It is proposed that the kinetic energy of a pigeon's wing, in hovering or slow forward flight, is transferred to the air at the end of the downstroke by a mechanism involving temporary storage of additional energy in bent primary feather shafts. Estimates of the amounts of energy which can be stored and recovered in this way are compared with the requirements of the theory. The hypothesis is not rejected, as far as present evidence goes. If is is correct, high-velocity pulses of calculable magnitude should be detectable in the wake.