Your so-called Rayleigh distance, as vfone alludes, is the range at which the quadratic phase error is less than 22.5 deg. or so. This assures that all field components will add 1) constructively, and 2) within cos(22.5 deg), or about 0.7 dB amplitude taper. That is, the incident field transmitted from the source arrives at the aperture of the antenna under test as a planar wavefront with uniform intensity. The distance is the minimum distance necessary to perform accurate farfield radiation pattern testing/measurement without sidelobe distortions, nulls filling in, cross-pol levels creeping up, etc.
This "nearfield" (the Fresnel zone) should not be confused with the "reactive nearfield", which is much closer in, and wherein any physical intrusion results in perturbation of the radiation pattern, impedance response, or both.
The Fresnel nearfield is a region in which the test probe and antenna under test are essentially de-coupled, but radiation pattern measurement in this range requires relative translation instead of rotation of the two, as well as corrections for the radiation characteristics of the test probe.
As a final note, the 2D²/λ; farfield criterion really only applies to apertures larger than 2.5λ;
For apertures less than λ/3, the farfield criterion is r > 1.6λ;
For apertures between λ/3 and 2.5λ;, the farfield criterion is r > 5D!
The attached paper by Rajeev Bansal that appeared in Applied Microwave & Wireless discusses this in more detail...