1: \begin{abstract}

2: Linear wavelet density estimators are wavelet projections of the

3: empirical measure based on independent, identically distributed

4: observations. We study here the law of the iterated logarithm (LIL)

5: and a Berry-Esseen type theorem. These results are proved under

6: different assumptions on the density $f$ that are different from

7: those needed for similar results in the case of convolution kernels

8: (KDE): whereas the smoothness requirements are much less stringent

9: than for the KDE, Riemann integrability assumptions are needed in

10: order to compute the asymptotic variance, which gives the scaling

11: constant in LIL. To study the Berry-Esseen type theorem, a rate of

12: convergence result in the martingale CLT is used.

13: \end{abstract}