4 Comparison of the new data to other data sets

Good agreement between photometric data coming from different observers is an essential issue when studying the structure and characteristics of star-forming regions, both on the basis of photometric diagrams or via the photometry-derived physical stellar parameters. Observing stars from previous lists helps to get more clear insight on the homogeneity of the existing photometric data sets. This allows one to evaluate the errors in the derived stellar parameters when using photometry from different sources. Furthermore, it will eventually permit to homogenize to some extent the existing $uvby\beta$ photometric data for the field under investigation. In this paragraph we compare our data to other sources. Tables 3 and 4 present this comparison for the H$\beta$ photometry and for the Vuvby photometry, respectively. The tables give the mean differences between the individual sources and the new data, together with the corresponding standard deviations (from the mean difference) in $\beta$, V, b-y, m₁ and c₁ for each source. The first columns show the number of stars in common involved in the comparison. The sources of the corresponding photometric data sets are listed in the last columns.

The comparison of the new data to the homogenized data from the Hauck & Mermilliod (1998) compilation in the sense (HM minus this paper) is given in Fig. 1. Figures 2 to 5 present the comparison of the data obtained in this paper to the individual sources used in the HM compilation for the considered field, in the sense (others minus this paper). The agreement between H$\beta$ photometry from HM and the present paper is in general good (Fig. 1). The mean differences Delta $\beta$ = ${\rm H}\beta - {\rm H}\beta$(SAT) listed in Table 3 are below 0.010 mag for the majority of the sources and the comparison to the individual sources present small systematic trends of $\pm 0.03$mag only in two cases. For some sources the V magnitudes show a systematic shift in the zero point of as much as 0.03 mag (Fig. 2). This shift is more significant towards the brightest stars (V<7 mag) and does not obviously depend of the colour term (b-y) (Fig. 3). The largest disagreements are with Hill & Perry (1969) and Eggen (1983): 0.032 and 0.019 mag, respectively. The agreement in (b-y) is within the photometric errors, although for some sources a systematic difference of 0.01 mag may exist (Fig. 4). The m₁ and c₁colour differences obtained in the present paper show a good agreement with Crawford et al. (1971), Schneider & Weiss (1988) and Kaltcheva & Georgiev (1994) (Fig. 5). There is a discrepancy of up to -0.02 in m₁ and up to 0.02 mag in c₁ with the other sources. Dealing with O and B luminous stars, one should keep in mind the variability in light among them (Abt 1957), which also contributes to the scatter between the different sets. Spectral anomalies can also cause large mean errors in magnitude or colour indices (Young 1974; Manfroid 1985; Franco 1994).

The comparisons in Tables 3 and 4 suggest differences in the precision and accuracy of the different data sets. The consistency of different uvby data sets has been already discussed by Manfroid & Sterken (1987), who stressed the need for a more strictly defined standard system and for closely matching instrumental system. The indication of a small systematic difference in m₁ and c₁ is slightly worrying. A similar difference in c₁ in a large data set has been previously noticed by Franco (1994). Recently, Crawford (1999a) also presented a comparison between uvby data from different publications, showing differences in m₁ and c₁ up to $\pm0.05$mag. He also pointed out the reasons that can lead to systematic errors (Crawford 1999b). An additional source may be not applying a correction of -0.008 in c₁ to a large set of the southern standards (cf. Olsen 1983). In the case of photometry in star-forming regions, the discrepancy may also be due to the difficulties in selecting a suitable set of primary and secondary uvby standards among luminous and reddened OB stars (cf. Paper I).

The systematic differences between the existing data sets lead to systematic differences in the photometry-derived stellar parameters. The ideal case to compare data only internally is often not possible. In this event the possible inconsistence should be carefully estimated. If the photometric diagrams including c₁ and m₁ are used, the differences in the photometry, mentioned above, can be easily misinterpreted in terms of luminosity or metallicity. In case of calculating the reddening, a systematic differences of 0.01 in b-y and 0.02 in c₁ lead to a difference of 0.01 in E(b-y). In calculating the absolute magnitude M_V, which is a function of $\beta$ and c₀, the $\beta$ indices are the most critical parameter. A systematic difference of 0.03 mag in $\beta$, leads to a systematic differences in the corresponding M_V values of 0.3 mag. A systematic difference from the standard system of +0.02 $\pm$ 0.02 mag in c₁ alone leads to overestimation of M_V of 0.15 mag and to overestimation of the distance of about 5%.

The structure of the Carina Spiral Feature field based on photometry presented here will be discussed in a forthcoming paper.

Acknowledgements

This research has been supported by the Danish Natural Science Research Council. This research has made use of the Simbad database, operated at CDS, Strasbourg, France.