Particle orbits in two-dimensional (2-D) kinetic equilibrium solutions for the magnetotail are considered. A comparison is made between the orbits in the equilibrium field and those found in previous studies of the orbits where the main component of the magnetic field (Bx) was approximated as either a hyperbolic tangent or a linear function of z with the normal field (Bz) taken to be a constant (one-dimensional models). Here, the plane z=0 defines the neutral plane. The orbits in the 2-D field are found to be significantly different from those in the one-dimensional models. The particles are generally reflected in much shorter distances and the loss region is much smaller than in the one-dimensional fields. This holds true independent of &rgr;/L, where L is the current sheet thickness and &rgr; is the Larmor radius in the asymptotic field. For &rgr;/L>0.1, the fraction of integrable orbits in the equilibrium fields also becomes different from that in the one-dimensional fields. As &rgr;/L becomes larger than 0.3 (with the exact value depending on the eqilibrium model and particle position in the tail), the phase space becomes dominated by integrable orbits, whereas in the one-dimensional fields the phase space remains divided into integrable and chaotic regions.

Another major difference between the orbits in the 1-D and 2-D fields lies in the fact that the equilibrium orbits have a net drift in the y direction whereas the orbits in the 1-D models have zero net drift. In the presence of an electric field Ey, the equilibrium orbits can be accelerated to much larger energies and behave quite differently from the orbits in the 1-D fields. The differences between the orbits in the equilibrium and one-dimensional field are mainly due to the neglect of the variation of Bx with x in the one-dimensional fields. In fact, it is found that even a weak variation of Bx with x can alter the orbits significantly. The effect of a constant-shear field on the orbits is also examined. Finally, it is shown that the border between the integrable and chaotic orbits is in general ''sticky''; i.e. if a stochastic orbit comes close to an integrable region, it can get trapped. Since the orbit is stochastic, it will escape eventually, but the detrapping time can be very long. Thestickiness occurs where there are islands embedded in phase space. Thus, care must be taken in applying the diffusion formalism to such regions. The relevance of the above findings to the problem of magnetic reconnection in the tail is discussed. ¿American Geophysical Union 1990