Why is the J/psi suppression enhanced at large transverse energy ?

We study the ratio of $J/\psi$ over minimum bias in $Pb$ $Pb$ collisions at SPS energy. The NA50 data exhibit a sharp turn-over at $E_T \sim 100$ GeV (close to the knee of the $E_T$ distribution) followed by a steady, steep decrease at larger $E_T$. We show that this behaviour can be explained by the combined effects of a small decrease of the hadronic $E_T$ in the $J/\psi$ event sample (due to the $E_T$ taken by the $J/\psi$ trigger), together with the sharp decrease of the $E_T$ distributions in this $E_T$ region (tail). This phenomenon does not affect the (true) ratio $J/\psi$ over $DY$ (obtained by the NA50 standard analysis), but does affect the one obtained by the so-called minimum bias analysis. A good agreement is obtained with the data coming from both analysis -- as well as with the ratios of $J/\psi$ and $DY$ over minimum bias -- in the whole $E_{T}$ region.