Erratum: Discovery of Microsecond Time Lags in Kilohertz QPOs

In the Letter ``Discovery of Microsecond Time Lags in Kilohertz QPOs'' by B. A. Vaughan, M. van der Klis, M. Méndez, J. van Paradijs, R. A. D. Wijnands, W. H. G. Lewin, F. K. Lamb, D. Psaltis, E. Kuulkers, and T. Oosterbroek (ApJ, 483, L115 [1997]), the reported time lags, while correct in magnitude, have a sign that is incorrect. We reanalyzed the data and checked the sign of our results using the hard lags in Cyg X-1 and GX 339-4 and the soft lags in the accreting millisecond pulsar SAX J1808.4-3658, as well as by using test signals. The true time delays for 4U 1608-52 are in the sense of the soft photons lagging the hard ones. Any lags in the reported energy and frequency ranges for 4U 0614+091 are between -80 and +15 μs and for 4U 1636-53 are between -50 and +25 μs (95% confidence), respectively, where a positive sign indicates a hard lag. This strengthens the conclusion of the Letter that in simple scattering models any time lags due to inverse Compton scattering are small and imply very small (<~1-10 km) scattering geometries. The time-lag data provide no independent evidence for inverse Compton scattering affecting the X-rays in these sources; another mechanism, perhaps related to the generation of the QPOs, must be operating to produce the soft lags in 4U 1608-52.