Values of osmotic water permeability (P _osm) calculated from steady-state osmotic flow in response to a gradient of a probe molecule tend to be underestimates, because of three unstirred-layer (USL) effects. These are: dissipation of the probe's gradient by diffusion in USL's; reduction of the probe's gradient, due to the sweeping-away effect of water flow gencrated by the probe itself; and solute polarization (creation of an opposing gradient of an initially symmetrically distributed solute by the sweeping-away effect). These errors increase with probe permeability. USL thickness,P _osm, and concentration ratio of symmetrically distributed solute to probe, and vary inversely as the fractional area available for water flow (e.g., lateral intercellular space width). The form of an osmotic transient, and the possibility of extracting a trueP _osm value from the transient, depend on the relative values of three time constants: those for solute diffusion in USL's, for solute polarization by water flow in USL's and for measuring water flow. Sweeping-away effects cause major underestimates (by one or more orders of magnitude) in epithelialP _osm determinations, as shown by apparent streaming potentials during osmotic flow and by transiently reversed flows after removal of the probe. TrueP _osm values for leaky epithelia probably exceed 10⁻³ or 10⁻² cm/sec·osm. The necessary conditions for resolving osmotic transients are set out.

I illustrate the difficulties in deciding what fraction of transepithelial water flow is via the cells, and what fraction via the junctions. There is no existing method for answering this question.

Controversies about the validity, or need for modification, of the standing-gradient theory are discussed.