The way we conceptualize rain is fundamental in many branches of science since it provides the basis not only for rain modeling notably in meteorology and hydrology, but also for interpreting rain data (from gauges and radars). In order to empirically address this question, we use stereophotographic data to measure the positions and volumes of raindrops from similar to10 m(3) regions containing 5000-15 000 of these drops. By determining the drop statistics in spheres of increasing size, we conduct a basic continuum mechanics thought experiment. We show that-presumably due to turbulence-there is no microscale-macroscale separation. We find that the large particle number (N) limit in rain is not a homogeneous continuum, but rather it is nonclassical, strongly inhomogeneous, and approaching a multifractal discontinuum.