I wish people here would pay attention to details I previously posted?
- The thermistors used were not sealed for contaminated environments, and that type of thermistor is known to be sensitive to reducing atmospheres.
- The thermistors were sealed inside the balls and never post-calibrated
- H vs D would have differential diffusion through the permeable encapsulation
- The high temp conditioning period looks likely to cause gas absorption by the thermistors, and is noted as being necessary to cause the temperature difference.
- Therefore we'd expect a long-term apparent temperature difference due to calibration change
The paper posted here had details enough (including the thermistor type) to see this as a possible mechanism, no details of any check that would rule it out.
For me, this was an experiment that bugged me till I worked out this mechanism