The velocity distributions observed in space have too many fast particles,
by Maxwell's standards. This ubiquitous property raises doubts about the va
lidity of models based on a set of fluid equations whose closure requires t
he distributions to be nearly Maxwellian. I discuss here two generic cases:
bound structures and winds. Near rapidly rotating magnetised planets, part
icles channelled along co-rotating magnetic field lines are acted on by the
field-aligned component of the centrifugal force, which exceeds the gravit
ational attraction beyond a few planetary radii. With dipolar magnetic fiel
ds, this tends to trap particles near the equator and produce torus-shaped
structures, whereas gravitational confinement occurs closer to the planet.
These confining forces act as high-pass filters for particle speeds, so tha
t the temperatures are rising with distance from the potential wells, if th
e velocity distributions are not Maxwellian - in sharp contrast to classica
l isothermal equilibrium; and the density profiles fall off less steeply th
an a Gaussian - just as the velocity distributions fall off less steeply th
an a Maxwellian. While these bound structures are shaped along closed magne
tic field lines, winds can brow along open field fines. A suprathermal tail
in the electron velocity distribution increases the electric field which e
nsures the balance of ion and electron fluxes, and should thus increase the
wind speed above the value predicted by classical hydrodynamic escape. (C)
2001 Elsevier Science Ltd. All rights reserved.