ar X iv : a st ro - p h / 96 01 17 7 v 1 3 0 Ja n 19 96 Test for Time Dilation of Intervals Between Pulse Structures in GRBs

If γ-ray bursts are at cosmological distances, then not only their constituent pulses but also the intervals between pulses should be time-dilated. Unlike time-dilation measures of pulse emission, intervals would appear to require negligible correction for redshift of narrower temporal structure from higher energy into the band of observation. However, stretching of pulse intervals is inherently difficult to measure without incurring a timescale-dependent bias since, as time profiles are stretched, more structure can appear near the limit of resolution. This problem is compounded in dimmer bursts because identification of significant structures becomes more problematic. We attempt to minimize brightness bias by equalizing signal-to-noise (s/n) level of all bursts. We analyze wavelet-denoised burst profiles binned to several resolutions, identifying significant fluctuations between pulse structures and interjacent valleys. When bursts are ranked by peak flux, an interval time-dilation signature is evident, but its magnitude and significance are dependent upon temporal resolution and s/n level. It might naively be thought that time dilation of intervals between peaks or pulses in γ-ray bursts (GRB) should be free of the energy-dependent effects that plague measures of time dilation of temporal structures (1,2). However, at least two effects are expected to give rise to systematic biases that make attempts to determine the actual measure of time dilation of pulse intervals difficult: (1) Structure appearing at limit of temporal resolution. For the present purpose , define " interval between pulses " to be the interval between two discernible peaks of emission, desired significance being adjustable. The average width of GRB pulses in long (T 90 > 2 s) bursts is ∼ 100–500 ms, dependent on energy band. However, there is a large dispersion in pulse width (3). Since the timescale for intervals between pulses is also of this order, there is often a high degree of pulse overlap. Consequently, for the 64-ms resolution data which we employ, time dilating a burst profile by a factor S ∼ 2 will result in the ap