Spread-wing postures and the water repellency of feathers: a test of Rijke's hypothesis