Previous
abstract 
Next
abstract
\def\Pd{{\dot P_{\rm orb}}}
The rate of change of the orbital period ($\Pd$) is arguably the most important diagnostic of the physical processes driving accretion in LMXBs. $\Pd$ is measured by fitting a non-linear ephemerides to the O-C residual of a fiducial phase in the orbital light curve. Small values of $\Pd$ can be measured inly by accumulating timings over decades. Only a few non-pulsing X-ray binaries have measured $\Pd$, including EXO0748-676, 4U1820-30, 4U1822-37, and Cyg X-3. The O-C method assumes that the uncertainties in the residuals are uncorrelated measurement errors. Variations in the O-C residuals which mimic orbital period evolution can be caused by intrinsic or correlated variability in the fiducial timing mark. We have recently developed statistical tests which can distinguish between these two cases (Hertz, Wood, \& Cominsky 1995, ApJ, 1 January issue, in press).
We will present the statistical tests which we have developed, show the significance of the evidence against orbital period evolution in EXO0748-676, 4U1820-30, and 4U1822-37, and comment on improved methods for measuring changes in the orbital period for both X-ray binaries and cataclysmic variables. Orbital timing of EXO0748-676, 4U1820-30, and 4U1822-371 are all consistent with no orbital period evolution. In EXO0748-676 the observed variability is probably due to variations in the atmosphere of the secondary star, which serves as the occulting edge in eclipse timings. In 4U1820-30 and 4U1822-37, the observed variability probably arises from changes in the shape of the light curve. Only Cyg X-3 is inconsistent with no orbital period evolution; however the reported non-constant rate of change in the orbital period ($\ddot P_{\rm orb}$) is not required.
