I have looked at your data, and man, isn't this just the greatest thing! I thoroughly enjoy this!
I have a science background, a decade ago, but is kinda not up to skills nowadays - however some things do die hard.

My point in short is that Gay-Lussac law of gases, state that P/T = constant if volume is constant and mass is constant. (P=pressure, T= any of them, but I have used T_GlassIn)

When the device was started up the other day for calibration runs, the P/T (as from the data in the cvs) showed a definitively NON constant behaviour. The result is very much 1/x with a bias, as I plotted P/T vs T, P seem not to follow T or the other way around.

So what gives?

I would very much someone to tell me : "You fool! It is xyz. And dont you forget it! Fool ..."
Insert your own very plausible explanation instead of xyz.

The heat transport properties of hydrogen are very non-ideal. It linearly increases with temp to about 200 C and then the rate of increase starts to decline and become a bit erratic until about 500C. I don't know why. When thermal or catalytic conditions start to create atomic hydrogen, then there is a another increase in thermal conductivity. Mixtures of helium are also challenging. A simple model of ideal gas laws quickly losses it's accuracy with very small mass atoms.

jdk

The calibration method shohuld probably be changed. The present method assumes a constant physical, mechanical and chemical environment in the reaction tube. I think the hydrogen can undergo partial dissaciation under some conditions. This would greatly increase the thermal conductivity inside the gas cell and disrupt the previous calibrations. The gas composition is not constant in this model.
We have programmable power supplies. They can be set to deliver power at a desired flow level. We then try to measure all the relavant nodes of sixth order differential heat flow equation relying on only a couple of temperatures under hopefully similar conditions. Changes to the thermal transport mechanisms from the calibration time will lead to errors.
During operation, we could heat with a power level composed of a DC static component and an AC variable componet. For instance, we could set the DC to 47 watts. The AC component could be set at a very low frequency (a very long period) and a small power change of say +/- 1 watt. The total power would thus range from 46 to 48 watts. This changing power level should give rise to a change in the monitored outer temperature. The change in the temperature should go up and down corresponding to the delta power. The coefficient of ( delta T/dlelta watts) will be new coefficient of the themal opeartion of the calorimeter. It is a contiuous calibration.
This was inspired by the calibration in Paul Hunts ' air flow calorimiter device
For a review of modulation calorimetry refer to
144.206.159.178/ft/849/46359/14264965.pdf
An even better scheme is buried in there.

jdk

I think these questions will be addressed soon.

B

With regard to the proposal of Ascoli65 (in the Jan 26 Follow blog) for density compensation of P_xs calculation:

1. The data filter on his first graph chooses only data at points of relative stability. As a consequence transient events aren't included in the P_xs calculation. If there were avalanche reactions of short duration at random intervals, the P_xs data would be "peaky" and some significant activity would be excluded. Can the calculations be done without the filter?

2. H2 follows the ideal gas law in that its thermal conductivity is nearly invariant across the pressures (and thus densities) in the active cells. Perhaps the hypothesis is that higher density and thus higher specific heat results in more net thermal transfer from convective flows in the cell. Can we test this?

The effect of pressure on T_GlassOut was shown and discussed in the 30Nov blog, without finding a good explanation. Pressure is a factor, but I don't think we fully know why yet.

references:
http://tinyurl.com/apjh9yq
http://tinyurl.com/a5swea8

Thanks to the author for writing the post, it was quite necessary for me and liked it. I wrote a note on the oxbridge essays reviews about this. I will be happy if you read it and accept it. Thank you for your concern.