\begin{table}
\par
  \caption{\label{tab:datasets}Characterization of the different datasets used in the fit. }
  %\centerline
{\small
  \begin{tabular}{ l r r r r r r }
    \hline \hline
    Source$^a$ & \# of Transitions$^b$ & Max. $J$ & Max $K_{\rm a}$ &
    \multicolumn{2}{c}{Frequency [GHz]} & $\sigma$ \\
    &  &  &  & \multicolumn{1}{ p{3em} }{\raggedleft Min.} &  Max. &  \\ \hline
    Synthesizer (Cologne)          &  344 (259) & 54 & 19 &  39 &   56 & 1.02 \\
    BWO (Cologne)                  &  586 (273) & 70 & 27 &  69 &  897 & 0.90 \\
    BWO + SL (Cologne)             & 1663 (635) & 63 & 27 & 263 & 2099 & 0.95 \\
    Multiplier chain (JPL)         & 2234 (435) & 67 & 30 & 780 & 1665 & 1.08 \\ \hline
    \citet{Niide2003}              &   16 ( 13) &  7 &  2 &  12 &   25 & 1.77 \\
    \citet{Durig1976}              &   20 ( 15) &  4 &  1 &  29 &   36 & 0.50 \\
    \citet{Lovas1979}              &  191 (157) & 22 &  6 &   9 &  112 & 0.74 \\
    \citet{Neustock1990}           &   79 ( 66) & 24 &  5 &  63 &  222 & 0.96 \\
    \citet{Groner1998} (Klystron)  &  962 (702) & 25 &  6 &  94 &  544 & 1.08 \\
    \citet{Groner1998} (FASSST)    &  735 (386) & 49 & 10 & 263 &  340 & 0.85 \\ \hline
    % 
  \end{tabular}}
\par
\medskip
$^a$ The first four rows summarize
    transitions measured within this work.\\
$^b$ The number of assigned
    transitions is given followed by the number of different frequencies
    in parentheses. The last column shows the unitless standard deviation
    of the dataset in the fit.
\end{table}