If M is a compact 3-manifold, it is known that M can be foliated by 2-manifolds. Topological obstructions are given to the geodesibility of such a foliation IF; that is, to the existence of a Riemannian metric on M making each leaf a totally geodesic submanifold. For example, wil(M) must be infinite, and hence the Reeb foliation of S3 is not geodesible. The study of foliations from the point of view of differential topology has made tremendous progress in recent years; the excellent survey article by Lawson [3] describes many of the most fruitful areas of research. Foliations are also of fundamental importance in differential geometry, particularly in the study of fiber bundles and connections, but the geometric aspects of foliations per se have received considerably less attention. This note considers a geometrization of these topological structures, namely totally geodesic foliations. That is, all leaves are required to be totally geodesic submanifolds. Such foliations arise naturally in Riemannian submersions [51; also, a flat connection on a principal bundle yields a totally geodesic foliation for a suitable metric on the total space [2]. Two basic questions in this realm are: Q1: Given a Riemannian manifold M, does it admit a totally geodesic foliation of a given codimension? Q2: Given a foliation 1Y on a manifold M, is there a Riemannian metric on M such that IF is totally geodesic; namely is '5Y geodesible? If the dimension of ff is one, H. Gluck has recently made significant inroads into these questions [1]. In particular, Gluck has shown that any closed orientable 3-manifold has a geodesic flow, and has characterized those flows on 2-manifolds that are geodesible. However, we show below that the codimension-one case is considerably more restrictive. THEOREM 1. If M is a compact 3-manifold that admits a codimension-one totally geodesic foliation for some Riemannian metric on M, then ir,(M) is infinite. As a particularly interesting special case: COROLLARY. The Reeb foliation of S3 is not geodesible. Received by the editors October 9, 1978 and, in revised form, December 30, 1978. AMS (MOS) subject classifications (1970). Primary 57D30; Secondary 53B25, 53C20.
Read full abstract