A number of obstruents in English are traditionally analysed (ie thought of, transcribed as) fortis. This talk aims at proving that many of these are in fact lenis, so much so that clusters of two (or more) fortis obstruents are ruled out by a phonotactic constraint in English. So in a word like act either the first ([agt]) or the second ([akd]) plosive must be lenis. We will present arguments for this surprising claim and try to explain what makes us analyse all these lenis obstruents as fortis.