For the problem of propagation of density waves in a preexisting gravitational field, the advantages of the deDonder gauge over the commonly used synchronous gauge are outlined. In a background matter substratum withp ρ as equation of state there are in the deDonder gauge only decaying modes of the perturbation density contrast with arbitrary large spatial extension, whereas in synchronous gauge there is one growing mode (calculated for vanishing spatial divergency of the perturbation in the 4-velocity, i.e.,usk(1),j/j≡0). The calculations are extended to the case of finite spatial extensions of the density perturbations. This is done by expanding all perturbations in a power series of the inverse square of the speed of light with the result of getting a recursive set of differential equations in both gauges for the equation of motion of the density perturbations. The lowest orders of this equation are the same in both gauges, but only in the deDonder gauge is the correct Newtonian limit of propagation of density waves in an expanding universe obtained. The correction by the next higher orders in the deDonder gauge are dependent explicitly on the spatial extension of the perturbations; whereas in synchronous gauge this is not the case. For attaining the Newtonian limit this dependence is a necessary condition. At appropriately large spatial extensions, however exact, this dependence in deDonder gauge leads ultimately to a decaying of density contrast modes growing in zeroth order (at least forp=0 andpρ/3 as equations of state for the background matter substratum). Hence, there are upper boundaries in the spatial extensions of instable growing modes of density contrast.