One of the characteristics of this whole debate is that people can't seem to believe a 1-2% heat loss can produce the "FPHE". Failing to accept the idea is an example of forcing experimental results to fit preconceived notions. While we certainly have theory to guide us, we should never assume we have it "all". Error bar magnitude must be determined based on the impact of all components of variation, not just the ones we want to be important.
The facts remain that a small calibration constant variation on data obtained in a highly efficient calorimeter can produce a (non-random in this case) variation many times the baseline fluctuation. That is not speculation, that is published fact. General principles present in defining how that happens (such as needing to use a minimum of 2 zones to describe cell/calorimeter function) can be extrapolated appropriately to other scenarios.
In your first paragraph you appear to be replying to me but you're not, which is fine. In your second paragraph you explain that the general idea of a non-random fluctuation is not speculation, which is also fine, and which is not what I was addressing. I'm saying the part about 2 or more zones still being subject to possible CCS, possibly because of the positioning of feedthroughs, is hand-wavy. Do you disagree? Can you suggest a concrete experiment using just resistance heaters, or possibly forced on-cathode recombination (using bubblers perhaps), that will show or falsify a CSS "signal" in a setup specifically with several thermal barriers?