The photoelectric observations were collected at two sites, with different telescopes and equipment: the ZEISS 60cm telescope of PDO-LNA-CNPq with a single-channel TEXAS photometer and the Danish 50cm telescope (SAT) at ESO with a 6-channel photometer. Although the uvby photometric system was used in all runs, it is natural to expect differences between the two sets of data, caused by differences in the equipment (e.g. filter transmission) and in the site properties (altitude and humidity).
During 35 nights (10 from Jun. 16 to Aug. 17, 1989, 9 from Jul. 30
to Aug. 15, 1990, 16 from May 23 to Jun. 18, 1991) V 3903Sgr was observed
at PDO (LNA-CNPq, Brasópolis, Brazil), with the 60 cm telescope and
a single-channel photometer equipped with a photon counting system. A
slightly elliptical diaphragm with 
major axis was used
(Cunha 1990). No other star was detected
inside this diaphragm with an image intensifier placed at the
eyepiece, despite the richness of the field close to V 3903Sgr.
Extinction corrections were based on the nightly coefficients from the
four comparison stars and other constant stars. When needed, linear or
quadratic corrections were applied for eventual
instrumental drifts, since the temperature of the cooled
photomultiplier was kept 
C below the (variable) ambient
temperature and/or for variations in the atmosphere transparency during
the night. The dead time of the RCA 8850 tube (spectral response 116,
RCA Photomultiplier Tubes Catalogue, 1971) was accounted for in the
reduction procedure.
HD 
, HD 
, HD 
(7 Sgr) and HD 
,
all close (maximum projected distance in the sky of 
) to V 3903 Sgr, were used as comparison stars and observed
alternately between the measurements of V 3903 Sgr.
All four stars
were found to be constant within the observational accuracy throughout
the observing periods. The observations of C2, C3
and C4 were transformed to C1 by
means of the constant difference of magnitude between them and
C1 obtained from all nights. For completeness we repeat
here Table 1 (click here) from
Vaz et al. (1997), with information for V 3903 Sgr and
the comparison stars.
The light curves V 3903 Sgr - HD in the instrumental system
u (478 points), v (532), b (544), y
(537), are shown in Fig. 1 (click here) and are accessible in
electronic form at the CDS as Table 2. Typical rms errors of one
magnitude difference between the comparison stars were found to
be: 
, 
, 
, 
. Most of the phases were covered at
least twice.
Figure 1: uvby magnitude differences V 3903 Sgr-HD165999
obtained at PDO, with the theoretical light curves (Vaz et al.
1997)
V 3903 Sgr was also observed, during 32 nights (3 from May 04 to
17, 1990, 3 from Apr. 12 to 25, 1991, Jun. 06 and 18, 1993 and 24 from
Jun. 06 to Aug. 09 1994), with the 50cm Strömgren Automatic Telescope
at ESO, La Silla, Chile. The
instrument was equipped with the six-channel spectrograph-photometer
(four channels for simultaneous uvby photometry and two channels for
simultaneous measurements of 
narrow and wide filters)
and the photon counting system described by Nielsen et al.
(1987). In the measurements taken from 1990 to 1993 a circular
diaphragm of 
diameter was used, but in 1994 observations
a 
diameter diaphragm was used.
The comparison stars used in the PDO observations, (Table 1 (click here)), were adopted, and they were constant within the observational accuracy throughout the observing period. Each night several variables were observed, and the nightly extinction corrections were based on coefficients determined from all the comparison stars. As happened in the reduction of PDO measurements, linear or quadratic corrections for instrumental drift and/or changes in sky transparency during the night were applied when needed. The dead times of the six EMI 9789QA uncooled photomultipliers, which have S11 spectral response, were accounted for in the reduction procedure.
Typical rms errors of one magnitude difference between the comparison
stars in the ESO measurements were found to be: 
, 
, 
, . Most of
the phases were covered at least twice. The light curves are shown in
Fig. 3 of Vaz et al. (1997) and the observations are available
in electronic form at the
CDS, as Tables 3 and 4. Table 3 contains 180 magnitude differences
V 3903 Sgr - HD in the instrumental system for 1990, 1991
and 1993, obtained with the diaphragm of diameter, and
Table 4 contains the 1994 observations (507 points,
instrumental system), which were obtained with a diaphragm of
diameter.