Abstract
The streamer and return stroke concepts in positive corona breakdown are reviewed, and examined critically for the case of short (<5 cm) gaps in ambient air. It is shown that return strokes normally should not be observed, but that return streamers may occur in longer or highly overvolted streamer channels. A new physical picture is presented of the secondary streamer, as one or several high field domains formed by Douglas-Hamilton–Mani-type attachment instability of the residual channel created by the passage of the primary streamer. In this way, both the erratic origin and development of the secondary streamer in air are easily explained, as well as the fact that spark breakdown follows if and when the secondary streamer fills the complete residual channel length. Further, it follows that the secondary streamer is not the cause of breakdown, but only a secondary side effect.