Accurate time measurement is a crucial element of seismic data collection. For data collected before the 1980s and especially before 1960, the technologies involved are no longer familiar to most researchers. I outline how reliable time has been obtained for seismology and describe the histories of master clocks, local clocks, time transfer, time comparison, and uniform motion for visual recording. A compendium of station data for 1921 gives a snapshot of early seismological timekeeping. I present an overview of subsequent developments, with suggestions on how to weight observed times using descriptions of the timing system used.