4.2. Recombination lines in nebulae
More precise estimates of helium abundance come from observations of emission lines in gaseous nebulae (planetary nebulae and H II regions), where both hydrogen and helium lines are formed predominantly by recombination of H+ and He+ for which a precise theory exists (Brocklehurst 1972) and adequate signal: noise can be obtained for 4472 (43D - 23Po), 5876 (33D - 23Po) and 6678 (33D - 23Po) using panoramic linear detectors (photon counters or CCD's) on nebulae with sufficiently high surface brightness. For Galactic H II regions, radio recombination lines can also be used (Thum, Mezger & Penkonin 1980; Thum 1981; Peimbert et al. 1988). However, Galactic H II regions have rather large heavy-element abundances, which makes extrapolation to pregalactic values uncertain, and planetary nebulae are additionally affected by the internal evolution of their central stars. Thus the most favourable objects for the estimation of Yp are extragalactic H II regions in dwarf galaxies (or the outer parts of spirals) where the heavy-element abundances are low. Bright examples of this class are seen in a few spirals (e.g. M101), in nearby irregular galaxies like the Magellanic Clouds, in a subset of blue compact galaxies discovered on direct photographs by Zwicky and Haro, and in H II galaxies, dominated by emission lines, and mostly discovered in objective prism surveys in Armenia (Markarian) and Chile (Palomar, Michigan and Cerro Tololo surveys).
Peimbert & Torres-Peimbert (1974, 1976) noticed a small but significant trend for helium abundance to increase with heavy-element abundance in the order I Zw 18, SMC and II Zw 40, LMC, Orion Nebula, and accordingly proposed that Yp could be found by looking at H II regions with different heavy-element abundances, plotting a linear regression of the form
and extrapolating to Z = 0.(Z 25(O/H).) They carried out this programme (Lequeux et al. 1979) and derived Yp = 0.23 and dY / dZ = 3, the latter quantity turning out to be rather large compared to expectations from the theory of stellar evolution (Maeder 1984; Serrano 1986). A later survey by Kunth & Sargent (1983) showed no clear evidence for a dY / dZ slope, but this is largely due to the high weight given by them to the 5876 line in II Zw 40, an object heavily reddened by dust in the Milky Way, which is probably affected by absorption due to Galactic NaI (French 1980). The remainder of Kunth & Sargent's data actually show a steep, if ill-defined, slope (cf. Peimbert 1985). Various other data, mostly of inferior quality, quoted, for example, by Boesgaard & Steigman (cf. Pagel 1989b), led to some doubt as to whether a dY / dZ correlation actually exists, but since then the situation has been somewhat restored as a result of more careful work by Peimbert, Shields, Terlevich, Pagel and associates (see Pagel 1991).