In this paper we quantitatively investigate the hypothesis proposed by Mich
el (Exp. Physiol. 82, 1-30, 1997) and Weinbaum (Ann. Biomed. Eng. 26, 1-17,
1998) that the Starling forces are determined by the local difference in t
he hydrostatic and colloid osmotic pressure across the endothelial surface
glycocalyx, which we propose is the primary molecular sieve for plasma prot
eins, rather than the global difference in the hydrostatic and oncotic pres
sure across the capillary wall between the plasma and tissue, as has been u
niversally assumed until now. A spatially heterogeneous microstructural mod
el is developed to explain at the cellular level why there is oncotic absor
ption at low capillary pressures in the short-lived transient experiments o
f Michel and Phillips (J. Physiol. 388, 421-435, 1987) on frog mesentery ca
pillary, but a small positive filtration once a steady state is achieved. T
he new model also predicts that the local protein concentration behind the
surface glycocalyx can differ greatly from the tissue protein concentration
, since the convective flux of proteins through the orifice-like pores in t
he junction strand will greatly impede the back diffusion of the proteins i
nto the lumen side of the cleft when the local Peclet number at the orifice
is >1. The net result is that the filtration in the capillaries is far les
s than heretofore realized and there may be no need for venous reabsorption
. (C) 1999 Academic Press.