4 Comparisons with other VBLUW observations

  

Table 2: Calibration of HD 151515 by Van Genderen et al. (1984) and Van Houten (unpublished) and the corrections applied to the photometry of Van Genderen et al. (1984)

  

Table 3: Comparison of the average differences between the present material and four other surveys; in the sense this survey minus others. The internal accuracy of each entry is about 0.001

To judge the accuracy of the photometry given here it is important to compare the present values with lists of observations published earlier, and which have stars in common with our Table 4. Four of such lists were compared:
1. Walraven & Walraven (1977) (denoted here as WW77) published a list of VBLUW measurements on southern stars. They are not all OB-stars. This list has 80 stars in common with the present material. The average differences, defined as: Table 4 minus WW77, are given in Table 3. The V-B values of 1959 were corrected by means of Eq. (3). The material in W77 was observed in (approximately) 1961-1962 and is thus closest in time to the material presented here.
2. Graham (1968) published the results of an extensive investigation of OB-stars in Carina, observed in 1964. Average differences: Table 4 minus Graham (1968) are given in Table 3.
3. Pel (1976) and Lub & Pel (1977) have 23 stars in common with our material. Their material was collected in 1971 and 1972. The average differences are again shown in Table 3; the sign convention is as for the comparisons 1 and 2. For this comparison our V and V - B values were corrected with expression (2).
4. Van Genderen et al. (1984), observing the Sco OB1 association, have 51 stars in common with the present material. Comparison with the 1959 values at first showed rather large differences. Van Genderen et al. used the star HD 151515 as a local substandard for their photometry. This star was used as a standard star by the first author in his (unpublished) survey of southern OB-stars, extending from 1965 to 1978. The average of the calibrations of HD 151515 made by the first author in 1969, 1974 and 1977 is entered in Table 2, together with the values used by Van Genderen et al. (1984). Using the 1969-1977 calibration instead of the one used by Van Genderen et al. considerably reduced the difference with the 1959 material. It is the difference with the latter values which is listed in Table 3. Inspection of Table 3 shows that the comparison of our material with Walraven & Walraven (1977), Pel (1976) and Lub & Pel (1977) and Van Genderen et al. (1984), after applying the correction of these data, mentioned above, all show about the same pattern, whereas the comparison with Graham (1968) is different, especially in the ultraviolet. The differences can be interpreted such that in the 1959 material the U channel intensity was systematically 0.008 too bright. This is the same effect as found for the standard star HD 210934, although in the latter case the excess is larger. This suggests that the adoption of an incorrect value of the U channel intensity of one of the standard stars introduced a systematic effect in the colours of the program stars. If this is correct the systematic error can be expected to be dependent on the right ascension of the stars, since for stars with a right ascension very different from that of HD 210934, the latter can have played no part in their photometry. The comparison with WW77, which extends over all right ascensions, shows this indeed to be the case. For stars which have $2^{\rm h}<\alpha<10^{\rm h}$ no systematic effect is present, whereas the other stars have their 1959 brightness in the U channel about 0.010 too bright. Using the spectral classification schemes published by Walraven & Walraven (1960) it is seen that this effect will introduce errors in classification of one tenth of a spectral class, and of about half a magnitude in M_v. Obviously it would have been better to reduce the 1959 material anew, with improved values of the standard stars, but the material is so large that the amount of work involved is prohibitive, and if the systematic effect derived above is taken into account the data published here are thought to be still useful.

The values in Table 4 were not corrected by means of Eq. (2) or otherwise. In all comparisons of Table 3 the values of the primary standard stars from Table 1 are not included. The standard deviation of one difference in Table 3 is, on the average, $\Delta\log I = 0.011$. If it is supposed that the accuracy of all series is equal, then this accuracy amounts to $\Delta\log I = 0.008$, or 0.02 magnitudes.