\begin{table}%t2 
 \caption{\label{12745tb2}Relative differences between metal measurements made with
 the three EPIC instruments MOS1, MOS2 and {\it pn} computed over the
 whole sample. }
\small%\centerline
{
  \begin{tabular}{l c c c}   
\hline\hline    
{\it mekal} model & & & \\
\hline
Metal $X$ & $(X_{\rm MOS1}-X_{\rm MOS2})/X_{\rm MOS2}$ & $(X_{\rm MOS1}-X_{pn})/X_{pn}$ &
$(X_{\rm MOS2}-X_{pn})/X_{pn}$ \\    
\hline
    Fe                & $~~0.02\pm 0.01$ & $0.07\pm 0.01$ & $0.02\pm 0.02$ \\
$+3\%$ syst.err.      & $~~0.02\pm 0.02$ & $0.04\pm 0.02$ & $0.01\pm 0.02$ \\
\hline
    Si                & $~~0.01\pm 0.02$ & $0.19\pm 0.03$ & $0.10\pm 0.03$ \\
$+3\%$ syst.err.      & $ -0.03\pm 0.04$ & $0.11\pm 0.04$ & $0.08\pm 0.04$ \\
\hline
    Ni                & $ -0.12\pm 0.08$ & $0.14\pm 0.11$ & $0.01\pm 0.09$ \\
$+3\%$ syst.err.      & $ -0.13\pm 0.08$ & $0.12\pm 0.11$ & $0.00\pm 0.10$ \\
\hline
& & & \\
& & & \\
\hline\hline                        
{\it apec} model & & & \\
\hline
Metal $X$ & $(X_{\rm MOS1}-X_{\rm MOS2})/X_{\rm MOS2}$ & $(X_{\rm MOS1}-X_{pn})/X_{pn}$ &
$(X_{\rm MOS2}-X_{pn})/X_{pn}$ \\       
\hline
    Fe                & $~~0.03\pm 0.01$ & $0.07\pm 0.01$ & $0.02\pm 0.02$ \\
$+3\%$ syst.err.      & $~~0.02\pm 0.02$ & $0.04\pm 0.02$ & $0.01\pm 0.02$ \\
\hline
    Si                & $~~0.01\pm 0.03$ & $0.22\pm 0.03$ & $0.13\pm 0.04$ \\
$+3\%$ syst.err.      & $ -0.03\pm 0.05$ & $0.13\pm 0.04$ & $0.11\pm 0.04$ \\
\hline
    Ni                & $ -0.16\pm 0.09$ & $0.13\pm 0.14$ & $0.01\pm 0.11$ \\
$+3\%$ syst.err.      & $ -0.17\pm 0.10$ & $0.12\pm 0.14$ & $0.01\pm 0.11$ \\
\hline
\end{tabular}}
\smallskip
For each element in the first line only statistical
 errors are accounted for, in the second line a $3\%$ systematic error
 is added in quadrature to each abundance measurement (see
 discussion in Sect.~4.3).
 \end{table}