Disclaimer: This was a very rough test and there are many thing to improve 
Both the chip and a thermal probe ( IDM99IV RS PRO) were stuck to the side of a plastic 3-4L container with a drop of hot glue. It was filled with hot tap water. While it cooled I took measurements aprox. every 2 minutes.
It gets messy at 39 C, since that’s pretty close to ambient I had to start adding ice cubes. I completely ignore the importance of stirring until 34 C when I finally come to my senses. That could partially explain the chip’s behavior on that range but still… suspicious?
The not-so-slight constant offset could be attributed to the chip having more surface area in contact with the plastic container wall.
Anyway, that was 1 hour and 30 minutes of my life, not counting the 3 previous attempts that failed in multiple ways.
I will be doing it again someday with a smaller container to make things faster. A stir bar and more regular measurements.
I wonder how I can avoid sticking the chip to a wall. A waterproof version of this maybe?
Edit: My impressions?
Excluding the chaos under 40C, I’m kinda neutral about it. I remember @JennyMcLane , I think, posted something similar about the xBT and it didn’t look great. I can’t find it 
but based on that I expected worse.
But I also hoped for better. I don’t know how much my crappy test conditions influenced the measurements but I don’t see the point of giving us two decimal points when you have almost a degree of variation compared to a presumably accurate probe. I don’t see either what kind of meaningful statistics we could do with it other than here’s when you had fever and here’s when you were in hypothermia…
But I’ll wait until I’ve done a test I can be proud of before taking any conclusions
Edit: The full test is on tape if someone wants it. I won’t be posting it online.
Edit again: Notice how consistently it flattens out at exactly the human body temperature? Sure I forgot to stir but really? Thats the suspicious part to me… Unless 37C was the ambient temperature at that moment, that could also be a thing.