It seems to me that Horace Heffner is talking about two separate issues here. They may both be happening. Or perhaps only one, or neither. It is better to test them separately.
I do not see how a heat exchanger is needed to test either one. Horace wrote: "The problem is simulating the nature of the manifold, with its very high thermal conductivity and large contact area between hot and cold sides. Taping together two hoses does not do this." To test the air pocket hypothesis, you need to add a body into the pocket which is much hotter or colder than your target. I had a body 34°C colder which I think is sufficient. The temperature gap is not as large as with Rossi's system, but it would have registered something, I believe. You do not need to conduct the heat from the cold hose to the TC via metal. The point is to find out if the cold air trapped in the pocket will do this by convection. In other words, I was not simulating the problem of conduction from one hose to the other, but only the air in the pocket convecting from one body to the other. Granted, a little heat would conduct from one pipe to the other, since they were touching. Can the cold air caught in the pocket coming from the cold pipe a short distance away affect TC performance? Answer: Nope. Not enough to measure. One could almost as easily test the conduction problem without involving air pockets, and without a manifold or what-have-you. Just use a copper pipe, water flowing through it, and a blowtorch. Or maybe just an electric heater or a gas grill. - Jed
