There has been increasing interest in the role played by small--scale motions in transporting and mixing thermal and chemical inhomogeneities in the core. When the heavy constituents (mainly Fe) of core fluid freeze onto the ICB, latent heat and the light constituents are released. Moffatt [1989] proposed that this light, hot fluid would congregate into ``blobs'' at the ICB that would, when large enough, break away from the ICB and rise through the core, stirring it as it did so, and possibly retaining their identity until they reach the CMB, where some may remain to form a light layer (see § 4 below). This idea has been pursued by Ruan and Loper [1993], Loper and Moffatt [1993] and Moffatt and Loper [1994]; see also Bush et al. [1992, 1994], Loper et al. [1994]
Braginsky and Meytlis [1990] argued that core turbulence is totally
unlike classical turbulence of Kolmogoroff type, in which energy `cascades' from large
to small eddies.
Nor is it related to classical MHD turbulence, where a reverse cascade may create large-scale magnetic fields by turbulent dynamo action, possibly through a turbulent 
effect.
[The 
effect is the creation of a mean electromotive force (emf) parallel to the mean magnetic field.]
They argue that the microscale fields are so tiny that they produce neither a significant turbulent 
effect, nor an enhancement in the mean field diffusivity.
Nevertheless, the turbulent diffusivities of the mean thermal and chemical inhomogeneities are enormously greater than their molecular counterparts, at least in some directions.
Because of the strong influence of Coriolis and Lorentz forces on motions of all scales,
the turbulence is highly anisotropic, forming `plate-like' eddies that have their long
dimensions parallel to the rotation axis (Oz) and to the mainly zonal (
)direction
of the prevailing toroidal field; the short dimension is in the s-direction, away from the rotation axis.
Turbulent diffusion is represented by one tensor, the same for both heat and composition.
The elements of that tensor corresponding to diffusion in the z and 
directions are large, of the same order as the molecular magnetic diffusivity, 
;
turbulent diffusion in the s-direction is comparatively weak. Braginsky--Meytlis theory
has recently been taken further by Braginsky and Roberts [1994a];
it still contains ad hoc elements.
It is the highly dispersive character of rotating fluids that led to this unusual picture of core turbulence. The effect of that dispersion on the Moffatt mechanism has recently been studied by St. Pierre [1994a, b] who argues, on the basis of computer simulations, that a blob released at the ICB will be stretched, laminated in plates, and absorbed into its surroundings before it can rise far into the FOC; see also St. Pierre and Roberts [1994]. It is difficult in the laboratory to mimic core conditions in which the effective diffusivities of heat and composition are (see above) the same. Experiments have however been performed by Cardin and Olson [1992].