The distribution of time lags to corona inception is examined theoretically and experimentally for the case where a double-exponential positive impulse voltage waveform is applied, at a fixed repetition rate, to a rod/plane gap. The influence of atmospheric humidity, temperature and pressure on the conditions for corona formation are discussed. Experimentally, the spread of the time-lag distribution is studied primarily as a function of humidity. The results are related to the variation, with humidity, of the negative-ion density prior to each impulse. Models for the conditions for detachment of electrons from negative ions, and for corona formation, which are derived from experiments using small air gaps, have been shown to agree with the results obtained in large air gaps.