JM,

> Except that in this case he is totally wrong. :-)

Well, at least that will probably guarantee his return to explain himself
;-)

> It is perfectly acceptable to have a model running on 21/79 for
> on/offgassing and using 50 to measure O2 toxicity/exposure at the same
> time. I'm ready to take bets that this is exactly what the computer is
> doing. Don't take 50/79 litteraly.

After a few failed attempts at cerebral activity, I arrived at that
conclusion myself. I follow the simple logic of that, and I am pretty sure
Bob arrived at the same conclusion (but almost certainly faster than  I
did).

But I am not sure Bob is wrong. What I hear him say is that, while
superficially it makes sense to be as conservative as possible on both
counts, this dicks around with the underlying algorithm in a way which
renders it unstable and perhaps unpredictable thereafter. The best analogy I
can think of is how you CAN dive successively deeper, and the computer will
"let" you do that, but the algorithm is f...ed up.

I'm not afraid to admit that I am getting into - for me - uncomfortably deep
water, so I'll sit back and watch what promises to be an interesting
thread - in more ways than one ;-)

Bjorn