The Observational Basis for Central Engines in $\gamma$ Ray Bursts

We review the observational differences between gamma-ray bursts occurring on a single shell (such as in the external shock model) and multiple shells (such as in the internal shock model). We do not argue against external shocks or for internal shocks, as such, but rather against a single shell and for a central engine that is small enough to produce the typical GRB time history without violating causality or other kinematic limits. The expected profile and average spectral evolution from a single shell is compared to the average of many bursts and found to be different. The presence of gaps in many gamma-ray bursts is a strong argument against a single shell because an observer should see many causally disconnected regions at any one time.The rapid variability is also difficult to explain from a single shell because of the large number of causally disconnected regions. The pulse width as a function of time in a burst should increase because there is a one-to-one relationship between arrival time and the off-axis angle of emission. The observations show that the pulse width does not increase with time. Finally, in GRB990123 there is evidence for deceleration from the simultaneous optical observations, yet the gamma-ray pulses show no lengthening of their pulse structure. We conclude that gamma-ray bursts are caused by a relatively small central engine