3 Some consequences

$\bullet$ If the high energy spectrum is due to quasi-thermal Comptonization, it will be sensitive to the amount of seed photons, which is bound to increase for successive shells. This will in turn increase the cooling, lower the particle temperature and thus produce a softer spectrum. This can qualitatively explain the hard-to-soft behavior of GRB emission and the (weak) correlation duration vs. hardness.

$\bullet$ If the intrinsic effective temperature is of the order of 50 keV, the observed Comptonized spectrum extends to $\sim 10 \Theta_{-1}\Gamma_2/(1+z)$ MeV at the beginning of the burst. One would also expect that during the initial phases a Wien peak at the electron temperature would be formed for (plausibly powerful) bursts with $\tau_{\rm T} \geq$ 3-5.

$\bullet$ If the progenitors of GRBs are hypernovae, the density in the vicinity of the central source is dominated by the pre-hypernova wind. This can lead to optical depths around unity at $R\sim 10^{12}-10^{13}$, where shell-shell collisions are assumed to take place. There is then the possibility that the GRB events are due to shocks with this material. Therefore: i) in the case of shocks between shells and the pre-hypernova wind the large densities involved suggest that inverse Compton emission is favored with respect to the synchrotron process; ii) if the (still unshocked) interstellar material has total $\tau_{\rm T} \geq$ 1, photons will be down-scattered, introducing a break in the emergent spectrum at the observed energy $511/[\tau_{\rm T}^2\,(1+z)]$ keV. Furthermore, the interstellar matter will act as a "mirror'', sending back the scattered photons, thus increasing the amount of Compton cooling.