SECT. 6] 



TURBULENCE 



819 



separation" theory (1926, referred to on page 805). A number of tests made in 

 the sea with parsnip floats, paper markers and dye spots verified the relation 



F{1) = €^4/3 (29) 



between the "neighbour separation" I and the "neighbour diflfusivity" F{1) in 

 the range 1 = 25 cm to Z=100 m (Richardson and Stommel, 1948; Stommel, 

 1949). Olson (1952) made use of drift-card observations to extend the tests to 

 values of I up to 10 km. More recently Olson and Ichiye (1959), by including 

 drift-bottle observations in Japanese waters (Ichiye, 1951), showed that the 

 relation 



F{1) = 0.0246Z4/3 (30) 



fitted the whole series of data over the range Z=10to 10^ cm. The same authors 

 have also shown that the Z^/s law may be derived from the general theory of the 

 turbulent diffusion of two particles (Ichiye and Olson, 1960). Fig. 2 shows 



Fig. 2. Neighbour diffusivity F{1) as a function of neighbour separation I. (Taken from 

 Ichiye and Olson, 1960, Fig. 1. By permission of the German Hydrographic Institute, 

 Hamburg.) 

 Sources of data : x Stommel (1949). V Olson (1952). 



n Platania (quoted by Olson, O Ichiye (1951), A, open sea; 

 1952). B, near shore. 



the data on F{1) plotted as a function of I. Experiments with paper floats have 

 also been described by Ozmidov (1957), who found that the Z^/s law was valid 

 provided I > IQh, where h is the depth of water. In another experiment Ozmidov 

 (1958) used pairs of indicators of different diameters d, and found that their 

 rate of dispersion was consistent with the postulate that only eddies of wave 



