Apparently the question was answered in the comments:  Edmonds proves a more general theorem for a family of matroids.  If the matroids are all the same and if that matroid is a vector matroid, then it is exactly the question stated.