It's a test to see if the variables associated with and are of any value.
By that we mean, do these x's help in explaining y any better than the first three terms in your model.
Note that increases as we introduce more terms, but the mean squares may not.
The mean squares, in particular MSE is SSE divided by n minus the number of parameters in the model.
SSE will decrease as you add terms, but so will the denominator of MSE.