In this thesis the solidification behaviour of different Al-Ni alloys is studied by means of the electromagnetic levitation (EML) technique. Of special interest is an anomaly of the growth behaviour, a decreasing solidification velocity for increasing undercooling. However, according to theoretical considerations the growth velocity should increase with increasing undercooling. In order to study the anomaly, levitation experiments on earth (1g) as well as levitation experiments using the electromagnetic levitation facility on board the International Space Station (ISS) are carried out. The electromagnetic levitator on board the ISS (ISS-EML) provides a unique processing environment in microgravity g where external influences are severely reduced. The new results obtained in g and 1g show no difference in terms of the growth velocity. However, the high-speed video data used to capture the solidification show an unexpected behaviour. The visible front consists of circular features which grow and consecutively form, referred to as scales. The measured front velocity is therefore a superposition of formation of new and growth of the existing scales. Accompanying microstructure analyses of samples processed on earth show that each scale corresponds to a nucleation event. It is concluded that the observed front is not a dendritic growth front, but a nucleation front. This resolves the contradiction between experimental results and theoretical considerations since the theoretical approaches are valid only for dendritic growth fronts. For a better understanding of its behaviour, the nucleation front is studied in greater detail. The number and size of the scales is measured. It is found that for an increasing undercooling fewer, yet larger scales form. The loss of nuclei is not compensated by the larger area, and therefore leads to a decrease of the front velocity. The velocity of the dendrites cannot be measured due to the opacity of the melt.