In this thesis two different projects are described dealing with different aspects of light scattering. In the first we are examining the origin of backward scattering as manifest in Rayleigh superradiance. Here we have studied the onset dependence on the sign of the probe detuning. In the second project, we have studied coherent forward scattering in the form of a memory experiment. In such an experiment we convert the input light pulse to an atomic excitation, and at a later time convert back the atomic excitation into the retrieved light pulse.

In the first project, we investigate the source for the detuning sign difference in the onset of Rayleigh superradiance. We find a difference of up to a factor of three between red and blue detuning when using the D1 line in Rubidium 87. We model this by adding a detuning dependent loss term to a rate equation description of the superradiance. With a microscopic description of the loss term due to light assisted collisions followed by radiation trapping, we find a reasonable quantiative agreement between model and experiment.

In the second project we have realized off resonance Raman memory in an ultracold thermal sample in a magnetic trap, with total efficiency of 15%. In addition we have imaged the retrieved signal using a detection system that can distinguish between 30 independent modes, using balanced homodyne imaging.

The goal with the memory experiment, as presented in this thesis, is a first step towards multimode memory utilizing the high optical depth of the ultra-cold sample. Here we find that due to magnetic dephasing of the ground levels the coherence time is 7ms, and that as we increase the optical depth or the drive light power we get a reduction of the total efficiency contrary to our expectations of saturating the total efficiency.