Anisotropic fibres may be analysed in terms of equivalent isotropic fibres. This gives a clear physical picture of propagation, particularly the tunnelling loss mechanisms of leaky modes and the effect of a finite cladding on such modes. Slightly tilting the cladding optical (stress) axis relative to that of the core axis greatly amplifies the leakage reported previously for aligned optical axes, e.g. about 60 dB per metre for a 3° tilt instead of about 4 dB per kilometre. Thus, in theory, tilting can produce a highly effective polarising fibre (with a typical finite cladding) without relying on the polarisation dependence of bending losses.