Thanks Jim,
I don't have Matlab so I can't do those nice plots 
but just visually looking at the figures I could see that they all appeared to be in the same quadrant.
Which as we both surmised is probably caused by an inaccuracy between the "pose" (orientation) vector from RTIMULib and the real direction of gravity.
It looks like the RTIMULib supports the LSM9DS0, but I'm not sure if it would work for your LSM303 its probably worth a shot
I'll post my revised Processing data visualation / collection program later (its on my other PC), when I get chance, in case its any use to you
Cheers
Roger