Proportional: y=mx
Linear: y=mx+b
That's the mathematical definition. Some people may misuse the term or define it poorly in colloquial usage.
The graph is not proportional. The flow at 8 cmH2O is not 2x the flow at 4 cm. The graph is fairly linear, but not quite. I'm tempted to try to fit a curve to the data, but since the airflow has to drop to zero at zero pressure, I think I don't have enough data points at low pressure to mean much.
I would have thought the flow would be proportional to pressure, but apparently not. The flow increases more slowly than proportionally. Apparently air through an orifice is not proportional to pressure. This page is a bit interesting for airflow through an oriface.
http://www.aircompressorworks.com/airfl ... fices.html
I'm guessing it's a question of turbulent airflow being nonlinear. I guess I gots myself edumacated today.
It would be neat if it were possible to have the flow rate be constant with pressure, because that would waste less air and water at higher pressures, but I don't see a reasonable way to do that in a mask without having something like a pressure setting on the mask itself.