We selected twelve Miras with bright minima found close to
the celestial south pole in order to be able to observe them over the whole
year.
For homogeneity reasons all the LPV in our sample are oxygen-rich Miras.
Since the spectral type of W Cha was not known before this work we first
observed it with a Boller & Chivens spectrograph at the ESO 1.52-m telescope
equipped with a CCD detector. A spectrum (shown
in Fig. 1 (click here)) was collected in the
Å wavelength region with a dispersion of 32 Å/mm. The
absence of any bands of the 
and CN molecules and the detection of
strong absorption bands of TiO confirm that W Cha actually is an
oxygen-rich star. Furthermore the strongest band of the ZrO 
system
(5551.7 Å) being absent, W Cha is not a S-type star and actually
a M-type one.
Figure 1: The spectrum of the oxygen-rich Mira W Cha
The selected Miras have a period in the range of
180-400 days which allowed us to observe them for at least three
consecutive cycles over a 4-year period.
We can guess that the properties of these stars are
rather different.
Indeed Feast (1963) showed that Miras with different periods exhibit distinctive
kinematic properties. This was confirmed by Foy et al. (1975).
A break in the properties of LPV (metallicities,
initial masses, etc.) is also suspected
around 
days (Little et al. 1987 and Jura & Kleinmann 1992).
On the other hand none of these Miras have been observed in
OH-maser lines except R Oct with no detection (Le Squeren 1996).
We have found that six stars in our sample have been classified
from their IRAS Low Resolution Spectrum. All of them have a period
larger than 300 days (confirming Feast 1986, who pointed out that Miras
with a smaller period were not detected by IRAS). They have been classified
as oxygen-rich sources with blue continuous energy distribution.
The infrared spectra of V Cha, RY Hyi, R Oct and RU Oct are quite similar
between 8 and 13 
m: they exhibit no emission or absorption features.
X Hyi is a star showing a 10
m silicate band
emission and R Cha a silicate absorption. These two stars thus differ
by the optical thickness of their oxygen-rich envelope, R Cha having
the thickest one.
Finally R Oct is a ``o2" star in the Valinhos classification
(Fouqué et al. 1992).
The main characteristics of the selected stars found in the literature
are listed in Table 1 (click here).
We give there the mean period of their light-curve in the visible, the
V magnitudes of the brightest maximum and the faintest minimum ever observed,
their spectral type at the extrema, the IRAS LRS spectral
classification and finally the number of observations we made.
The observations were collected at the European Southern Observatory
(La Silla, Chile) from September 1991 to January 1995.
We used the ESO 50-cm Cassegrain
telescope equiped with the single channel photometer and the thermoelectrically
cooled Hamamatsu R943-02 photomultiplier tube. This photomultiplier
was chosen because of its almost constant sensitivity over the whole spectral
domain studied (see for instance Fig. 4.2 of Schwarz & Melnick 1993).
We used standard 
ESO filters.
A neutral density was placed in front of the 
filter to avoid to
saturate the photomultiplier for the brightest stars.
Transformation coefficients were obtained by observing
Cousins's E and F-region photometric standard stars chosen from the list given
by Menzies et al. (1989).
All the observations were reduced with the ESO
photometric reduction program SNOPY.
Given the precision of the standards data and extinction and
transformation errors we have rejected each measurement with an error larger
than 0.1 mag. Finally the published data have a mean accuracy better
than 0.05 mag in the five colours.
Figure 2: The observed visible and colour light-curves of the twelve Miras.
The different symbols correspond
to the successive cycles n that we observed. First cycle (observations in
late 1991): filled circle (
); n = 2 (one period later): filled
star (
); n = 3: open circle (
); n = 4: asterisk (
);
n = 5: plus (+); n = 6: open rhombus (
) and n = 7:
open square (
)
The journal of the observations and all the 
data are compiled
in Table 2 (click here).
It was not possible to get a sampling of the light-curves
as good as desired. The distribution of allocated nights,
those lost due to bad weather conditions,
and first of all the technical problems with the ESO 50-cm telescope are
responsible for the periods without any measurements, some of them
as long as a few months.
We also report in Table 2 (click here) the derived spectral types of the Miras
in the observed phases of the light-curve.
The M-spectral subtypes are defined from the intensity of the
molecular absorptions of the TiO and VO molecules. Celis (1986a, 1986b
and references therein) showed that these subtypes can be obtained from the
(V-
) and (
) colour indices.
This is confirmed by model atmospheres of cool giant stars and particularly
the estimation of the effective temperature from the (
) colour index
(Plez et al. 1992). Furthermore
since our phase coverage is quite good the use of colours to derive
probable spectral types is worthwhile. On a practical point of view
Celis (1986b, Fig. 1) showed that the
distribution of the M Miras conforms to a sequence in a (V-
, 
)
colours diagram. He then derived two spectral types-colour index scales.
The derived spectral types reported in Table 2 (click here) are the mean of the
ones that have been calculated from these two relations.
They are consistent with the previous determinations reported in
Table 1 (click here) and discussed in Sect. 3.3 (click here).
Table 3: New extrema of the visible magnitudes
deduced from the observations and Table 1; and observed amplitude of the
spectral type variations
The obtained visible and colour light-curves are plotted in
Fig. 2 (click here).
A period search was performed for V Cha since it was not known. We
found 
days over the three observed cycles. The evaluation of the
period of the other stars is consistent with the previous determinations
reported in Table 1 (click here).
We have thus reconstructed the cycles by assuming that the periods
have not changed since 1985 and are
constant over the 4-year period of the observations.
The data belonging to different cycles
are plotted using different symbols in order to illustrate the
cycle-to-cycle variations.
The maximum of each cycle (
) corresponds to the
largest V magnitude we observed. Since the sampling of the light-curves
is actually not regular there might be some small discrepancies
between the maximum of the plotted light-curves and the real maximum of
luminosity of the stars that was perhaps not observed. Finally the
observations cover between three and seven successive cycles.
