Torsion of thin tubes: a justification of some classical results