Spermatogenesis and auxospore development were studied in the freshwater centric diatom Hydrosera triquetra. Spermatogenesis was unusual, lacking depauperating cell divisions within the spermatogonangium. Instead, a series of mitoses occurred within an undivided cell to produce a multinucleate plasmodium with peripheral nuclei, which then underwent meiosis. 32 or 64 sperm budded off from the plasmodium leaving a large residual cell containing all the chloroplasts. Similar development apparently occurs in Pleurosira, Aulacodiscus, and Guinardia, these being so distantly related that independent evolution of plasmodial spermatogenesis seems likely. After presumed fertilization, the Hydrosera egg cell expanded distally to form a triangular end part. However, unlike in other triangular diatoms (Lithodesmium, Triceratium), the development of triradiate symmetry was not controlled by the "canonical" method of a perizonium that constrains expansion to small terminal areas of the auxospore wall. Instead, the auxospore wall lacked a perizonium and possessed only scales and a dense mat of thin, apparently entangled strips of imperforate silica. No such structures have been reported from any other centric diatoms, the closest analogs being instead the incunabular strips of some raphid diatoms (Nitzschia and Pinnularia). Whether these silica structures are formed by the normal method (intracellular deposition within a silica deposition vesicle) is unknown. As well as being more rounded than vegetative cells, the initial cell is aberrant in its structure, since it has a less polarized distribution of the "triptych" pores characteristic of the species.