Wolf-Rayet stars as starting points or as endpoints of the evolution of massive stars?

The paper investigates the evidence for the two interpretations of Wolf-Rayet stars suggested in the literature: (1) massive premain-sequence stars with disks and (2) massive stars which have lost most of their H-rich layers in a stellar wind is investigated. The abundance determinations which are done in two different ways and which lead to different conclusions are discussed. The composition is solar, which would suggest interpretation (1), or the CNO abundances are strongly anomalous, which would suggest interpretation (2). Results from evolutionary calculations, stellar statistics, the existence of Ofpe/WN9 transition stars and W-R stars with evolved companions show overwhelming evidence that W-R stars are not premain-sequence stars but that they are in a late stage of evolution. Moreover, the fact that W-R stars are usually in clear regions of space, whereas massive premain-sequence stars are embedded in ultracompact H II regions also shows that W-R stars are not young premain-sequence stars.