The question of how external ULF power in the Pc 3-4 frequency range (15--50 mHz) travels within the magnetosphere to drive the commonly observed resonant harmonics is still unresolved. The conventional approach has assumed direct transmission of compressional waves across field lines from the equatorial magnetopause, but increasing attention is focused on the high latitude ionosphere. Engebretson et al. [1991a] suggested that modulated Birkeland currents and/or particle precipitation moving toward the cusp/cleft ionosphere from the magnetospheric boundary may provide an efficient although roundabout means of driving toroidal pulsations throughout the outer dayside magnetosphere. Observations at South Pole, Antarctica by Engebretson et al. [1994b] have provided quantitative evidence that the large amplitude Pc 3-4 pulsations observed there can be driven by modulated electron precipitation near and slightly equatorward of the magnetospheric boundary, while Olson and Fraser [1994], using multistation and conjugate data, suggested that cusp entry was a more important source of Pc 3-4 wave energy. A recent study by Anderson and Engebretson [1995] using AMPTE CCE satellite data also supports the importance of an alternate wave excitation pathway. They found a poor correlation between compressional and azimuthally polarized wave power in the Pc 3-4 band as a function of L shell (i.e., location on a specific magnetic dipole field shell, parameterized by its distance from Earth at the magnetic equator, in units of Earth radii) and local time, and suggested that this surprisingly poor coupling between wave modes could in fact be less important than an indirect, nonlocal source as the means of stimulating azimuthally polarized Pc 3-4 waves.
Such an alternative source becomes of even more interest because of
the continuing lack of evidence for global cavity mode waves,
proposed a decade ago as a promising means of establishing field
line resonant waves with the discrete frequencies often seen on the
ground (cf. the review by Hughes [1994]). As part of a new
modeling effort Fujita and Patel [1992] reproduced many of the
results of earlier global mode calculations, but also pointed out
several idealizations in the boundary conditions used in such
calculations. Neglect of energy losses down the magnetotail,
outward through the magnetopause, and into the ionosphere in most
models might lead to significant overestimates of the importance of
such modes. Preliminary work by Hughes and Singer [1993] and
Engebretson and Anderson [1993] showed no evidence of such
cavity mode waves, down to amplitudes of 
0.1-0.3 nT, in data from
the Combined Release and Radiation Effects Satellite (CRRES) and the
Geostationary Operating Environmental Satellites (GOES), and AMPTE
CCE, respectively.
Attention has now shifted toward waveguide modes, which appear to better fit some of the constraints provided by high frequency (HF) radar observations of early morning long-period pulsations [e.g., Ruohoniemi et al., 1991; Walker et al., 1992]. Several studies using the Goose Bay HF radar have noted field line resonances with a set of fixed frequencies at 1.3, 1.9, 2.7, and 3.3 mHz, often in association with auroral arcs and large field-aligned currents. These well-documented observations of pulsations with surprisingly steady frequencies support in general the picture of ULF pulsations which are initiated by disturbances in the solar wind perturbing the magnetopause, but these authors claim that existing theories of ULF wave excitation cannot explain some of the important features of their observations.