The heliocentric velocities, velocity dispersions (b -values), and Ca II column densities were obtained for each velocity component by means of a Marquardt least-squares fitting program (described in more detail by Vallerga et al. 1993), and these are given in Table 2. The errors quoted on these values were obtained by an interactive estimation of the range of each parameter about the least-square value which is consistent with an acceptable overall fit. Although somewhat subjective, these error estimates are at least conservative (in most cases they are two to three times the formal 1 errors obtained by the least-square fitting program which, in order to reduce the computation time to a manageable level, made the simplifying assumption that the errors on each component can be treated independently).
Table 2. Line profile parameters for the interstellar Ca K lines shown in Fig. 1, compared with other published values; a colon indicates that a previously published value is uncertain. is the total equivalent width (i.e. summed over all velocity components; 2 errors). Column 7 indicates whether a component falls within the velocity range occupied by the LIC or G clouds (Table 1; a question mark indicates that the component falls just outside the errors quoted on the cloud velocities), and also identifies possible circumstellar components (see text).
Table 2 also compares our results with those obtained by other authors at lower spectral resolution. It will be seen that, as expected, the new results mainly concern the discovery of velocity structure within components previously thought to be single, and the accurate measurement of velocity dispersions for components not properly resolved in earlier work. In this section we outline these new results, deferring a more detailed discussion of some of their implications to Sect. 4.
3.1. Equivalent widths
The equivalent widths were measured using the DIPSO spectral analysis program (Howarth, Murray & Mills 1993), and generally agree well with previously published values. This is reassuring because it indicates that the present observations have not been significantly affected by uncorrected scattered light, which would have caused our measurements to underestimate the actual values. This problem was noted in the earliest observations obtained with the UHRF, but it appears that the steps taken to correct it (discussed in Sect. 7.3 of Diego et al. 1995) have been successful.
3.2. Velocity structure
In Table 2 we use curly brackets to group together velocity components which we have resolved, but which lie within what appeared to be single components when observed at lower resolution. Six such cases were identified, towards four of the eight stars: (i) Cen.
The -8.1 km s-1 component observed by Lallement et al. (1986) with a resolution of 3 km s-1 is here resolved into two components, at -9.2 and -7.7 km s-1. (ii) Cen.
Earlier observations, obtained with a resolution of 3.6 km s-1 (Crawford 1991), identified two velocity components towards this star, whereas the present observations reveal five. The component previously identified at -12.5 km s-1 is here split into two (-12.9 and -8.1 km s-1), which may plausibly be identified with the G and LIC clouds (see Sect. 4.1). In addition, the earlier -0.5 km s-1 component is here resolved into three discrete components (at -1.9, +0.3, and +3.4 km s-1). Also, the unusually large b -value obtained for the +3.4 km s-1 component (4.1+-10:9:1km s-1 ; Table 2) may indicate the presence of additional unresolved velocity structure. (iii) Oph.
The single component observed at -31.1 km s-1 by Lagrange-Henri et al. (1990) is here resolved into components at -33.0 and -29.9 km s-1. This structure will be discussed more fully in Sect. 4.2. (iv) 51 Oph.
The Ca K line towards 51 Oph was observed at 3 km s-1 resolution by Lagrange-Henri et al. (1990), who identified three discrete velocity components (cf. their fig. 5). However, the higher-resolution observations presented here show that at least two of these are actually double, making a minimum of five components in all (Table 2). Lagrange-Henri et al.'s -26.9 km s-1 component is resolved into two (at -29.0 and -25.2 km s-1), which correspond to the velocities expected for the G and LIC clouds (see Sect. 4.1). In addition, the narrow central component (at -21.3 km s-1) is found to be double, and this will be discussed in Sect. 4.2.
3.3. Velocity dispersions
The velocity dispersions (b -values) found here are generally consistent with other published values (Table 2), when allowance is made for the lower resolution of earlier work and the previously unresolved blends discussed above. However, it is important to stress that the values measured here are much more accurately determined, owing to the order-of-magnitude higher resolution employed. As the observed line profiles are given by a convolution of the intrinsic profiles with the instrumental resolution, line profile modelling is insensitive to intrinsic velocity dispersions much smaller than the instrumental b -value (). The previously published measurements referenced in Table 2 were all performed with km s-1 (). This is comparable to the intrinsic widths of the lines being studied, and renders the resulting measurements somewhat uncertain. In contrast, the observations presented here were obtained with km s-1, with the result that all the velocity components have been fully resolved and therefore have well-determined velocity dispersions.
© European Southern Observatory (ESO) 1997
Online publication: July 8, 1998