OK, shrunken duplicate approach isn't giving me joy - the shape I want to use has concave portions, and scaling down causes the inner surface of the duplicate part to breach the concave surface of the "glass" part. Of course, I could use a "shell" type boolean to shrink proportionately on all surfaces, I suppose, but that would cause thinner portions of the shape to not have any particles at all.
Ugh, my head hurts.