We present the quantitative methods used for selecting candidate metal-poor stars in the Hamburg/ESO objective-prism survey (HES).The selection is based on the strength of the Ca ii K line, B − V colors (both measured directly from the digital HES spectra), as well as J − K colors from the 2 Micron All Sky Survey. The KP index for Ca ii K can be measured from the HES spectra with an accuracy of 1.0 Å, and a calibration of the HES B − V colors, using CCD photometry, yields a 1-σ uncertainty of 0.07 mag for stars in the color range 0.3 < B − V < 1.4. These accuracies make it possible to reliably reject stars with [Fe/H] > −2.0 without sacrificing completeness at the lowest metallicities. A test of the selection using 1121 stars of the HK survey of Beers, Preston, and Shectman present on HES plates suggests that the completeness at [Fe/H] < −3.5 is close to 100% and that, at the same time, the contamination of the candidate sample with false positives is low: 50% of all stars with [Fe/H] > −2.5 and 97% of all stars with [Fe/H] > −2.0 are rejected. The selection was applied to 379 HES fields, covering a nominal area of 8853 deg 2 of the southern high Galactic latitude sky. The candidate sample consists of 20 271 stars in the magnitude range 10 B 18. A comparison of the magnitude distribution with that of the HK survey shows that the magnitude limit of the HES sample is about 2 mag fainter. Taking the overlap of the sky areas covered by both surveys into account, it follows that the survey volume for metal-poor stars has been increased by the HES by about a factor of 10 with respect to the HK survey. We have already identified several very rare objects with the HES, including, e.g., the three most heavy-element deficient stars currently known.

We determine the metallicity distribution function (MDF) of the Galactic halo by means of a sample of 1638 metal-poor stars selected from the Hamburg/ESO objective-prism survey (HES). The sample was corrected for minor biases introduced by the strategy for spectroscopic follow-up observations of the metal-poor candidates, namely "best and brightest stars first". Comparison of the metallicities [Fe/H] of the stars determined from moderate-resolution (i.e., R ∼ 2000) follow-up spectra with results derived from abundance analyses based on high-resolution spectra (i.e., R > 20 000) shows that the [Fe/H] estimates used for the determination of the halo MDF are accurate to within 0.3 dex, once highly C-rich stars are eliminated. We determined the selection function of the HES, which must be taken into account for a proper comparison between the HES MDF with MDFs of other stellar populations or those predicted by models of Galactic chemical evolution. Although currently about ten stars at [Fe/H] < −3.6 are known, the evidence for the existence of a tail of the halo MDF extending to [Fe/H] ∼ −5.5 is weak from the sample considered in this paper, because it only includes two stars [Fe/H] < −3.6. Therefore, a comparison with theoretical models has to await larger statistically complete and unbiased samples. A comparison of the MDF of Galactic globular clusters and of dSph satellites to the Galaxy shows qualitative agreement with the halo MDF, derived from the HES, once the selection function of the latter is included. However, statistical tests show that the differences between these are still highly significant.

We determine the metallicity distribution function (MDF) of the Galactic halo based on metal-poor main-sequence turnoff-stars (MSTO) which were selected from the Hamburg/ESO objective-prism survey (HES) database. Corresponding follow-up moderateresolution observations (R 2000) of 682 stars (among which 617 were accepted program stars) were carried out with the 2.3 m telescope at the Siding Spring Observatory (SSO). Corrections for the survey volume covered by the sample stars were quantitatively estimated and applied to the observed MDF. The corrections are quite small, when compared with those for a previously studied sample of metal-poor giants. The corrected observational MDF of the turnoff sample was then compared with that of the giants, as well as with a number of theoretical predictions of Galactic chemical evolution, including the mass-loss modified simple model. Although the survey-volume corrected MDFs of the metal-poor turnoff and the halo giants notably differ in the region of [Fe/H] > −2.0, below [Fe/H] ∼ −2.0, (the region we scientifically focus on most) both MDFs show a sharp drop at [Fe/H] ∼ −3.6 and present rather similar distributions in the low-metallicity tail. Theoretical models can fit some parts of the observed MDF, but none is found to simultaneously reproduce the peak as well as the features in the metal-poor region with [Fe/H] between −2.0 to −3.6. Among the tested models only the GAMETE model, when normalized to the tail of the observed MDF below [Fe/H] ∼ −3.0, and with Z cr = 10 −3.4 Z , is able to predict the sharp drop at [Fe/H] ∼ −3.6.