Ooops,
The angle was assigned wrongly in the post "A Bloom Crosses Over" dated 10 Aug 2015.
It should have been,
β=90o−θ
and so,
tan(θ1)tan(θ2)=ε2ε1
as the cone spreads, β increases but θ decreases. And when,
ε1<ε2
θ1>θ2
polarization decreases.
This is then consistent with post "On Reflection Radar Theory 101" dated 16 Aug 2015, where the horizontal component of E attenuates and θ decreases.
Which proves common sense is rare.