I was reading a good article on ball lightning today in a recent issue of Science News
(February 9, 2002 - "Anatomy of a Lightning Ball by Peter Weiss, vol. 161 p 87-89).
After years of trying scientists have now been
able to duplicate some of the phenomena in ball lightning in the lab. The newest
theory is that a cluster of nano-sized particles (often from dusty material) is ignited
by an electrical spark such as other lightning or common friction. This burns hot
for a few seconds and then disappears, although pictures of fireballs (the one in the
article lasted 5 minutes in Australia) which last much longer have been taken.
If the clustering of material is correct, it would naturally lead to fireballs being associated
with faulting. The fault is continuously emitting nanosized particles and a slight movement
could create the spark necessary to ignite the ball.