1: \begin{abstract}

2: We establish global rates of convergence of the Maximum Likelihood Estimator (MLE) of 

3: a multivariate distribution function on $\RR^d$ in the case of (one type of) ``interval censored''

4: data.  The main finding is that the rate of convergence of the MLE in the Hellinger metric

5: is no worse than $n^{-1/3} (\log n)^{\gamma}$ for $\gamma = (5d - 4)/6$. 

6: \end{abstract}

7: