OBSERVATIONAL CONSTRAINTS ON FU ORI WINDS

We have searched for velocity shifts in the photospheric lines of four FU Orionis objects that could show that their winds originate from the surfaces of pre-main sequence accretion disks. We confirm the pattern of photospheric line shifts in FU Ori first observed by Petrov and Herbig, in which increasingly strong photospheric lines are increasingly blueshifted as the two absorption components move closer together in wavelength. We show that this pattern is precisely that predicted by a disk wind model: the profiles of the weaker lines exhibit mostly Keplerian rotation, and thus should show nearly symmetric double-peaked absorption, while the stronger lines, formed farther out in the expanding wind, are more blueshifted and less doublepeaked. These observations provide at present the only direct evidence for a continuously accelerating wind originating from the surface of a pre-main-sequence disk. Similar velocity shifts are not observed in the photospheric lines of V1057 Cyg, V1515 Cyg, and Z CMa, even though all exhibit strong blueshifted absorption in the Na I and Hβ line profiles. We suggest that the Na I and Hβ lines are generally saturated and thus not sensitive measures of mass loss rates, and that V1057 Cyg and V1515 Cyg have smaller mass loss rates than FU Ori. Z CMa may have a mass loss rate similar to that of FU Ori, but photospheric velocity shifts may be less apparent because Z CMa has a higher projected rotational velocity. Our estimates of mass loss rates do not support the "extreme" wind model, in which all of the angular momentum needed for accretion is removed by the wind.