration of dust and high column density ($\sim {10}^{23}\;{{\rm{cm}}}^{-2}$), optical identification is difficult (Gallo & Fender 2002) and its nature as a HMXB or LMXB, inclination, and distance remain unknown. However, the high absorption, position near the Galactic center and presence of bipolar jets, all favor a distance of ∼8.5 kpc and disfavor a face on geometry (Natalucci et al. 2014).

Periodic modulation has been detected and interpreted as an ${P}_{{\rm{orb}}}\sim 12.7$ days, suggesting the object could have a red giant companion (Smith et al. 2002a). However, Marti et al. (2010) have reported a candidate IR counterpart that would exclude the red giant companion possibility. A much longer periodicity has also been reported by Ogilvie & Dubus (2001), thought to be related to cyclic transitions between a flat and warped disk, similar to what is observed in Cyg X-1 and LMC X-3.

The transient Swift J174510.8−262411 (or Swift J1745−26) was discovered by Swift/BAT in 2012 (Cummings et al. 2012). During this time a variable IR counterpart was identified (Rau et al. 2012b) and an optical counterpart with Hα emission was discovered (Muñoz-Darias et al. 2013). Muñoz-Darias et al. (2013) argue a ${P}_{{\rm{orb}}}\leqslant 21$ hr and a spectral type of the counterpart as A0 or later.

Spectral and timing observations (by Swift and INTEGRAL) suggest that Swift J1745−26 was an LMXB BH system that never left the hard states for the duration for the outburst (Belloni et al. 2012; Grebenev & Sunyaev 2012; Tomsick et al. 2012; Vovk et al. 2012; Krimm et al. 2013c; Sbarufatti et al. 2013), making Swift J1745−26 a "hard-only" outburst source (Curran et al. 2014). A radio counterpart was detected by Miller-Jones & Sivakoff (2012) and both Corbel et al. (2012) and Coriat et al. (2013a) find a spectral index suggestive of optically thick synchrotron emission from a partially self-absorbed compact jet. See Curran et al. (2014) and Tetarenko et al. (2015d) for complete analysis of the evolution of the jet throughout the outburst from radio through sub-mm frequencies.

The new Galactic source IGR J17454−2919 was discovered by the INTEGRAL/JEM-X in 2014 October (Chenevez et al. 2014b) and labelled a new transient XRB source (Chenevez et al. 2014a). Spectral and timing analysis with NuSTAR revealed a hard power-law index, high energy cutoff, and level of variability consistent with that of an accreting BHXB in the hard state (Tendulkar et al. 2014). This result paired with the fact that no evidence for bursts or pulsations were found make the source a probable BHC (Tendulkar et al. 2014).

The transient source 1A 1742−289 was discovered by the Ariel V satellite in 1975 (Eyles et al. 1975). Branduardi et al. (2001) observe similarity in spectral behavior and timescales with transient BH source 4U 1543−475 indicating that 1A 1742−289 contains a compact object (based on high X-ray luminosity) with variable mass transfer from a low mass companion of spectral type M–K. Davies et al. (1976) find a radio counterpart and observe changes in intensity at radio and X-ray wavelengths in 1A 1742−289 similar to behavior observed in Cyg X-1 and 1A 0620−00 (Elvis et al. 1975; Owen et al. 1976). Maeda et al. (1996) give an estimate of ${P}_{{\rm{orb}}}\sim 8.4$ hr. While the nature of the compact object is not known, the source is included in the BHC list of McClintock & Remillard (2006) and therefore is included in our sample.

The new transient source CXOGC J174540.0-290031, located only 0.1 pc in projection from Sgr A*, was discovered by Chandra in 2004 (Muno et al. 2005b). This source has also been detected by XMM-Newton (Belanger et al. 2005). The 7.9 hr orbital modulation detected in the Chandra X-ray light curve and upper limit on the magnitude of the infrared counterpart of $K\gt 16$, coupled with its low peak X-ray luminosity, suggest that it is an LMXB that is being viewed nearly edge-on (Muno et al. 2005a). While the bright radio outburst observed (coincident with the X-ray) suggests that it contains a BH. See Bower et al. (2005) for a detailed analysis of the radio emission.

H 1743−322 was discovered during a bright outburst by the Ariel V (Kaluzienski & Holt 1977) and HEAO–1 (Doxsey et al. 1977) satellites in 1977. The source was classified as a BHC based on its very soft spectrum (White & Marshall 1984).

H 1743−322 was detected in 1984 with EXOSAT (Reynolds et al. 1999), in 1996 with Mir-Kvant (Emelyanov et al. 2000), and again in 2003 by INTEGRAL (Revnivtsev 2003) and RXTE (Markwardt & Swank 2003). In 2003, QPOs of typical BHC frequencies were observed (Homan et al. 2003b) and the system followed the typical "turtlehead" pattern, transitioning through hard and soft accretion states (Capitanio et al. 2005; McClintock et al. 2009).

Detections of the radio (Rupen et al. 2003a) and optical (Steeghs et al. 2003) counterparts followed quickly thereafter. H1743−322 is classified as a micro-quasar, as jets have been detected at both radio and soft X-ray wavelengths (Rupen et al. 2004; Corbel et al. 2005). There exists numerous studies of this source at radio wavelengths (Kalemci et al. 2006; Jonker et al. 2010; Coriat et al. 2011; Miller-Jones et al. 2012). An additional outflow, in the form of an accretion disk wind, has also been detected in this source (Miller et al. 2006b; Ponti et al. 2012).

Since 2003, H 1743−322 has undergone 13 observed outbursts, 3 of which where the source never reached the softer states, we label H 1743−322 as a "hard-only" outburst source. Despite being one of the most well studied BHXBs in the Galaxy, no dynamical confirmation has ever been made on the system (Motta et al. 2010). We adopt the distance estimated by Corbel et al. (2005) from the proper motion of the jet, 10.4 ± 2.9 kpc, for the purpose of our analysis.

XTE J1748−288 was discovered in 1998 by the ASM on board RXTE (Smith et al. 1998b) and BATSE on board CGRO (Harmon et al. 1998). General spectral and timing properties and their evolution during the outburst were typical of BH XRN (Revnivtsev et al. 2000c). Revnivtsev et al. (2000c) and Brocksopp et al. (2007) suggest that the outburst began in the SPL state. As the outburst continued, the source then passed through the SDS and made the transition back to the HCS. In addition, an iron emission line was detected by Kotani et al. (2000) and Miller et al. (2001a).

The optically thin radio counterpart was discovered soon after by Hjellming et al. (1998a), confirmed to be associated with XTE J1748−288 (Fender et al. 1998; Hjellming et al. 1998d) and resolved by the VLA (Rupen et al. 1998). Follow-up work revealed a jet with a velocity $\gt 0.93c$ (Hjellming et al. 1998e), making XTE J1748−288 only the third known Galactic source that has displayed superluminal motion (Brocksopp et al. 2007).

IGR J17497−2821 was discovered with ISGRI on board INTEGRAL in 2006 (Soldi et al. 2006). Assuming a distance of 8 kpc, Kuulkers et al. (2006a) estimated a peak 2–200 keV luminosity of $\sim {10}^{37}\;{\rm{erg}}\;{{\rm{s}}}^{-1}$. Given this luminosity and a source position closely projected toward the Galactic center, this strongly suggests that IGR J17497−2821 was an XRB (Rodriguez et al. 2007). Spectral analysis and the FRED type light curve observed further implied that the source was a BHC in the HCS (Kuulkers et al. 2006b; Walter et al. 2007). The source never left the HCS during this outburst, classifying it as a "hard-only" outburst source (Rodriguez et al. 2007; Walter et al. 2007; Paizis et al. 2009).

Given a refined Chandra position, Paizis et al. (2007b) identified two possible optical/IR counterparts consistent with a B-type star (i.e., HMXB) or a K-giant (which would make the system a symbiotic LMXB). However, given that neither is consistent with the standard LMXB light curve of this source, it is probable that the true counterpart has not yet been identified. No radio counterpart has been found for this source (Rodriguez et al. 2007).

SLX 1746−331 was discovered by the SpaceLab 2 X-ray Telescope in 1985 (Skinner et al. 1990) and detected by the RASS in 1990 (Motch et al. 1998). Both Skinner et al. (1990) and White & van Paradijs (1996) speculated that SLX 1746−331 may harbour a BH based on both its transient nature and its soft spectrum. Further evidence was added to this claim when the source was once again detected in 2003 by the RXTE/PCA bulge scan, finding a very soft spectrum modeled by blackbody emission at ∼1.3 keV (Markwardt 2003b; Remillard & Levine 2003) and by INTEGRAL/ISGRI that found an additional hard component to the spectrum that only contributed at most ∼10% of the flux (Lutovinov et al. 2003a). Spectral and timing analysis showed that the source had made the transition from the soft state back to the hard state, commonly associated with BHCs (Homan & Wijnands 2003).

SLX 1746−331 has since been detected twice more, once in 2007/2008 by both RXTE/PCA (Markwardt & Swank 2007) and INTEGRAL/JEM-X (Kuulkers et al. 2008) and once in late 2010 by MAXI (Ozawa et al. 2011). Both outbursts showed the typical very soft spectrum, commonly associated with BHCs in the soft state.

The X-ray transient XTE J1752−223 was discovered by the ASM on board RXTE in 2009 (Markwardt et al. 2009c). RXTE, MAXI, and Swift monitoring suggested that the source was a BHC as variability was indicative of an imminent state transition (Markwardt et al. 2009a; Nakahira et al. 2009; Remillard 2009; Shaposhnikov et al. 2009, 2010; Shaposhnikov 2010). Later works agreed (Munoz-Darias et al. 2010; Stiele et al. 2011; Reis et al. 2011; Nakahira et al. 2012a).

Optical/NIR observations revealed a possible counterpart (Torres et al. 2009b, 2009d). However, given the crowded field (Russell et al. 2010a), it remains unclear whether this is the true counterpart. Brocksopp et al. (2009) discovered the radio counterpart with a flat spectrum consistent with a compact jet in the hard spectral state. The source appeared to stay in the hard state for an extended period of time and finally transitioned to the softer states (Homan 2010; Curran et al. 2010; Nakahira et al. 2010; Shaposhnikov 2010; Shaposhnikov et al. 2010; Chun et al. 2013). Radio observations of ejection events, often associated with state changes, supported this result (Brocksopp et al. 2010a; Yang et al. 2010a, 2011). There has also been evidence to suggest that XTE J1752−223 contains X-ray jets (Yang et al. 2011). For in-depth radio analysis see Brocksopp et al. (2013). Using correlations between spectral and variability properties with GRO J1655−40 and XTEJ1550−564, Shaposhnikov et al. (2010) estimated a distance of 3.5 ± 0.4 kpc and a BH mass of $9.6\pm 0.9\;{M}_{\odot }$, which we adopt for the purpose of our analysis.

Swift J1753.5−0127 is an X-ray transient discovered in outburst by Swift/BAT in 2005 (Palmer et al. 2005). Soon after, the source was detected in UV, optical, NIR and radio bands (Fender et al. 2005; Halpern 2005; Still et al. 2005b; Torres et al. 2005). The radio observations by Fender et al. (2005) indicated likely compact jet activity. The hard X-ray spectrum, up to ∼600 keV (no NS system has been detected past ∼200 keV; Cadolle Bel et al. 2007), and the detection of QPOs (Morgan et al. 2005), has provided strong evidence that the system harbours a BH (Cadolle Bel et al. 2007).

The system has not returned to quiescence since 2005, and therefore we treat Swift J1753.5−0127 as a persistent source in our analysis. It is usually seen in the HCS, making it a "hard-only" outburst source (Soleri et al. 2012; Shaw et al. 2013; Froning et al. 2014). However, it has been observed to make occasional transitions to the intermediate states before returning back to the hard state, similar to the incomplete state transitions of Cyg X-1 (Soleri et al. 2012).

Swift J1753.5−0127 follows the lower track in the X-ray/radio luminosity plane (Cadolle Bel et al. 2007), along with an increasing number of BH sources (Coriat et al. 2011; Corbel et al. 2013), and it could be the BH with the second shortest orbital period of ∼3.2 hr, according to Zurita et al. (2008). For complete multi-wavelength analysis, see Froning et al. (2014), Cadolle Bel et al. (2007) and for spectral and timing analysis, see Miller et al. (2006a), Soleri et al. (2012), Mostafa et al. (2013).

XTE J1755−324 was discovered by the ASM on board RXTE in 1997 (Remillard et al. 1997). The spectrum was fit with a multicolor disk blackbody at ∼0.7 keV and a hard power-law tail extending to ∼20 keV (Remillard et al. 1997). Given the spectrum, the FRED type light curve behavior (Revnivtsev et al. 1998a), typical of XRN (Tanaka & Shibazaki 1996), and the fact that no Type-I X-ray bursts or pulsations were observed throughout the course of the outburst (Goldoni et al. 1999), the source was suggested as a good BHC (Goldoni et al. 1999). For complete X-ray spectral analysis, see Revnivtsev et al. (1998a) and Goldoni et al. (1999). Goldoni et al. (1999) did not detect any quiescent X-ray emission coincident with the position of the source in the ROSAT All-Sky Bright Source Catalog. No radio counterpart has ever been found (Ogley et al. 1997).

H 1755−338 was discovered by the UHURU satellite when it was active in 1970 (Jones 1977). It had an unusually soft spectrum (White & Marshall 1984; White et al. 1984) and a hard X-ray tail (Pan et al. 1995), which was suggestive of a BHC (Kaaret et al. 2006). H 1755−338 shows X-ray dips indicating a high inclination and a ${P}_{{\rm{orb}}}\sim 4.4$ days (White et al. 1984; Mason et al. 1985). The source was still active in 1993 (Church & Balucinska-Church 1997), but in quiescence in 1996 (Roberts et al. 1996). Therefore the source likely remained active for at least 23 years. We classify this source as persistent (Kaaret et al. 2006).

The distance to the source is likely $\gt 4\;{\rm{kpc}}$, as the optical counterpart, which was identified during outburst by McClintock et al. (1978), was not detected in quiescence (Wachter & Smale 1998) and $\lt 9\;{\rm{kpc}}$, suggested by the low level of visual extinction (Mason et al. 1985). Angelini & White (2003) found a linear structure in the X-ray, roughly symmetric, about the position of the source and extending outwards by ∼3', suggesting the presence of X-ray jets, see Park et al. (2005) and Kaaret et al. (2006).

The hard X-ray source GRS 1758−258 was discovered with GRANAT in 1990 (Mandrou 1990; Sunyaev et al. 1991a). GRS 1758−258 displays a hard power-law spectrum with photon indices ${\rm{\Gamma }}\sim 1.4$–1.9 and a high energy cutoff above ∼100 keV (Kuznetsov et al. 1999; Main et al. 1999; Lin et al. 2000) and strong short term variability up to 10 Hz (Smith et al. 1997; Lin et al. 2000), making it one of only three BHCs that not only remain persistently near their maximum luminosity, with the exception of a few dim states that can last up to several months (Pottschmidt et al. 2006), but also spend most of their time in the HCS. Sometimes a weak soft excess is seen in the spectrum, observed in conjunction with a slightly reduced X-ray flux, thought to be characteristic of the source transitioning into the intermediate states (Mereghetti et al. 1994, 1997; Lin et al. 2000; Heindl & Smith 2002).

From the X-ray properties and radio double sided jet structure (Rodriguez et al. 1992), GRS 1758−258 is classified as a micro-quasar (Pottschmidt et al. 2006). An additional outflow, in the form of an accretion disk wind, has also been detected in this source (Ponti et al. 2012).

Three possible counterparts have been identified, a K0 III giant and two main-sequence A stars (Marti et al. 1998; Eikenberry et al. 2001; Rothstein et al. 2002). The former has been suggested as the most likely counterpart given the ${P}_{{\rm{orb}}}\sim 18.5$ days (Smith et al. 2002a).

Numerous observational campaigns, in both the soft (Mereghetti et al. 1994, 1997; Smith et al. 1997; Main et al. 1999; Smith et al. 2001, 2002a, 2002b; Pottschmidt et al. 2006) and hard (Gilfanov et al. 1993; Kuznetsov et al. 1999; Pottschmidt et al. 2006) X-rays have been undertaken over the years, as well as a multi-wavelength study by Lin et al. (2000).

The X-ray transient XTE J1812−182 (or XMMU J181227.8−181234) was discovered by XMM-Newton in outburst in 2003. After reprocessing of data, this source was found in the RXTE/ASM data as well. Cackett et al. (2006) searched the two Micron All-Sky Survey (2MASS) All-Sky Catalog of Point Sources and the USNO-B1.0 catalog within 6'' of the source position and found no counterpart. They reason that this is most likely due to the large absorption ($\sim {10}^{23}\;{\mathrm{cm}}^{-2}$) in the direction of the source.

The spectrum of the source is fit equally well with an absorbed power law (${\rm{\Gamma }}\sim 2.5$) or a multi-color disk blackbody (∼2 keV) and no pulsations were detected in the timing analysis. A color–color diagram, along with the high absorption found in the direction of the source suggests an HMXB system. However, the power-law spectral index is more typical of an LMXB (Cackett et al. 2006).

The source was detected again in 2008 with RXTE/PCA, and confirmed to be XMMU J181227.8−181234 (Markwardt et al. 2008d; Torres et al. 2008b). The spectrum was again consistent with a highly absorbed power law. While Cackett et al. (2006) speculated that the source was an HMXB, Markwardt et al. (2008d) interpreted the spectrum and variability behavior as being a BHC in a soft state. Following Markwardt et al. (2008d), we include XTE J1812−182 in our sample as a possible BHC.

The hard X-ray transient IGR J18175−1530 was discovered by INTEGRAL in 2007 (Paizis et al. 2007a) and also detected during RXTE/PCA scans of the region and designated XTE J1817−155 (Markwardt et al. 2007). Cheung (2007) discuss the detection of a radio source that may be associated with IGR J18175−1530.

XTE J1817−330 was discovered by RXTE in 2006 (Remillard et al. 2006) and shown to have a very soft spectrum, dominated by the accretion disk component, typical of transient BHCs in the soft state (Sala et al. 2007) Both the radio counterpart (Rupen et al. 2006) and a probable optical (Torres et al. 2006) counterpart were identified. Sala et al. (2007) suggest a spectral type of K–M for the optical counterpart. An outflow, in the form of an accretion disk wind, has also been detected in this source (Ponti et al. 2012).

At the peak of the outburst, the source was in the SDS and then later transitioned back to the HCS as the source intensity gradually decreased (Gierlinski et al. 2008; Roy et al. 2011). QPOs associated with the intermediate states have also been detected in this source (Homan et al. 2006a; Roy et al. 2011). For complete analysis of QPOs present during the outburst of XTE J1817−330, see Sriram et al. (2012).

XTE J1818−245 was discovered by the ASM on board RXTE in 2005. The hardness ratio indicated a very soft spectrum, typical of BHCs (Levine et al. 2005b) and no pulsations were detected (Markwardt et al. 2005). Soon thereafter, the optical (Steeghs et al. 2005b) and radio (Rupen et al. 2005d) counterparts were discovered. The spectral type of the optical counterpart could not be identified as the optical emission was found to be dominated by the accretion disk (Zurita Heras et al. 2011).

Spectral parameters showed behavior typical of the SDS and intermediate states seen in BHXBs, including the usual decrease in disk temperature, increase in inner disk radius and decrease in disk flux as the high-energy flux became stronger, and radio flares associated with discrete ejecta (Cadolle Bel et al. 2009). Based on the above analysis and the observed light curve behavior, Cadolle Bel et al. (2009) concluded that XTE J1818−245 is most likely a LMXB and a BHC.

They estimate a distance of 2.8–4.3 kpc using the model developed by Shahbaz et al. (1998) and the outburst light curve properties of the source. We assume a uniform distribution in this range for the purposes of our analysis.

In 1999 the transient SAX J1819.3−2525 was discovered by BeppoSAX (in't Zand et al. 1999) and RXTE (Markwardt et al. 1999b) with a position consistent with variable star V4641 Sgr. Its optical (Stubbings 1999) and X-ray (Smith et al. 1999a, 1999b) flux increased rapidly and then began to decline within two hours. Emission lines found in both optical and infrared spectra, during this bright X-ray flare (Ayani & Peiris 1999; Charles et al. 1999; Djorgovski et al. 1999; Liller 1999), were reminiscent of accretion onto a compact object.

Shortly thereafter, the radio counterpart was discovered by Hjellming et al. (1999a). This counterpart which declined on timescales of hours to days (Hjellming et al. 1999a, 1999b) and showed ejecta moving with relativistic motion (Hjellming et al. 1999c), which led to SAX J1819.3−2525 being classified as a possible micro-quasar. For a discussion of the rapid X-ray variability occurring at super-Eddington luminosities during this flare, see Wijnands & van der Klis (2000) and Revnivtsev et al. (2002).

Since 1999, major outbursts have been observed in 2000 (Hjellming 2000), 2002 (Uemura et al. 2004), 2003 (Buxton et al. 2003; Maitra & Bailyn 2006), 2004 (Swank 2004) and 2015 (Yoshii et al. 2015). Each outburst was much shorter than typical compact transient systems (Maitra & Bailyn 2006). Weaker flare-like activity has been reported in 2000 (Hjellming 2000; Uemura et al. 2004), 2007 (Cackett & Miller 2007), 2008/2009 (Yamaoka et al. 2008), 2010/2011 (Yamaoka & Nakahira 2010; Yamaoka et al. 2010a, 2010b) and 2014 (Tachibana et al. 2014). An outburst in 1978 was found on photographic plates at the Sternberg Astronomical Institute (Barsukova et al. 2014).

Optical spectroscopy and photometry during quiescence allowed Orosz et al. (2001) to measure a $f(M)\quad =2.74\pm 0.12\;{M}_{\odot }$ and ${P}_{{\rm{orb}}}\sim 2.8$ days, estimate a distance between 7.4 and 12.3 kpc and mass between ∼8.7 and $11.7\;{M}_{\odot }$, and classify V4641 Sgr as a B9III star, in turn making SAX J1819.3−2525 a confirmed BH (Revnivtsev et al. 2002).

Its optical companion, has a mass estimated at ∼5.5–$8.1\;{M}_{\odot }$ (Orosz et al. 2001). However, MacDonald et al. (2011) argue the maintenance of the LMXB label for two reasons. The first, mass transfer occurs via Roche lobe overflow in the system. The second, the optical counterpart, is not more massive than the probable BH in the system.

More recently, MacDonald et al. (2014) have compiled and subsequently separated 10 years of data on this source into passive and active states. They find the passive state data to be dominated by ellipsoidal variations and stable in the shape and variability of the light curve. By fitting ellipsoidal models to the passive state data they find an improved set of dynamical parameters including $i=72\buildrel{\circ}\over{.} 3\pm 4\buildrel{\circ}\over{.} 1$, ${M}_{\mathrm{BH}}=6.4\pm 0.6\;{M}_{\odot }$, and an updated distance of 6.2 ± 0.7 kpc.

MAXI J1836−194 was discovered in August of 2011 simultaneously by MAXI (Negoro et al. 2011b) and Swift/BAT (Ferrigno et al. 2011). Follow-up observations led to the discovery of the optical (Kennea et al. 2011a) and radio (Miller-Jones et al. 2011c) counterparts. The relatively strong radio and IR emission observed was associated with a jet (Miller-Jones et al. 2011c; Trushkin et al. 2011).

Strohmayer & Smith (2011) classified the source as a BHC from a spectrum consistent with a power law of photon index ${\rm{\Gamma }}\sim 1.8$, the presence of an iron line, and a transition from the HCS to the IMS with RXTE/PCA. Ferrigno et al. (2011) found that the source never made the transition from the HCS to the SDS, thereby classifying MAXI J1836−194 as a "hard-only" outburst source (also see Reis et al. 2012). For full multi-wavelength analysis of the outburst see Russell et al. (2013) and Russell et al. (2014a).

Using optical/UV signatures of an accretion disk (Hα, HeII 4686) found in the optical spectra, Russell et al. (2014b) infer a plausible inclination range for the system between 4° and 15°. In addition, Russell et al. (2014b) are also able to place mass and radius constraints on the counterpart, and in-turn derive an upper limit on the orbital period of $\lt 4.9$ hr, using stellar evolution models and the quiescent luminosity limits found from optical photometry.

Swift J1842.5−1124 was discovered with Swift/BAT in 2008 (Krimm et al. 2008d) and studied in detail (Krimm et al. 2008a, 2008c; Racusin et al. 2008). The spectrum of the source was fit with a blackbody (∼0.9 keV) and power-law model (${\rm{\Gamma }}\sim 1.5$), where the blackbody component only contributed ∼6% of the total 2–40 keV flux. Strong QPOs near ∼0.8 Hz were also observed (Markwardt et al. 2008a). A QPO at 8 Hz and a hard spectrum suggested that the source was transitioning from the HCS to the SDS (Krimm et al. 2013c).

As the hard X-ray peak preceded the soft X-ray peak by ∼10 days in the light curve, a behavior also seen in BH sources Swift J1539.2−6227 (Krimm et al. 2011b) and GRO J1655−40, (Brocksopp et al. 2006), suggested the system was a BHC (Krimm et al. 2013c). Swift J1842.5−1124 underwent a later, very weak outburst in 2010 February. Follow-up optical/near-IR observations performed in 2008 revealed a possible candidate counterpart (Torres et al. 2008a).

EXO 1846−031 was discovered by EXOSAT in 1985 (Parmar & White 1985). Parmar et al. (1993) observe an X-ray spectrum well fit with a multi-color disk blackbody and power-law component extending to ∼25 keV as well as significant variability in this hard component, suggesting that EXO 1846−031 is a BHC. While a search for the optical counterpart was performed, no significant conclusions were made due to technical issues that affected the source positions derived (Parmar et al. 1993).

The hard X-ray transient IGR J18539+0727 was discovered by INTEGRAL in 2003 (Lutovinov et al. 2003b). Lutovinov & Revnivtsev (2003) observe an X-ray spectrum fit well with a power law and a fluorescent line at ∼6.4 keV, typical of XRBs in the HCS (Gilfanov et al. 1999). IGR J18539+0727 shows strong flux variability on timescales of tenths to tens of seconds and a break frequency in the power spectrum at $\lt 0.1\;{\rm{Hz}}$. Although most NS systems do not demonstrate a break with this low of a frequency (Wijnands & van der Klis 1999), Linares et al. (2007) have observed an accreting milli-second pulsar that exhibited BH-like X-ray variability, including a break frequency below 0.1 Hz. As such, alone these power spectra properties should not be taken as strong evidence for a BH primary. Given the spectrum and observed properties in the power spectra of the source, Lutovinov & Revnivtsev (2003) suggest IGR J18539+0727 is a BHC.

XTE J1856+053 was discovered by RXTE/PCA in 1996 (Marshall et al. 1996). The RXTE/ASM light curve showed two peaks, separated by ∼4.5 months. The first in April displayed a symmetric shape, and the second in September displayed a FRED pattern (Remillard 1999). Both outbursts are classified as successful, especially when the HIDs are also considered (Sala et al. 2008).

In 2007, XTE J1856+053 was again detected by RXTE (Levine & Remillard 2007). Much like the 1996 outburst, two peaks were observe but this time they were separated by only a few weeks and therefore are counted as only one outburst (Sala et al. 2008). Sala et al. (2008) observed that the X-ray spectrum of the 2007 outburst was dominated by emission from the accretion disk, consistent with a BHC in the soft state (McClintock & Remillard 2006). Our algorithm detected a fourth outburst in 2009. This short outburst, lasting ∼24 days, is not found in the literature.

More recently, in 2015 March, XTE J1856+053 was observed in outburst again, this time undergoing a short-lived "hard-only" outburst. Originally detected by MAXI (Suzuki et al. 2015), only basic spectral analysis from Swift/XRT is available for this outburst (Sanna et al. 2015). Similar to the 1996 and 2007 events, a second outburst, occurring ∼70 days after the onset of the first outburst, was also observed. During this outburst, which was brighter than the first, the source appears to have entered the soft state (Negoro et al. 2015a).

Sala et al. (2008) classify XTE J1856+053 as a LMXB based on its non-detection at IR wavelengths, therefore ruling out a massive companion. They also suggest a BH primary based on the low temperature of the accretion disk and their rough estimate of an accretor mass range between 1.3 and $4.2\;{M}_{\odot }$.

XTE J1859+226 was discovered by the ASM on board RXTE in 1999 (Wood et al. 1999). Follow-up RXTE/PCA observations exhibited a hard power-law spectrum and the existence of QPOs of frequency 0.45 Hz (Markwardt et al. 1999a). BATSE observations confirmed the hard spectrum extending up to ∼200 keV and revealed that the hard X-ray flux peaked while the soft X-ray flux was still rising (McCollough & Wilson 1999). A series of soft flares were seen (Focke et al. 2000), in which QPOs were detected, at 6–7 Hz and 82–187 Hz (Cui et al. 2000). Further analysis of the outburst (Brocksopp et al. 2002; Casella et al. 2004; Farinelli et al. 2013) suggested a likely BH primary in the system.

Radio (Pooley & Fender 1997) and optical (Garnavich et al. 1999) counterparts were discovered. Brocksopp et al. (2002) found a series of radio ejections occurred simultaneously with spectral hardening of the source, suggesting a disk/jet connection. Garnavich & Quinn (2000) and Sanchez-Fernandez et al. (2000) searched the optical photometry finding a potential ${P}_{{\rm{orb}}}\sim 9.2$ hr. Filippenko & Chornock (2001) determine a mass function for XTE J1859+226, indicating the BH nature of the primary.

More recently, Corral-Santana et al. (2011) perform optical photometry and spectroscopy of XTE J1859+226 and find an ${P}_{\mathrm{orb}}\sim 6.6$ hr, a companion spectral type of K5–7 V, and a $f(M)=4.5\pm 0.6\;{M}_{\odot }$ which, while lower than the original estimate, still requires a BH primary in the system. Corral-Santana et al. (2011) fit a star-only model to find an $i=60^\circ$. As their data is consistent with the passive state (Kreidberg et al. 2012), we adopt the Corral-Santana et al. (2011) value for the inclination and use it to calculate ${M}_{{\rm{BH}}}$.

The distance to XTE J1859+226 remains problematic. Zurita et al. (2002) estimate 11 kpc based on the brightness of the outburst and the quiescent optical counterpart. However, this estimate is based on assumed values for both the orbital period and companion spectral type (Filippenko & Chornock 2001). Markwardt (2001) use a combination of spectral and timing information to estimate ∼5–13 kpc and Hynes et al. (2000a), using fits to optical-UV spectral energy distribution, find an estimate of ∼4.6–8.0 kpc. Due to the uncertainty that still exists in the system parameters, we follow Hynes (2005) in believing that the distances estimated during outburst are more reliable and thus adopt their distance estimate of 8 ± 3 kpc.

XTE J1901+014 was discovered by the ASM on board RXTE in 2002 (Remillard & Smith 2002). This outburst lasted between 2 minutes and 3 hr. Later, when reanalyzing RXTE/ASM data, Remillard & Smith (2002) found a previous outburst of the source occurring in 1997, lasting between 6 minutes and 8 hr.

Karasev et al. (2007) performed spectral and timing analysis on the 1997 and 2002 outbursts, arguing that they are not type I X-ray bursts. They fit the spectrum with a power law of photon index ${\rm{\Gamma }}\sim 2.3$, finding no cutoffs at energies between 20 and 30 keV (typical of pulsars; Filippova et al. 2005), or any emission lines in the spectrum. They conclude that such a spectrum, indicates that the primary is in fact a BH. Karasev et al. (2007) suggest that the outbursts are similar in spectral and timing properties to those observed in Galactic BH source SAX J1819.3−2525 (Stubbings & Pearce 1999) and to a lesser degree, the outbursts of fast transients like SAX J1818.9−1703 (Grebenev & Sunyaev 2005), concluding that XTE J1901+014 may be a fast X-ray transient containing a BH.

XTE J1901+014 once again became active in 2006. Karasev et al. (2008) found a power-law spectrum consistent with accreting XRBs. They failed to find the optical counterpart, estimating upper limits on the magnitude of ∼23.5 in the $r^{\prime}$ band and ∼24.5 in the I band. In 2010, Swift/BAT detected XTE J1901+014 in outburst for the fourth time. his outburst lasted at least 2.5 minutes. The spectrum was consistent with the other three outbursts (Krimm et al. 2010). It is interesting to note that this source is generally detected as a low-level persistent source in the BAT survey.¹⁰

XTE J1908+094 was discovered serendipitously in observations of SGR 1900+14 with RXTE/PCA in 2002. The spectrum was consistent with an absorbed power law of photon index ${\rm{\Gamma }}\sim 1.6$ and no pulsations were detected suggesting that XTE J1908+094 was an XRB containing a BH primary (Woods et al. 2002). Subsequent BeppoSAX observations (in't Zand et al. 2002a) confirmed the hard spectrum, extending up to ∼250 keV, and the high Galactic absorption. Detailed spectral and timing analysis by in't Zand et al. (2002c) and Gogus et al. (2004), which confirmed that the source passed through both the HCS and SDS during outburst, agreed with Woods et al. (2002) that XTE J1908+094 is a BHC. A second outburst from this source was detected in 2013/2014 (Coriat et al. 2013b; Krimm et al. 2013a, 2013b; Miller-Jones et al. 2013a).

The radio counterpart was discovered by Rupen et al. (2002). Jonker et al. (2004) analyzed simultaneous X-ray and radio observations during the outburst decay and discuss the X-ray/radio correlation.

A likely near-IR counterpart was detected (Chaty et al. 2002; Garnavich et al. 2002; Wagner & Starfield 2002). Once the source had returned to quiescence Chaty et al. (2006) performed NIR observations with CFHT, finding two possible counterparts (separated by 08) within the X-ray error circle. They postulate that the companion star could be either (i) an intermediate/late type main-sequence star of spectral type A–K, located between 3 and 10 kpc; or (ii) a late-type main-sequence star of spectral type K or later, located between 1 and 3 kpc. They favor the former due to an independently determined lower limit on distance of 3 kpc derived by in't Zand et al. (2002c) from the peak bolometric flux. We follow the suggestion by Chaty et al. (2006) and take the distance to be in the range 3–10 kpc.

Swift J1910.2−0546 was simultaneously discovered by Swift/BAT (Krimm et al. 2012) and MAXI (as MAXI J1910−057; Usui et al. 2012) in 2012. Rau et al. (2012a) detected the optical/NIR counterpart. Lloyd et al. (2012) and Casares et al. (2012b) report possible periodic variations in the optical light curve, which could be attributed to orbital variations, of ∼2.2 hr and ∼4 hr, respectively.

The complex light curve (Krimm et al. 2013c) of the source, as well as spectral analysis from MAXI (Kimura et al. 2012; Nakahira et al. 2012b) and INTEGRAL (King et al. 2012a), show progression through state transitions and the mirrored behavior of the hard and soft X-ray flux. For this reason Krimm et al. (2013c) tentatively suggest that Swift J1910.2−0546 is a BHC.

The Galactic micro-quasar SS 433 was first discovered in a survey of stars exhibiting Hα mission (Stephenson & Sanduleak 1977) and later identified as a variable X-ray source in 1978 by the UHURU satellite (Margon 1980). SS433 is unique among XRBs. The main property that distinguishes it from other "normal" XRBs, is that the system remains in a continuous regime of supercritical accretion onto the compact object. Basically all photometric and spectroscopic properties of the system are determined by the accretion disk and its orientation. See Fabrika (2004) for an extensive review of the system.

In the X-ray, SS 433 is a weak source, generally not observed past ∼30 keV (Nandi et al. 2005). The highly erratic spectral and temporal behavior of the system combined with the internal complexity of the system (Brinkmann et al. 1989) makes the nature of the compact object and companion uncertain (Fabrika & Bychkova 1990).

The basic picture involves an evolved binary undergoing extensive mass-transfer. The secondary feeds an enlarged accretion disk around a compact object (NS or BH; Blundell et al. 2001). Some of this mass is directed through the disk toward the oppositely facing relativistic jets. We observe red and blueshifted optical lines, indicating material is accelerated by the jets (Fabian & Rees 1979; Milgrom 1979; Gies et al. 2002).

Margon (1984) successfully fit a precessing jet model to these lines, finding that the jets moved near constant velocity of $\sim 0.26c$ and had a precession periodicity of 162.15 days. Radio imaging of the source, showed twin processing jets with structure on scales ranging from milliarcseconds to arcseconds (Hjellming & Johnson 1981; Vermeulen et al. 1987, 1993; Fejes et al. 1988). The system is known to have a 13 day orbital period. This photometric periodicity, originally identified as orbital by Crampton et al. (1980), Crampton & Hutchings (1981), is further supported by both the observation of eclipses at 13 days (Cherepashchuk 1981), and the X-ray dimming observed during the optical eclipses (Stewart et al. 1987).

SS 433 is believed to be a binary system consisting of a compact object and an O or B type star (Margon 1984). Extensive arguments for the nature of the compact object being a BH (Zwitter & Calvani 1989; Fabrika & Bychkova 1990; Lopez et al. 2006; Hillwig & Gies 2008; Kubota et al. 2010; Goranskij 2011) and a NS (Filippenko et al. 1988; D'Odorico et al. 1991; Goranskij 2011) have been made. The currently favored distance to the source is 5.5 kpc. Originally estimated by Hjellming & Johnson (1981) (and later Blundell & Bowler 2004) using the proper motion of the jets, this distance is now also supported by observations of HI absorption occurring as a result of a gas cloud interacting with the related supernova remnant W50 (Lockman et al. 2007).

GRS 1915+105 was discovered in 1992 by the WATCH ASM on board GRANAT (Castro-Tirado et al. 1992) and has remained active ever since (Belloni & Altamirano 2013). The system exhibits very peculiar behavior, in the form of complex structured variability, such as QPOs ranging in frequency from ${10}^{-3}$ to 67 Hz and patterns of dips and rapid transitions between high and low intensity in the light curve (Greiner et al. 1996; Morgan et al. 1997; Belloni et al. 2000; Klein-Wolt et al. 2002; Fender & Belloni 2004; Hannikainen et al. 2005). Currently, there are 14 known variability classes (Belloni et al. 2000; Klein-Wolt et al. 2002; Hannikainen et al. 2005). Modeling of this X-ray variability has led to major insights into the connection between the accretion disk and relativistic jets in XRBs (Belloni et al. 1997; Klein-Wolt et al. 2002). Its luminosity is estimated to be near Eddington (Mirabel & Rodríguez 1999).

The probable optical counterpart was discovered by Boer et al. (1996) and the radio counterpart was found with the VLA (Mirabel et al. 1993). Further radio monitoring revealed structures travelling at superluminal speed (Mirabel & Rodríguez 1994), making GRS 1915+105 the first superluminal source in the Galaxy. GRS 1915+105 was suggested to harbour a BH based on its similarity with GRO J1655−40, the second Galactic source to exhibit superluminal motion for which the dynamical mass estimate proved the presence of a BH (Bailyn et al. 1995b). GRS 1915+105 also exhibits a second type of outflow in the form of an accretion disk wind (Ponti et al. 2012). Studies of the accretion disk wind in GRS 1915+105 have suggested that these winds could act as the jet suppression mechanism in the soft states (Neilsen & Lee 2009).

The binary system parameters for GRS 1915+105 remained elusive, despite extensive observational effort (Castro-Tirado et al. 1996; Mirabel et al. 1997; Eikenberry et al. 1998; Marti et al. 2000a; Greiner et al. 2001b; Harlaftis et al. 2001), until Greiner et al. (2001a) obtained a radial velocity curve, ${P}_{{\rm{orb}}}$ and therefore $f(M)$.

Inclination has been estimated based on the orientation of the jets ($i=70^\circ \pm 2^\circ$ from Mirabel & Rodríguez 1994 or $i=66^\circ \pm 2^\circ$ from Fender et al. 1999b). These parameters, along with the mass ratio estimated by Harlaftis & Greiner (2004), allowed for a dynamical mass estimate confirming the BH accretor. A trigonometric parallax measurement by Reid et al. (2014) on the system yielded a direct distance measurement and a revised estimate for BH mass.

Cygnus X-1, one of the brightest X-ray sources in the sky, was discovered in the X-rays by the UHURU satellite in 1971 (Tananbaum et al. 1972). The X-ray emission has been shown to exhibit strong variability on timescales from millliseconds to months (Priedhorsky et al. 1983; Miyamoto & Kitamoto 1989).

Numerous spectral and timing studies over the years (Holt et al. 1979; Ling et al. 1983; Barr & van der Woerd 1990; Belloni & Hasinger 1990; Kitamoto et al. 1990a; Ubertini et al. 1991a; Miyamoto et al. 1992; Gierlinski et al. 1997, 2010; Zdziarski et al. 2002; Pottschmidt et al. 2003; Wilms et al. 2006; Grinberg et al. 2013) have shown Cyg X-1 to spend most of its time in the HCS, resulting in its X-ray spectrum never fully being disk dominated (Grinberg et al. 2013). Cyg X-1 often undergoes "incomplete" state transitions, never fully transitioning to the softer states (Pottschmidt et al. 2003). This extended hard state of Cygnus X−1 shows weak and persistent radio emission, which has been resolved to be a steady jet (Stirling et al. 2001). In fact, the "incomplete" transitions exhibited by this source are thought to be connected to the radio jet, as jet activity is thought to be quenched in the soft accretion states (Tananbaum et al. 1972).

With this being said, the source has been occasionally observed in the soft state (e.g., see Grinberg et al. 2011, 2014b; Yamada et al. 2013; Jourdain et al. 2014) and there is evidence for the presence of a weak compact jet in the softer states of Cyg X-1 (Rushton et al. 2012). For a complete list of X-ray studies see Table 14. For studies of Cyg X-1 at radio wavelengths, see Hjellming & Wade (1971b), Hjellming (1973), Tananbaum et al. (1972), Stirling et al. (2001), Gleissner et al. (2004), Fender et al. (2004, 2006).

This system is known to contain a O9.7Iab type supergiant companion (Gies & Bolton 1986), which orbits around a compact object with a period of ∼5.6 days (Holt et al. 1979). Over the past 44 years, many estimates on the mass of the compact object have been made (Orosz et al. 2011a). While there exist several low-mass models (Trimble et al. 1973; Bolton 1975), all conventional models, which assume an O-type supergiant companion (Paczynski 1974; Gies & Bolton 1986; Ninkov et al. 1987; Caballero-Nieves et al. 2009), find a large (and uncertain) mass of the compact object exceeding $\sim 3\;{M}_{\odot }$, therefore confirming a BH primary. (Kalogera & Baym 1996).

The large range of these mass estimates is mainly due to the large uncertainty in distance to the source (Reid et al. 2011). With the more recent trigonometric parallax distance measurement calculated by Reid et al. (2011), Orosz et al. (2011a) was able to build a complete improved dynamical model for Cyg X-1, including ${M}_{{\rm{BH}}}$, q and i. We make use of this dynamical model for the purposes of our analysis.

4U 1957+115 was discovered by UHURU in 1973 (Giacconi et al. 1974). Despite having an X-ray brightness that is comparable to (and occasionally larger) than the well-studied persistent BH sources LMC X-1 and LMC X-3, little is known about the nature of the system (Nowak et al. 2008).

Optical spectra reveal a power-law continuum with Hα, Hβ and He ii 4686 Å emission lines (Cowley et al. 1988; Shahbaz et al. 1996a), typical of systems dominated by an accretion disk (Thorstensen 1987). Long-term variations in the optical light curve have revealed modulations (interpreted as evidence for an accretion disk with a large outer rim (possibly due to a warp) that is seen close to edge on; Hakala et al. 1999) with a 9.33 hr period, generally believed to be the orbital period of the system (Thorstensen 1987).

The short orbital period, indicative of a late-type, main-sequence star as a companion, and the absence of X-ray eclipses, has allowed for an upper limit estimate of inclination between 70° and 75°, which is consistent with the model of optical variability (Hakala et al. 1999).

4U 1957+115 was first classified as a possible BHC in 1984 when EXOSAT observations revealed a very soft spectrum (Ricci et al. 1995). Observations by Nowak et al. (2008), who analyzed the complete set of available data from RXTE, and more recently, Nowak et al. (2012), reveal that the X-ray spectrum is a pure disk spectrum ∼85% of the time, with the remaining ∼15% involving some non-thermal component.

This dominant soft spectrum, coupled with the observed low fractional variability (Nowak & Wilms 1999; Wijnands et al. 2002; Nowak et al. 2008), both characteristic of the soft state, make 4U 1957+115 one of only three BHCs that not only remains persistently in outburst but also spends most of its time in the soft disk-dominated accretion state (Nowak et al. 2008). Further evidence for this behavior is implied from the recent radio non-detection (Russell et al. 2011b), as jet production is believed to be quenched in the soft state. While 4U 1957+115 may not show evidence for a relativistic jet, this source has been observed to exhibit an outflow in the form of an accretion disk wind (Ponti et al. 2012). As there exists no dynamical mass or distance measurements for the system, X-ray and optical observations, analyzed at different times, have been used to argue whether the compact object is a BH (Wijnands et al. 2002; Nowak et al. 2008, 2012; Maitra et al. 2013; Gomez et al. 2015) or NS (Yaqoob et al. 1993; Ricci et al. 1995; Robinson et al. 2012). Regardless of the uncertainty, we include this system in our sample as a possible BHC.

GS 2000+251 was discovered by the ASM on board GINGA in 1988 (Tsunemi et al. 1989). The source has been observed to exhibit spectral and temporal characteristics similar to other X-ray nova systems believed to contain BH primaries (Tanaka & Lewin 1995; van Paradijs & McClintock 1995). Hjellming et al. (1988) found a transient radio source associated with GS 2000+25 exhibiting a spectrum that was fit well with a synchrotron model, similar to the radio emission observed in 1A 0620−00 (Owen et al. 1976), suggesting the possibility of a radio jet in the system. Optical photometry after this outburst revealed that the system had a ∼8.3 hr orbital period (Chevalier & Ilovaisky 1993). Dynamical measurements first obtained by Filippenko et al. (1995) and later improved upon by Casares et al. (1995a) and Harlaftis et al. (1996) revealed a $f(M)=5.01\pm 0.12$, subsequently confirming the BH nature of the primary. The distance to this source is estimated by Barret et al. (1996b).

Ioannou et al. (2004) has performed the most extensive study of ellipsoidal variability. They measure an inclination of $54^\circ \lt i\lt 60^\circ$, which is consistent with estimates by Callanan et al. (1996) and Beekman et al. (1996). However, while Kreidberg et al. (2012) conclude that the source was passive during their observations, they suggest their inclination measurement is depressed due to the binning of the light curves, which can slightly decrease the amplitude of ellipsoidal variations. Following Kreidberg et al. (2012) we adopt $55^\circ \lt i\lt 65^\circ$ from Callanan et al. (1996), and assume a uniform distribution across this range to calculate ${M}_{{\rm{BH}}}$.

The transient X-ray source XTE J2012+381 was discovered with the ASM on board RXTE in 1998 during outburst (Remillard et al. 1998). The light curve exhibited FRED like behavior (Chen et al. 1997) and the X-ray spectrum was well described by a combination multicolor disk blackbody ($T\sim 0.76$ keV) and power law with photon index ${\rm{\Gamma }}\sim 2.9$ (White et al. 1998), characteristic of BHXBs (McClintock & Remillard 2006). In addition, extensive spectral and timing analysis of this outburst reveal the source exhibited the typical "turtlehead" evolution seen in BHXBs, transitioning between the hard and soft states (Vasiliev et al. 2000; Campana et al. 2001). The radio counterpart was discovered by Hjellming et al. (1998b), while Hynes et al. (1999) identified a probable optical counterpart with a faint red star heavily blended with a brighter foreground star. Given the light curve behavior, spectral and timing characteristics, XTE J2012+381 is considered a BHC.

GS 2023+338 (V404 Cyg) was discovered with the GINGA satellite in 1989 during outburst (Makino 1989). V404 Cyg is one of the most well-known transient X-ray sources due to both its high X-ray luminosity and levels of variability at many different wavelengths in outburst and quiescence (Tanaka & Lewin 1995; Hynes et al. 2002). The many X-ray observations (Kitamoto et al. 1989; in't Zand et al. 1992; Miyamoto et al. 1992, 1993) that exist have shown that, despite its high X-ray luminosity, no soft component exists in the spectrum (Oosterbroek et al. 1997), suggesting that V404 Cyg is a "hard-only" outburst source (Brocksopp et al. 2004). While this fact has been disputed by Zycki et al. (1999), who claim that the source spent a short period of time in the soft state, we consider the 1989 event as a possible "hard-only" outburst. Two additional outbursts of this source have been found on photographic plates at the Sonnenberg Observatory (Richer 1987). V404 Cyg began its fourth recorded outburst in June 2015 (Kuulkers et al. 2015; Negoro et al. 2015b).

The source has also been detected at radio (Hjellming & Han 1989) wavelengths. The persistent radio emission displays a flat spectrum, which is indicative of a self-absorbed synchrotron jet (Gallo et al. 2005; Miller-Jones et al. 2008). In addition, another type of outflow in the form of an accretion disk wind has also been observed in this source (Oosterbroek et al. 1997). The optical counterpart was identified by Wagner et al. (1989) as Nova Cygni 1938. Further optical observations led to a calculation of the mass function by Casares et al. (1992), which was later refined by Casares & Charles (1994) to be $f(M)=6.08\pm 0.06\;{M}_{\odot }$, confirming the BH nature of the primary. Shahbaz et al. (1994b) modeled ellipsoidal variations of the source obtaining a ∼6.5 day orbital period. More recently, Miller-Jones et al. (2009) obtained a distance to V404 Cyg of $d=2.39\pm 0.14\;{\rm{kpc}}$ using trigonometric parallax, making this the first accurate parallax distance measurement to a BH system. We make use of this distance, which is significantly lower than the previously accepted values, for the purposes of our analysis.

The light curve of GS 2023+338 exhibits strong aperiodic variability making it difficult to obtain precise inclination measurements for the system. Wagner et al. (1992) and Shahbaz et al. (1994a) constrained the inclination to $50^\circ \lt i\lt 80^\circ$ and $45^\circ \lt i\lt 83^\circ$, respectively. Wagner et al. (1992) obtained their lower limit from observations of Balmer spectral lines and the upper limit based on the lack of eclipses. While Shahbaz et al. (1994a) obtain their range by fitting a star only model, which provided a poor fit due to incorrect color correction. In addition, while Sanwal et al. (1996) fit a star only model to the IR data, the authors note hour timescale variability in the light curve, suggesting the source is most likely in the active state. We therefore adopt the inclination calculated by Kreidberg et al. (2012), which makes use of the inclination estimate from Sanwal et al. (1996) of $i\gt 62^\circ$ to obtain a corrected inclination estimate to calculate a ${M}_{\mathrm{BH}}$.

The X-ray source Cyg X-3 was discovered in 1967 (Giacconi et al. 1967). It is the only known Galactic XRB containing a compact object and Wolf–Rayet counterpart (van Kerkwijk 1993; van Kerkwijk et al. 1996; Fender et al. 1999c). Despite being an HMXB, Cyg X-3 has an unusually short orbital period of only 4.8 hr (van Kerkwijk et al. 1996). The distance to the system is estimated at ∼7–9 kpc (Dickey 1983; Predehl et al. 2000; Ling et al. 2009).

Cyg X-3 is a persistently accreting X-ray source that can be observed in both the hard and soft spectral states (Hjalmarsdotter et al. 2008; Szostek & Zdziarski 2008; Szostek et al. 2008; Koljonen et al. 2010). In addition, it is also the brightest radio source among XRBs (McCollough et al. 1999), exhibiting both strong radio flares as well as resolved jets (Watanabe et al. 1994; Marti et al. 2001; Mioduszewski & Rupen 2001). The true nature of the compact object in the system remains uncertain (e.g., BH or NS; Hanson et al. 2000; Vilhu et al. 2009; Zdziarski et al. 2013) due to the lack of reliable estimates for the mass function and inclination. However, a BH is favored over a NS given that (i) the X-ray spectral evolution resembles the typical "turtlehead" pattern through BH accretion states, and (ii) the radio/X-ray correlation closely corresponds to that found in BHXB systems (Szostek et al. 2008).

In 2010, the emission line Be star MWC 656 was found within the error circle of the gamma-ray source AGL J2241+4454 with the AGILE satellite (Lucarelli et al. 2010; Williams et al. 2010). Williams et al. (2010) noted the similarity of its spectral properties and rapid Hα variations (unusually fast for Be stars) to the gamma-ray binary LSI +61°303 and analyzed available optical photometry, finding a variation with a period of 60 days, which they interpreted as an orbital modulation of the flux from the disk surrounding the Be star. The distance to the star is known to be 2.6 ± 0.6 kpc (Williams et al. 2010).

Performing a radial velocity study, Casares et al. (2012a) test the binary hypothesis, estimate an inclination of $i\gt 66^\circ$ and a compact object mass. Later, with new optical observations Casares et al. (2014) improve upon the previous radial velocity curve and determine a more precise set of binary parameters including a spectral classification of the Be companion of B1.5-B2 III, a mass of the companion star of 10–16 M_⊙, and a compact object mass of 3.8–6.9 M_⊙. This evidence makes MWC 656 the first know Be XRB to contain a BH.

The source has also been detected at X-ray (Munar-Adrover et al. 2014) and radio wavelengths (Dzib et al. 2015). Both radio and X-ray luminosities of the source agree with the behavior of BHXBs in the hard and quiescent state, and the source occupies the same region of the radio/X-ray plane that the faintest known BHs do (e.g., 1A0620−00 and XTE J1118+480), making it a strong BHC.

The RXTE (Swank 1997) satellite was perhaps the most important vehicle for the study of transient phenomena in the last decade due to the wide-sky coverage of the ASM, high sensitivity of the PCA and its overall fast response time (McClintock & Remillard 2006). The ASM (Levine et al. 1996) on board RXTE, made up of three wide-field proportional counters, operated in the 1.5–12 keV band from 1996–2012 and had the ability to cover ∼90% of the sky every orbit, which took about 90 minutes, with a sensitivity between ∼10–20 mCrab (integrating all orbits over a full day)(McClintock & Remillard 2006).

The BAT X-ray monitor (Krimm et al. 2013c) has provided near real time, wide-field (2 steradians) coverage of the X-ray sky in the 15–150 keV energy range, with an energy resolution of 5% at 60 keV, since 2005. The BAT has the ability to observe 80%–90% of the sky every day with a sensitivity of 16 mCrab (integrating scans over 1 day) and arcminute positional accuracy. One of the key characteristics of the Swift satellite is the ability to "swiftly" (≲90 s) and autonomously repoint itself after detection by BAT to bring the source within the field of view of the sensitive narrow-field X-ray and UV/optical instruments that are also on board the observatory.

MAXI (Matsuoka et al. 2009), mounted on the International Space Station (ISS) has the ability to scan 85% of the sky every 92 minutes (one orbit/rotation period of the ISS) with its wide field of view providing near real-time coverage with a positional accuracy of $\lt 6$ arcmin and a daily sensitivity of 9 mCrab. The Gas Slit Camera (GSC; Mihara et al. 2011) detector, one of the two ASMs on board MAXI, contains a proportional counter that covers the 2–20 keV energy band with its large detection area (5000 ${{\rm{cm}}}^{2}$) and an energy resolution of 18% at 6 keV.

The INTEGRAL (Winkler et al. 2003; Kuulkers et al. 2007) Monitoring Program has provided periodic scans of the Galactic Bulge since 2005. Data is taken approximately every 3 days (the length of one orbit) and is provided in the form of single observations, consisting of 7 pointings in a hexagonal pattern of spacing 2°, and lasting ∼1.8 ks total. The INTEGRAL satellite contains three coded mask imagers. One of which is the Integral Soft Gamma-Ray Imager (IBIS/ISGRI; Ubertini et al. 2003), which has a primary energy range of 20–60 keV, an energy resolution of 8% at 100 keV, and a field of view of 29 square degrees. The other two are the Joint European X-ray Monitor (JEM-X; Lund et al. 2003) X-ray detectors, which have a primary energy range of 3–35 keV, an energy resolution of 9% and 13% at 30 keV, and a circular field of view of diameter 10°. Note that within each observation, only one JEM-X unit is used at a time. Each observation covers 0.1% of the sky with JEM-X and 2% of the sky with ISGRI, with a sensitivity of 3–9 mCrab.

The Galactic Bulge Scan Survey, which used the PCA (Jahoda et al. 1996; Swank & Markwardt 2001) on board RXTE (see Section 3.3.2), provided periodic scans of the Galactic bulge region in the 2.5–10 keV energy band between 1999 and 2011. Each scan covered ∼8% of the sky with a sensitivity of ∼3–7 mCrab (estimated from real observations; see Table 1 for details).

HEXTE (Rothschild et al. 1998) on board RXTE provided high energy coverage of the X-ray sky in the 15–250 keV energy band, with an energy resolution of 15% at 60 keV, from 1996–2012. HEXTE consisted of two clusters of detectors, each of which contained four NaI(Tl)/CsI(Na) phoswich scintillation counters. Each cluster had the ability to "rock" along mutually orthogonal directions, providing background measurements 15 or 30 away from the source every 16 to 128 s. Overall, HEXTE was capable of measuring a 100 mCrab X-ray source to 100 keV or greater in 10³ live seconds. The field of view per cluster was 1°, and all 8 detectors in both clusters covered a net open area of 1600 ${\mathrm{cm}}^{2}$.

The PCA (Jahoda et al. 1996) on board RXTE consisted of an array of 5 proportional counters, which operated in the 2–60 keV¹⁶ range, with an energy resolution of 18% at 6 keV, between 1996 and 2012. It had a total collecting area of $6500{{\rm{cm}}}^{2}$, a field of view of 16 square degrees and a sensitivity of 1mCrab (estimated from observations; see Heinke et al. 2010).

The purpose of data cleaning is two-fold. As we have found the background subtraction performed by the surveys to be inadequate and the quoted errors to be underestimated across all four telescopes, it becomes necessary to perform a further background subtraction on the data and include an additional factor in the treatment of the errors. The following analysis is performed on daily averaged data. If the data is collected in the form of orbital data, it is first converted to daily average data by calculating the weighted mean count rate (and uncertainty) over each day of data available (see Bevington & Robinson 2003).

In the case of the transient sources, the algorithm must first be able to identify times when a source was in quiescence for each energy band and instrument combination. To do so, it begins by calculating a background rate ${R}_{{\rm{bg}}}$ using an iterative bi-weight²³ method. This method, which makes use of σ-clipping, allows for a determination of both the location and scale (i.e., mean and standard deviation) of all the long-term light curve data for the source. This task is performed in part using the astLib.astStats²⁴ python module, which makes use of methods described in Beers et al. (1990) to provide robust estimations of location and scale (i.e., robust equivalents of mean and standard deviation). From here, quiescence is defined as times in which the count rate is less then $3\sigma$ above the background rate, ${R}_{{\rm{bg}}}$.

Generally, if the quoted errors ${\sigma }_{{\rm{quo}}}$ are correct, you would expect the quiescent data to follow a Gaussian distribution (Bevington & Robinson 2003). However, this is not observed in our data sets, which becomes a serious problem when the algorithm begins to identify individual outbursts (see Section 3.7.2). To remedy this problem, an ad hoc method is employed, defining a correction factor ${C}_{\sigma }$ as the standard deviation of a Gaussian distribution fit to the quiescent non-detection data. ${C}_{\sigma }$ then acts as a multiplicative factor, scaling up the quoted errors appropriately to be ${\sigma }_{{\rm{corr}}}={C}_{\sigma }{\sigma }_{{\rm{quo}}}$.

To ensure that a transient source is not in fact being detected during times that the algorithm has defined as quiescence (which would effectively decrease the value of our errors), we have performed numerous trials comparing the algorithm results to pointed observations with more sensitive instruments found in the literature for the three most well studied recurrent transient sources in the Galaxy: 4U 1630−472, GX 339−4, and H 1743−322.

While the above analysis is adequate for the transient sources, the persistent sources must be handled differently. In this case, the algorithm makes use of background estimates and error corrections (i.e., the standard deviation of the distribution of observed quiescence count rates) reflective of the local transient source population. Here it takes the average background rate and average error correction for all the transient sources within a 16° radius of each persistent source, per energy band, to be the background rate and error correction for each persistent source, respectively. The 16° radius was specifically chosen to ensure that at least two transient sources are used in the background/error correction estimation of each persistent source. The LMC sources are the obvious exception to this criterion as there are no nearby transients from our sample. In their case the algorithm makes use of all transient sources in our sample to compute the background/error correction estimates.

The detection stage begins by first performing a second background subtraction (via the method outlined in Section 3.7.1) on the now error-corrected data, followed by differentiating the data into two separate catagories, (i) detection: in which the count rate ${R}_{c}\geqslant {R}_{{\rm{bg}}}+3{\sigma }_{{\rm{corr}}};$ and (ii) non-detection: in which ${R}_{c}\lt {R}_{{\rm{bg}}}+3{\sigma }_{{\rm{corr}}}$. Note that negative count rates, which happen on occasion due to over-subtraction of the background are included in the non-detections. From here the algorithm produces lists of individual outbursts detected, in every given energy band (see Table 14), based on the minimum criteria that there must be at least 2 detections occurring every 8 days to be counted as an outburst. We set a minimum number of detections to eliminate a large number of spurious "outbursts." This criterion is justified given that we do not currently know of any outbursts from BH transients lasting less than 2 days. Lists of outbursts detected per energy band are then combined, creating outburst lists per instrument, then per telescope, and lastly into a final outburst list for each individual source, taking into account detections from all four telescopes in our data sample (where available).

The detection stage of the algorithm is equipped to deal with (i) situations in which large gaps in the data exist (largely due to Sun constraints), there is a lack of continuous daily coverage (e.g., survey instruments), and there is known down time (e.g., Space Shuttle docked at the ISS affecting MAXI coverage); and (ii) complicated non-trivial behavior exhibited during outburst such as double (or multiple) peak features (e.g., XTE J1550−564; Kubota & Makishima 2004), extended flare-like activity (e.g., 4U 1630−472; Tomsick et al. 2005), and prolonged outburst periods (e.g., GX 339−4; Zdziarski et al. 2004).

To deal with the complicated non-trivial behavior the algorithm repeats the data cleaning (Section 3.7.1) and detection (this section) stages on both weekly (8-day averaged) and monthly (24-day averaged) data with the minimum criteria for an outburst being at least 2 detections occurring every 24 days and at least 2 detections occurring every 72 days, respectively. This is followed by combining the produced results (1-day average, 8-day average, and 24-day average) into a final list of outbursts detected in an individual source.

Given the variable data quality at times, to ensure that the outburst detector (Section 3.7.2) is catching "real" outbursts rather than artificial flare-like/dip-like profiles, which can be the product of unexpected background increases/decreases (i.e., Sun glints) or instrumental errors, the algorithm implements a sensitivity limit.

During the sensitivity selection stage, a weighted mean method (see Bevington & Robinson 2003) is used to calculate the mean count rate (${\mu }_{{\rm{rate}}}$) and error on the mean count rate (${\sigma }_{{\rm{rate}}}$) per outburst. This calculation is followed by the application of a σ-clip, whereby only outbursts with ${\mu }_{{\rm{rate}}}\gt 10{\sigma }_{{\rm{rate}}}$ are considered "real" outbursts. The $10{\sigma }_{{\rm{rate}}}$ limit has been determined to be the optimal value through numerous trials comparing the results from the outburst detector to literature detections of the well studied transient sources 4U 1630−472, GX 339−4, and H 1743−322. In the hardness computation stage the algorithm computes hardness ratios using nine different combinations of energy bands, each of which are listed in Table 2.

Table 2.  Empirical Outburst Classification Criteria

Telescopea	Hard Band	Soft Band	${C}_{{\rm{soft}}}$ b	${C}_{{\rm{hard}}}$ b
ID	(keV)	(keV)
SM	15–50	4–10	0.2846	0.3204
SR	15–50	3–12	0.1646	0.2675
SRp	15–50	2.5–10	0.5597	0.8601
SI	15–50	3–10	0.3884	0.5751
RR	5–12	3–5	0.3843	0.4220
II	18–40	3–10	0.3579	0.5449
HR	15–30	3–12	0.3717	0.6890
HRpp	15–30	4–9	0.2246	0.4938
RRpp	9–20	4–9	0.4620	0.6433

Notes.

^aSM: Swift  and  MAXI, SR: Swift  and  RXTE/ASM, SRp: Swift  and  RXTE/PCA, SI: Swift  and  INTEGRAL/JEM-X, II: INTEGRAL/ISGRI  and  INTEGRAL/JEM-X, HR: RXTE/HEXTE  and  RXTE/ASM, HRpp: RXTE/HEXTE  and  RXTE/PCA, RRpp: RXTE/PCA  and  RXTE/PCA. ^b ${H}_{X}$ boundaries defining the HCS and SDS (See Sections 3.7.4 and 3.7.7).

Download table as:  ASCIITypeset image

The algorithm assumes the form of the hardness ratio H_X to be the hard band flux density (in crab units) divided by the soft band flux density (in crab units) and computes the quantity using a Markov Chain Monte-Carlo (MCMC) method via the emcee python module (Foreman-Mackey et al. 2012). As this problem involves two-dimensional uncertainties, the standard linear formulation (i.e., $y={mx}+b$) is ill-suited. As such, the algorithm makes use of an alternative model, originally proposed by Hog et al. (2010), which involves parameterizing the slope m in terms of the angle θ that the generative locus (line) makes with the x-axis.

Adapting this technique, it parametrizes H_X (slope m) in terms of hard band flux density (y-variable) and soft band flux density (x-variable) such that,

$$\begin{eqnarray}&&{H}_{X}=\mathrm{tan}\theta =\displaystyle \frac{{f}_{{\rm{hard}}}}{{f}_{{\rm{soft}}}},\end{eqnarray} \tag{ 3 }$$

yielding a log likelihood of the form (Hog et al. 2010),

$$\begin{eqnarray}&&\mathrm{ln}{ \mathcal L }=K-\displaystyle \sum _{i=1}^{N}\ \displaystyle \frac{{({f}_{{\rm{hard}}}\mathrm{cos}\theta -{f}_{{\rm{soft}}}\mathrm{sin}\theta )}^{2}}{{\sigma }_{{\rm{soft}}}^{2}{\mathrm{sin}}^{2}\theta +{\sigma }_{{\rm{hard}}}^{2}{\mathrm{cos}}^{2}\theta }.\end{eqnarray} \tag{ 4 }$$

Initialization is performed using the following prescription,

$$\begin{eqnarray}&&{x}_{0}=(\pi /4)+(0.1)r,\end{eqnarray} \tag{ 5 }$$

where r is a random number in the range $(0,1]$, (E. Rosolowsky 2015, private communication) in combination with a 100 step "burn in" phase.

After likelihood maximization is performed on θ (via Equation (4)), a probability distribution function (PDF) of H_X is obtained by taking the tangent of the resulting PDF for θ found via the MCMC algorithm. The algorithm takes the median of the resulting distribution to be the accepted value of H_X at the particular time t, and defines the $1\sigma$ confidence interval as the upper and lower limits on H_X.

Within our data sets there exist three separate situations. On any given day t there may be a (i) detection in both hard and soft bands; (ii) detection in the hard band and non-detection in the soft band, or; (iii) detection in the soft band and non-detection in the hard band.

In case (i), both upper and lower limits on H_X are tabulated. However, in cases (ii) and (iii) the situation is more complex. Here we must be particularly cautious as there is a possibility for large errors on the non-detection data. As such, the algorithm disregards data with large errors by applying a σ-cut. Here only non-detection data points in which the errors are within $2\sigma$ of the mean error value for the given band are considered valid.²⁵ Note that this σ-cut is not applied to detection data (i.e., times in which we have detections in both the hard and soft bands). In a case (ii) situation, where there is only a hard detection, only lower limits on H_X are tabulated. This indicates the source is most likely in the HCS, assuming similar sensitivity in hard and soft bands. Similarly, in a case (iii) situation, where there is only a soft detection, only upper limits on H_X are tabulated. This indicates the source is most likely in the SDS.

The last part of the hardness computation stage involves a calculation of a total hardness range for each outburst, followed by the placement of a "classification flag" on each outburst indicating whether or not it meets the "minimum data requirement." This requirement indicates whether or not the algorithm has enough data on the outburst to later confidently classify it (via procedure outlined in Section 3.7.7). To receive a "classification flag" the outburst must consist of at least 5 data points in which H_X has been computed. In addition to cases where there are not enough H_X data points, an outburst will also fail to receive a "classification flag" when data are only available in one energy band (making the calculation of H_X impossible).

By modeling each day's flux of a BHXB as a combination of a soft disk blackbody spectral component and hard Comptonized spectral component, assuming a Crab-like spectrum in each given energy band and a known distance (from the literature), it is possible to obtain X-ray luminosity ${L}_{{\rm{X}}}$ (in any energy band) for a source on a given day t. To accomplish this task we use an algorithm that relies on X-ray hardness to determine the relative dominance of the disk and Comptonized spectral components in the spectrum. Here we have assumed a standard two component spectral model in Xspec, diskbb+comptt, representing (i) the soft disk component of the spectrum with a ${T}_{\mathrm{in}}=1\;{\rm{keV}}$ multi-color disk blackbody, typical of BHXBs in the SDS (McClintock & Remillard 2006), and (ii) the hard Componized component with the analytical comptt model corresponding to a plasma temperature of ${T}_{e}=50\;{\rm{keV}}$, a soft disk photon temperature of ${T}_{\mathrm{in}}=1\;{\rm{keV}}$, and an optical depth $\tau =1.26$, which has been calculated to roughly match a typical hard state photon index of 1.7 for the 3–20 keV range, as is often found in BHXBs (McClintock & Remillard 2006; Done 2010).

Several sources have been shown to soften as they fade into quiescence (e.g., see Plotkin et al. 2013; Wijnands et al. 2015), but we are not capable of following a source into quiescence with the instruments that we use (i.e., our sensitivity limit is a few times 10³⁵ $\mathrm{erg}\;{{\rm{s}}}^{-1};$ see Figure 17). Thus, this trend is not relevant to our analysis.

Each day's flux for a chosen source and energy band can then be modeled as

$$\begin{eqnarray}&&{f}_{X}=a\;{m}_{1}+b\;{m}_{2},\end{eqnarray} \tag{ 6 }$$

where m₁ and m₂ are the flux of the diskbb and comptt models for the energy band in question, respectively. Provided that there is one energy band available that can act as a hard band, one available energy band that can act as a soft band, and that at least one of these bands exhibits a detection of the source on the day in question, the algorithm estimates the flux of a source on this day in the chosen energy band using an MCMC method, via the emcee python module (Foreman-Mackey et al. 2012), to fit for the normalization parameters a and b. For a list of the m₁ and m₂ constants corresponding to each energy band used in this study see Table 3. In this case, the log likelihood takes the form

$$\begin{eqnarray}&&\mathrm{ln}{ \mathcal L }=K-\displaystyle \sum _{i=1}^{N}\displaystyle \frac{{({y}_{i}-a{m}_{1}-b{m}_{2})}^{2}}{{\sigma }_{{y}_{i}}^{2}}.\end{eqnarray} \tag{ 7 }$$

Initialization is accomplished by starting the MCMC "walkers" in a very compact grid in parameter space around the point that is expected to be close to the maximum probability point, in combination with the implementation of a 500 step "burn in" phase.

Table 3.  Spectral Fitting Constants

Instrument	Energy Band	diskbb Flux ${m}_{1}$ a	comptt Flux ${m}_{2}$ b
	(keV)	(${\rm{erg}}\;{{\rm{cm}}}^{-2}\;{{\rm{s}}}^{-1}$)	(${\rm{erg}}\;{{\rm{cm}}}^{-2}\;{{\rm{s}}}^{-1}$)
ASM	3–5 (B)	$4.08\times {10}^{-12}$	$5.08\times {10}^{-8}$
ASM	5–12 (C)	$1.49\times {10}^{-12}$	$1.43\times {10}^{-7}$
ASM	3–12 (B+C)	$5.57\times {10}^{-12}$	$1.94\times {10}^{-7}$
BAT	15–50	$4.99\times {10}^{-16}$	$3.23\times {10}^{-7}$
GSC	4–10	$2.92\times {10}^{-12}$	$1.36\times {10}^{-7}$
GSC	4–20	$2.95\times {10}^{-12}$	$2.80\times {10}^{-7}$
HEXTE	15–30	$4.99\times {10}^{-16}$	$1.69\times {10}^{-7}$
ISGRI	18–40	$4.24\times {10}^{-17}$	$2.12\times {10}^{-7}$
ISGRI	40–100	$7.29\times {10}^{-26}$	$7.06\times {10}^{-8}$
JEM-X	3–10	$5.55\times {10}^{-12}$	$1.60\times {10}^{-7}$
JEM-X	10–25	$2.95\times {10}^{-14}$	$2.00\times {10}^{-7}$
PCA	2.5–10	$9.83\times {10}^{-12}$	$1.80\times {10}^{-7}$
PCA	2–4	$6.91\times {10}^{-12}$	$4.42\times {10}^{-8}$
PCA	4–9	$2.88\times {10}^{-12}$	$1.17\times {10}^{-7}$
PCA	9–20	$7.01\times {10}^{-14}$	$1.63\times {10}^{-7}$

Notes.

^aThe flux of the diskbb model in the energy band specified. ^bThe flux of the comptt model in the energy band specified.

Download table as:  ASCIITypeset image

The algorithm computes where this compact grid is located in any particular case by fitting for the a and b parameters from Equation (6) using a bounded least-squares fitting algorithm (leastsqbound²⁶ ), as bounds are necessary (both a and b must be greater than zero) to obtain a "physical" flux (i.e., non-negative).

Once a and b are obtained from leastsqbound, a suitable range for each parameter is found by performing a simple grid search, splitting the parameter space from $0\to a$ and $0\to b$ into four equal fractional sections (i.e., 1/8, 3/8, 5/8, and 7/8 multiples of the parameter). The grid point with the minimum ${\rm{\Delta }}{\chi }^{2}$ is applied as a symmetric error for each parameter.

After likelihood maximization is performed, the algorithm takes the median and $1\sigma$ confidence intervals of the resulting PDFs created via the MCMC algorithm as the accepted value, and upper and lower limits on each of the normalization parameters a and b, respectively.

Once a and b are found for a given day, the algorithm can then compute the flux in the 2–50 keV band on that day. This band limited flux is then converted to bolometric flux (0.001–1000 keV) by multiplying each component (disk and comptonized) by a derived bolometric correction from the Xspec models. Finally, the disk and comptonized flux (and in turn luminosity, ${L}_{{\rm{X}}}$) components are obtained via

$$\begin{eqnarray}&&{f}_{{\rm{bol}}}=a\;{m}_{1,\mathrm{bol}}+b\;{m}_{2,\mathrm{bol}}={f}_{{\rm{disk}}}+{f}_{{\rm{comp}}},\end{eqnarray} \tag{ 8 }$$

and a disk fraction, ${d}_{f}={f}_{{\rm{disk}}}/{f}_{{\rm{bol}}}$, is computed.

We stress that this algorithm assumes one spectral model is valid for all systems. For example, if the disk was in fact hotter then expected (i.e., when a source is in the SDS), then the algorithm would overestimate the Comptonized component and underestimate the thermal disk component in the spectrum. Similarly, if the disk was cooler then expected (i.e., when a source is in the HCS), then the algorithm would underestimate the Comptonized component and overestimate the thermal disk component in the spectrum. This assumption also breaks down when a source has a power-law index Γ that is significantly steeper then the chosen ${\rm{\Gamma }}=1.7$ value. This situation could occur if a source enters a SPL state or other anomalous high luminosity variability states (e.g., GRS 1915+105 and IGR J17091−3624; see Sections 2.2.70 and 2.2.26 for references).

To determine which outbursts are likely affected by these problems we (i) observe the evolution of the disk fraction over each algorithm-classified outburst, looking for deviations from the expected trend (i.e., lower disk fraction in the HCS and higher disk fraction in the SDS), and (ii) check for appearances (via spectral/timing analysis in the literature) of SPL state behavior (which could be a possible explanation for uncharacteristic behavior in the disk fraction) in outbursts that deviate from the expected disk fraction trend. Overall, this allows us to quantify the problem and eliminate affected outbursts from our analysis (see Sections 4.6–4.9 and 5.2 for a detailed discussion on the impact of this problem).

With the computed bolometric X-ray luminosities ${L}_{{\rm{bol}}}$, the algorithm next builds an X-ray Luminosity Function (XLF), obtains a time-averaged bolometric luminosity (over the last 19 years), and derives a mass-transfer history for a source.

For transient systems, the algorithm assumes ${L}_{{\rm{bol}}}=0$ during all quiescent (non-detection) periods. It then uses the days in which data was available (and thus those days an estimate of ${L}_{{\rm{bol}}}$ exists via Section 3.7.5) during outburst to interpolate a ${L}_{{\rm{bol}}}$ for the missing days in between. Using the ${L}_{{\rm{bol}}}$ estimates for every day a source was in outburst, the algorithm creates a complete transient XLF. To accomplish this task, the algorithm checks if an ${L}_{{\rm{bol}}}$ exists for every day during an outburst. If no estimate exists, the algorithm takes the nearest days bracketing the missing day(s) on either side (call them t₁ and t₂), that have estimates and performs a linear interpolation using an MCMC method via the emcee python module (Foreman-Mackey et al. 2012).

While the log likelihood may take the standard linear form in the linear interpolation MCMC method, the situation is more complex due to the fact that the flux values have asymmetric errors. We thus add a conditional statement to the likelihood function. If the fit (the flux calculated from fitted parameters m and b) obtains a flux value greater than that of the input data point (i.e., the fit is above the data), the upper error bar is used in $\mathrm{ln}{ \mathcal L }$. In contrast, if the fit obtains a flux value less than that of the input data point (i.e., the fit is below the data), the lower error bar is used in $\mathrm{ln}{ \mathcal L }$ (E. Rosolowsky 2015, private communication). This asymmetric error situation complicates the MCMC initialization procedure. To set the initialization of the "walkers" in the linear interpolation MCMC method the algorithm again makes use of the idea of starting the "walkers" in a very compact grid in parameter space around the point that is expected to be close to the maximum probability point in combination with the implementation of a 500 step "burn in" phase. However, to compute the compact grid, the procedure presented in Section 3.7.5, which makes use of the bounded least squares algorithm, must be modified. Performing a Monte-Carlo sampling for each data point (flux f_X) in the range bracketed by its asymmetric error bars, allows the algorithm to compute the 1σ confidence intervals of the resulting distributions and in turn a suitable range for each parameter to create the compact grid.

After likelihood maximization, the algorithm takes the median of the PDF distributions created via the MCMC algorithm as the accepted values of m and b and the $1\sigma$ confidence intervals as the upper and lower limits on each parameter. Knowing the values of m and b across the time interval ${t}_{1}\to {t}_{2}$, it can then interpolate ${f}_{{\rm{bol}}}$ (and in turn calculate ${L}_{{\rm{bol}}}$) for the days missing estimates in this time interval and as such calculate a time-averaged bolometric luminosity over the last 19 years via

$$\begin{eqnarray}&&{L}_{{\rm{bol,avg}}}=\displaystyle \frac{{\displaystyle \sum }_{i=1}^{N}\;{L}_{{\rm{bol}},i}}{{t}_{{\rm{tot}}}}.\end{eqnarray} \tag{ 9 }$$

The situation is more involved for the persistent systems as, unlike transient systems, many of these sources never fully return to quiescence. Instead persistent sources occasionally transition to a state, often occurring in between long bright outburst periods, in which a low-flux level is maintained. To deal with this situation the algorithm first estimates a detection limit, ${L}_{{\rm{det}}}$, for each particular source, by fitting the non-detection data (using the procedure outlines in Section 3.7.5) to find the lowest non-detection luminosity upper limit. Whenever ${L}_{{\rm{bol}}}\gt {L}_{{\rm{det}}}$, then ${L}_{{\rm{bol}}}$ is assumed to be the true luminosity on that day.

In contrast, when ${L}_{{\rm{bol}}}\lt {L}_{{\rm{det}}}$ the algorithm is equipped to deal with three separate cases: when a source, on a particular day, in a given energy band, is detected (i) above where the source is normally detected during that time period (e.g., when the Sun is near the Galactic Center or when a nearby bright source goes into outburst), (ii) below where the source is normally detected during that time period (e.g., a clear decrease in luminosity), or (iii) where the source is normally detected during that time period within error (e.g., no discernible increase/decrease in luminosity).

To quantify the limits on ${L}_{{\rm{bol}}}$ when a source is below ${L}_{{\rm{det}}}$, we begin by estimating a local sensitivity limit ${f}_{{\rm{sl}}}$ in each available energy band for the day t in question. This is done by finding the weighted mean (and corresponding error) of the local data points corresponding to the closest two days bracketing t on either side, in each available instrument/energy band (in crab units). From here we then compare the flux (in crabs) of the source on the day in question, ${f}_{{\rm{t}}}$, to this sensitivity limit in that band. If at least one instrument/energy band has a lower limit of ${f}_{{\rm{t}}}$ greater then the upper limit of ${f}_{{\rm{sl}}}$ (i.e., case i), then we use the linear interpolation method, described above for transients, to estimate ${L}_{{\rm{bol}}}$ of the source on this day. If none of the data satisfies the above condition, then the algorithm checks if any of the available data show ${f}_{{\rm{t}}}$ consistent (within error) of ${f}_{{\rm{sl}}}$ (i.e., case iii). If at least one instrument/energy band satisfies this condition then we assume ${f}_{{\rm{t}}}={f}_{{\rm{sl}}}$ on that day for each available energy band, and then use these new estimates for flux to fit for ${L}_{{\rm{bol}}}$ using the spectral fitting algorithm discussed in Section 3.7.5. Finally, if all available instrument/energy bands have an upper limit of ${f}_{{\rm{t}}}$ less then the lower limit of ${f}_{{\rm{sl}}}$ (i.e., case ii), then we assume the flux on that day is a uniform distribution between zero and the lower limit of ${f}_{{\rm{sl}}}$ in each available energy band, and then fit for ${L}_{{\rm{bol}}}$ using the spectral fitting algorithm discussed in Section 3.7.5. We stress that, if all requirements for spectral fitting are not met on any particular day (e.g., we do not have at least one energy band which can act as a soft band and one energy band which can act as a hard band, see Section 3.7.5 for detail) or no data exists on this day, then the algorithm will default to using the linear interpolation method, described above for transients, to estimate ${L}_{{\rm{bol}}}$ of the source using the closest available luminosity estimates bracketing the day in question.

Because we are dealing with non-detection data, there is a possibility of large errors in flux. To deal with this possibility we use the same method used in our X-ray hardness computation algorithm, whereby we test whether a band limited flux has an error greater then $2\sigma$ of the mean error value for the given band (see Section 3.7.4). If this condition is satisfied in a given band, we ignore the data and instead interpolate the flux on this day using the algorithm described above for the transient sources and the local flux data points corresponding to the closest two days bracketing the day in question on either side. Using this interpolated flux we then follow the the same three step procedure described previously to estimate ${L}_{{\rm{bol}}}$.

Following the estimation of ${L}_{{\rm{bol}}}$ on every day over the past 19 year period, a time-averaged bolometric luminosity is calculated for each persistent source using Equation (9). Finally, using this calculated ${L}_{{\rm{bol,avg}}}$ for all transient and persistent sources, the algorithm estimates a long-term averaged mass-transfer rate for each individual source via

$$\begin{eqnarray}&&{\dot{M}}_{{\rm{avg}}}=\displaystyle \frac{{L}_{{\rm{bol,avg}}}}{{c}^{2}\eta },\end{eqnarray} \tag{ 10 }$$

where η denotes accretion efficiency.

To provide an accurate estimate of ${\dot{M}}_{{\rm{avg}}}$, the algorithm uses Monte-Carlo simulations to take into account the uncertainties which come into the calculation in the form of errors in distance (d), and spectral modeling (fit parameters a and b). Accretion efficiency (η) is fixed at 0.10, corresponding to radiatively efficient flow through a thin accretion disk (Frank et al. 2002) Once again, the algorithm takes the median of the resulting distribution of ${\dot{M}}_{{\rm{avg}}}$ to be the accepted value and the $1\sigma$ confidence interval as the upper and lower limits on this value.

We note that if the true accretion efficiency differs from the chosen value of 0.1 at any point during outburst (e.g., during the lower-luminosity hard state, which is thought to be dominated by radiatively inefficient flows; Narayan & Yi 1994; Meyer-Hofmeister 2004; Knevitt et al. 2014), then the resulting ${\dot{M}}_{{\rm{avg}}}$ will be affected to a certain degree. See Section 5.1 for a thorough discussion of the factors that have the potential to affect the long term $\dot{M}$ balance of BH systems.

In the final stage, the algorithm makes use of the empirical hardness ratio (H_X) parameter to categorize outburst behavior into one of three classes: "successful," "indeterminate," and "hard-only."

The classification procedure begins by differentiating data for each outburst into hard, soft, and intermediate states based on critical hard ${C}_{{\rm{hard}}}$ and critical soft ${C}_{{\rm{soft}}}$ hardness values. As these critical values will differ depending on the telescopes involved in H_X, the algorithm makes use of 10 calibration sources (found in Table 4) to set these baseline critical values. These calibration sources have been specifically chosen based on the criteria that they have exhibited (proven via spectral and/or timing analysis) either "hard-only" outbursts, or a combination of "successful" and "hard-only" outbursts over the last 19 years. The literature classification is then used to find the baseline critical values for each of the nine H_X combinations. See Table 2 for the critical values corresponding to each H_X combination.

The criteria for a source to be in the soft state requires at least one upper error bar on H_X, ${\sigma }_{{\rm{H,high}}}\lt {C}_{{\rm{soft}}}$. In turn for a source to be in the hard state, all lower error bars on H_X, ${\sigma }_{{\rm{H,low}}}\gt {C}_{{\rm{hard}}}$. If the observation does not fall into either category, then the source is classified to be in an intermediate state.

${C}_{{\rm{hard}}}$ was found by taking the minimum ${\sigma }_{{\rm{H,low}}}$ for each "hard-only" calibration outburst, followed by finding the absolute minimum of these values across all calibration sources, yielding the softest a source can be while still remaining in the hard state. ${C}_{{\rm{soft}}}$ is found by taking the minimum ${\sigma }_{{\rm{H,high}}}$ for each "successful" calibration outburst (thus fulfilling the minimum requirement for a source to reach the soft state that at least one ${\sigma }_{{\rm{H,high}}}\lt {C}_{{\rm{soft}}}$), followed by finding the absolute maximum of these values across all calibration sources, yielding the hardest a source can be while still being in the soft state.

There are multiple telescope pairs involved in this process (between 1 and 9 separate pairs). As such, each pair of observations (two bands involved in H_X) will indicate whether the source is currently hard, soft, or intermediate. The algorithm must be able to take into account all possible combinations of hard, soft, and intermediate classifications and logically combine them to classify the outburst as a whole. If a classification flag exists on the outburst the algorithm will perform the following procedure. If all observations during the outburst classify the state of the source as hard, then the outburst is classified as "hard-only." If any observations during the outburst classify the state is soft, then the outburst is classified as "successful." If neither of these conditions are met and/or if no classification flag exists, the outburst is classified "indeterminate," which indicates that the outburst was detected by the algorithm but that there was not enough data available to confidently determine whether or not the source reached the soft state during the outburst.

We recommend caution when interpreting both outburst classification and accretion state when only RXTE/ASM hardness ratios (i.e., 5–12 keV/3–5 keV; "RR") are available. We have found that when using the ratio of 3–5 keV and 5–12 keV RXTE/ASM fluxes, if the disk is hotter than we would typically assume for a source in the soft state, then the hardness ratio will mimic a ratio typical of the HCS, leading to mis-classification of the accretion state of the source on a given day and in-turn the possibility for the outburst as a whole being classified wrong by our algorithm. We have also found that if an unusually power-law dominant SPL state occurs, then the algorithm will incorrectly label this state as the HCS as we do not have the ability to differentiate the SPL state from the HCS state due to the limited available spectral information. If this situation occurs, it is possible for our "successful"/"hard-only" dichotomy to be confused. See Section 4.3 for specific examples of sources that display this behavior and further discussion on the impact these issues have on our "successful" and "hard-only" outburst detection rates.