Basic stellar structure dictates that stars of ∼ 100 M⊙ or more will be close to the Eddington limit, with luminosities in excess of 106 L⊙, and radiation pressure contributing prominently to the support against gravity. Although it is formally possible to generate static structure models of even more massive stars, recent studies of dense clusters show there is a sharp cutoff at masses above ∼ 150 M⊙. This talk examines the role of extreme mass loss is limiting the masses of stars, emphasizing in particular that continuum driving, possibly associated with structural instabilities of radiation dominated envelope, can lead to much stronger mass loss than is possible by the usual line-scattering mechanism of steady stellar winds.

However, population studies of very young, dense stellar clusters now suggest quite strongly that there is a sharp cutoff at masses above ca. 150 M⊙ (see, e.g., the talk by Sally Oey, in this JD 05, p. 206). This is sometimes attributed to a mass limit on star formation by accretion processes, though there are competing formation scenarios by binary or cluster merging that would seem likely to lead to formation of even higher mass stars (see talks in JD14 and S237).

So given the above rough coincidence of the observational upper mass limit with the Eddington-limit domain of radiation-pressure dominance, it seems associated instabilities in stellar structure might actually be a more important factor in this upper mass limit, leading to extreme mass loss in LBV and/or giant eruption events, much as inferred from circumstellar nebulae observed around high mass stars like eta Carinae and the Pistol star.