\begin{table}%t2 \caption{\label{Table2}Normalized reflectivity gradient $S_{\rm ir}'$ and depth of the 1.52 and 2.02 ~$\mu$m bands \textit{D}, as defined in the text, for all the considered infrared spectra.} %\centerline { \small \begin{tabular}{l c c c c c} \hline\hline\noalign{\smallskip} Name & $^{a}S_{\rm ir}'$ & $D_{1.52}$ & $^{b}\varepsilon_{1.52}$ & $D_{2.02}$ & $^{b}\varepsilon_{2.02}$\\ \hline Aug. 2005 & 1.07 & 31.13 & 7.2 & 49.7 & 13.5 \\ % \hline sp 2006all & --1.40 & 34.0 & 4.8 & 53.8 & 6.6 \\ % \hline 2006 phase1 & 1.61 & 33.9 & 6.7 & 55.8 & 10.0 \\ % \hline 2006 phase2 & --1.77 & 33.3 & 5.6 & 51.3 & 10.2 \\ % \hline 2006 phase3 & --2.12 & 38.3 & 7.0 & 54.6 & 10.6 \\ %\hline 2006 phase4 & --2.32 & 32.4 & 9.2 & 49.5 & 13 \\ %\hline 2006 phase5 & --1.97 & 33.1 & 7.8 & 59.8 & 10.6 \\ \hline \end{tabular}} \medskip $^{a}$ Uncertainties of 2\%/1000~\AA{} in the slope are typical due to systematic errors (see text for details); $^{b}~\varepsilon$ are the errors associated with~\textit{D}. \end{table}