A gene expression oscillator called the segmentation clock controls the periodic formation of somites in vertebrate embryos. In zebrafish, negative autoregulation of the transcriptional repressor genes her1 and her7 is thought to control the clock’s oscillations. Delays in this negative-feedback loop, including transcriptional delay (the time taken to make each her1 or her7 mRNA) should thus control the clock’s oscillation period. On p. 444, Ertuǧrul Özbudak, Julian Lewis and colleagues report that, unexpectedly, mutants in which only her1 or her7 is functional have almost identical segmentation clock oscillation periods – because the her1 and her7 genes are very different lengths, the researchers had anticipated that the two mutants would have different transcriptional delays and thus different oscillation periods. The researchers resolve this paradox by showing that the RNA polymerase II elongation rate is extremely fast in zebrafish embryos. They suggest, therefore, that the time taken for her1 and her7 transcript elongation is relatively insignificant, and that other sources of delay (e.g. splicing delay) may instead determine the oscillation period.