Soft 404
I always wondered why ice cubes in water don't grow bigger. Maybe its because thermal conductivity of water is higher than ice.
Cool! Excuse the pun. Even in the flask, that lasted longer than I thought it would.
.
.
If a pool noodle is what I think it is (it isn't a common expression in the UK!), foam-based floating support, then I'd think it would make an excellent insulator with all the air trapped in it. That was a good experiment and I'm surprised to see that ambient warms to such a degree - 1.34W seems like a lot. I know of some transistors that wouldn't hold up as well as that ice.
.
It's easy to forget that the air pockets make for a large surface area. Your graph, for the 4 degree point looks fairly linear - was it like that all the way? If not your thermal resistance is just a snapshot in time isn't it? There is another option: most domestic fridges are set at around 4 degrees (perhaps 5) but it may not be clear or exact (mine has a circular knob that can move between "cool" and "eco" but without degree C markings.) However, it should be constant as long as you don't keep opening the door. If you can measure from within there - e.g. stuff your DMM, thermistor and flask in the fridge then ambient would be controlled - you could also use a second thermistor to measure how constant that fridge ambient was (how much hysteresis does it exhibit?): you could verify that thermal resistance!
If nothing else, these are 'outside of the box' experiments with thermistors and good fun.
.
LM399
Continuing on the tangent: I have a Keithley DMM6500 with an LT SL40057 reference. Rumours say it's an LM399 subset for (selected by) Keithley.
image: blurry photo from my device