$\Upsilon(nl)$ decay into $ B^{(*)} \bar B^{(*)}$

We have evaluated the decay modes of the $\Upsilon(4s), \Upsilon(3d), \Upsilon(5s), \Upsilon(6s)$ states into $B\bar B, B\bar B^*+c.c., B^* \bar B^*, B_s \bar B_s, B_s \bar B^*_s +c.c., B^*_s \bar B_s^*$ using the $^3P_0$ model to hadronize the $b\bar b$ vector seed, fitting some parameters to the data. We observe that the $\Upsilon(4s)$ state has an abnormally large amount of meson-meson components in the wave function, while the other states are largely $b\bar b$. We predict branching ratios for the different decay channels which can be contrasted with experiment for the case of the $\Upsilon(5s)$ state. While globally the agreement is fair, we call the attention to some disagreement that could be a warning for the existence of more elaborate components in the state.