1: \begin{abstract}

2: \noindent

3: We present an improved analytic calculation for the tidal radius of

4: satellites and test our results against N-body simulations.

5: 

6: The tidal radius in general depends upon four factors: 

7: the potential of the host galaxy, the potential of the satellite, the

8: orbit of the satellite and {\it the orbit of the star within the

9:   satellite}. We demonstrate that this last point is 

10: critical and suggest using {\it three tidal radii} to cover the

11: range of orbits of stars within the satellite. In this way we

12: show explicitly that prograde star orbits will be more easily

13: stripped than radial orbits; while radial orbits are more easily

14: stripped than retrograde ones. This result has previously been

15: established by several authors numerically, but can now be understood

16: analytically. For point mass, power-law (which includes the

17: isothermal sphere), and a restricted class of split power law

18: potentials our solution is fully analytic. For more general

19: potentials, we provide an equation which may be rapidly solved

20: numerically.

21: 

22: Over short times ($\simlt 1-2$ \,Gyrs $\sim 1$ satellite orbit), we

23: find excellent agreement between our analytic and numerical models. Over longer

24: times, star orbits within the satellite are transformed by the tidal

25: field of the host galaxy. In a Hubble time, this causes a convergence

26: of the three limiting tidal radii towards the prograde stripping

27: radius. Beyond the prograde stripping radius, the

28: velocity dispersion will be tangentially anisotropic.

29: 

30: \end{abstract}

31: