Mean Square (MS) is also called variance; approximated by the sum of squares (SS) divided by the degress of freedom (i.e., n-1); the MSE from the empty model can be thought of as roughly the average squared deviation.