I just ran a side by side comparison of C3 vs C4.1 with the same temperature probe. They were giving identical readings.
I can't think a situation in which the firmware could lead to measurement errors like that. 3 degrees is small enough that it could be because of noise of equipment grounding. 10 degrees is quite a bit.
I looked in our server logs and found the time period when you upgraded. There was quite a gap in time between the two readings:
Tue, 01 Dec 2009 01:45:39 GMT - Channel 1: 16.8 C
Tue, 01 Dec 2009 01:45:44 GMT - Channel 2: 31.3 C
Tue, 01 Dec 2009 19:39:00 GMT - Upgrade to C4.1
Tue, 01 Dec 2009 19:41:15 GMT - Channel 1: 9.3 C
Tue, 01 Dec 2009 19:41:19 GMT - Channel 2: 26.9 C
You know your server better than I do obviously, but is there a chance the temperature actually changed that much over the 18 hour gap?