A preliminary examination of the synthetic line profiles showed that the value of microturbulence, , had little effect on the profiles of the hydrogen lines. This is to be expected since the thermal velocities of the hydrogen ions are higher than our typical microturbulent velocities and any small differences are masked by the large intrinsic Stark broadening. The ionised helium lines also showed little dependence on microturbulence. In this case, although the thermal velocities are comparable to , the lines are very weak and close to the linear part of the curve of growth. There is therefore only a small dependence which is again masked by the strong Stark profiles. An exception is the strongest He II line at 4686Å, which has a more strongly developed Doppler core (relative to the Stark wings) than the lines at 4200 and 4542Å. However, in the case of the stronger neutral helium lines, some very significant changes in both line profile and equivalent width were found. It might be expected therefore, that the parameters derived predominantly from H I and He II lines (namely and ) would be insensitive to the adopted microturbulence, while the derived helium fraction would depend on this parameter.
In order to investigate the consequences of these changes in the line profiles, we have estimated atmospheric parameters (, and y) for Ori (B0 Ia) and Ori (B0.5 Ia), initially adopting a zero microturbulent velocity, and then relaxing this constraint by using a value estimated from the metal lines. This procedure should be valid if the metal lines form at depths comparable to the helium lines and we discuss this point further below.
All synthetic profiles were convolved with both a Gaussian filter to account for instrumental broadening, and a rotational velocity broadening function (80 & 90 kms-1 for Ori & Ori respectively - as determined from their metal line profiles). Our analysis is similar to that described in Lennon et al. (1991), whereby a simultaneous estimation of , and y is sought.
4.1. Hydrogen and helium lines considered
The He II lines at 4200, 4542 and 4686Å were observed in both stars. However, the usefulness of the line at 4686Å may be compromised by wind emission (see Gabler et al. 1989, Lennon et al. 1991, Herrero et al. 1992) and hence it was not used as a temperature indicator. The line at 4199.83Å is also affected by absorption due to N III at 4200.02Å (as noted by Smith & Howarth 1994 and as shown in Fig. 2), while that at 4542Å appears to be unaffected by such problems and hence was given highest weighting in our fitting procedure.
Our spectra include the Balmer hydrogen lines H , H , H , H , H6 and H7. H is often affected by core emission from the stellar wind and this effect was observed in our spectra, while H was also excluded due to the presence of an interstellar line of calcium in its blue wing. The higher lines in the Balmer series were rejected as they are close to the blue edge of our spectra and had lower signal to noise ratios. Therefore we have used H and H as gravity indicators, in all cases giving greater priority to H , which is less affected by blending with metal lines.
Neutral helium lines were available at 4026, 4387, 4437, 4471, 4713, 4922, 5015 and 5047Å and we have tried to fit this set in its entirety. (4009 & 4121Å were also observed. The former was excluded as our line profile computations did not include it while the latter was not used due to blending problems associated with lines of O II.)
4.2. The value of microturbulence
The microturbulent velocity was estimated from the relative strengths of metal lines with the species O II, N II and Si III being considered. Having determined preliminary atmospheric parameters from the hydrogen and helium spectra, the microturbulence, , was freely varied in order to ensure that the derived abundances were not a function of strength (using a method very similar to that described in Smartt et al. 1997).
The close proximity of many of the O II lines led to blending problems and poorly resolved features were not considered. For N II lines, only the lines at 3995, 4236 & 4242Å were resolved, and hence this species was not used to estimate the microturbulence. By contrast, the Si III multiplet near 4560Å had three well-resolved lines covering a substantial range in line strength, with good quality line-strength measurements.
The abundance - equivalent width plots for Ori are presented in Fig. 1. A considerable scatter is found in the abundance estimated from lines due to O II. Also we note that for this ion, any single multiplet has a small range in equivalent width and hence the value of is not well constrained. Therefore, whilst the lines of oxygen taken as a whole indicate a large microturbulence (of greater than 20 kms-1), errors inherent in our plot (that may be due to atomic data errors or inadequate allowance for non-LTE effects) prevent our making any definite conclusions. The Si III lines, which come from a single multiplet, imply a microturbulent velocity of = 12 kms-1 for both and Ori (with an error estimate of 3 kms-1 probably being appropriate) and this value is adopted in the subsequent H and He line calculations. Comparing the depth of formation of the silicon lines (see Fig. 5) we see that they form at depths comparable to the He I lines at 4387, 4713, 5047 and 4437Å. Thus our procedure should be adequate for these lines, but may possibly underestimate for those lines forming further out in the wind, if microturbulence does in fact mimic the effects of the wind.
© European Southern Observatory (ESO) 1998
Online publication: December 8, 1997