1: \begin{abstract}

2: The paper deals with kernel density estimates of~filtering densities

3: in the particle filter. The convergence of the estimates is investigated

4: by means of Fourier analysis. It is shown that the estimates converge 

5: to the theoretical filtering densities in the mean integrated squared error

6: under a certain assumption on the Sobolev character of the filtering 

7: densities. A~sufficient condition is presented for the persistence of 

8: this Sobolev character over time. Both results are extended to

9: partial derivatives of the estimates and filtering densities.

10: \end{abstract}