... stars[*]
Based on observations collected at the European Southern Observatory (La Silla, Chile), using Director Discretionary Time. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... transition[*]
Such a dependence had been found by Stenflo & Lindegren (1977) for the parameter they used to characterize the width of the Fe I lines analysed in their study. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... pulsation[*]
Pulsation may appear as line broadening provided that the integration time used to record the analysed spectrum is long compared to the pulsation period. This is not unlikely for rapidly oscillating Ap stars, considering that their periods are short (5-21 min). Nevertheless for these stars another complication may arise, because different lines of a given ion may have different pulsation amplitudes (e.g., Kurtz et al. 2005). In such cases, the line-to-line variation of the pulsational broadening does not have the same simple dependence as the other Doppler contributions to the line widths. Unless the measured values of $\ensuremath {R_{I}^{(2)}(\lambda _I)} $ can be corrected (using the results of a detailed pulsation study) prior to the magnetic analysis, its presence may render the latter impossible. In particular, although this was not recognised at the time, the effect just described is almost certainly the reason why Cowley & Mathys (1998) could only derive a nearly meaningless upper limit of 8.9 kG in their attempt to determine the quadratic field of HD 101065. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
...).[*]
We do not consider here the upper limit of \ensuremath{v~\sin i} derived from the analysis of the lines of Fe I, because the smaller size of the sample of these lines likely makes the results obtained from their consideration less reliable (see also Sect. 4.2). 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
...)[*]
For the Fe II lines of HD 91375, since a2 is negative, the upper limit of the quadratic field given in Table 4 corresponds to $3\sigma(a_2)$ (for the form of the regression including a constant term). 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... observation[*]
Note that a field modulus of this order is fully consistent with the absence of any visible distortion of the profile of the line Fe II  $\lambda~6149$ in the high-resolution spectrum of this star shown in Fig. 1 of Mathys (1990a). 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Copyright ESO 2006