A central theme in the history of the turbulence problem is about the metho
d of 'closure' in the models and 'theories' which have been proposed. Closu
re has invariably been by empirical calibration with experimental data. In
this note we draw attention to a paper by Morris, Giridharan and Lilley, in
which for the first time empiricism is obviated. For the turbulent mixing
layer, this is accomplished by including in its description the mechanism f
or production of turbulent shear stress (i.e. turbulent momentum transfer),
by large-scale instability waves. Some implications for the theory of turb
ulent shear flows are discussed.