Let r (or v^T) be that linear combination of the row basis vectors such that the same linear combination of the rows of M is 0.
Then we have r M = 0.
This means that for any column vector w we must have r (M w) = 0 or r ^T M w = 0.