Past exam of the mathematics course of the University of Cambridge / 2003 / iii / Paper 17 / 4 / c

False even when both base and fibre are closed and oriented. Take the Hopf fibration $h:S^3\to S^2$, and define the smooth fiber bundle $p:S^3\times S^1\to S^2$ by $p(x,\theta )=h(x)$. Its local trivializations are the Hopf trivializations times $S^1$, and its fibre is $S^1\times S^1=T^2$.

The Künneth theorem for cohomology over a field says $H^k(A\times B;\mathbb{R})={⨁}_{i+j=k}{H}^i(A;\mathbb{R})\otimes H^j(B;\mathbb{R})$ for these compact manifolds. Here it gives $H^2(S^3\times S^1;\mathbb{R})=0$. If a symplectic form $\omega$ existed on this closed four-manifold, it would be exact, $\omega =d\eta$. Since $d\omega =0$,by Stokes theorem. But ${\omega }^2$ is a positive volume form in the symplectic orientation, so its integral is positive. This contradiction proves the claim fails. The mere bundle structure does not ensure that a fibrewise area class extends to the total space.