[Arcturius]

The text also states that they use the same testbed for rad testing. Also recall that they are using only one liter of water. If you look at the graphs, they level off pretty rapidly, given the small amount of coolant.

After inspecting more graphs, their testing time varies--they seem to stop testing whenever they feel the temperature has leveled off for about 10m. This _seems_ to require a temp increase of 0.3C or less, over that 10m, as near as I can tell.

I added some linear trendlines in Excel, and over the 20 data points previous to temp drops (when the heat source appears to have been turned off), I seem to be getting slopes of less than 0.02C/interval (min=0.0126, max=0.0189), for the four spreadsheets I played with quick.