Individual extinction coefficients determined from a nights repeated observations of several primary standards passing through a wide airmass interval were used for the reduction of nearly all the data; well established mean values were used for the remaining 3-4 nights. The coefficients are listed in Table 1 (click here) and may well be useful for other projects carried out at La Silla during the three observing periods; see also Burki et al. (1995) and references therein. When needed correction for drift through a night from temperature changes (uncooled photomultipliers) or gradually changing transparency was applied.
For the transformation to the standard system we have adopted the three uvby
regions defined by Olsen (1983) and the two 
regions defined
by Grønbech & Olsen (1977). Mean scale and
colour coefficients, derived for each of the three observing periods,
were used in combination with individual nightly zero points. The
scale and colour coefficients are listed in Table 2 (click here), adopting the
notation by Grønbech et al. (1976) and
Grønbech & Olsen (1977).
The alternative approach of using individual nightly
transformations, i.e. both zero points and scale and colour
coefficients, was tested. As expected this approach leads to nearly
identical standard indices, since a complete set of primary standards
was observed on (almost) every night.
The uvby transformations may be compared with those determined by Olsen (1993, Table 7) for observations 1985-1987 with the same instrumentation as used here. In general the scale and colour terms agree well. Slight differences are noticed for the 1995 BAF transformation. This is probably due to the fact that only standards with b-y > 0.25 were used during this period; we will return to this matter later. Jønch-Sørensen (1993) also used SAT but arrived at uvby transformations somewhat different from ours.
The scale coefficients of the 
transformations agree well
with those obtained for SAT 1991-92 by Jønch-Sørensen
(1993) except for the 1993 coefficient for the late group (A) which
appears slightly low. Gray & Olsen (1991) determined coefficients of
1.226 (B) and 1.298 (A) from observations with SAT 1987-88. Olsen
(private communication) reports 1.316 (A) from 1988-89 data.
For each observing period the mean of the transformed indices for
the primary standard stars have been compared with their standard
values, and residual plots generally show that the
transformations listed in Table 2 (click here) do in fact bring the data on
the standard system. A few comments on the b-y transformations
are, however, relevant. Some indication of small systematic
trends is noticed in b-y for the 1993 and 1994 BAF
transformations, which cover the interval -0.08 to 0.40, but not
for the 1995 transformations where the blue end below 0.25 was
not included. Briefly the transformed b-y for stars below -0.05
are in average about 0.005 larger than their standard values, and
between about 0.10 and 0.20 they are about 0.005 smaller. A
second order correction of 
would lead to a more random distribution of the
residuals through the b-y interval, but the improvement is
marginal. We have noticed a similar effect for
the BAF group in the much larger observing program carried out at
SAT by Olsen (1993, Table 8), who included standard stars
from this region although his catalogue contains only G5-type
type HD stars outside the b-y range discussed here.
The reason for the small systematic trends is not clear, and in order not to introduce a correction which might be artificial, we have decided not to adjust the two b-y transformations. In case a future detailed study confirms that a second order term is needed in b-y for transformation from the SAT instrumental system to the standard system our results can easily be adjusted through the approximation given above.
Mean values of the transformed 
indices for the primary
standards observed 1993-1995 are listed in Tables 3 (click here) and
4 (click here), and
compared with the standard indices in Figs. 1 (click here) and 2 (click here).
Table 4: Catalogue of transformed 
indices for the observed
primary standard stars. 
is the transformation region,
N is the number of observations; m.e. is the internal rms errors of one observation.
The last colum gives the differences d = standard value - transformed value in units
of 0.001 mag