Strange gap action

Afternoon all,

    Just recently I have noticed my spark gap doing something different that I
have never encountered before. As the coil is running, the spark gap fires
the electrodes as normal and then about every second or two, it fires the
complete length of the electrode assembly ( 2.5" ). Gap type is a single
set ( 2
electrodes tungsten 1/8" dia set approx.  .5" apart, the tungsten is set into
1/4-20 bolts (( used for adjusting gap width)) these bolt/electrode assemblies
are mounted in a 2" pvc pipe that has been heated and "squished" to form an
shape ). this is the reason for the 2.5 " opening. Air is drawn over the gap
assembly via a shop vac. Obviously the system is not suffering from a lack
of air
flow  :) . btw  main system is  4 NST's -at- 15KV -at- .030 A   all parallel wired
outputs (( 15KV -at- .120 A ))  primary is 3/8 Cu tubing -at- 15 degrees 13 turns .25
bewteen turns. Secondary is 8"X30" (30" of windings)22awg on acrylic
tubing. Caps
are 1 homemade P.E. & Al plate for a measure of .0224 mF ( vertical form .060 "
PE plates)
run in series with a power correction cap from the power company. ( it is rated
at 18KV
1mF ) did this to relieve voltage stresses on homemade cap. Toroid consists of
1-at- 4X10
and on top of that one is a 6X16  ( both are aluminum flex duct)  and for
grounding  2 Cu clad rods 8' apart joined by flattened 3/8 soft Cu tubing.

back to the problem ...  why do I get these long arcs ( from furthest points
inside of tube)
when the actual electrode gap is only .5 "  and btw  these long arcs are above
the gap set
by at least .5" also.  ( very loud sharp cracking noise  alot louder than the
normal gap noise)

Is it possibly a least resistant path due to slower airflow? or is there some
other mystical reason??

hope i included enuf info

Scot D