1: \begin{abstract}

2: Our main result is a generalization of Cappell's 5-dimensional

3: splitting theorem. As an application, we analyze, up to internal

4: $s$-cobordism, the smoothable splitting and fibering problems for

5: certain 5-manifolds mapping to the circle. For example, these maps

6: may have homotopy fibers which are in the class of finite connected

7: sums of certain geometric 4-manifolds. Most of these homotopy fibers

8: have non-vanishing second mod 2 homology and have fundamental groups

9: of exponential growth, which are not known to be tractable by

10: Freedman--Quinn topological surgery.  Indeed, our key technique is

11: topological cobordism, which may not be the trace of surgeries.

12: \end{abstract}

13: