Modern micrometeorology is an amalgam of two traditions: one arises from field measurements in the open atmosphere and another from wind tunnel experiments. Early micrometeorological field work, and even relatively recent studies (viz. the von Karman constant experiments by Frenzen and Vogel, 1995), were done with precision cup anemometers, which can only provide a scalar speed measurement. Most of the early flux/gradient measurement programs (see Dyer, 1974) used scalar wind measurements, and derived quantities such as friction velocity (u*), logarithmic wind profiles, and Monin-Obukhov Similarity Theory (MOST) relationships based on these measurements. Businger et al. (1971) made a distinction between the mean horizontal (scalar) wind speed and the magnitude of the mean horizontal wind vector. However, Businger (1973) reverted to the old habit of referring to the mean (scalar) horizontal wind speed u (the underbar signifies an averaged or mean quantity) in his work on turbulent transfer in the atmospheric surface layer. Meanwhile, velocity component and Reynolds stress experiments done in wind tunnels relied primarily on hot-wire anemometry (see, for example, Bradshaw, 1971), which provided orthogonal velocity component measurements. With the advent of sonic anemometry (sonics), atmospheric measurements of alongwind u, crosswind v, and vertical w velocity components have become routine. In consequence, eddy correlation techniques taken from the laboratory tradition are used in computation of a “vector” momentum flux Fn and derived quantities such as friction velocity u* and Obukhov length L. Other flux computations are also possible, including the scalar, gross, and alongwind flux. Given the capacity to compute momentum flux in a variety of different ways, which way is most appropriate for MOST applications?