\begin{table}%t4
\par
\caption {\label{tab:abundances}Element source abundances for Model~III-C and comparison with HEAO-3 results.}
%\centerline
 {
\begin{tabular}{ccll}\hline\hline
  $Z$    & Element &  $ 10^{22}\times q_Z$ &  ~~~~~HEAO-3 \\
	     &                     &  ~(m$^3$~s~GeV/n)$^{-1}$ &  (Engelmann et al.)\\\hline
	6    &          C          &  $148.5 \pm 3.		      $   &  $164.9 \pm 4.7		 $\\
	7    &          N          &  $ 8.1 \pm 0.6	      $   &  $ 9.9 \pm 3.4	   $\\
	8    &          O          &  $185.  \pm 3.        $  
	 &  $204.  \pm 2.2  $\\
	9    &          F          &  $ 3.67~\pm 0.05      $   &  $ 3.67~\pm 0.05  $\\
	10   &          Ne         &  $ 24.1 \pm 0.4 	    $   &  $22.5 \pm 1.3 	 $\\
	11   &          Na         &  $ 1.88 \pm 0.08      $   &  $ 1.25 \pm 0.5   $\\
	12   &          Mg         &  $  40.3 \pm 0.6       $   &  $ 40.3 \pm 1.0   $\\
	13   &          Al         &  $ 3.69~\pm 0.1       $   &  $ 3.02 \pm 0.6   $\\
	14   &          Si         &  $ {\bf 38.8} \pm 0.5 $   &  $ {\bf 38.8} \pm 0.5   $\\
\hline
\end{tabular}}
\par
\smallskip
Note 3: {Isotopic fractions of elements $q_i/q_Z$ are fixed to Solar System ones \citep{2003ApJ...591.1220L}
and $q_Z\equiv\sum_{i=\rm isot.} q_i$. The best-fit value is obtained from the best-fit model (on B/C),
where source abundances were rescaled to match HEAO-3 data at 10.6~GeV/n. The uncertainty
in $q_Z$ originates from the same rescaling, but arising from all combinations of
parameters satisfying the~68\% CL on the $\chi^2$ distribution. For HEAO-3, the numbers are taken
from Table~7 of \citet{1990AetA...233...96E}, and have been rescaled to
$q_Z({\rm Si})=38.8\times10^{-22}$~(m$^3$~s~GeV/n)$^{-1}$  (boldface) to ease the comparison.}
\end{table}