HD 965 has never been studied in detail before. Since magnetically resolved lines have been discovered in this star, in November 1993, its mean magnetic field modulus seems to have been monotonically decreasing (Fig. 5 (click here)). It appears highly probable that the rotation period of HD 965 is much longer than 2 years, although given the small number of measurements, one cannot definitely rule out a shorter period.
The observation of resolved magnetically split lines in this star was
first reported in Paper II. The only value of the period that
adequately accounts for all the longitudinal field measurements
(Babcock 1958; Wolff 1975; Mathys &\
Hubrig 1996) is 
d, in good agreement with Wolff's
(1975) original estimate of 525 d. The longitudinal field is always
negative, varying approximately between -950 and -400 G.
The profile of 
in HD 2453 is very clean, mostly free from
blends, so that we expect this star to be one of those where we can
measure the magnetic field modulus with the best accuracy. As discussed
in Sect. 6, the measurement uncertainty should then be of
the order of 30 G. The standard deviation of our 9 measurements, 60 G
(see Table 3 (click here)), significantly exceeds this value, which indicates
that we are likely detecting actual field variations. As a matter of fact,
when plotted against rotation phase, our data seem to lie along a
nearly sinusoidal curve (see Fig. 6 (click here)), suggesting that the field
modulus of HD 2453 may vary with a peak-to-peak amplitude of the order
of 160 G. The maximum of the mean field
modulus variations seems to coincide roughly with the largest
negative value of the longitudinal field, and the minimum of the mean
field modulus apparently occurs approximately when the longitudinal
field is closest to 0 (compare Fig. 6 (click here) with Fig. 7 (click here)).
Additional measurements, allowing a better
sampling of the rotation phases of HD 2453, will be useful to confirm
that this description of the variations of the field
is indeed correct.
Figure 6: Mean magnetic field modulus of HD 2453 against rotation
phase. The meaning of the symbols is as given in Table 5 (click here)
Figure 7: Mean longitudinal magnetic field of HD 2453 against rotation
phase. Data are from Babcock (1958; open triangles),
Wolff (1975; crosses: Lick data; open squares:
Mauna Kea data), and Mathys & Hubrig (1996; filled
circle)
Figure 8: Mean magnetic field modulus of HD 9996 against heliocentric
Julian date. The meaning of the symbols is as given in Table 5 (click here)
Figure 9: Same spectral region as in Fig. 1 (click here), as observed in
HD 9996 on the dates indicated next to each tracing (
)
In Paper II, we had reported the first observation of resolved magnetically split lines in HD 9996. From the discovery spectrum, we had measured a mean field modulus of 3.8 kG. This is significantly larger than the estimate of 2.2 kG (Preston 1971a) and the upper limit of 2.5 kG (Scholz 1983) obtained from the analysis of the differential broadening of unsplit lines. The resulting suspicion that the magnetic field modulus of this star may show large variations is fully confirmed by our new measurements. From the consideration of Fig. 8 (click here), where our data are plotted against Julian date, it seems that HD 9996 may just have passed the time of maximum of its mean field modulus. The latter probably occurred at the end of 1993 or the beginning of 1994. Adopting for the rotation period the most recent estimate of 21 yr (Rice 1988), the phase of the maximum field modulus would appear to coincide roughly with that of the negative extremum of the mean longitudinal (or effective) field, to the accuracy with which the latter can be inferred from the published measurements (Babcock 1958; Preston & Wolff 1970; Scholz 1978, 1983). According to the latter, the range of variation of the longitudinal field is from -1200 G to +300 G.
Figure 10: Mean magnetic field modulus of HD 12288 against rotation
phase. The meaning of the symbols is as given in Table 5 (click here)
HD 9996 was already famous for its extreme spectroscopic variations (Preston & Wolff 1970). The results presented here indicate that the variation of its magnetic field modulus also is unusually large. Indeed, we find that the field reaches 5 kG (or more) at maximum, while from the above-mentioned estimates of Preston (1971a) and Scholz (1983), at minimum it does not exceed 2.5 kG. Thus the ratio between its extrema is at least 2.0, which is greater than in any other star with magnetically resolved lines studied until now.
From the consideration of the line profiles at different phases, it
furthermore appears that not only the field strength but also its
orientation with respect to the observer vary remarkably with phase.
This is illustrated in Fig. 9 (click here), where portions of our oldest and of
our most recent spectra of the star are plotted. The very unusual,
triangular shape of the line 
in the former had
already been stressed in Paper II. Comparing the two observations
shown in Fig. 9 (click here), one can see that this triangular shape can to a
large extent be attributed to the fact that the 
components of the
line are remarkably weak. The contribution of these components to the
line is seen to have increased quite noticeably between our first and
last observations. Even in the latter, the strength of the 
components relative to that of the central 
components remains
smaller than in most other stars studied (see Figs. 2 (click here) to
4 (click here)). Similarly, the outermost 
components of the quadruplet
are weak too, even almost not visible at all
in our first spectrum. This seems to indicate that the magnetic field
of HD 9996 at the considered phases is predominantly transversal, at
least on the regions of the stellar surfaces where this field (which
appears to be unusually inhomogeneous) is strongest.
Note also that HD 9996 is a spectroscopic binary with an orbital period (273 d) much shorter than its rotation period (Preston &\ Wolff 1970).
Figure 11: Periodogram obtained in a frequency search on the
measurements of the mean magnetic field modulus of HD 14437. The
ordinate is the reduced 
of the fit of the field modulus
measurements by a sine wave with the frequency given in abscissa
Analysing our 20 measurements of the mean field modulus of this star,
where magnetic line splitting had first been observed by Preston
(1971a), we find that the most probable value of its
rotation period is
This is in excellent agreement with the value 
d that
had been derived by Wolff & Morrison (1973) from photometric
observations. These photometric observations also allow us to rule out an
alternative value of the period, 286, which appears almost as
satisfactory from the consideration of our magnetic data
alone. Conversely, our measurements are inconsistent with the aliases
close to 1 day that appeared marginally plausible from the
photometry.
The variation of the mean magnetic field modulus of HD 12288 through its rotation cycle is shown in Fig. 10 (click here). The amplitude of the variations, of the order of 1 kG peak-to-peak, is much smaller than the difference between the extreme values of 6.1 and 8.6 kG of the field modulus reported by Preston (1971a). However, this discrepancy, the origin of which is unclear, does not cast doubt on the reliability of the rotation period derived here.
The split components of the line 
show some evidence of
distortion due to rotational Doppler effect. This is well seen
in the portion of the spectrum shown in Fig. 2 (click here), from the fact that
the red component of the line is sharper and narrower than the blue
component. This is the opposite of the characteristic asymmetry
corresponding to the partial Paschen-Back effect regime of formation
of this line, which is actually observed when rotation is negligible
(see Paper I for details).
We have obtained three measurements of the longitudinal magnetic field
of HD 12288 through H
photopolarimetry: two null values at phases
0.875 and 0.974, and a marginal detection (
kG) at phase
0.129.
Finally, our study reveals that the radial velocity of HD 12288 is
variable, thus that it is a spectroscopic binary.
The amplitude of the
variation is at least 16 
. The timescale on which it occurs is
long: the orbital period is unlikely to be much shorter than
4 years. The determination of the orbital parameters, and more
generally, the detailed study of the radial velocity of this and of
the other stars considered in this paper are beyond its scope. They
will be the subject of a separate work.
The detection of resolved magnetically split lines in HD 14437 has been reported in Paper III. Its mean field modulus appears to vary with an amplitude of the order of 1 kG peak-to-peak. The period cannot be unambiguously determined from our present data. In a periodogram covering the range comprised between 2.5 and 1000 d (Fig. 11 (click here)), three groups of periods clearly stand out, around 15 d, 30 d, and 350 d. The latter seems to be ruled out by the fact that we observe apparently significant variations on timescales of a few days. The longitudinal field is also reported by Glagolevskij et al. (1982) and Glagolevskij et al. (1986) to vary on similarly short timescales between -2.3 and -0.4 kG. This is consistent with our own determinations of the longitudinal field (3 measurements over one
Figure 12: Mean magnetic field modulus of HD 29578 against heliocentric
Julian date. The meaning of the symbols is as given in Table 5 (click here)
week, through H
photopolarimetry, ranging between
-1.4 and
-1.9 kG). We do not have enough data yet to decide
between the two groups of periods around 15 d and around 30 d, nor to
resolve the strong aliasing within these groups.
In the first spectrum (taken in October 1990) of HD 18078 where we
observed magnetic splitting in 
(Paper II), this doublet was just
barely resolved. Its components were better separated in January 1992.
But they were hardly resolved in November 1992 and October 1993: the
field modulus values derived on these dates are rather uncertain. In
February 1994, splitting was not seen, but the total width of the line
was still consistent with a field of the order of 3 kG. The doublet
was again resolved in our last observation of this star, in January
1995. From these observations, the mean magnetic field modulus of
HD 18078 appears to undergo quite significant variations (the ratio
between the extrema is at least 1.4). The split components of 
are
unusually broad. Whether this is due to rotational Doppler effect or
to a particularly inhomogeneous magnetic field distribution on the
stellar surface is at present unclear. Indeed, the distribution in
time of our observations, which was inspired by Wolff & Morrison's
(1973) statement that the rotation period might exceed 10 years,
does not allow us to decide whether the period is indeed very long or
whether it might be short enough so that rotational Doppler effect
could significantly contribute to the line profiles. Our future
observations will be planned so as to resolve this ambiguity.
Figure 13: Mean magnetic field modulus of HD 50169 against heliocentric
Julian date. The meaning of the symbols is as given in Table 5 (click here)
Figure 14: Mean magnetic field modulus of HD 55719 against heliocentric
Julian date (a time span of 700 days has been extracted from the whole
dataset). The meaning of the symbols is as given in Table 5 (click here)
The observations of magnetically resolved lines in this otherwise
not well known Ap star is reported here for the first time. The
standard deviation of 59 G of our 9 measurements of the mean magnetic
field modulus of a star where the purity of the 
profile
probably allows us to determine this quantity with a significantly
better accuracy (see Sect. 6)
indicates that we are almost certainly detecting
low-amplitude variations. A plot of the measurements against Julian
dates (Fig. 12 (click here)) suggests that the period may be long (significantly
exceeding
2 years), but more observations are required to establish this more
reliably.
Figure 15: Mean magnetic field modulus of HD 55719 against rotation
phase, computed assuming that the rotation period is 847 d. The phase
origin is HJD 2447160.0. The meaning of the symbols is as given in
Table 5 (click here)
Figure 16: Mean magnetic field modulus of HD 55719 against rotation
phase, computed assuming that the rotation period is 775 d. The phase
origin is HJD 2447346.0. The meaning of the symbols is as given in
Table 5 (click here)
Figure 17: Mean longitudinal magnetic field of HD 55719 against rotation
phase, computed assuming that the rotation period is 847 d (data from
Bonsack 1976). The phase origin is HJD 2447160.0
Figure 18: Mean longitudinal magnetic field of HD 55719 against rotation
phase, computed assuming that the rotation period is 775 d (data from
Bonsack 1976). The phase origin is HJD 2447346.0
On the other hand, the radial velocity of HD 29578 is definitely
slowly variable. A maximum has been reached beginning of 1994, which
differs by more than 16 
from the lowest value recorded so far, in
our last observation of the star, in August 1995: HD 29578 appears to
be a spectroscopic binary with an orbital period significantly longer
than 2 years.
Since our discovery of resolved magnetically split lines in this star
in March 1991 (Paper II) until our second last observation in February
1995, its mean field modulus has monotonically increased from 4.4 to
5.1 kG (Fig. 13 (click here)). Our last observation in March 1995 yielded a
somewhat smaller field value (just below 5 kG), which most likely
is spurious (possibly due e.g. to an unrecognized cosmic ray hit
affecting the profile of 
). However, it
may also indicate that maximum field
strength has just been passed. This will have to be confirmed by
future observations, especially because it is not quite
consistent with Preston's (1971a) estimate of the field
strength (5.6 kG) from differential magnetic broadening of spectral lines.
In any case, the period of HD 50169 appears to be
much longer than 4 years.
This is consistent with the slow variation of the longitudinal field reported by Babcock (1958). The only recent determination of this field moment, +1.3 kG (Mathys & Hubrig 1996) falls within the range of values derived by Babcock.
Between our first and last observations of HD 50169 (4 years apart),
its radial velocity has monotonically increased, by about 2 
:
thus HD 50169 appears to be a spectroscopic binary with an orbital
period significantly longer than 4 years.
HD 55719 is one of only three magnetic Ap stars known to be SB2. This property has been discovered by Bonsack (1976), who has determined the orbital parameters and has also obtained 24 measurements of the longitudinal magnetic field. From 22 of them, ranging roughly from +1 to +2 kG, he derived possible values of the rotation period of 3039 and 3648. Regardless of the period, Bonsack's (1976) remaining two longitudinal field measurements (-1.1 and -0.6 kG) cannot be reconciled with the bulk of his magnetic data.
We have already discussed the properties of HD 55719
in Papers I (where the discovery of resolved magnetically split lines
was first reported) to III. We now have 29 measurements of
its mean magnetic field modulus. They show no
significant correlation with the phases computed using the two
values of the period indicated by Bonsack (1976). Since
Bonsack's (1976) data
marginally indicated that the period would not be very long, we
first tried to fit our mean field modulus measurements with periods
in the range 
, without success: the corresponding
periodogram looks like pure noise. On the other hand, plotting against
Julian Date our
data of the two consecutive observing seasons during which we have
observed the star most intensively (from October 1992 to May 1994)
suggests the existence of a
systematic modulation with a periodicity of the order of 2 years or
more (see Fig. 14 (click here)). This urged us to extend the frequency analysis
to longer periods. We found, indeed, that in a
periodogram covering the range 
, a double peak clearly
stands out, indicating that the rotation period of HD 55719 may well
have one of the two values:
or
The magnetic field modulus is plotted against the phases computed with
these two possible values of the period in Figs. 15 (click here) and
16 (click here). No choice between them can be made on the basis of the
presently available data.
Figures 17 (click here) and 18 (click here) show phase diagrams of Bonsack's (1976) mean longitudinal field measurements obtained with the tentative periods derived from the mean field modulus data. Except for the two negative field values (unexplained by any plausible period), and to a lesser extent, for the third smallest measurement (+760 G, which tends also to be an outlier with Bonsack's values of the periods), these measurements are not violently inconsistent with the long periods proposed here. They do, in fact, seem to show some trend with these periods (especially 847 d). Three more recent measurements of the longitudinal field, performed from observations carried out at ESO with the 3.6 m telescope and the Zeeman analyzer of CASPEC by Mathys & Hubrig (1996), yielded values between 450 and 950 G, that is, smaller than the bulk of Bonsack's data. However, such inconsistencies between longitudinal field data obtained at different sites and with different instruments are not unusual (see e.g. Mathys 1991). Only when more measurements are obtained with CASPEC can they be used (possibly in combination with Bonsack's data) to constrain the value of the rotation period.
That the rotation period of HD 55719 is long receives further support from the fact that the star undergoes no significant photometric variations over a timescale of the order of one month (Heck et al. 1987). However, more data need to be obtained before its value can be indisputably established.
It should finally be noted that the components of the line
in HD 55719
have unusual, quite asymmetric shapes (see Fig. 2 (click here)), with rather
steep edges toward the line centre and more extended wings outwards
(on the red side of the red component and on the blue side of the blue
component). This distortion cannot be
attributed to the contribution of the secondary component of the
binary to the observed spectrum (see Papers I and II). It hints at a
rather unusual structure of the magnetic field.
HD 59435 is also an SB2. This property has been discovered by North (1994). It is probably the most interesting of the three SB2s known to comprise a magnetic Ap star, since both components have rich, sharp-lined spectra, so that both of them can be studied in great detail. This system has been thoroughly discussed in a separate paper (Wade et al. 1996b). Here we shall just summarize some of the main results of this work.
Figure 19: Mean magnetic field modulus of HD 59435 against heliocentric
Julian date. The meaning of the symbols is as given in Table 5 (click here)
The orbital period of HD 59435 is 1387 d. The primary is a G giant, while the secondary is the Ap star, which must be close to the end of its main-sequence life.
The line 
of the Ap component is resolved into its magnetically
split components. At many orbital phases, this line is blended with
the same or another line of the other component. This complicates
the determination of the mean field modulus: the spectrum of the Ap
star first has to be rebuilt by removing the contribution of the other
component. Details of the procedure are given by Wade et al.
(1996b).
The mean field modulus of the Ap component is plotted against Julian date in Fig. 19 (click here). A slow variation is very clearly seen. The data obtained until now seem to indicate that the rotation period may be slightly longer than 1000 d. But this must be taken with caution, since the magnetic field measurements performed so far do not quite cover this time span. In any case, it is already clear that the amplitude of variation of the mean field modulus of HD 59435 is remarkably large: the ratio between the maximum and minimum strengths found until now, 1.8, is among the largest found in any star with resolved lines known. Also, the field modulus at minimum (hardly more than 2200 G) is the smallest one measured in any of the stars considered in this paper.
No measurements of the mean longitudinal magnetic field of HD 59435 have been published. But Babcock (1967) had reported it to be magnetic on the basis of observations of circular polarization in spectral lines.
Our most recent discovery of a star with magnetically resolved lines
is HD 61468. Very little is known about this faint, cool Ap
star. Until now, we have obtained four measurements of its mean
magnetic field modulus, spanning 119 days. These measurements are
consistent with a slow monotonic decrease of the field intensity
during that time interval. The radial velocity was also found to vary,
with an amplitude of at least 45 
, so that the star definitely is
SB. The orbital period cannot be determined yet, of course, but it
must be significantly shorter than 119 days.
Figure 20: Mean magnetic field modulus of HD 65339 against rotation
phase. Asterisks represent Huchra's (1972) data; the
meaning of the other symbols is as given in Table 5 (click here)
HD 65339 (= 53 Cam) is possibly the Ap star whose magnetic field has
been most
studied. The presence of resolved lines in its spectrum has been first
reported by Preston (1969b), and Huchra (1972)
has been the first one to take advantage of it to try to model the magnetic
field structure. The most extensive sets of measurements of the
longitudinal field have been published by Babcock (1958) and by
Borra & Landstreet (1977). According to the latter authors,
the longitudinal field varies between -5.4 and +4.2 kG along the stellar
rotation period of 80267. When phased together according to this period,
Huchra's (1972) measurements of the mean magnetic field
modulus and ours agree well (see Fig. 20 (click here)).
The extrema of the field modulus variations coincide, to the
achieved accuracy, with those of Borra & Landstreet's (1977)
longitudinal field curve. Our data for 53 Cam show a unusually large scatter
around a smooth variation curve, reflecting the relatively large
errors of our measurements. Indeed, the line 
appears poorly
suited to the diagnosis of the field modulus of that star: the
combination of rotational Doppler effect and of particularly strong
blending on the blue side makes the separation of its components
(especially the wavelength of its blue component) very difficult to
determine accurately. In particular, near phase 0.1, the components of
can hardly be recognized, and we derive a very
discrepant field value.
Accordingly, our new data at best confirm that
Borra & Landstreet's (1977) value of the period adequately
matches data spanning a timebase of nearly 40 years. But they do not
significantly contribute otherwise to an improved knowledge of the
magnetic field of 53 Cam.
Figure 21: Mean magnetic field modulus of HD 70331 against rotation
phase, computed assuming that the rotation period is 30308.
The phase origin is HJD 2446987.10.
The meaning of the symbols is as given in Table 5 (click here)
Figure 22: Mean magnetic field modulus of HD 70331 against rotation
phase, computed assuming that the rotation period is 36515.
The phase origin is HJD 2447000.60.
The meaning of the symbols is as given in Table 5 (click here)
Figure 23: Mean magnetic field modulus of HD 81009 against rotation
phase. The meaning of the symbols is as given in Table 5 (click here)
Let recall that HD 65339 is a spectroscopic binary, which has been studied in detail by Scholz & Lehmann (1988). Its orbital period is long: 2432 d.
HD 70331, in which the observation of magnetically resolved lines is
reported here for the first time, is one of the hottest stars
presently known to have this property. Due to this high temperature,
the line 
is rather weak. It is furthermore quite distorted by
rotational Doppler effect, so that the diagnosis of the mean magnetic
field modulus cannot be done with high accuracy. This probably
explains at least in part why we were not able to determine
unambiguously the rotation period of this star. There is little doubt
that the latter is short, thus that the star is seen almost pole on.
Possible, but still very questionable values are
d (see Fig. 21 (click here)) or 
d
(see Fig. 22 (click here)).
The only determination of the longitudinal magnetic field of HD 70331 performed so far yielded a value of -2.8 kG (Mathys & Hubrig 1996).
We report here for the first time the presence of magnetically
resolved lines in HD 75445. The standard deviation of the 9
measurements of the mean magnetic field modulus of this star
is 42 G. This somewhat exceeds the estimated uncertainty of
our magnetic measurements: the latter should be at most 30 G,
given the good S/N ratio of our spectra and the purity of the profile
of 
. It seems likely that we have recorded small but real
variations of the field. But for the time being, we do not have enough
data to describe them fully: it only appears that the
rotation period of the star might be significantly shorter than the 450
days covered by our observations.
HD 81009 has already been extensively discussed in Papers I through
III. The presence of resolved magnetically split lines in its spectrum
had first been noticed by Preston (1971a). The value of its
rotation period derived from our 39 magnetic measurements alone,
is in perfect agreement with that obtained from
photometry by Waelkens (1985). The phase coverage of our
Figure 24: Mean magnetic field modulus of HD 93507 against rotation
phase. The meaning of the symbols is as given in Table 5 (click here)
observations is excellent. The variation of the mean field modulus shows some anharmonicity (Fig. 23 (click here)): the shape of the variation curve near maximum is almost triangular, while it is broader and flatter around minimum.
18 determinations of the longitudinal field, well distributed throughout
the rotation cycle, have been recently performed by G. Hill and
D. Bohlender, using the technique of H
photopolarimetry (Hill,
private communication). The variation curve is definitely anharmonic,
with 
raising slowly from minimum (of the order of 600 G, close to
phase 0.0) to maximum (about 2000 G around phase 0.65), and a much
steeper slope back
from maximum to minimum. Mathys & Hubrig's (1996) only
measurement of 
(1900 G at phase 0.706) agrees well with the data
of Hill & Bohlender.
HD 93507 is another star that has been very little studied until now,
for which the resolution of magnetically split lines is reported here
for the first time. Its rotation period can be determined without
ambiguity from our 28 measurements of the mean field modulus:
The variation of the field modulus with this period is shown in
Fig. 24 (click here). It is probably anharmonic, in particular with a
triangular shape of the curve
at field minimum. The larger scatter of the
measurements close to field maximum is due to the fact that 
is
considerably weaker around this phase, and that its blue component is
heavily blended, making its wavelength very difficult to determine
accurately (see Fig. 25 (click here)).
Figure 25: Same spectral region as in Fig. 1 (click here), as observed in HD 93507
at the phases indicated next to each tracing. Note that the 
\
lines 
and 
are significantly weaker
around phase 0.5 than around phase 0.0
Mathys & Hubrig (1996) have determined the mean longitudinal field of HD 93507 finding values of 1.6 and 2.6 kG at phases 0.609 and 0.911, resp.
HD 94660 has been observed for more than 7 years within the framework of this programme since the discovery of its having magnetically resolved lines (Paper I). The corresponding measurements of its mean field modulus are shown in Fig. 26 (click here). From their consideration, the rotation period of the star appears to be of the same order as the length of the time interval during which we have followed it. This is consistent with Hensberge's (1993) suggestion, based on photometric observations, that this period may be close to 2700 d. In any case, the variation of the mean field modulus appears very anharmonic.
The four measurements of the longitudinal field of HD 94660 carried
out by Mathys (1994b) and Mathys & Hubrig
(1996) from ESO CASPEC circular polarization observations span a
time interval of 2500 d, similar to that covered by the field modulus data.
The longitudinal field does not appear to have varied significantly during
that time, when it was always of the order of -2.0 kG. This implies that,
if the stellar rotation period is indeed of the order of 7 years, the
longitudinal field of HD 94660 must be essentially constant. No
variation of this field moment was found either by Bohlender et
al. (1993), from 4 H
polarimeter measurements spanning
1200 days, which all yield values close to -2.5 kG. Since these
measurements are contemporaneous with those of Mathys (1994b)
and Mathys & Hubrig (1996), the difference between the two
sets can be safely ascribed to the different measurement techniques. Such
discrepancies between H
and CASPEC measurements are not unusual
indeed (Mathys 1991). Similarly, the -3.3 kG
Figure 26: Mean magnetic field modulus of HD 94660 against heliocentric
Julian date. The meaning of the symbols is as given in Table 5 (click here)
value derived
by Borra & Landstreet (1975) through H
photopolarimetry is, given its estimated uncertainty of 0.5 kG, only
marginally different at most from the measurements of Bohlender et
al. (1993), all the more because there may be some scaling error
between the H
and H
data.
The radial velocity of HD 94660 is definitely variable. The determination of the orbital elements of this spectroscopic binary is beyond the scope of the present paper. But it can be noted that its orbital period should not be much longer than 2 years: that is, it is much shorter than the stellar rotation period.
The 4 observations obtained so far of HD 110066, in which the presence of magnetically resolved lines has first been reported in Paper III, cover only about 1/5 of its probable rotation period of 4900 d (Adelman 1981). They are therefore mostly inconclusive as far as the field variability is concerned, all the more because two of them were performed with AURELIE and the other two with the KPNO coudé spectrograph (the existence of systematic differences between AURELIE data and measurements made with other instruments is discussed in relation with other stars). It can just be noted that the four measurements cluster around 4.1 kG, while Preston (1971a) had estimated a surface field of 3.6 kG from differential line broadening - again, there may be systematic differences between this technique and measurements of line splitting.
Figure 27: Mean magnetic field modulus of HD 116114 against heliocentric
Julian date. The meaning of the symbols is as given in Table 5 (click here)
Five measurements of the longitudinal field, obtained over an interval
of 2 years, have been published by Babcock (1958). None of them
exceeds 300 G. A null measurement was also obtained recently through
H
photopolarimetry.
As for HD 110066, the discovery of resolved magnetically split lines in HD 116114 has been reported in Paper III. The 18 measurements of the mean field modulus performed since the discovery time are shown in Fig. 27 (click here). For a long time, it seemed that this field moment did not undergo any significant variation. But recent observations show hints of a slow increase, suggesting that we have just witnessed the field minimum in a star with a period much longer than 3 years. The standard deviation of the whole set of measurements, 50 G, is rather large for one of the stars where the field diagnosis is easiest and should be achievable with the best accuracy. This view is furthermore supported by the fact that if the 1995 data are excluded from the set of measurements, the standard deviation of the latter is reduced to 31 G.
The only determination of the longitudinal magnetic field of HD 116114 obtained so far yielded a value of -1.9 kG (Mathys & Hubrig 1996).
The radial velocity of HD 116114 has been slowly, monotonically
increasing from 4 to 7 
between our first and last observations:
the star appears to be a spectroscopic binary with a period longer
than 3.5 years. The low amplitude of the variations and the long
timescale over which they occur probably explain
why they were not detected by Abt & Willmarth (1994).
HD 116458 is the first star with resolved magnetically split lines that has been discovered within the framework of the present programme (Papers I and II). Its rotation period, 1479, has been determined by Hensberge (1993) from photometric observations and shown to be consistent with
Figure 28: Mean magnetic field modulus of HD 116458 against rotation
phase. The meaning of the symbols is as given in Table 5 (click here)
the longitudinal field data of Albrecht et al. (1977) and of Mathys (1991). The revision of the latter by Mathys (1994b) does not question this conclusion, which is also consistent with the more recent measurements of Mathys &\ Hubrig (1996). The longitudinal field appears to vary between -1.4 and -2.4 kG.
In contrast, the field modulus shows no significant variation. The standard deviation of our 15 measurements of this quantity is only 30 G, one of the smallest of the whole sample of stars studied here. It can be seen in Fig. 28 (click here) that these measurements are well distributed throughout the stellar rotation cycle.
It may also be noted that this star is a spectroscopic binary with an orbital period (12618, Dworetsky 1982) shorter than its rotation period.
We report here for the first time the observation of resolved
magnetically split lines in HD 119027. It is the faintest star
having this property presently known. Furthermore, it is one of the
stars of the present sample with the weakest 
lines.
For these reasons, our field modulus determinations for HD 119027
are probably somewhat less accurate than those of other stars that we
have studied. But there is little doubt that the large standard
deviation (162 G) of our twelve measurements of this star reflects the
observation of actual variability. Our data are insufficient to
establish the period of these variations. However, significant changes
seem to occur on timescales of the order of one month. On the other
hand, HD 119027 is a rapidly oscillating Ap star in which Martinez et
al. (1993) did not detect any significant photometric modulation due
to rotation over 19 days. It seems that the only way to reconcile this
with our own data is to assume that the star may have a period of
rotation of the order of a few weeks, but that its surface brightness
is fairly homogeneous.
HD 126515 is also known as Preston's star, since Preston's
(1970) discovery that it has resolved magnetically split lines. This
is one of the four stars for which, before this work, the variations of the
mean field modulus had been studied throughout the rotation
period. The latter, as a matter of fact, had been determined by
Preston (1970) from his field modulus measurements, spanning a
timebase of 12.5 years. The value that he had found, 1300, is
consistent with the value of 
d determined by
North & Adelman (1995) from Geneva and Strömgren
photometric observations collected between 1971 and 1994.
When we plot Preston's (1970) and our mean magnetic field modulus measurements together against phase, there appears to be a systematic intensity shift between the two sets. In order to obtain a quantitative estimate of it, we fitted the variation with phase of each set separately by a sinusoid, using North & Adelman's (1995) value of the rotation period. The mean value over the stellar rotation period of the field modulus determined in that manner from Preston's (1970) data is 13485 G, while from our measurements it is 12735 G: hence the two sets differ systematically by 750 G. Such a difference, which is of the order of 6% of the mean field value, can probably be explained by systematic effects (e.g., difference of instrumental polarization) between the measurements performed on different spectral lines with different telescopes and spectrographs.
After subtracting 750 G from Preston's (1970) data, we
combined them with our data and performed a period search on the whole set.
The value of the period derived in that way, from measurements
spanning 38 years, is:
which is slightly more accurate but fully consistent with North &
Adelman's (1995) value. Both Preston's (1970)
measurements (after subtraction of 750 G) and our data are plotted against
the phase computed with this period in Fig. 29 (click here). Figure 30 (click here)
is a phase diagram, based on the same period, of all the published
measurements of the mean longitudinal field (Babcock 1958;
Preston 1970; van den Heuvel 1971; Mathys
1994b; Mathys & Hubrig 1996). A very remarkable
property of HD 126515 appears from the comparison of these two figures: the
curve of variation of its longitudinal field is extremely asymmetric, while
the variation of its mean field modulus is much less anharmonic. This
behaviour, which is similar to that of HD 81009 (see Sect. 5.15), is at
odds with those of other stars studied in this paper for which enough
Figure 29: Mean magnetic field modulus of HD 126515 against rotation
phase. Asterisks represent Preston's (1970)
measurements, shifted by -750 G. The meaning of the other symbols is as
given in Table 5 (click here)
Figure 30: Mean longitudinal magnetic field of HD 126515 against
rotation phase. Data are from Preston (1970; open
squares), van den Heuvel (1971; crosses),
Mathys (1994b; filled circles) and
Mathys &\
Hubrig (1996; open circles)
longitudinal field determinations are available: in all of them, the longitudinal field varies nearly sinusoidally, while the field modulus may show significant anharmonicity.
Figure 31: Mean magnetic field modulus of HD 134214 against rotation
phase, computed assuming that the rotation period is 41456. The phase
origin is HJD 2447018.10. The meaning of the symbols is as given in
Table 5 (click here)
We had reported in Paper III the observation of resolved magnetically
split lines in HD 134214. Our 26 measurements of the magnetic field
modulus of this star, obtained at ESO and at KPNO,
are all clustered around 3.1 kG, with a standard
deviation of 65 G. Since our spectra of this fairly bright star have a
good S/N ratio and since 
is not heavily blended, we estimate
that the uncertainty of our field modulus determinations is
significantly lower than this standard deviation, hence that we are
detecting actual variations. Kreidl et al. (1994) were unable
to establish the rotation period of this rapidly oscillating Ap star from
their photometric observations. A period search conducted over our
entire dataset yielded a most probable value of 12844. It is
obviously spurious, as it would correspond to a very peculiar
distribution in phase of the observations: a cluster of scattered
measurements over 0.2 cycle, and a field constant throughout the rest
of the period. However, given the small amplitude of the variations,
the period search could easily be frustrated by small systematic
instrumental effects. To probe this eventuality, we repeated it on the
ESO 
data alone. The standard deviation of these 24
measurements is 63 G. The 12844 period still appears, but a second
peak corresponding to
also stands out in the periodogram. This period gives a much more
convincing match to the ESO data, but in order for the two KPNO
measurements to be consistent with it, one must assume that they are
systematically smaller than those of ESO by about 100 G (see
Fig. 31 (click here)). This cannot be definitely ruled out, but is not easily
reconciled with the absence of large systematic difference between the
KPNO and ESO data for the other stars observed from both sites. It can
be noted that if a shift of +100 G is applied to the KPNO
measurements, the standard deviation of the whole set is slightly
reduced, to 61 G.
In any event, the value of the period suggested above must be regarded at best as tentative. Should it be the actual period of rotation of HD 134214, that we see magnetically resolved lines would imply that the star's rotation axis makes a very small angle with the line of sight, a possibility also contemplated by Kreidl et al. (1994). On the other hand, the period analysis performed on the ESO data alone does not allow one to exclude even shorter rotation periods. Nevertheless, the constraint that they would impose on the geometry of the observation would be even more stringent. In relation with this, it may also be stressed that no significant Doppler effect is definitely seen in the line profiles.
Two attempts by Mathys & Hubrig (1996) to measure the longitudinal magnetic field of HD 134214 gave null results.
HD 137909 (= 
CrB) is one of the brightest and best studied Ap
stars. The presence of magnetically resolved lines in its spectrum has
first been reported by Preston (1969c). Systematic
determinations of the mean magnetic field modulus throughout the rotation
period have been carried out by Wolff & Wolff (1970). They
have subsequently been used by various authors, in combination with
longitudinal field measurements, to derive constraints on the magnetic
geometry of this star (see e.g. Landstreet 1980 and references
therein).
We have obtained a new set of measurements of the mean magnetic field
modulus of 
CrB from observations performed at OHP with the
AURELIE spectrograph and at KPNO. In Fig. 32 (click here), they are plotted
together with Wolff & Wolff's (1970) data against the
rotation phase calculated according to Kurtz's (1989)
ephemeris. Our measurements are seen to be in good agreement with those of
Wolff & Wolff (1970) close to magnetic maximum, but the
amplitude of the variations seemed larger from the latter than it appears
from our data. This discrepancy is probably mostly spurious, as indicated by
the large scatter of Wolff & Wolff's (1970) data around
field minimum. This scatter likely results from the difficulty of measuring
the rather weak field of 
CrB around that phase on photographic
spectra restricted to the blue region (where the Zeeman splitting is smaller
than in the red).
Figure 32: Mean magnetic field modulus of HD 137909 against rotation
phase. Asterisks represent Wolff & Wolff's (1970)
measurements. The meaning of the other symbols is as given in
Table 5 (click here)
Figure 33: Mean longitudinal magnetic field of HD 137909 against rotation
phase. Data are from Mathys (1994b; filled squares)
and Mathys & Hubrig (1996; open squares)
Considering our observations alone in Fig. 32 (click here), the mean field modulus variations are seen to be strongly anharmonic, with a rather sharp raise toward a fairly narrow maximum and a smoother decrease toward a broader minimum. However, an inconsistency between the KPNO and OHP measurements, significantly larger than the internal scatter of each of these separate sets about a smooth variation curve, is apparent. Because the KPNO measurements were obtained contemporaneously with the OHP data interleaved with them, the discrepancy cannot be attributed to an inaccuracy of the value of the period used nor to the occurrence of secular intrinsic changes in the stellar field. Also the good internal consistency of each of the datasets taken independently (for instance, as we mention in Sect. 6, the rms deviation of the AURELIE data about a fit by a sine and its first harmonic is 28 G) rules out the possibility of short timescale intrinsic variations of the stellar magnetic field. The most likely interpretation of the observed different behaviour of the measurements conducted at OHP and at KPNO is that it results from a systematic instrumental effect. As this effect appears to depend on the rotation phase of the star, it is most probably related to instrumental polarization. This is of course worrisome, since if this explanation is correct, the shape of the curve of variation of the mean field modulus may depend on the instrument used to obtain it, which limits its usefulness for the diagnosis of the geometrical structure of the stellar magnetic field. This point is discussed in more detail in Sect. 6.
Table 6: The mean longitudinal magnetic field of HD 137949
Notwithstanding this, there is no doubt that, as already pointed out
by Wolff & Wolff (1970), there is a very significant phase
lag between the extrema of the mean field modulus and those of the
longitudinal field: compare Fig. 32 (click here) with Fig. 33 (click here), where
the longitudinal field measurements of Mathys (1994b) and of
Mathys &\
Hubrig (1996) have been plotted against rotation phase. This implies
that the field of 
CrB is not symmetric about an axis passing
through the centre of rotation of the star: this star shows an
extreme example of this property which appears widespread among Ap
stars (e.g., Mathys 1993). Modeling of the geometry of its
field using the data presented here as well as Leroy's (1995b)
broadband linear polarization measurements is under way, following the
approach pioneered by Leroy (1995a). Preliminary results have
been presented by Wade (1995).
The mean magnetic field modulus of HD 137949 has not significantly varied since our discovery that this star has magnetically resolved lines in March 1991 (Paper II). Our 13 determinations of it over almost 4.5 years, all performed from spectra taken at ESO with the CAT and the CES, have a standard deviation of only 23 G. This gives a good idea of the accuracy and reproducibility that are achieved in the measurements reported in this paper.
The absence of detectable variation of the mean field modulus is consistent with the lack of photometric variability in more than 7 years of observation (Deul & van Genderen 1983), with the lack of detectable rotational sidelobes in the pulsation frequency spectrum of this rapidly oscillating Ap star (Kurtz 1991), and with the constancy of its broadband linear polarization over 3 years (Leroy 1995b).
The 2326 rotation period inferred by Wolff (1975) from the
consideration of her and van den Heuvel's (1971) longitudinal
field measurements is probably not significant. Indeed, the 5
measurements of van den Heuvel (1971), spanning 97 days,
have a standard deviation of
285 G, of the same order as their probable error: in other words, no
significant variation is detected. Similarly, although Wolff
(1975) does not quote the uncertainty of the nine longitudinal field
determinations that she obtained in one year, their standard deviation
of 149 G is of the same order as the typical uncertainty of such
measurements based on photographic spectra. The average and the
standard deviation of all the longitudinal field determinations
performed by various authors for HD 137949 are summarized in
Table 6 (click here). The columns are, in order, the reference of the paper
where the data are published, the first and last date of observation, the
number of measurements found in the considered reference, their
average, and their standard deviation. For Babcock's (1958)
single measurement, the latter is replaced by the estimated uncertainty.
Under the hypothesis that indeed no short-term variation is detected
in van den Heuvel's (1971) and Wolff's (1975)
data, all the existing measurements of the longitudinal field of HD 137949
are consistent with a slow monotonic increase of this quantity, at a rate of
G/yr, since its first determination by Babcock (1958).
This, combined with the other pieces of evidence mentioned above,
strongly supports the view that HD 137949 is rotating very
slowly. Under the assumption that its longitudinal field varies
roughly sinusoidally (which is not unusual among Ap stars), its
rotation period might be of more than 75 years: thus it would be of
the same order as or longer than the period of 
Equ, until now
the longest one of any known Ap star (see Sect. 5.38). Further
measurements, in particular of the longitudinal field, of HD 137949,
will be most useful to confirm the suggestion made here.
Figure 34: Mean magnetic field modulus of HD 142070 against rotation
phase. The meaning of the symbols is as given in Table 5 (click here)
Magnetically resolved lines have been discovered in HD 142070 in
1994. That this star is rotating rather quickly was soon recognized
from the contribution of Doppler effect to the profile of 
\
(clearly seen in Fig. 3 (click here)) and
the occurrence of significant variations of its mean field modulus
from night to night. To this date, 22 measurements of the latter have
been accumulated. Their frequency analysis unambiguously shows that
the rotation period of HD 142070 is
This makes HD 142070 the star with resolved magnetically split lines
with the shortest unambiguously determined
rotation period. This implies that the
star must be observed almost pole-on. Indeed, from the width of the
resolved line components, 
is constrained to be smaller
than 5 
, approximately. Assuming that the stellar radius is at
least 
, we derive for the inclination of the
rotation axis on the line of sight an upper limit 
.
That in these conditions the
modulus of the magnetic field varies by almost 10% of its average
value over a rotation cycle indicates that this field
must be very inhomogeneous. Note also the very asymmetric shape
of the variation curve of the the mean field modulus (see Fig. 34 (click here)).
Between our first and last observations of HD 142070, its radial
velocity has smoothly varied with a total observed amplitude of
2.5 
. The amplitude of the variation that has taken place since our
first observation must be larger, though, as the radial velocity
must have
passed through a minimum between July 1994 and February 1995,
when we have no spectra. In any case, our observations of this
spectroscopic binary only cover part of its orbital period.
Accordingly, this period must be significantly longer than 500 days.
Figure 35: Mean magnetic field modulus of HD 144897 against rotation
phase. The meaning of the symbols is as given in Table 5 (click here)
Figure 36: Mean magnetic field modulus of HD 150562 against heliocentric
Julian date. The meaning of the symbols is as given in Table 5 (click here)
Very little is known about HD 144897, a star in which the presence of
magnetically resolved lines is reported here for the first time. The
measurement of the field modulus in this star is difficult and not
very accurate, due to heavy blending of 
on both its blue and red
wings. Nevertheless, thanks to the fairly large amplitude of variation
of the field modulus, the rotation period can be unambiguously
determined from our 26 measurements of this quantity:
The corresponding phase diagram is shown in Fig. 35 (click here). Due to the
above-mentioned difficulty of measuring the field modulus accurately,
the shape of its curve of variation is not well defined. It however
appears less anharmonic than that of other stars.
One measurement of the longitudinal magnetic field of HD 144897 by Mathys &\ Hubrig (1996) yielded a value close to +2.0 kG at phase 0.794.
About all that was known of HD 150562 until now is that it is rapidly oscillating (Martinez & Kurtz 1994). We report here that it has spectral lines resolved into their magnetically split components. We have observed this star for a little more than one year. During this time, its mean field modulus has shown a slow, monotonic increase (Fig. 36 (click here)). Thus the rotation period must be significantly longer than 1 yr.
The peculiarity of HDE 318107 has been discovered by North & Cramer
(1981) through a study of the open cluster NGC 6405 in Geneva
photometry. The observations reported here seem to be the first
spectroscopic observations of this star, from which we report the
detection of magnetically resolved lines. Due to its faintness,
most of our spectra of this star are rather noisy (the S/N ratio is
most often of the order of 70). Furthermore, 
is rather severely
blended, especially on the blue side. Accordingly, our measurements of
the mean magnetic field modulus are not very accurate: we estimate
that their uncertainty is of the order of 300 G. This is
still significantly
less than the standard deviation of the whole set of measurements,
1065 G: the latter must reflect real variations. They appear
inconsistent with the value of the rotation period favoured by North
(1987) from a study of photometric variability. Reanalyzing the
photometric data (kindly made available by P. North) together with
our magnetic measurements, we were unable to find a single period
consistent with both types of observations.
Obviously, more data are required. Consideration
of the longitudinal field may prove very helpful in that respect. One
measurement of that moment has been reported by Mathys & Hubrig
(1996): +2.0 kG.
Figure 37: Mean magnetic field modulus of HD 165474 against rotation
phase, computed assuming that the rotation period is 254065. The phase
origin is HJD 2447000.0. The meaning of the symbols is as given in
Table 5 (click here)
Resolved magnetically split lines have been discovered in HD 165474 by
Preston (1971a), who determined that the field modulus was
7.2 kG from the observation of Zeeman doublets. This value seems
significantly larger than our own measurements, which range from 6.1 to
6.8 kG. However, from magnetic broadening of unresolved lines, Preston
(1971a) had derived a field strength of 6.6 kG, more consistent with
our data. The latter have a standard deviation of 139 G. This almost
certainly indicates that we are detecting real field variations, since
HD 165474 appears as one of the most favourable cases for the accurate
measurement of the mean field modulus. Our measurements are
inconsistent with the two values of the rotation period favoured by
Leroy (1995b) from his broadband linear polarization
measurements (2338 and 1202). Accordingly, we performed a frequency
analysis on our data alone. One value of the rotation period seems to stand
out rather clearly:
The corresponding phase diagram is shown in Fig. 37 (click here). One
measurement is seen to show a large deviation: it corresponds to a very
noisy spectrum, which may explain its behaviour. On the other hand, the ESO,
OHP and KPNO data all appear reasonably consistent with this period. But
because of the small amplitude of the variations, and because the
visibility of magnetically resolved lines requires a rather peculiar
geometry of observation with such a short period, we feel that this
value of the period must be confirmed by additional observations
before it can be regarded as definitive. Leroy's (1995b)
linear polarization data are not useful in that respect, due to their
unsuitable distribution in time. Published longitudinal field
measurements, on the other hand, are somewhat puzzling: while
Babcock (1958) had
reported a non-negligible positive value (+900 G), three null
measurements have been obtained by Mathys (1994b) and by
Mathys &\
Hubrig (1996), the first one in 1988 and the last two in 1992, at an
interval of two days. This appears to be purely coincidental:
preliminary visual examination of a spectrum recorded recently (end
of July 1996) in both circular polarizations clearly indicates the
unquestionable
presence of a quite sizeable longitudinal field. Thus 
in
HD 165474 turns out to show significant variability. Additional
determinations of this field moment should consequently prove most
useful to derive a definite value of the rotation period.
Figure 38: Mean magnetic field modulus of HD 166473 against heliocentric
Julian date. The meaning of the symbols is as given in Table 5 (click here)
We report here the discovery of the presence of resolved magnetically split lines in the rapidly oscillating (Kurtz &\ Martinez 1987) Ap star HD 166473. Our measurements of the magnetic field modulus of this star are plotted against Julian date in Fig. 38 (click here). It can be seen that the field was at its maximum at the time of discovery of the resolved lines. It remained fairly constant during our first observing season, and since then, it has been steadily decreasing. The relative amplitude of the variation is definitely one of the largest among the stars considered here. Minimum does not seem to have been reached yet: the rotation period must significantly exceed the time spanned by the observations obtained until now (3.2 yr).
Three measurements of the longitudinal field over a time interval of 64 days are reported by Mathys & Hubrig (1996): all yielded a field close to -2.1 kG.
HD 177765 is another star about which almost nothing is known. Since we found that it has magnetically resolved lines, we have obtained six measurements of its mean field modulus, spanning almost two years. Their standard deviation is only 19 G: the field has shown no variation during the interval covered by the observations. HD 177765 may well be another star with a very long period: this should be established by future observations.
Figure 39: Mean magnetic field modulus of HD 187474 against rotation
phase. The meaning of the symbols is as given in Table 5 (click here)
Didelon (1987, 1988) has discovered the presence of resolved magnetically split lines in HD 187474. This star has a period of rotation of 2345 d, derived by Mathys (1991) from his own and Babcock's (to be also found in Mathys 1991) longitudinal field measurements. This field moment varies between -1.8 (at phase 0.5) and +1.8 kG (at phase 0.0). The value of the period derived from its consideration is supported by photometric observations (Hensberge 1993). Our magnetic field modulus data are plotted against the phase computed using Mathys' (1991) ephemeris in Fig. 39 (click here). The anharmonicity of the variation curve is remarkable, all the more because the variation of the longitudinal field is almost perfectly sinusoidal (see Fig. 33 (click here) of Mathys 1991). Observations in the next two years will be important, as they will fill the gap between phases 0.37 and 0.66 in the field modulus curve, and as they should in particular constrain the maximum of this quantity.
Note also that HD 187474 is a spectroscopic binary with an orbital period of 690 d (Leeman 1964), much shorter than its rotation period.
Figure 40: Mean magnetic field modulus of HD 188041 against rotation
phase. The meaning of the symbols is as given in Table 5 (click here)
Preston (1971a) had been the first to observe magnetically resolved lines in HD 188041. Our 15 measurements of its field modulus have a standard deviation of 37 G, part of which appears to be due to the detection of real variations of very low amplitude (probably not much more than 50 G peak-to-peak). Indeed, there is some hint in Fig. 40 (click here) at a systematic variation of the mean field modulus with the phase computed with the rotation period of 2239 derived by Hensberge (1993). The field modulus appears to be maximum close to phase 0, and minimum close to phase 0.5. These extrema are close to those of the longitudinal field (see Fig. 36 (click here) of Mathys 1991). Measurements of the latter have been performed by Babcock (1954, 1958), Wolff (1969), Mathys (1994b), and Mathys & Hubrig (1996). It is always positive and varies with a peak-to-peak amplitude not much larger than 0.5 kG, which is qualitatively consistent with the low amplitude of the field modulus variations. Note also that significantly variable broadband linear polarization is measured in this star (Leroy 1995b).
From his broadband linear polarization measurements, Leroy (1995b) has derived a value of 64186 for the rotation period of HD 192678, a star in which the detection of magnetically resolved lines has first been reported in Paper II. Our magnetic field modulus measurements show a definite trend with this period, but there appears to be a systematic shift between those obtained at OHP with AURELIE and those obtained at KPNO. Since both sets are well distributed throughout the rotation cycle, this difference can be estimated just from the comparison of the mean of the measurements of each set. For the 20 OHP measurements, this mean is 4831 G, while for the 14 KPNO points, it is 4668 G. Accordingly, in order to bring all the data to the same scale, we subtracted 163 G from each OHP measurement.
Figure 41: Mean magnetic field modulus of HD 192678 against rotation
phase. The meaning of the symbols is as given in Table 5 (click here). The
AURELIE data have been shifted by -163 G
We then performed a period search on the whole set of our
measurements, modified as just explained. Two values of the period
stand out: 56237, which is quite inconsistent with Leroy's
(1995b) linear polarization data and must thus be ruled out, and
d, which is in excellent agreement with Leroy's
(1995b) value. The latter is used to compute the phases of our
observations. The resulting phase diagram is shown in Fig. 41 (click here).
As for other short-period stars discussed in this paper, the axis of
rotation of HD 192678 must make a small angle to the line of sight, so
that resolved lines can be observed despite the rather fast
rotation. As a matter of fact, Doppler distortion is definitely seen
in the split components of 
. A detailed analysis of the magnetic
geometry of this star, taking into account not only the present field
modulus data and Leroy's linear polarization measurements, but also
the longitudinal field determinations of Babcock (1958) and
Glagolevskij et al. (1986), is presented in a separate paper
(Wade et al. 1996a).
The presence of resolved magnetically split lines in HDE 335238 has first been reported by Preston (1971a). The distribution of our 16 measurements of the mean field modulus of that star along its rotation period appears particularly unfortunate, since while 14 of them lie between 7.8 and 8.8 kG, the other two determinations yield much higher values: 11.2 and 11.7 kG. Although this unusual distribution of the measurements may lead one to wonder whether the two ``discrepant'' high field values are not spurious (which might seem even more plausible because most of our spectra of this fairly faint star are rather noisy), their correctness is unquestionable. Indeed, in the two spectra from which a field of more than 11 kG is derived, all spectral lines do consistently show larger magnetic splitting than in our 14 other observations. As a matter of fact, Preston (1971a) had already reported field modulus values ranging from 9.1 to 11.8 kG in HDE 335238, in excellent agreement with the measurements presented here.
As a result of the poor phase distribution of our data, our attempts to determine the star's rotation period from their consideration suffer from severe aliasing problems. It appears that the period of HDE 335238 must undoubtedly be between 40 and 50 days, but there are several
Figure 42: Mean magnetic field modulus of HDE 335238 against rotation
phase, computed assuming that the rotation period is 440. The phase
origin is HJD 2447000.0. The meaning of the symbols is as given in
Table 5 (click here)
possibilities in that range between which one cannot
unambiguously choose. The value
seems marginally more probable than the others (the phase diagram
obtained with this period is shown in Fig. 42 (click here)). But one cannot
definitely rule out the aliases 494, 479, and 428.
Obviously, more observations of this star are required to settle the
issue of its period.
Babcock (1967) had mentioned that HDE 335238 is magnetic, probably on the basis of spectra taken in circular polarization, but the only published determination of its mean longitudinal field has been performed by Mathys & Hubrig (1996), who found a value of -1.3 kG.
The first report in the literature of the presence of resolved
magnetically split lines in HD 200311 is Adelman's (1974)
mention of a private communication from Preston. However, this property of
HD 200311 had apparently been forgotten until it came serendipitously
to our attention (Paper II). The star brightness is at the limit for
observation with a 0.9 m telescope, so that our KPNO spectra of it are
rather noisy. Furthermore, since it is hot, the line 
is fairly
weak. Therefore our mean field modulus measurements, especially those
performed from KPNO spectra, are not quite so accurate as for most
stars studied here. However, the amplitude of the field variations is
large enough so that the stellar rotation period can be uniquely
determined from the field modulus data. We find:
Figure 43: Mean magnetic field modulus of HD 200311 against rotation
phase. The meaning of the symbols is as given in Table 5 (click here)
We used this period to build the phase diagram shown in Fig. 43 (click here). One clearly sees in this figure that the KPNO data are more scattered than those from OHP about a smooth variation curve. This reflects the limits of accuracy mentioned above.
Like HDE 335238, HD 200311 was one of the stars reported to be magnetic by Babcock (1967). More recently, measurements of its longitudinal field have been performed at the Special Astrophysical Observatory of the USSR Academy of Sciences (Glagolevskij et al. 1986) and at the Elginfield Observatory of the University of Western Ontario (to be published separately). They range, roughly, between -2.5 and +2.5 kG: HD 200311 appears to have a strong, polarity reversing longitudinal field.
We have confirmed in Paper I Scholz's (1979) probable
detection of resolved magnetically split lines in HD 201601 (=
Equ). It is well known, mostly from the longitudinal field
measurements, that this star has a period that must exceed 70 years (see
e.g. Mathys 1991). It has sometimes been questioned whether
this period really corresponds to the stellar rotation, or if some
other effect might be involved. A compelling argument in support of
the interpretation in terms of stellar rotation has recently been
presented by Leroy et al. (1994), on the basis of their
observations of broadband linear polarization.
Figure 44: Mean magnetic field modulus of HD 201601 against heliocentric
Julian date. The meaning of the symbols is as given in Table 5 (click here)
The behaviour of the mean magnetic field modulus over the 7 years during which we have repeatedly measured it is also consistent with the long period mentioned above: indeed, we mainly observe a slow, monotonic increase (Fig. 44 (click here)). There is some hint of a flattening of the curve in the last measurements, which may suggest that we are getting close to the field maximum. This is difficult to ascertain though, all the more because like in other stars discussed in this paper (e.g., HD 137909, HD 192678), there also seem to be some systematic differences between measurements obtained with different spectrographs (with in particular the AURELIE data somewhat larger than the others).
In the region of 
, the spectrum of HD 208217, in which the
presence of magnetically resolved lines is reported here for the first
time, bears some resemblance with that of 53 Cam (HD 65339): the lines
are affected by a fairly large rotational Doppler effect, but the
magnetic field is strong enough so that magnetic line splitting is
observed. However, as in 53 Cam, rotation seriously complicates the
determination of the mean magnetic field modulus. The difficulty is
even aggravated by the fact that the blue component of 
is heavily
blended. These circumstances severely limit the accuracy achievable in
the diagnosis of the field modulus.
Figure 45: Mean magnetic field modulus of HD 208217 against rotation
phase. The meaning of the symbols is as given in Table 5 (click here)
A first determination of the rotation period of HD 208217 has been
made by Manfroid & Renson (1983) from Strömgren
photometry. New photometric observations, combined with the magnetic data
reported here, have allowed Manfroid & Mathys (1996) to
refine this determination. They found
a value which we use here to compute the phases of our
observations. The resulting phase diagram for the mean magnetic field
modulus is shown in Fig. 45 (click here).
Though as a result of the measurement intricacies mentioned above, it
is rather noisy, there is a clear indication of a double wave in the
curve of variation of the field modulus (Manfroid & Mathys
1996 even found the contribution of the first harmonic to be
marginally more significant than that of the fundamental). HD 208217 is the
first star with resolved magnetically split lines known to show this
behaviour, which should be widespread if the magnetic fields of Ap stars
generally were centred dipoles. (This does not imply that the field of
HD 208217 is a centred dipole.)
Our observations furthermore show that the radial velocity of
HD 208217 is variable, with a peak-to-peak amplitude which is at least
of the order of 20 
. The orbital period of this spectroscopic
binary is presently unknown. A value of 751 cannot be definitely ruled
out, but it appears more likely that the period is at least
of the same order as the time span of our observations (2 years).
For a long time after the discovery that HD 216018 has resolved magnetically split lines (reported for the first time here), no variation of the mean field modulus of this star has been observed. Only during the last observing season has it started to show some trend to increase. This is illustrated in Fig. 46 (click here). It appears probable that HD 216018 has a very long rotation period (much longer than the 3 years covered by our observations) and that we have first observed it close to its magnetic minimum. More observations should allow us to confirm this.
Figure 46: Mean magnetic field modulus of HD 216018 against heliocentric
Julian date. The meaning of the symbols is as given in Table 5 (click here)
However, the long period hypothesis already receives some support from the three determinations of the longitudinal field performed by Mathys &\ Hubrig (1996), 2 in 1992 on 2 consecutive nights and 1 almost one year later, which all yield essentially the same result (+1.3 kG).
The radial velocity of HD 216018 shows a behaviour very similar to the
mean field modulus: it has remained essentially constant over most of
our observations, and it has started a slow decrease (by a
couple of 
) during the last observing season.