4.1. From Observed to Primordial Abundances
Even when heavy element abundances are low, it is difficult to prove that abundances are primordial. Arguments include the following.
Helium is observed in the ionized gas surrounding luminous young stars (H II regions), where O abundances are 0.02 to 0.2 times those in the sun. The 4He mass fraction Y in different galaxies is plotted as a function of the abundance of O or N. The small change in Y with O or N is the clearest evidence that the Y is almost entirely primordial (e.g. [7] Fig 2). Regression gives the predicted Yp for zero O or N [44]. The extrapolation is a small extension beyond the observed range, and the deduced primordial Yp is within the range of Y values for individual H II regions. The extrapolation should be robust [45], but some algorithms are sensitive to the few galaxies with the lowest metal abundances, which is dangerous because at least one of these values was underestimated by Olive, Skillman & Steigman [46].
For deuterium we use a similar argument. The observations are made in gas with two distinct metal abundances. The quasar absorbers have from 0.01 to 0.001 of the solar C/H, while the ISM and pre-solar observations are near solar. Since D/H towards quasars is twice that in the ISM, 50% of the D is destroyed when abundances rise to near the solar level, and less than 1% of D is expected to be destroyed in the quasar absorbers, much less than the random errors in individual measurements of D/H. Since there are no other known processes which destroy or make significant D (e.g. [4], [47]), we should be observing primordial D/H in the quasar absorbers.
Lithium is more problematic. Stars with a variety of low heavy element abundances (0.03 - 0.0003 of solar) show very similar abundances of 7Li ([48] Fig 3), which should be close to the primordial value. Some use the observed values in these ``Spite plateau'' stars as the BBN abundance, because of the small scatter and lack of variation with the abundances of other elements, but three factors should be considered. First, the detection of 6Li in two of these stars suggests that both 6Li and some 7Li was been created prior to the formation of these stars. Second, the possible increase in the abundance of 7Li with the iron abundance also indicates that the 7Li of the plateau stars is not primordial. If both the iron and the enhancement in the 7Li have the same origin we could extrapolate back to zero metals [49], as for 4He , but the enhanced 7Li may come from cosmic ray interactions in the ISM, which makes extrapolation less reliable. Third, the amount of depletion is hard to estimate. Rotationally induced mixing has a small effect because there is little scatter on the Spite plateau, but other mechanisms may have depleted 7Li . In particular, gravitational settling should have occurred, and left less 7Li in the hotter plateau stars, but this is not seen, and we do not know why. More on this later.
The primordial abundance of 3He is the hardest to estimate, because stars are expected to both make and destroy this isotope, and there are no measurements in gas with abundances well below the solar value.