An alternative interpretation to our breakout model is that the sub-caldera magma chambers grew to an elliptical shape due to the presence of pre-existing, NW-SE striking structures or zones of weakness in the older volcanic rift fill and basement ("transverse tension zones" of McCall, 1968). Many structural attributes of the Kenya rift, such as border fault geometry and basin segmentation, have been shown to be at least partially controlled by basement structures (McConnell, 1972; Smith and Mosley, 1993; Hetzel and Strecker, 1994). Dunkley et al. (1993) have identified NW-SE trending shear zones in the Precambrian basement complex of the eastern rift shoulder that project across the rift valley at the position of the Emuruangogolak and Namarunu volcanoes. However, no such corresponding features are known for Paka, Silali, or the Barrier (Kakorinya).

If the orientations of the rift axis calderas were controlled by pre-existing basement structures, then this would fail to explain why the older Suswa caldera formed elongate ENE-WSW, when the extension direction was approximately the same, while the northern calderas are elongate progressively more NW-SE, according to their age of formation. If pre-existing faults or fractures were the dominant factor, then it would seem more likely that calderas should be elongate parallel to the profuse system of rift-parallel structures, or approximately N-S.

As discussed above, aligned minor craters and pyroclastic cones of early to mid-Pleistocene age throughout the rift valley axis are generally oriented ~N-S, whereas in the latest Pleistocene to Recent, NE-SW alignments appear, such as in the Black Hills of Silali. This shows that despite the very strong N-S structural and volcanic fabric of the inner rift, feeder dikes have been forced to change to the new, mechanically most efficient orientation. When borehole breakouts were first recognized in wells in the 1960’s and 1970’s, they were interpreted to represent elongation parallel to joints and fractures intersected while drilling (e.g., Babcock, 1978). However, after detailed observational and theoretical studies, it was established that the elongations are related to the state of stress in the borehole wall, and not to pre-existing fractures (Bell and Gough, 1979; Plumb and Hickman, 1985; Zoback et al., 1985).

Fortunately, the catastrophic Kenya rift valley caldera collapse events occurred during prehistoric times, prior to the colonization of large populations along the rift axis. Detailed studies of Quaternary stress field evolution may help assess the likelihood of future catastrophic magma loss and caldera collapse at other Holocene volcanoes.

Acknowledgments - Funding for travel and field work was provided by SFB 108 and Marathon Oil Company. Discussions with Martin Smith were greatly appreciated. John Platt made helpful comments on the manuscript. Win Means first directed one of the authors (WB) toward potential uses of the cylindrical hole problem many years ago during his work as a doctoral student.