There are several far-field approximations (see Stutzman and Thiele pg 24):
Rff>2*D^2/lambda
Rff>>D
Rff>>lambda
D=maximum dimension of antenna
You pick the longest of the three.
Rff>>lambda is used for electrically small antennas, such as an HF ferrite rod, or can also be used for dipoles, patches, monopoles, etc.. Although I don't think you would measure a 1 MHz ferrite antenna at a distance of 3 km.
Rff>>D is used when the antenna D is on the order of 1*lambda. Maybe for low directivity arrays such as a couple of dipoles.
Rff>2*D^2/lambda is used for arrays where D>>lambda. If you had a 30 GHz planar array that was 1m^2 you would not encounter the far field at 10*lambda=10cm. Make sense? 10m may be too close. 200m as given by the last equation is good, but do you have a 200m chamber? This is why near-field measurements are made on arrays.
To summarize, the old rule of thumb (Rff=>10*lambda) is no good for large arrays, hence the last equation.