1 Introduction

The nice agreement between the spectral evolution of afterglows and an adiabatically and relativistically expanding spherical shell that emits through synchrotron (sy) has established the SSM for the description of the late stages of the evolution of the fireball that is believed to give rise to the main GRB event and its afterglow (e.g., [Wijers et al. 1997]; [Galama et al. 1998]). These considerations imply that sy is the dominant cooling mechanism for the electrons of the flow during at least the afterglow phase and therefore the magnetic field B is close to its equipartition value (see also [Wijers & Galama 1997]). On the other hand, spectral fitting of time resolved BATSE spectra of bright GRBs ([Preece et al. 1998]) has yielded a significant number of cases with low energy photon indices $\alpha$ ($n_{\nu} \sim \nu^{-\alpha}$) exceeding -2/3; this is inconsistent with the SSM and $\alpha =-2/3$ has been referred to as its "death line''. One way for the SSM model to overcome this difficulty is if radiation became self absorbed in the BATSE window for a portion of the burst.

Under the reasonable extrapolation that the MeV range spectrum of the GRB proper is also due to the sy component, I calculate observed spectral evolution series for GRB pulses. The relativistic motion of the region may cause the observed spectra to have a different shape from the intrinsic ("co-moving'').

I use values that are appropriate for the description of internal shocks ([Mészáros & Rees 1994]) and where the ${\rm e}^-$ come from the ionization of the baryonic material. Equipartition values of the physical parameters for a flow of total luminosity per unit solid angle L= 10⁵² L₅₂ erg/s sr, expanding at constant Lorenz factor $\Gamma =300$ and with intrinsic variability timescale $t_{\rm var}$ are $B\simeq 2\ 10^3 \sqrt{L_{52}}/t_{\rm var}$ G and $n_{\rm e} \simeq n_{\rm p} \simeq 10^8 L_{52}/t_{\rm var}^2 \,\rm{\rm cm}^{-3}$. I present broad band spectra and discuss the effect that pairs may have on the BATSE component.