Originally Posted by
ijsbrand
Also, air resistance only begin to play a role at speeds > 32 km/h [20 m/h]
Not strictly true. It plays an increasingly larger role at higher speeds, but on a flat road without head wind energy expenditure for overcoming air resistance dominates over other sources of friction at any speed you're likely to find a randonneur riding under such conditions (i.e. even at 20 km/h).
Play around with a cycling power calculator such as
this one, it gives some interesting data. If we assume air resistance to be negligible at 5 km/h, without any air resistance your power output at 20, 25 and 30 km/h should 4, 5 and 6 times greater. In fact it is 9.2, 16.5 and 24.4 times greater. This suggests that air resistance accounts for roughly 57%, 70% and 75% of power output at 20, 25 and 30 km/h respectively (12.5, 15.5 and 18.6 mls/h), if my math is correct (riding on the hoods, box rim 32 spoke wheels, training tyre, 10 kg bike weight).