Understanding the molecular mechanism controlling the nanofabrication of the diatom shell and explaining the generation of the stunning diversity of patterns is a major goal in the field. Based on the amphiphilic properties of long-chain polyamines, a phase separation model has been proposed that is able to explain the nanopatterning of biosilica (Sumper, 2002). Repeated phase separation events during wall biogenesis are assumed to produce self-similar silica patterns in smaller and smaller scales. Based on this single assumption, it is possible to predict the apparently complex patterns found in the valves of the diatom genus Coscinodiscus. High resolution SEM analysis of valves in statu nascendi from different Coscinodiscus species supports the conclusions derived from the model.