Is the dynamics of Jupiter family comets amenable to Monte Carlo modelling?

In order to explore the range of applicability of Monte Carlo modelling, we reconsider in depth the Monte Carlo simulations of Froeschlé & Rickman aimed at the study of the orbital evolution of Jupiter family comets. We first analyse the computation of perturbation distributions, and find that to get statistically significant results, the sample size of these distributions must be larger than several times 10⁵. We then show the necessity of introducing a procedure, referred to as the Spurious Perturbations Removal (SPR) procedure, to clean the perturbation distribution of any spurious contribution due to comets undergoing a close encounter with Jupiter at the beginning or/and the end of the perturbation computations. The partition of element space into 30 boxes in the (Q, q) plane (4 ≤Q≤ 13 au and 0 ≤q≤ 7 au) does not lead to dynamically homogeneous boxes, as is required for Monte Carlo simulations; these inhomogeneities are due to various reasons. Different ways to make such a partition are discussed. The strong influence of mean motion resonances on the dynamics of Jupiter family comets implies that resonances must be taken into account in any Monte Carlo simulation.