data raised from msd/molecule command

Hi;
When i drew diagram of msd command’s data toward timestep, I faced with an irregular graph so that i cant compute slope of this graph in order to obtain diffusion coefficient. I dont know how i could make careful my statistics so that diagram of MSD toward time becomes a straight line?
I read in manual that the slope of the mean-squared displacement (MSD) versus time is proportional to the diffusion coefficient of the diffusing molecules
But when i draw this diagram, even approximately it isnt a straight line!?
I dont know what i should do? Dose my system have problem?
Please help me
Thanks in advance
Samira;

Over long enough times and enough diffusive movement,
the slope of MSD should be a constant straight line.
If you don’t have enough statistics, it will be noisy.

Steve