There is a theory call the Central Limit Theorem that tells us that even if the original data is not normal, the process of averaging a set of data makes the mean quite close to the normal shape, especially for larger n values (at least 30).