Observations indicate that a circumstellar disk is formed around a Be star while the stellar rotation is below the break-up velocity. I propose a working hypothesis to explain this mystery by taking account of the effect of leaky waves upon angular momentum transfer.
In B-type stars near the main sequence, low-frequency nonradial oscillations are excited by the κ-mechanism in the iron bump. They transport angular momentum from the driving zone to the surface. As a consequence, the angular momentum is gradually deposited near the stellar surface. This results in a gradual increase in the “critical frequency for g-modes”, and g-modes eventually start to leak outward, long before the surface rotation reaches the break-up velocity. This leads to a substantial amount of angular momentum loss from the star, and a circumstellar disk is formed. The oscillations themselves will be soon damped owing to kinetic energy loss. Then the envelope of the star spins down and angular momentum loss stops soon. The star returns to being quiet and remains calm until nonradial oscillations are newly built up by the κ-mechanism to sufficient amplitude and a new episode begins.
According to this view, the interval of episodic Be-star activity corresponds to the growth time of the oscillation, and it seems in good agreement with observations.