This study uses evidence from early English poetry to determine when certain sound changes took place in the transition from Old to Middle English. It builds on the premise that alliteration in early English verse reflects faithfully the identity and similarity of stressed syllable onsets; it is based on the acoustic signal and not on the visual identity of letters. Examination of the behaviour of onset clusters leads to new conclusions regarding the causes for the special treatment of sp-, st-, sk-, and the chronology and motivation of cluster reduction.