Just How Noisy are Accelerometers and Gyros from Nunchuck and Wii Motion Plus?

In evaluating accelerometers from Nunchuck and gyroscopes from WM+ , among other things, I wanted to see what kind of data they will produce if they are not moving. I set 'em up as level as I could, plugged them into Arduino and collected the data for little longer then an hour, taking care not to move them at all. I ended up using this data for code in acquiring Nunchuck and WM+ data from pass-through port. I repeated this couple of times, performed simple statistical analysis, and got very consistent results.
On Wii Motion Plus data was collected for all three gyros. As you can see there is about 30 or so "counts" uncertainty in Wii gyroscopes.

.	Mean	Std. Dev.	Variance	Slope	Intercept
Yaw	8163.42	6.06	36.74	-0.000078	8163.65
Pitch	7856.99	4.58	20.98	-0.000097	7857.27
Roll	8533.91	4.48	20.06	-0.000065	8534.10

For Nunchuck data was collected for all three accelerometers. X and Y accelerometers show values for 0g (or close to it) and z accelerometer data shows close to 1g data.

.	Mean	Srd. Dev.	Variance	Slope	Intercept
ax	523.01	0.09	0.01	0	523.02
ay	521	0.07	0.01	0	521.01
az	757.17	0.38	0.14	0	757.31