In a discussion of the turbulence characteristics of patches of ''microstructure'' in the ocean, the hypothesis advocated by Gibson (1982), that the patches are produced by very rare but extremely powerful turbulence-generating events which usually have ''fossilized'' before their observation, is contrasted with the hypothesis of a turbulence field driven at the time and scale at which it is observed. In this ''continuous creation'' notion, by no means original here, the driving energy is converted to turbulence kinetic energy in such a way that the observed overturning thickness scale LT is linearly related to the length scale (&Egr;/N3)1/2, where &Egr; is the kinetic energy dissipation rate and N is the buoyancy frequency. (This relationship does not hold in boundary layers, where another length scale, the distance from the boundary, is imposed.) If the time scale of the largest vertical eddies is N-1, the parameters of turbulence and its effects can be estimated by the measurement of N and LT. For example, the kinetic energy dissipation rate would be proportional to LT2N3, and the vertical eddy diffusivity would be proportional to LT2N. Careful attention must be paid to the sampling process and its assumptions. ''Fossilized'' regions are expected, in the sense that these regions had previously been the sites of turbulence stronger than that present at the time of observation, but consideration of the fossilization process is not necessary for the interpretation of microstructure data.