A frequency glitch in an accreting pulsar

Frequency glitches have been observed so far only in radio pulsars and anomalous X-ray pulsars. Here we present evidence of a glitch in a neutron star accreting from a Be companion. The transient KS 1947+ 300 reappeared in 2000 October as a moderately strong X-ray source that exhibited 18.7 s pulsations, leading to an identification with the BATSE source GRO J1948+32, last detected in 1994. We have analyzed Rossi X-Ray Timing Explorer (RXTE) observations taken during the 2000 - 2001 outburst, as well as additional observations taken during a smaller outburst in 2002 July. Orbital Doppler shifts are apparent in the temporal variation of the pulse frequency. A joint fit of the RXTE data together with BATSE measurements from an outburst in 1994 yields the orbital period P-orb = 40.415 +/- 0.010 days, the projected orbital radius a(X) sin i = 137 +/- 3 lt-s, and the eccentricity e = 0.033 +/- 0.013. This degree of eccentricity is unexpectedly low for such a wide orbit. Pulse timing results also show that the intrinsic pulse frequency increased from 53.30 to 53.47 mHz at a rate approximately proportional to the X-ray flux. This is about the degree of spin-up expected from the accretion torques that must be present when the X-ray luminosity reaches similar to 10(38) ergs s(-1). On one occasion during the 2000 - 2001 outburst, the pulse frequency increased by similar to 1.8 x 10(-6) Hz in less than or similar to10 hr over and above the mean trend seen around that time, without any indication of a correspondingly large increase in X-ray flux. The fractional change in frequency of 3.7 x 10(-5) during this event is significantly larger than the values observed in the glitches in radio pulsars and anomalous X-ray pulsars. We discuss other similarities and differences between these events.