The peculiar infrared counterpart of GX 17 + 2

We discuss the nature of the infrared (IR) counterpart of GX 17 + 2, one of the most luminous of the persistently bright X-ray binaries. Chandra HRC-S astrometry is consistent with either NP Ser (the original counterpart of GX 17 + 2 proposed by M. Tarenghi & C. Reina in 1972) or star "A" of E. W. Deutsch et al. as the counterpart of the X-ray source. However, we present Keck K-band observations that reveal a bright counterpart in the radio error circle of Deutsch et al. 0″.9 north of NP Ser itself. Furthermore, the position of this counterpart is consistent with that of star A to within 0″.1, implying an amplitude of variation of ∼25-33 between the Keck observations and the Hubble Space Telescope (HST) measurements of Deutsch et al. Subsequent Keck imaging also reveals star A in an "IR-faint" state (K = 18.3 mag, with a corresponding amplitude of variability of ∼22). In addition, archival Cerro Tololo Inter-American Observatory observations provide evidence for K-band variability, albeit of smaller amplitude. The HST and Keck K-band variations, however, do not appear to be accompanied by any changes in the overall X-ray luminosity of GX 17 + 2 as measured by contemporaneous (but not simultaneous) Rossi X-Ray Timing Explorer All-Sky Monitor observations. We propose instead that the large radio outbursts observed when the source is in the horizontal branch of its "Z" state are likely to give rise to synchrotron flares in the IR. The amplitude of the radio flares is in agreement with this scenario. Such IR variability, unrelated (directly) to X-ray reprocessing and the gross characteristics of the mass accretion rate, may be present in the IR flux of other low-mass X-ray binaries but harder to see owing to the intrinsically brighter IR fluxes of the longer period systems.