Drag reduction, or the mean velocity increase in a turbulent flow at a fixed pressure drop through the addition of tiny amounts ͑several parts per million͒ of high-molecular-weight polymers ͑Thoms effect͒, has been known already for more than 60 years. Long ago it was understood that this effect is related to supramolecular structures formed in the flow. Recent experiments by Chu, Shaqfeh, and associates, where the motion of supramolecular structures was directly observed, made it possible to understand and quantify the dynamic interaction of the polymeric structures with the solvent ͑water͒ flow. These results lead to the construction of a mathematical model of the Thoms effect, based on the Kolmogorov ͑1942͒-Prandtl ͑1945͒ semiempirical theory of shear flow turbulence. This is the subject of the present letter.Let us first introduce the basic equations of the present work. Turbulent shear flow is considered, so the mean velocity field is assumed to beThe density of the solution is indistinguishable from the water density ; therefore under zero mean mass force the averaged momentum balance equation has the same form as in the absence of the mass force,In these relations the coordinate x 1 is reckoned along the mean flow and the axis x 2 is directed perpendicularly to the wall x 2 = 0. Furthermore, T 12 is the only nonzero component of the Reynolds shear stress, is the shear stress assumed to be constant across the flow, and averaged viscous stress is neglected. The bars denote probability average ͑mean͒ values and the turbulent fluctuations are denoted by primes. The quantity u ء = ͑ / ͒ 1/2 is the dynamic, or friction, velocity.The equation of the turbulent energy balance in the shear flow takes the formHere, unlike in the customary derivations, the mass force is taken into account: F =0; FЈ is the mass force fluctuation. Furthermore, ⑀ is the mean dissipation rate of the turbulent energy into heat,͑4͒ D is the symmetric part of the strain-rate tensor, is the fluid kinematic viscosity, and the summation over repeated Greek indices from 1 to 3 is assumed. The derivation of Eqs. ͑1͒ and ͑2͒ follows the general lines of Ref. 1.The turbulent flow is a cascade of vortices of various scales. The basic Kolmogorov hypothesis 2 ͑see also the paper by Prandtl 3 ͒ can be presented in the following form: At large Reynolds numbers the vortex cascade is self-similar. Hence, all dimensionless flow field properties are universal. According to this hypothesis, all kinematic flow properties are determined by two kinematic properties with different dimensions. Kolmogorov 2 and his direct followers ͑see Ref.1͒ selected as such quantities the local mean turbulent energy per unit mass, b = ͑u 1 Ј 2 + u 2 Ј 2 + u 3 Ј 2 ͒ / 2=u ␣ Јu ␣ Ј / 2, and the local mean length scale of vortices, or the local external length scale of turbulent flow l, which is proportional to it. This hypothesis forms the basis of the ͑b , l͒ semiempirical model. The other possibility is the ͑b , ⑀͒ model which also came into wide use. ͑I used here the origina...