noise noise noise.....
what would we do without noise.....
OK....
Capacitors have a 'reactance' or a resistance based on the frequency of voltage applied. Under DC conditions, the capacitor (without getting complicated) looks like an open circuit - looks like there is nothing there. So, in other words, the low frequency DC component of the signal coming down the temperature sensor wire is not effected by it (again, without getting too complicated).
Now, when you up the frequency, the capacitor starts behaving like a resistor (believe it or not), that is to say will convert electrical energy to heat. So, the higher frequency noise 'energy' that I think is effecting the temperature probe / probe wires, is now going into the capacitor and being dissipated as heat (killed off) instead of going into the unit and confusing the AtoD's at the microprocessor inside.
You can think of this as a low pass filter. This should be in the bloody design of the thing and looks like this simple filter trick is missing..... I could be wrong of course, do some more testing see if it really does work long term.
Better still is to use a series resistor (about 4K7) and then couple this to gnd using the 10nF.
Andy