2. Observations and data reduction
The H data have been obtained during two observing runs at the Danish 50 cm telescope located at ESO-La Silla (Chile): 1982, April 20 - May 2 (12 nights, observer NV) and 1982, July 11 - 31 (20 nights, observer WWW), using the two-channel H photometer described by Grönbech & Olsen (1977). A beam splitter in front of the field lens in the direct beam reflected 20% of the light via a spherical mirror through a second field lens. The two photomultipliers of type EMI 6256SA were operated at ambient temperature. The filters were placed immediately in front of the cathodes, the narrow filter in the direct beam and the wide filter in the reflected beam. Thus, the fluxes falling on the cathodes were approximately equal. The dead-time of the photoelectron counting system was 70 nanoseconds. In order to avoid large dead-time corrections we limited our observations to stars V 3 mag.
The standard stars were taken from Tables 3 and 7 in Grönbech & Olsen (1977) which are based on the system defined by Crawford & Mander (1966). In order to control possible sensitivity changes in the photomultipliers we measured about 5 - 10 standard stars at the beginning and at the end of each night, and 1 - 2 standards every 2 hours during the night. Some individual standard stars were observed 2 - 3 times during the same night at different air masses, but no significant effects on atmospheric extinction and/or telescope position were found.
As program stars we selected all Ap stars with V 8.5 mag in Table IV of Bidelman & McConnell (1973) within the right ascension range accessible during our observing runs (approx. 0h 30 min 4 h and 8 h 23 h). The main goal of our observations was to obtain precise values of each program star in at least two different observing nights. In order to reach an accuracy of 0.3% per measurement a total net count rate of about 300000 in each channel was necessary. For practical reasons, the integration time unit was fixed to 20 seconds for standard stars and to 100 seconds for program stars. The integration was repeated as often as necessary to reach the desired total of 300000 counts per channel.
The reduction procedure was rather simple and straightforward: first, we determined the mean instrumental value of each standard star, as an average from all observing nights within one run (April/May or July resp.). Subsequently, the individual deviations = - of all standard stars versus time were considered for determining drifts and zero point corrections of each observing night. Applying these corrections, all measurements (standard and program stars) were converted into a uniform instrumental system. Finally, the transformation into the standard system was performed by means of a simple linear least squares fit
based on the mean values of the standard stars. While the classical procedure (Crawford & Mander 1966) divides the stars into two groups (B-type and late type stars) yielding separate -transformations for them, our data could be reduced for both groups together, since no increase of the scatter around the regression line was found compared to the analysis of the individual groups.
The HD identifications, the mean values and the total number N of observations obtained are listed in Table 1. The resulting internal mean errors of the April/May run are rather small (0.003 mag) while those of the July run are larger (about 0.005 mag in average), probably due to worse weather condition which prevented, in some cases, to reach the above mentioned accuracy requirements. On the other hand, in the July run we obtained more individual observations (up to 6 per star). Therefore, the accuracy in the final mean values in each observing run should be comparable. Fig. 1 shows the error distribution of the observed H values. About 95 % of our program stars have a 0.02 mag.
Table 1. (continued)
© European Southern Observatory (ESO) 2000
Online publication: March 21, 2000