The diffusion of silica particles with radii ranging from 12 to 510 nn in dilute solutions of carboxymethyl cellulose (M-w = 180 to 1200 kg mol(-1), c(CMC) = 5 to 1000 mg l(-1)) was investigated by means of dynamic light scattering at pH 5 in 0.01 mol l(-1) NaCl. The viscosity of the polymer solution as experienced by the silica probes (the "microscopic" or effective viscosity, eta(eff)) differs from the viscosity as determined by capillary viscometry (eta(P)) For small particles eta(eff) nearly equals the viscosity of the solvent (eta(0)). The effective viscosity increases with the size of the probe particles and the polymer concentration but remains less than eta(P) The effective viscosity is interpreted in terms of a model in which the particle is surrounded by a layer of polymer free solution (eta = eta(0)). The thickness of the polymer-free layer is assumed to be equal to the thickness of the depletion layer (lambda(d)). Applying this model, a decrease in h, as a function of CMC concentration is observed. At low concentration lambda(d) equals the radius of gyration. The hydrodynamic layer thickness (delta(h)) of cellulose derivatives (carboxymethyl cellulose and hydroxyethyl cellulose) adsorbed on inorganic oxide surfaces (alpha-Fe2O3 and SiO2) is also investigated by dynamic light scattering. Upon using eta(P) maxima in delta(h) are found. However, these maxima are a consequence of an incorrect choice of the viscosity. When the viscosity is used as obtained from inert probe diffusion, no anomalies are observed.