\begin{table}%t5
\caption{Estimated bandwidth for the kernel method and mean integrated squared error: for each subsample, the size~$n$ (number of stars) is given, as well as the estimated $\hat{h}$ according to \citet{ShrJos91}, computed in the logarithmic velocity scale, and the variability band width~$\epsilon$ (Eq.~(\ref{epsilon})).}
\label{sjeqd}
%\centering
\small
\begin{tabular}{lrrr}
\hline
\hline
\multicolumn{1}{c}{Subsample} & \multicolumn{1}{c}{$n$} & \multicolumn{1}{c}{$\hat{h}$}  & \multicolumn{1}{c}{$\epsilon$} \\
\hline
B9     & $125$ & $0.223$ & $0.0503$ \\
A0--A1 & $271$ & $0.147$ & $0.0421$ \\
A2--A3 & $258$ & $0.170$ & $0.0401$ \\
A4--A6 & $137$ & $0.152$ & $0.0583$ \\
A7--A9 & $141$ & $0.137$ & $0.0604$ \\
F0--F2 & $150$ & $0.149$ & $0.0561$ \\
\hline
\end{tabular}
\end{table}