R40 = HD6884 was discovered as an S Dor variable or LBV by Szeifert et al. (1993). The spectral type at minimum light around 1960 was B8Ie (Feast et al. 1960). Table 2 (click here) lists the global characteristics of the light and colour curves.
Figure 7: The schematic light and colour curves of R40, 1985-1994.
The insert shows the schematic light curve, 1957-1994
Figure 7 (click here) shows the light and colour curves of R40 covering the
time interval 1985-1994. The colour curves b-y, v-b and u-b
(note: not u-v) are relative to the two comparison stars A and
B in the sense variable minus 
. The error in brightness
and colours is of the order of 
.
The insert shows the schematic light curve 
for the time
interval 1957-1994. The dots are the scattered observations
collected from the literature by Szeifert et al. (1993) and
the smooth curve represents the underlying light curve from the large
diagram. The data sets separated from each other by gaps in time
are numbered 1-10. Brightness and colours have the same magnitude
scales.
Smooth curves have been sketched through the forest of
schematically indicated micro-variations ( amplitudes
; time scale 1-3 months). Visual observations by A. Jones
(for a description, see Sterken et al. 1996b)
during the same time interval have
a scatter of 
and globally show the same trend as the
curve. According to Jones' observations which continue until the
end of 1995 (ours run only until the end of 1994), the brightness still
rose another 01. There is excellent agreement between the of
the LTPV project made with filter systems 6 and 7 and the
transformed from the Walraven V. However, there is a systematic
difference between the latter and the observations made with filter
system 8 of the LTPV project. This follows from observations made
simultaneously on 13 nights. The correction applied to the filter 8
magnitudes equals 
.
The behaviour of brightness and colours from 1985 to 1994 is again
typical for an SD phase: the colours become redder, especially in the
u-b colour index which reddens by 
, while the rise in the
visual was only 06. Thus, it appears that by the time the
curve has reached the top of the ascending branch, the u magnitude
has dropped nearly to its minimum value. If this behaviour is typical
for LBVs, as shown by S Dor, HR Car and R127, it suggests that at the
end of our ascending branch, R40 is close to its visual maximum.
Figure 7 (click here) and its insert show that R40 underwent a low amplitude
maximum of 
between 1957 and 1982. It is not clear whether
this should be classified as a VLT- or a normal SD phase,
although the duration is 25y, which is the typical time
scale for VLT-SD phases of several other LBVs. Then, after a
shallow dip, the star brightens very steeply by 06 to 
within 7y. The only magnitude known prior to 1957 is the
photographic magnitude in the HDcatalogue: 107, based on plates
taken at the end of the 
century. Thus, knowing that in
minimum R40 has a B type spectrum and that the interstellar
extinction is relatively low (Szeifert et al. 1993), the
visual magnitude should then also have been about 107 at the end of the
19th century.
Since the observations of R40 are very numerous, we performed
a period search (with Sterken's (1977) algorithm) of the data for the data sets 1-4, 5-7 and 8-10 separately, corrected for
the increasing brightness of the SD phase with the aid of the curve
sketched in Fig. 7 (click here). Since the micro-variations are superimposed on
a progressively increasing mean brightness, it is of interest to
study a possible change of the period, i.e., whether R40 shows
a similar switch between the two types of micro-variations like
HR Car and R127. Further, one would expect that, if the brightness
rise were solely due to the expansion of the star, the pulsation
period changes smoothly as a result of the decrease of the mean
density.
The search made between 35d and 150d yielded as best periods: 462, 935 and 983 for the three data sets, respectively.
Thus, the change is relatively small, only a factor two, and obviously
not smooth at all. The period change is rather abrupt and appears to
happen within a time gap of 
. As in the case of HR Car and
R127 (Sect. 4.2) it is difficult to see how this change could be due to
a change in stellar radius only.
Figure 8: The phase diagram for the light and colour curves of the
micro-variations of R40 in data set 1-4 for a period of 462. The
curve also contains the results of the Walraven system; the
colours are from the Strömgren system only
Separate phase diagrams were constructed for the three
data sets. Figures 8 (click here), 9 (click here) and 10 (click here) show them for
sets 1-4, 5-7 and 8-10 with periods of 462, 935 and 983,
respectively, for and the Strömgren colours b-y, v-b and
u-b relative to the two comparison stars.
Figure 9: The same as Fig. 8 (click here), but now for data set
5-7 and a period of 935
Some conclusions from Figs. 8 (click here)-10 (click here) are as follows:
,
is of intrinsic nature and is largest for . That is, there is
a large variation in the light-curve shape from cycle to cycle. Evidently, a
variety of time scales may be involved and more data might show complex
periodicities.
The scatter in the u-b curves is significantly larger than that in the
b-y and v-b curves, pointing to additional fluctuations in the Balmer
continuum and Balmer jump. We suspect that this may be caused by
non-thermal photospheric velocity fields (like in the LBV HD160529,
studied by Wolf et al. 1974) and/or by atmospheric turbulence
(van Genderen 1991). A large intrinsic scatter in colour
indices which include an ultraviolet pass band has also been noticed in the
behaviour of other 
variables observed in the Walraven
system (van Genderen 1991).
Figure 10: The same as Fig. 8 (click here), but now for data set
8-10 and a period of 983