In February 1992, the Ulysses spacecraft passed through the Io plasma torus. A preliminary estimate of electron density and temperature was obtained by [Hoang et al. 1993] from the spectrum of the plasma quasi-thermal noise ([Meyer-Vernet and Perche 1989]) measured by the radio and plasma wave experiment ([Stone et al. 1992a]). This determination used both the upper-hybrid frequency peak, as in ([Stone et al. 1992b]), and a novel technique based on the polarization of Bernstein waves ([Meyer-Vernet et al. 1993]). A detailed analysis of the data has subsequently given the dispersion characteristics of these waves ([Moncuquet et al. 1995]), yielding more precise and extended results. The present paper studies these in situ values of the electron density and temperature as a function of latitude in the torus.
The Ulysses trajectory was basically north-to-south
(crossing the Jovian magnetic equator at about 8 
from
Jupiter), in contrast to Voyager which explored the torus near
the equatorial plane. Hence the present results provide the
first in situ measurement of the torus latitudinal structure;
they are unique, since Ulysses particle analyzers were not
operating near Jupiter. This may be an important key to
understand this medium, of which no self-consistent theory yet
exists, and whose energy balance is not fully understood (see
for example [Barbosa et al. 1983]; [Smith and Strobel 1985]; [Smith et al. 1988];
[Strobel 1989]).
We first briefly recall the principle of the measurement and its significance when the plasma is not in equilibrium. We then show that, instead of being constant along magnetic field lines - as currently assumed in the torus models ([Bagenal and Sullivan 1981]; [Divine and Garrett 1983]; [Bagenal 1994]) - the (bulk) temperature varies with latitude in anticorrelation with the density, following an approximate polytrope law with an exponent smaller than one.
We suggest a simple interpretation in terms of velocity filtration by the ambipolar electric field set up in the presence of plasma corotation, which tends to confine particles close to the centrifugal equator: since the more energetic electrons overcome more easily the confining potential, their proportion increases outside the equatorial region whereas the bulk density decreases; so that the density and temperature are anticorrelated. This mechanism only works if the velocity distribution is not Maxwellian. This basic property of a confining potential to act as a high-pass filter for particle energies - yielding anticorrelated density and temperature - was first noted by Scudder (1992a), who suggested that it should hold in many astrophysical contexts, and used it to explain temperature inversions in stellar coronae (Scudder 1992b). This allows us to build a simple model of the latitudinal density profile in the outer plasma torus, which we will compare with our observations.