We have investigated the growth, the pulse energy statistics, the spatial modes, and the spectra of the first and second Stokes light generated in transient stimulated Raman self-generator experiments in hydrogen and deuterium. In particular, we have observed two distinct regions of second Stokes growth. The first region is characterized by a low signal level due to 4-wave mixing. The second region appears to be one of stimulated growth seeded by spontaneous scattering at the second Stokes wavelength, an assumption supported by the rapid growth of the second Stokes signal and the observation of pulse energy statistics indicative of growth from noise. We have measured the second Stokes threshold as a function of gas pressure. At high pressures the threshold is consistent with measured first Stokes thresholds, indicating that the second Stokes is not being seeded by the second Stokes light generated from 4-wave mixing and is due entirely to stimulated scattering from quantum noise. At lower pressures, however, the second Stokes threshold is lowered, indicating some seeding of the stimulated process due to 4-wave mixing. This change in the effectiveness of seeding may be caused by changes in the coherence length of the non-phase-matched 4-wave mixing signal as the pressure is changed. Measurements of the first Stokes spectrum indicates that there is substantial spectral broadening at pump energies corresponding to the second Stokes threshold. This spectral broadening is larger for higher pressures. We postulate that the spectral broadening and shift are due to the AC Stark effect and that this may also reduce effective seeding by 4-wave mixing due to spectral mismatches.