InChaetoceros peruvianus, the two very long, delicately tapered setae (spine-like processes) of the “lower” valve curve downwards gently until they are often almost parallel, while those emerging from the “upper” valve curve sharply downwards until oriented almost in the same direction as the setae of the lower valve. This curvature creates a deep pit between the bases of the upper valve's setae, where they emerge from the valve. In live cells, extension of setae is rapid and very sensitive to disturbance. After cleavage the new silica deposition vesicle (SDV) appears in the centre of the furrow and expands outwards over it. A distinct microtubule centre (MC) appears directly on top of the SDV. Microtubules (MTs) from the MC ensheath the nucleus, and others fan out over the SDV and plasmalemma. A little later, the MC in the lower daughter cell moves off the SDV, and its MTs now appear to mould the plasmalemma/ SDV into the deep pit between the base of the setae. In the upper daughter cell, the MC remains on the SDV. Initiation of setae is first observed as protuberances of bare cytoplasm growing from the sides of the daughter cells, through gaps in the parental valve. Many MTs initially line the plasmalemma of these protuberances as they grow outwards and the SDV also expands over the new surface. As the setae get longer, a unique complex of three organelles appears. Just behind the naked cytoplasm at the tip of the seta, a thin flat layer of fibrous material lines the plasmalemma. This, the first of the complex, is called the “thin band”. Immediately behind this is the second, a much thicker, denser fibrous band, the “thick band”. At the front edge of the SDV, 5–6 “finger-like outgrowths” of silicified wall grow forwards. These are interconnected by the elements of the thick band which thus apparently dictate the polygonal profile of the seta. These also appear to generate the spinules (tiny spines) that adorn the surface of the seta; the spiral pattern of the spinules indicates that this whole complex might differentiate one after the next, in order. Further back from the tip, evenly spaced transverse ribs are formed. These are connected to the third organelle in the complex, the “striated band”; our interpretation is that the striated band sets up the spacing of the ridges that regularly line the inner surface of the setae. During seta growth, this complex is apparently responsible for controlling the delicate tapering curvature of the very fine silica processes. Since the complex is always seen near the tip of the seta, we conclude that it migrates forwards steadily as the tip grows. While the thin and thick bands could slide continuously over the cell membrane, the striated band must be disassembled and then recycled forward during extension if it is indeed connected to the ridges lining the inside of the setae. We could find no indication that turgor pressure drives extension of the setae, in which event the activity of these organelles is responsible for growth using the justformed silica tube as the base from which extension is generated.