The $\eta$ transitions of $\Upsilon(4S,5S)$ into $\Upsilon(1S,2S)$ are studied in the rescattering model by considering the final state interactions above the $B\bar B$ threshold. The width of the $\eta$ transition of $\Upsilon(4S)$ into $\Upsilon(1S)$ is found to be larger than that of the dipion transition, and the ratio of $\Gamma(\Upsilon(4S)\to\Upsilon(1S)\eta)$ to $\Gamma(\Upsilon(4S)\to\Upsilon(1S)\pi^+\pi^-)$ is predicted to be $R_4=1.8{ -}4.5$, which is about two orders of magnitude larger than the expectation of the conventional hadronic transition theory, and is supported by the new BaBar measurement. The widths of the $\eta$ transitions of $\Upsilon(5S)$ are found to be sensitive to the coupling constants $g_{ \Upsilon(5S)B^{ (*)}B^{ (*)}}$ due to a large cancelation between contributions from the $B\bar{ B}$, $B^*\bar{ B}+c.c.$, and $B^*\bar{ B}^*$ channels, and only a rough estimate $\Gamma(\Upsilon(5S)\to\Upsilon(1S,2S)\eta)=10{ -}200$ KeV can be given. The widths of the $\eta'$ transitions of $\Upsilon(4S,5S)$ are also discussed, and they could be much smaller than that of the corresponding $\eta$ transitions mainly due to the tiny phase space.
Categories: hep-ph hep-ex