Correlations of Lithium Abundance with T$_{eff}$ and [Fe/H] on the Spite Plateau
Session 22 -- Stellar Spectroscopy, Atmospheres, Models, Intrinsic Variables, Theory, Part II
Display presentation, Tuesday, June 13, 1995, 9:20am - 6:30pm

## [22.19] Correlations of Lithium Abundance with T$_{eff}$ and [Fe/H] on the Spite Plateau

T.C. Beers (Michigan State), S.G. Ryan (Anglo-Australian Obs), C.P. Deliyannis (Yale), J.A. Thorburn (Chicago)

We use a set of homogenized'' observational results for some 90 metal-deficient stars in the Galaxy to re-evaluate claims and counter-claims concerning the existence of correlations between the Li abundances for metal-poor stars, T$_{eff}$, and stellar metallicity, [Fe/H]. Molaro, Primas, \& Bonafacio (1995, preprint) argue that previously-reported slopes in the plateau as a function of these two variables (Thorburn 1994, ApJ 421, 318; Norris, Ryan, \& Stringfellow 1994, ApJ 423, 386) disappear when a subset of stars with temperatures based on Balmer line profiles is adopted. Upon closer examination, both of the Molaro et al. data and our own newly-expanded data, we find that these correlations persist, but several points are worth noting: (a) correlations are most apparent when subgiants are eliminated from the samples, (b) metallicity trends are evident only when stars of a wide range of metal abundances are included in the samples, especially the most metal- poor stars, (c) the tests must be performed in a multiple-regression environment (i.e., not when T$_{eff}$ or [Fe/H] is considered the only independent variable), and (d) the results survive when regression methods resistant to the presence of high-leverage outliers are applied. Our current best estimate of Li abundance as a function of T$_{eff}$ and [Fe/H] is:

$$A(Li) = -0.28 (\pm 0.30) + 0.043 (\pm 0.005) * {\rm T}_{eff}/10^2 + 0.09 (\pm 0.02) * {\rm [Fe/H]}$$

The slopes of this relationship are consistent, within expected errors, with the results of Thorburn (1994), who considered the existence of a multiple correlation of A(Li) with T$_{eff}$ and [Fe/H], and with the results of Norris et al. (1994) for an inhomogeneous sample which they divided into subsamples restricted by metallicity. The reported correlations appear to be real, in contradiction to the claim of Molaro et al. (1995).