All trunk segments in the pill millipede Glomeris marginata (Myriapoda: Diplopoda) are initially patterned genetically, (as visualized by the embryonic expression pattern of the even-skipped gene) and formed morphologically, (as visualized by 4-6-diamidin-2-phenylindol stained embryos) in a single segmental period. In addition, formation of every nascent trunk segment concerns ventral as well as dorsal segmental units. Only after the formation of the nascent posterior trunk segments, the dorsal segmental units of two adjacent segments fuse to form a single dorsal segmental unit that subsequently covers two ventral leg-bearing segmental units. The formation of a diplosegmental unit, or in short a diplosegment, is thus the result of dorsal fusion of embryonic tissue and not the result of any splitting-process or fusion of dorsal tergites. The new data also argue against heterochrony as a primary causative factor for the formation of the diplosegments during the formation of dorsal versus ventral segmental units. Furthermore, no evidence was found supporting the hypothesis that anterior trunk segments in diplopods represent degenerate diplosegments. Two possible scenarios arise from the ontogenetic data presented here, whether this represents an ancestral feature of the diplopods, or alternatively if they represent an isolated case only found in Glomeris (and close relatives). If the former is the case, my work may provide an impressive example of Haeckel's recapitulation theory.