The dot density symbology seems to yield hugely more dots on polygon boundaries than would be probable with a uniform probability distribution on the polygons. This is evident no matter the number or size of the dots. It is particularly evident, though, when the dot value is increased so that a large number of polygons have only one dot. In that case, a huge fraction of all dots are on boundaries.
The centers of these dots are apparently so close to the boundaries that the boundary usually appears to go through the center of the dot with the naked eye. Thus it becomes practically impossible to visually determine the actual number of dots in any given polygon and thus to estimate the value of the symbolized field.
Has anyone else noticed this? Is this a known bug? Does anyone know of a way I could fix this?