Lithium is observed in the solar system, the atmospheres of a wide variety of stars and in the ISM. Arnould & Forestini  review light nuclei abundances in a variety of stars and related stellar and interstellar processes, while halo stars are reviewed by , , , and .
Old halo stars which formed from gas which had low iron abundances show approximately constant 7Li / H 1.6 x 10-10 and little variation with iron abundance or surface temperature from 5600 - 6300 K. The lack of variation amongst these ``Spite plateau''  (references in ) shows that their 7Li is close to primordial.
Since the halo stars formed about ten times more 7Li has been produced in the inner Galaxy. Abundances of 7Li /H 10-9 are common, although some stars show more, presumably because they make 7Li . Stars typically destroy 7Li when they evolve, accounting for the low abundances, < 10-11, in evolved stars. Stars with deeper convection zones, such as halo stars with lower surface temperatures, show less 7Li , because they have burnt it in their interiors.
Here, and in the next section on 6Li , we will the following topics:
The recent homogeneous data on 22 halo stars with a narrow range of temperature on the ``Spite plateau'' have very small random errors and show that most (not all) stars with similar surface properties have the same 7Li / H . Earlier data showed more scatter, which some considered real (references in ), and hence evidence of depletion.
The Ryan, Norris, & Beers  sample shows a clear increase of 7Li with iron abundance, as had been found earlier. This trend appears to be real, because the data and stellar atmosphere models used to derive the 7Li abundance do not depend on metallicity. But it was not found by Bonifacio & Molaro , perhaps because of larger scatter in temperatures and iron abundance. This trend is not understood, and there are several possible explanations. It may have been established in the gas from which the stars formed, perhaps from cosmic rays in the ISM, or from AGB stars. Alternatively, we speculate that it might instead relate to depletion of the 7Li in the stars. In either case, the BBN 7Li will be different from that observed: smaller if the 7Li was created prior to the star formation, and higher if the trend is connected to destruction in the stars. More on this below.
Creation of 7Li in the ISM by cosmic ray spallation prior to the formation of the halo stars is limited to 10 - 20% because Be would also be enhanced by this process , .
A clear summary of arguments for and against significant depletion is given by Cayrel . There are two main reasons why depletion is believed to be small: the negligible dispersion in 7Li for most halo stars on the plateau, and the presence of 6Li . The main arguments for depletion are that it is expected, it clearly occurs in some stars, some halo stars on the plateau show differing abundances, and star in the globular cluster M92 which should have similar ages, composition and structure, show a factor of two range in 7Li .
Different depletion mechanisms include mixing induced by rotation or gravity waves, mass loss in stellar winds and gravitational settling. Some models predict either variation from star to star, or trends with temperature, which are not seen for the stars on the plateau. For example, the rotationally induced mixing model implies that stars with different angular momentum histories will today show different 7Li . Ryan, Norris & Beers  find that the small scatter in their data, especially after the removal of the correlation with the iron abundance, limits the mean depletion in these models to < 30%, much less than the factor of two needed to make 7Li agree exactly with the predicted abundance from low D/H.
Some stars which should lie on the plateau have very low 7Li , while others show a range of abundances (see ref. in ). Differences are also seen between halo field stars  and stars in the globular cluster M92 , , which show a factor of two spread in 7Li . These observations are not understood.
Gravitational settling (diffusion) of heavier elements reduces the 7Li in the atmospheres of stars. However, the depletion should be most in the hottest (highest mass) stars, which is not seen, and not understood. Vauclair & Charbonnel  proposed that small stellar winds might be balancing the settling. Vauclair & Charbonnel  noted that the peak abundances inside the stars are independent of both mass and iron abundance. Normal stellar models predict that these peak abundances will not be seen in the stellar atmospheres, because convection does not reach this far down into the stars. However they point out that if some mechanism does mix gas from the 7Li peak zone into the bottom of the convection zone then the stars on the plateau would have similar abundances as observed. Assuming that the observed abundances are those from the peaks inside the stars, they find that the initial abundance in the stars was 7Li /H = 2.2 ± 0.6 x 10-10, without free parameters, which is still below but statistically consistent with the prediction from low D of 3.5+1.1-0.9 x 10-10 .