ROTATION AND OXYGEN LINE STRENGTHS IN BLUE HORIZONTAL-BRANCH STARS

We have measured rotational velocities and oxygen abundances in 29 blu e horizontal-branch (BHB) stars in M13, 22 in M3, and 16 in NGC 288 fr om high-resolution spectra of the O I triplet at 7771-7775 Angstrom. H ere we outline the behavior of rotation and oxygen line strength with stellar effective temperature T-eff within each cluster and compare th e average values of each from cluster to cluster. The mean strength of the oxygen lines is greatest in NGC 288 and weakest in M3. Within eac h cluster, the oxygen abundances are nearly constant among stars of a given T-eff but may decrease with increasing T-eff. Among the BHB star s cooler than 10,000 K, there is no sign of the counterparts to the su per-oxygen-poor giants which dominate the bright end of the giant bran ch in M13. The stars hotter than T-eff = 11,500 K show no oxygen lines at all. While these faint blue stars have poorly exposed spectra and may be hot enough to ionize oxygen, the abruptness of the disappearanc e of the O I lines suggests that either the oxygen abundances are trul y low or diffusion sets in abruptly at this color. No star in any clus ter rotates more rapidly than 40 km s(-1). Thus [upsilon sin i] is muc h lower in BHBs than in the superficially similar Population I B8 V-A7 V main-sequence stars. In M13, there is no obvious dependence of upsi lon sin i on horizontal-branch (HE) color. The distribution of upsilon sin i values in that cluster can be matched only with a bimodal or ot herwise non-Gaussian distribution of upsilon values. Rotational veloci ties are largest in M13, where six stars have v sin i greater than or equal to 30 km s(-1), and are much smaller in M3 and NGC 288, with non e at this level. Since both M13 and NGC 288 have very blue HBs, this i ndicates that rotation and oxygen abundance are not enough by themselv es to determine HE color, either alone or together; another factor suc h as age must also be at work. Thus, the ''second-parameter'' problem involves more than one such parameter, whose relative influence varies among clusters.