Posted for Jim Sower <canarder@starband.net>:

It sounds to me like air introduced into solution is causing problems, and that
maybe it can be prevented from mixing into the water without eliminating air
altogether.  It seems to me that if there is room for a little "stagnant" air
above the highest point in the system, and if there is "stagnant" water below
it, the air won't mix.  I'm assuming that the pressure cap is at the high point
in the system.  That said, if the pressure cap is at a point of high flow and
turbulence, any air under the cap will tend to mix with the coolant.  If, on the
other hand, there is a column of coolant several inches high (like an upside
down "T") with the cap at the top, all the activity will take place below the
air pocket and separated from the air by the coolant in the stem of the "T".  If
that is already the case, and the air in the overflow tube that is drawn back
into the system still mixes with coolant, then this is not a solution.  If OTOH
there is no place for some "still water" beneath the air bubble, perhaps it
would be easier to provide some still water than totally eliminate the bubble.
Just a theory .... Jim S.