Explain exactly why networks of perceptrons with linear activation functions are uninteresting

$($that is$,$ networks of perceptrons where, for each perceptron, the output is some constant times the

weighted sum of the inputs$).$ Use equations if necessary.