For many years, speckle interferometry has proven its usefulness for measuring deformation and contours of objects with rough surfaces. Nevertheless, the influence of a parameter as important as the lens aperture of the imaging system on the result of the measurement is still not precisely known. Generally, it is stated that the speckles have to be resolved by the camera and that therefore the lens aperture number should be large enough (e.g. f/11 or f/16 for typical applications with a standard CCD). This contribution shows, theoretically as well as experimentally, that it is not necessary to resolve the speckles when making measurements in speckle interferometry. The joint probability distribution of the pixel modulation and the pixel background intensity, which determines the percentage of valid pixels, is investigated for different lens aperture numbers, i.e. for different numbers of speckles per pixel. The comparison between theory and the experimental counterpart allows a number of speckles per pixel to be determined for any given aperture number and pixel size. It appears that the average pixel modulation remains high enough even with a relatively large number of speckles per pixel. The results lead to the conclusion that--especially when the total camera noise is low enough--lens apertures up to f/2 may be used. Thus, measurements are less constrained by limited laser power and decorrelation effects.