The spectrophotometric data gathered for all program stars are presented in
Fig. 6 (available in electronic form). For each star there are three panels. The
first shows the observed time variability of the V, 
and D.
The second panel shows the (U,D), (V,D) and (
)
correlations and the third the (
) correlation. The symbols
used are: black dots for the BCD data, crosses for UBV data, stars for
UBVRI, triangles for Geneva photometry, diamonds for the uvby system and
squares for 13-colour data. The following comments for individual stars are
divided into two subsections according to the degree of known spectrophotometric
and spectroscopic variations.
This subsection deals with program Be stars, for which we have enough BCD
spectrophotometric observations to allow us: (a) to estimate the uncertainties
discussed in Sect. 2.6; (b) to assign not only a degree of reliability to
relations derived using only indirect (
) parameters, but
also to judge the likelihood of scatters and more or less systematic deviations
from the sketched mean relations. As for all these objects rather copious
spectroscopic observations exist, in this section we also summarize their main
spectroscopic variations over the last fifty years, more or less simultaneous
to photometric data which are analyzed in this paper. The qualitative
description of the correlated spectroscopic and spectrophotometric behaviours of
these objects can be used as a reference for the remaining program stars, for
which only spectrophotometric variations are presented. These stars are divided
into two groups: (1) Be stars most commonly seen in Be phases: 
Cas, X
Per, 11 Cam, 
Oph and 59 Cyg; (2) Be stars most frequently observed in
Be-shell phases: Pleione, 48 Lib and 88 Her.
Cas (HD 5394)
Observed since 1866, Cas firstly showed a long Be phase of strong
emission. Then, between 1932 and 1942, dramatic changes occurred in its
spectrum; two strong Be phases, each followed by a strong shell phase, have been
observed (Doazan et al. 1983 and references therein). In 1942, Cas
entered a quasi-normal B phase. Four years later, a subsequent Be phase
increased slowly and irregularly. Weak secondary minima in emission were
reported in 1956-1961 (Hubert-Delplace & Hubert 1979
hereafter referred to as
"Atlas''), end 1975, 1983-1984 (Horaguchi et al. 1994
and references therein) and
in 1990-1991 (Peters 1990, 1991); a "veiling'' effect of lines was seen from
the end of 1963 to 1972 (Atlas). According to Horaguchi et al. (1994),
H
intensity has oscillated with a time scale of several years.
Long-term far UV and visual variability seems to be associated
(Doazan et al. 1987).
It is interesting to note that the minima in the gradient
correspond to minima in emission strength and the maximum of in 1966 to a veiling in line spectra. Furthermore the total BD is strongly
negative at the epochs of strong outbursts (1932-1942). Somewhat scattered but
single (, D) and (, V) correlations with
slopes are present, although
photometric and spectrophotometric data correspond to epochs before and after a
number of Be 
B-normal Be-shell phase
transitions. We note, however, that the (V, D) relation has two branches in
dex. Let us finally note that the rapid photometric
variability which is commonly associated with surface stellar activity is still
debated for this object.
X Per (HD 24534)
Variable but generally weak emission was noted in the spectrum of X Per before
the end of the forties (Cowley et al. 1972). In 1951-1953, a strong emission
phase phase with a "veiling'' effect began; afterwards marked strengthening
(outburst) of emission lines (HI, HeI, FeII,
SiII) and of the "veiling'' effect from 1957 to 1961 has been reported,
with a maximum in 1961 (Cowley et al. 1972;
Wackerling 1972, Atlas). In 1974 the
emission decreased for 2 years, becoming bright again by the end of 1976.
In 1978, only very weak emission was seen in H, the spectrum was
consistent with a quasi-normal O9-B0 star (de Loore et al. 1979).
By the end of
1978, a new emission phase reappeared (Roche et al. 1993
and references
therein). Still fairly bright in February 1990, the H line was seen
converted to absorption in September 1990. A strong shell feature superposed on
the photospheric component of H and HeI6678 in Nov., Dec.
1990 and Jan. 1991 was reported by Norton et al. 1991. Still in absorption in
Sept. 1991 (Reynolds et al. 1992), weak emission was again present on H
in Oct. 1991 (Kaper & van Kerkwijk 1992).
Decrease and minima of line emission generally correspond to lower brightness
states. Single (, D) and (, V) correlations
are accompanied by bifurcated or double-fold (U, D), (V, D) relations with
slopes in D < D* corresponding to definite
Be phases. The stellar BD is D* = 0.050 dex (de Loore et al. 1979).
For 
incipient horizontal relations: 
are apparent which correspond to the shell
phase.
11 Cam (HD 32343)
11 Cam has always had strong emission observed up to high Balmer lines. Often
observed before 1975, this star has presented recurrent outbursts.
Schild (1973)
reported outbursts in 1916-1919, 1929-1931 and 1943-1949. From 1953 to 1974,
emission was always strong in Balmer lines with a weak minimum in 1961. Bright
and particularly clear FeII lines were reported from 1963 to 1971 during
maximum emission. Afterwards they seem to have disappeared (Atlas). Isolated
observations between 1980 and 1989 of the H emission line have revealed
a rather constant strength of this line except in October 1983
(Ballereau &
Chauville 1987) for which the line is very strong (no photometry available at
that time). Conversely no spectra are available at maximum brightness in
November 1977. The rapid photometric variability of this star is suspected
although not yet confirmed.
Single (U, D), (V, D), and (, V) relations are seen,
although they all have 
and 
slopes. The (, D) relation is scattered and
single, however it has a 
slope.
Oph (HD 148184)
Since 1919 Oph has been characterized by strong emission in Balmer lines
(Schild 1973; Slettebak 1982). A very strong emission in Balmer and FeII lines between 1953 and 1973 with a veiling effect seen up to 1969 (Atlas).
From 1979, FeII emission lines have decreased in intensity; they were
barely visible in 1975. Andrillat & Fehrenbach (1982)
and Dachs et al. (1981)
noted a steady decrease in the strength of H emission from 1972 to
1980. Then it seems to have slightly increased in 1982-1983, with a maximum
through March 1983 (Dachs et al. 1986), and afterwards fluctuated with maxima
in 1985, 1989 and 1993 (Hanuschik et al. 1996).
The decrease of the H emission line from 1972 to 1980 corresponds to an
increase of brightness and to reddening of the gradient . Total
Balmer discontinuity D varied strongly during this period with a strong
minimum
near 1983. D is at minimum when 
of H emission is at
maximum. Oph is a rapid photometric variable star (Balona 1995)
with a
moderate 
(< 130 km s-1).
The (U, D) and (V, D) relations are single for D <0 with slopes, but two branches with slopes seem to appear at 
. The (, D) and
(, V) relations are scattered and whilst the first is likely to
be single with a slope, the second
seems to be double valued with one of the branches with a slope.
59 Cyg (HD 200120)
Since 1904, 59 Cyg has presented active and quiescent phases in the behaviour
of its CE, successively producing spectra with the appearance of a quasi-normal
B, Be and Be-shell star. Long-term, rather stable, strong emission-line phases
alternate with phases of marked changes, generally characterized by slow
increase and shorter decreases of the strength of emission lines. A slow
increase of emission was observed from 1945 to 1950
(Merrill & Burwell 1949).
Afterwards, a quiescent phase set in from 1953 to 1970, the emission features
being relatively stable; emission was at maximum in 1956 and in 1961, and at
minimum in 1967 (Atlas). Short-lived outbursts were then observed from July to
the end of 1972 and from December 1973 to June 1974, each followed by a strong
shell phase of several months duration. After that, emission vanished and
almost completely disappeared in 1976-1977
(Hubert-Delplace & Hubert 1981). An
increase of emission started in the end of 1979 and ended near mid-1980.
Emission decreased again to a minimum in 1982, and again increased from 1983 to
1986. From 1979 to 1985 strong H emission strength changes occurred,
being correlated with the wind observed in the far UV
(Doazan et al. 1989). From
1986, a new, rather stable quiescent emission phase has
been established (Peters 1989-1992, 1994).
Strong emission lines are present when D < D* and the line emission
decrease corresponds to a progressive increase of the BD, which ended
at D =
D* in 1977 when the line emission almost disappeared. From the existing
photometric data, it is seen that after 1961 the mean V magnitude suddenly
began decreasing until 1977, fluctuating at intervals of 
yrs.
with a variable, damped-like amplitude whose maximum was nearly 0.4 mag. The
intrinsic continuum reddening seems to be strongest when the star is brightest
but not necessarily well correlated with line emission intensity as remains high even in 1967 at a relative minimum of line emission.
This reddening cleared up in 1977, so that became that of a
B1V normal star for D* = 0.116 dex shown by a dashed vertical line in Fig. 6.
Unfortunately, there are no photometric data corresponding to epochs of
short-lived shell phases in 1973 and 1974. It is seen, however, that the
(U,D), (V,D), () and () relations are
single with and slopes and that they are nicely conserved, even after
the shell phases and the periods of loss of emission characteristics. A
photometric microvariability of this star was observed, but there is no period
determination at the present time.
Pleione (HD 23862)
Notable long-term changes are reported in the literature, for this star, see
Hirata (1995) and references therein. Recently
this star was discovered by
Katahira et al. (1996) to be a binary with an orbital period of 218.0 days. We
give here only a summary of spectral variations. Observed in a Be phase
until 1904, Pleione entered a phase showing a rapidly rotating B normal-like star from
1905 to 1936. The first Be-metallic shell phase occurred in 1938 and ended in
1954. From 1954 to 1972, this star again entered a Be phase with strong Balmer
and FeII emission lines at maximum around 1960. Then gradual weakening
began; in 1972, the H emission line was very weak. In late 1972 another
metallic shell phase started and developed with a maximum near mid-1982. The
metallic lines faded and disappeared in 1988 (Hirata 1995). The star entered a
new Be phase; the H emission line gradually increased from 1982 (epoch
of maximum strength in shell lines) to 1993 and decreased after. The global
behaviour of spectral and light changes has been studied by many authors
(Hirata & Kogure 1976; Kogure & Hirata 1982;
Goraya & Tur 1988). A steep
decline in brightness occurred in 1938 and 1972 as each respective shell phase
started. Furthermore, a strong weakening in the U magnitude came with the
maximum of metallic shell lines in 1982-1983.
From our study it is shown that in 1972, at the epoch of the appearance of a
new Be-metallic shell phase associated with a rapid fainting in the UBV
colours, the gradient rapidly increased and total BD decreased
indicating the contribution of a secondary BD in emission. Between 1972 and 1988
the behaviour of the total BD was the same as the strength of the metallic shell
lines. It is interesting to note that the maximum value of the total BD reported
in 1982 corresponds to a maximum of the absorption strength of metallic lines.
In the short interval of D < D*, this star shows two (U,D) and (V, D)
relations which are continued in the region of D > D*, where normally in Be
shell phases there is a single relation. The (, D) and (, V) seem to be single, unless for D < D* possible double
relations are not well resolved. The (, D) relation is
slopeless, while (, V) seems to have a global slope. Rapid photometric variability was found in this
object (McNamara 1987).
48 Lib (HD 142983)
48 Lib is one of the best examples of shell stars. The variable shell episode
of 48 Lib started in 1935: RV cycles (8-13 yr), marked variations of intensity
and in line profiles. These phenomena become more striking with increase of the
amplitude of radial velocity (RV) (Aydin & Faraggiana 1978
and references
therein; Mon 1984; Hubert et al. 1987). In 48 Lib successive cycles exhibit
negative RV phases longer than positive ones. The strength of shell lines has
varied strongly since 1935. It was near or at maximum in 1943, 1965, 1970, 1974
and towards the end of 1993, according to recent observations published by
Hanuschik et al. (1996). Maxima in the strength of emission in the first Balmer
lines were generally seen at the same time or very close. Long-term photometric
variations associated with spectroscopic cycles, and mid-term quasi-periodic
oscillations (10-20 days) and sudden fadings (1-2 days) were
reported by
Mennickent et al. (1994) . These authors noted a brightness maximum in
1988 and
possible minima in 1982 and 1991. In compiled data presented in this study,
pronounced maxima of brightness were observed in 1976 and 1988. Balmer
discontinuity is respectively lower and higher at epochs close to minima and
maxima of the strength of shell metallic lines. Single relations are seen in
(U, D), (V, D), (, D) and (). The
(U, D) has a 
slope, while in all others there is
no slope. This star is a rapid photometric variable, which may in part explain
the width of spectrophotometric relations.
88 Her (HD 162732)
88 Her is well-known to have presented associated long-term changes in its
spectrum and in its light; in addition it was discovered by
Harmanec et al. (1972) to be a binary (P = 86.72 days). This star exhibited a Be phase with a
hydrogen and metallic shell from 1955 to 1960. The H emission line
gradually weakened from 1961 to 1971; the metallic shell lines, very weak from
1966, disappeared near 1970 (Atlas). In 1972-1976, the star was in a
quasi-normal B phase, the H emission line was very weak in 1976-1977,
and at the same epoch a minimum of the linear optical polarization was seen
(Arsenijević et al. 1987). Then, the H emission line again gradually
increased from 1977 to 1986. In addition, from 1978 to 1982, a new metallic
shell phase was increasing, followed from 1983 to 1986 by a decreasing metallic
shell and mild Be phase (Doazan et al. 1982a; Hirata 1995).
It was reported that
the gradual weakening of Balmer emission and metallic shell lines in 1966-1970
was associated with brightening of the star (Doazan et al. 1982b)
and that a
large and rapid drop of the luminosity occurred near 1978, just before the
development (1978-1982) of a new Be-shell metallic phase
(Harmanec et al. 1980;
Doazan et al. 1986; Hirata 1995). The polarization percentage rapidly increased
in 1978, followed by a slow decline from 1979 to 1985
(Arsenijević et al. 1987).
From this study, it is seen that the total BD was at minimum when the metallic
shell lines disappeared towards 1970 and during the quasi-normal B phase
(1972-1976). Then it gradually increased to mid-1982, in the same manner as the
strength of shell lines. On the other hand, the star became brighter as the
equivalent width of the H emission line increased (1978-1986),
according to Hirata (1995). It is worth noting that BCD observations of the
stellar BD from mid-1977 to the end of 1983 do not reveal any change in the
photospheric 
and log g parameters of this star during its phase
changes (Zorec et al. 1989). The difference of brightness in the U and V
magnitudes between the regions of D < D* and D > D* are clearly seen in
Fig. 6. Single and slopeless relations (U, D), (V, D), (, D) and (, V) are characteristic for D > D*. No rapid
photometric variability was found in this object.
HD 28497. A rapid photometric variable star. The emission is strong when D is small. The emission was low from 1989 to 1992; since then, D and V are increasing.
HD 35439. After a B normal-like phase, H emission appeared when D
was at its minimum value.
HD 37202. A shell and rapidly variable star. D is at minimum when the strength of the shell is at minimum, but in such a case the RV is not at minimum in this star.
HD 37967. Emission is strong and there is almost no variation.
HD 48917. A rapidly variable star. Only a few spectroscopic data exist for this star.
HD 50123. Interacting binary having a composite B spectrum with a giant K companion. "Ellipsoidal'' photometric variations with a period of P = 28.6 days were found by Sterken et al. (1994).
HD 56014. Star with a moderate rotation
(
km s-1)
showing at epochs a shell behaviour D > D*.
HD 56139. A rapidly variable star with a small (60 km s-1)
but sometimes showing the shell behaviour D > D*. The value of
for H is rather high (
). When D is smallest,
H emission is weak.
HD 58978. A helium-shell star. The strong spectroscopic variations in 1990 and in 1991 do not seem to be apparent for the spectrophotometric variations.
HD 60848. One of the hottest Be stars where the Balmer and Paschen line emission is well correlated with the Balmer continuum emission.
HD 60855. Star with moderate line emission showing B 
Be phase
transitions. There are no strong spectrophotometric variations.
HD 63462. Star showing slight variations in the H emission
intensity.
HD 65875. A rapidly variable star with moderate (150 km s-1)
and showing however phases with the shell behaviour D > D*. The star has a
high value of in H (
). Slopes
and
seem to exist for this star.
HD 68980. Star with a very moderate
(95 km s-1) and with
very strong H emission (
). D seems to be smallest
when H emission is strongest. Two relations between
U, V versus D and for against V are apparent.
HD 83953. A rapidly variable star where D is probably at maximum when the emission is low.
HD 89890. A star with a low rotation, 
km s-1, having
D > D*.
HD 91120. Emission is moderate with few variations.
HD 109387. A rapidly variable star. Emission is strong when D is low
and is high. There is a high dispersion in the (D,V) diagram
for 
in 1986.
HD 120991. Star with small (70 km s-1). Sometimes is high for H (
in 1993). D diminishes when the
emission increases. There are probably two correlations for U and V as a
function of D.
HD 131492. A temporary strong shell phase was observed by
Slettebak in 1980
which corresponded to a minimum in V and to a maximum in D
(Slettebak 1982).
This is responsible for the horizontal part in the diagram of
against D. The shell behaviour D > D* is clearly seen, although the star
apparently has moderate rotation ( = 100 km s-1).
HD 162428. The emission and the shell are variable.
HD 168797. This is a B Be phase variable star. Emission is
moderate.
HD 171780. A B Be variable star with moderate emission.
HD 178175. Star observed in spectroscopy very irregularly. The shell
behaviour D > D* is slightly present and the star has a moderate rotation
( = 120 km s-1).
HD 183656. A rapidly variable shell star. D is correlated with the V/R variations (the V/R and radial velocity curves are given in Koubsky et al. 1989). D is at maximum when the RV is at maximum.
HD 184279. A shell star, where D is correlated with the magnitude V and with the RV curve: D is at maximum as RV is maximum (in 1979).
HD 187811. A B Be variable star with rather moderate emission.
HD 191610. A rapidly variable star. It has been seen in B phase in 1955 to 1958, and in a Be phase since then. Emission is moderate.
HD 195325. A late Be star with a hydrogen and metallic shell.
HD 205637. A rapidly variable star where the hydrogen and metallic-line shell are variable.
HD 217050. A rapidly variable shell star. D is at maximum when the strength of the shell is at maximum.
HD 217543. Star with moderate emission showing B Be phase
transitions.
HD 218674. Star with rapid variability. It also shows strong emissions and a hydrogen-shell.
We may conclude that patterns in Fig. 6 show that spectrophotometric
changes in Be stars are characterized by 
and 
relations which differ on the emission/absorption phase and may
differ from star to star. More or less well-defined relations involving D <
D* correspond to definite Be-phases. They can be either single or bivalued,
mostly with and , but sometimes also with slopes and 
. On the contrary, the
most current shape of relation during SPh-shell phases where D > D* can be
summarized by: 
, and depending
on the stars 
or .
The constancy of the magnitude V is within 0.15 mag and sometimes even more.
However, for some Be-shell stars, variation in V can be as high as 0.25 mag.
In general, of those Be stars which have two correlations of U, V against
D, and, against V in Be phases (D < D*), nearly all have
strong Balmer emission lines. The fact that there are two relations or only one
is not related to particular values of .
in SPh-shell phases
It is noteworthy the behaviour of some stars like HD 56014, HD 56139, HD
65875, HD 131492, where all have D* determined in the BCD system, which
show both SPh-Be and SPh-shell phases. The mean deviation of these objects for
the SPh-shell phase is 
dex which is more
than 3 times the expected error 
given in Table 3. Most of these
deviations are depicted by data from the Geneva photometric system, which is
one of the most stable and uniform (Sterken & Manfroid 1991). These comments
are also relevant to HD 178175, where D* is only for the mean MK spectral
type and the 
dex deviations of the SPh-shell like phase are
established from the Geneva photometric data. In Fig. 6 we see that for most of
the above mentioned stars, the transition between a SPh-Be and a SPh-shell phase
is characterized by a change of slopes in the spectrophotometric correlations.
It would be difficult to understand such slope changes in terms of errors
affecting the BDs, as they would likely conserve one of the observed slopes.
Among those stars where both spectrophotometric behaviours: Be (
;
D < D*) and shell (
; D > D*) were seen, there are some
with small or moderate . We also note that two Be stars with low
were seen only in SPh-shell phases where 
(HD
89890 (
km s-1) and HD 178175 (
km
s-1). Low values of may correspond either to small 
or
to low velocity V. In the first case, the SPh-shell phase: D > D*, cannot
be explained by highly flattened CE seen pole-on. In the second case, we should
admit the existence of an important fraction of slowly rotating Be stars
(Mennickent et al. 1994). Nevertheless, for most Be-shell stars studied in this
paper it is 
300 km s-1.
On the other hand, it is also worth noting that in Be-shell stars where enough RV data existed to be correlated with D (HD 142983, HD 183656, HD 184279 etc.), the highest RV > 0 appear when the line shell phenomenon and the absorption in the BD due to the CE are strongest. This phenomenon might favor the formation of compact CE layers near the star.