For all the objects in Table 1, absolute magnitudes are derived according to the corresponding distance modulus. In Fig. 1, oscillation amplitudes are presented versus absolute magnitudes. The points marked with diamonds correspond to HD73575, HD73819 and HD28319. Strong evidence exists that the two former objects are no longer on the main sequence, but rather in or just after the phase of overall contraction associated with the exhaustion of hydrogen in the core. The latter is Tau. Its evolution stage is ambiguous and it is known to be a binary star.
This figure indicates the existence of a correlation between oscillation amplitude and absolute magnitude for stars on the main sequence. The relation can be fitted by the line of
The trend does not appear when considering the complete set of field stars given in Rodriguez et al. (1994). This might be explained by the following fact, in the sample stars belonging to clusters described in Table 1, the internal errors on absolute magnitudes are smaller. Error bars on the absolute magnitude determination are estimated to be approximately mag. As for error bars on oscillation amplitude, two factors are involved, the length of the available time series as already commented in Sect. 2 and the intrinsic measurement error which we estimate to be mag.
The amplitude correlation trend line predicts that the lowest luminosity of Scuti variable which can have non-zero amplitude is about . Corresponding to this luminosity on the Main sequence, the spectral class is about F5, which is five subclasses away from the observational cool border of Scuti stars on the main sequence F0V (Breger 1979). So according to our result, if we can improve the observational accuracy high enough, the observational cool border of Scuti on the main sequence can move beyond of F0 V.
If our statistical result is real, then this suggests that brighter stars should preferentially have larger amplitudes. How does this result affect the Scuti variables distribution in the instability strip? The following conclusion can be drawn. If the idea that all stars located in the instability strip are variable is true, then our result suggests that there is a greater possibility of detecting the variability in stars of higher luminosity, given present detection limits, and that there will be more observational nonvariables near the low or cool part, but this is in contradiction with the observational results (Breger 1979, Baglin et al. 1973). So our trend seems to support the existence of real nonvariables in the instability strip. Maybe, more accurate observations and statistical research on the subject is necessary.
Fast rotation is very common among Scuti stars and it is known that rotation significantly changes the estimation of the absolute magnitude (Michel et al. 1998). This phenomenon is probably the reason for a large amount of the dispersion observed in in Fig. 1. A further investigation of this aspect will probably be necessary to obtain a more precise description of the trend presented here. We here limit ourselves to the presentation of the oscillation amplitudes versus values (Fig. 2). A weak correlation might be noticed in Fig. 2.
Effective temperature is pointed out by linear stability analyses as a key parameter in deciding which modes are unstable (Houdek et al. 1995). Fig. 3 presents the oscillation amplitude versus the index, representative of the effective temperature for A and early F stars on the main sequence. No correlation is noticed.
© European Southern Observatory (ESO) 1999
Online publication: March 18, 1999