Modern abundance compilations and theories of stellar evolution, both owing much to the pioneering work of Russell, have led to a theory of the origin of the elements as a by-product of stellar energy-generation following the general lines described by B 2FH (1957). Other synthesis sites may be needed to account for the high abundance of helium. Evidence as to where and when the relevant reactions occurred, and were followed by the supply of the products to the interstellar medium through mass ejection, can be derived from studies of abundances in the oldest stars, especially those belonging to the “halo” population of the Galaxy. Such studies can be classified in three main categories: (i) general abundance of metals (chiefly Fe) deduced from low-resolution spectroscopy or photometry using a calibration from detailed spectral analyses; (ii) variations in the ratio M Fe of spectroscopically detectable elements to iron, deduced from high-dispersion spectra; (iii) helium abundances deduced from spectra or from comparison of masses, luminosities and effective temperatures with stellar models. Some results in the three categories are reviewed. Correlation of (i) with kinematic and evolutionary age data suggests that about half the metal content in a cylinder through the Sun, perpendicular to the galactic plane, is the result of an initial burst of star formation which was completed in a few times 10 8 years and which may have left a substratum of uniformly enriched interstellar material in the galactic disc. Additional enrichment may afterwards have occurred locally, e.g. through supernova activity in stellar associations, and this would account for the high metal content of certain stars and galactic clusters (with a wide range of age) in the disc. Detailed abundance studies needed for (ii) require careful attention to certain details in the technique of analysis by the differential curve-of-growth method, and a number of published spectroscopic data have been re-analysed to allow for this. Also some completely new analyses of halo stars have recently been carried out. The results show that many elements (Sc, Cr, Co, Ni, Zn, Sr, Y, Zr) are always present in nearly the same proportions (within an uncertainty of a factor of 2 or less) relative to iron, while others (C, V, Ba, Ce) show a marked over-deficiency only in stars having extremely low metal abundance, presumably because these are very old even by the standards of the halo population. The deficiencies of C, Ba and Ce form a “mirror image” of overabundances of the same elements found in two CH stars, so that possibly these elements were supplied to the medium by CH-star-like objects in a time scale of about 10 7 years, whereas the corresponding time scale for the lighter metals must have been still less. α-particle elements (Mg, Si, Ca, Ti) have been shown by Wallerstein to be slightly overabundant relative to iron in a group of dwarfs with moderate metal deficiency, but otherwise they behave like iron, whereas manganese is often over-deficient by a factor of about 2 when iron is deficient. This effect could be related to differences between the mass distributions of supernovae in the initial burst and in the disc associations. Overall, however, the heavy-element mixture in halo stars (from carbon onwards) is so similar to that found in disc stars that one would expect the same types of synthesis process to be responsible for both. Recent developments in cosmology make a cosmological origin for helium attractive, but the evidence from (iii) is rather confusing at present. Stars left over from the later stages of the halo phase seem to have about the same initial helium content as the Sun, but there is a deficiency in certain halo B-stars which may possibly require a galactic origin for helium. Stellar energy generation at the present rate is inadequate by a factor of about 30 to account for this, but it could have occurred under unknown conditions very early in the history of the Galaxy.