Antonio, Terry, and all,
It has been amply demonstrated that leaders and streamers are
initiated, and propagate, at lower E-fields within a diverging
positive E-field than within a diverging negative E-field. And, the
underlying propagation mechanisms are significantly different for
positive leaders versus negative leaders. Under sufficient fields, a
positive leader will grow more or less continuously, while a
negative leader proceeds in discrete jumps, forming bright,
isolated, leader-like "space stems" ahead of the negative leader.
Each space stem has streamers coming off BOTH ends - positive
streamers towards the advancing negative leader and negative
streamers on the other end. (BTW, this is also observed during
stepped leader growth during negative lightning propagation). And,
some of Peter's streak photos actually appear to show space stems. For example:
http://tesladownunder.com/HVRotMirrorPolarityNegLeader.jpg
Both Peter and Terry are presently using "rod-rod" electrode
configurations which complicate the shape of the initial E-field in
the gap. It would be interesting to see if they see different
results when using a rod-plane or a rod-wire configuration instead -
theory predicts that this should cause significantly more streamer
and leader branching in the gap. Of course, once leaders and
streamers start growing within the gap, previously injected space
charge and topload voltage reversals dramatically complicate things. =<:^O
At higher average E-fields and longer sparks, the propagation
"advantage" of a positive source tends to disappear, and negative
leaders appear to grow (macroscopically, at least) at fields similar
to positive leaders. And, branching occurs with either polarity as
the discharge propagates into weaker field areas. Similar branching
behavior is readily observed for both positive and negative lightning.
Branching of streamers and, (for long gaps) leaders, appears to be
inherent to dielectric breakdown of gaseous, liquid, and even solid
dielectrics (electrical treeing and Lichtenberg figures). Branching
angles fall between 30 - 40 degrees (i.e., 15 - 20 degrees on either
side of the previously unbranched leader direction). And there are
lots of interesting theories (and various simulations) that hint as
to why this is the case.
Bert
--