A SPECTROSCOPIC SEARCH FOR COLLIDING STELLAR WINDS IN O-TYPE CLOSE BINARY-SYSTEMS .3. 29-UW CANIS MAJORIS

We have investigated the orbital-phase variations in the optical emiss ion lines and UV P Cygni lines of the massive O-type binary 29 UW Cani s Majoris to search for evidence of colliding winds. We present high s ignal-to-noise ratio spectra of the Halpha and He I lambda6678 emissio n lines and give radial velocity curves for several features associate d with the photosphere of the more luminous primary star. The Halpha f eature consists of a P Cygni component that shares the motion of the p rimary, and which probably originates at the base of its wind, and a b road, stationary emission component. We propose that the broad emissio n forms in a plane midway between the stars where the winds collide. U sing a simple geometric model, we show that this placement of the broa d component can explain (1) the lack of orbital velocity shifts, (2) t he near-constancy of the emission strength throughout the orbit, (3) t he large velocities associated with the Halpha wings, and (4) the cons tancy of the velocity range observed. On the other hand, the phase var iations of the UV P Cygni lines (Si IV lambda1400, N V lambda1240, C I V lambda1550) appear to be more influenced by the primary's wind. The red peaks follow the motion of the primary, but with a smaller semiamp litude due to a focusing of the primary's wind toward the secondary an d/or line blending with emission from the secondary's wind. The Si IV lambda1400 red peak has maxima at the conjunctions which may result fr om the presence of the bow shock and an enhanced gas density between t he stars. The blue edge of the P Cygni absorption troughs displays orb ital-phase variations in velocity which we suggest reflect differences in turbulence in the wind in directions toward and away from the seco ndary. However, the shortward limit of the zero-intensity core of the line C IV lambda1550 (which we take as representative of the wind term inal velocity) shows no orbital variations. Because of the dominance o f the primary's wind, a bow shock wrapped around the secondary might b e expected, but we show that a planar bow shock between the stars is n ot inconsistent with the UV observations.