The neighbor diffusivity for pairs of diffused particles is determined from observations on drifts of various kinds of materials such as drift bottles in many parts of the ocean and a lake. Its values are expressed as the 4/3 power of separation of particles over a range from 10 to 108 cm.A recent work by Joseph and Sendner on horizontal diffusion in the sea seems to give promise of an eventual resolution of the Fickian versus neighbor diffusivity problem (1). Among other things, it has caused us to reexamine our own work on the subject. The purpose of this note is essentially to extend the range in which the 4/3 power law appears to be valid.It will suffice to consider only the one-dimensional aspect of diffusion. If v(x) is the concentration of particles at x, there will be v(x)dx particles between x and x+dx. By analogy to classical concentration, the neighbor concentration q(l) is defined as the number q(l)dl of pairs of particles whose separations are in the range I to l+?dl. The Richardson diffusion equation Oq a [F(l) ] (1) is analogous to the classical Fickian equation since F(l), the neighbor diffusivity, takes the place of the ordinary diffusivity K. Stommel postulated that the initial separation lo is large compared with hilo, where lh is the separation after time T (2). With this restriction, he derived the relation F[ (It1r )] =(/1-)where the bars indicate averages. Stommel checked the validity of this equation by using pieces of parsnip with spacings of the order of 25 to 200 cm (Richardson and Stommel, 3) and dye spots with spacings 1000 to 10,000 cm and sheets of mimeograph paper with spacings 40 to 1000 cm (Stommel, 2). These data are shown by crosses in Fig. 1. Later, Olson (4) showed that his drift card data (5) and Platania's (6) drift bottle data also seemed to satisfy Eq. 2 in spite of the severe deviation from the conditions imposed in deriving the equation. This may be seen in Fig. 1 where Platania's data are represented by a square and Olson's data by triangles.