In this paper we have analysed, using the spectral subtraction technique, high resolution echelle spectra of 16 chromospherically active binary systems. These spectra include all the optical chromospheric activity indicators from the Ca II H & K to Ca II IRT lines, and in some observing runs the Li I $\lambda$ 6707.8 line too.

H $\alpha$ emission above the continuum is observed in UX Ari and HU Vir, in the rest of the systems excess H $\alpha$ emission is clearly detected in the subtracted spectra. Filled-in absorption H $\beta$ line profiles have been found in all the stars of the sample except LR Hya and BL CVn, which also have a lower level of activity in the other chromospheric activity indicators.

Very broad and variable extended wings have been found in the subtracted H $\alpha$ profile of the very active star HU Vir. These line profiles are not well matched using a single-Gaussian fit, and have been fitted using two Gaussian components (narrow and broad). Similar behaviour has been found in other very active stars (Hatzes 1995; Jones et al. 1996; Stauffer et al. 1997; Papers I, II, Montes et al. 1998b). We have interpreted this broad component emission as arising from microflaring (high turbulence emitting plasma) activity that takes place in the chromosphere by similarity with the broad components found in the chromospheric Mg II h & k lines (Wood et al. 1996; Busà et al. 1999) and in several transition region lines of active stars observed with GHRS-HST (Linsky & Wood 1994; Linsky et al. 1995; Wood et al. 1996, 1997; Dempsey et al. 1996a,b; Robinson et al. 1996) and recently confirmed with STIS-HST observations (Pagano et al. 2000). In some cases the H $\alpha$ line is asymmetric and the fit is better matched when the broad component is blue-shifted or red-shifted with respect to the narrow component. These asymmetries are also observed during the impulsive and gradual phases of solar and stellar flares (Montes et al. 1996b, 1999; Montes & Ramsey 1999), and favour the interpretation of the broad component as due to upward and downward motions produced by microflaring in the chromosphere.

Excess absorption and emission is observed in the wings of the H $\alpha$ and H $\beta$ lines of several systems. Absorption features are detected by us in the blue wing of V1149 Ori and in the red wing of 12 Cam (also observed by Eker et al. 1995), IL Hya (also detected by Weber & Strassmeier 1998) and HU Vir (also seen by Strassmeier 1994a, and Hatzes 1998). We detected excess emission in the blue wing of FG UMa too. Similar red-shifted absorption features were already seen by other authors in the single star OP And (Fekel et al. 1986) and in the binaries VY Ari (Bopp et al. 1989); XX Tri (Bopp et al. 1993) and IN Vir (Strassmeier 1997). Blue-shifted emission was also detected in the single giant YY Men (Vilhu et al. 1991). Several dynamical processes, or a combination of them, could be the origin of these blue-shifted or red-shifted absorption features: plage- and prominence-like structures (Hall & Ramsey 1992, 1994; Neff 1996; Eibe 1998); continuous low-level mass infall (Walter & Byrne 1998; Walter 1999; Eibe et al. 1999); local velocity fields and mass motions due to magnetic field inhomogeneities possibly coupled with a loop-like geometry (Strassmeier 1994a); fluctuations of both the column density and temperature gradient within the chromosphere (Smith & Dupree 1988); stellar winds and anti-winds (Linsky et al. 1995).

Prominence-like extended material viewed off the limb has been detected in many stars of the sample according to the high ratios $E_{\rm H\alpha}/E_{\rm H\beta}$ obtained. Prominences viewed against the disk seem to be present at some orbital phases in the dwarfs OU Gem and BF Lyn.

The application of the spectral subtraction reveals that the He I D₃ line appears as an absorption feature (Fig. 24) in mainly all the giants of the sample as in the case of the stars analysed in Paper I. Total filling-in of He I D₃is observed in the very active star HU Vir, similar to the behaviour observed in EZ Peg (Paper II). These results are in agreement with the behaviour reported by Saar et al. (1997). These authors found that while for few active stars the He I D₃ line behaves "normally'', increasing in absorption with increasing rotation, and showing consistent correlations with other activity indicators, the behaviour clearly diverges (large filling-in) when stars become very active, suggesting that the line could be filled-in due to frequent low-level flaring. In the most evolved stars the behaviour is different as a consequence of the lower chromospheric densities of these stars.

Ca II H & K emission is observed in all the stars of the sample. Small emission in the close H $\epsilon$ line is also detected in some of the more active stars. In some systems like OU Gem, BF Lyn, LR Hya, DK Dra and MS Ser the emission from both components is clearly deblended in these lines. Self-reversed absorption core with red asymmetry ( $I({\rm K}_{2{\rm V}}) < I({\rm K}_{{\rm 2R}}$ )) is detected in the Ca II H & K lines of the giants 12 Cam, FG UMa and BM CVn. The self-reversed feature is a consequence of the line formation process in the chromosphere (depth variation of the line source function in an atmosphere having a chromospheric temperature rise). Asymmetries in these profiles provide information on velocity fields in the line formation regions. In these three giants we observed a small red asymmetry (indicative of outward mass flux, wind), contrary to the blue asymmetry (indicative of upward propagating waves, but not large wind) normally observed in giant stars hotter than spectral type K3 (Stencel 1978) and also observed by us (FFMCC) in the giants (V1817 Cyg and V1764 Cyg).

The Ca II IRT lines result to be a very useful chromospheric activity indicator too. We have found that all the stars analysed here show a clear filled-in absorption line profile or even notable emission reversal (UX Ari, OU Gem, BF Lyn, IL Hya, HU Vir, BQ CVn, BM CVn, MS Ser). Thanks to the higher resolution of our spectra, we were able to detect emission reversal in the Ca II IRT lines in some systems in which previous studies (Linsky et al. 1979; Dempsey et al. 1993a) only reported a filling-in. An increase in the level of activity of these stars could also be the cause of these detections. When both components of the binary system are active the excess emission detected in the Ca II IRT lines is much more easily deblended thanks to the large wavelength of these lines (see e.g. DK Dra, Fig. 18). We have found $E_{\rm 8542}/E_{\rm 8498}$ ratios in the range $\simeq~1-2$ which is indicative of optically thick emission in plage-like regions, in contrast with the prominence-like material inferred by the $E_{\rm H\alpha}/E_{\rm H\beta}$ ratios. These small $E_{\rm 8542}/E_{\rm 8498}$ ratios, also found by Chester et al. (1994) and Arévalo & Lázaro (1999) in other active binaries, indicate the existence of distinct sources of Balmer and Ca II IRT emission and suggest that the activity of these very active stars is not simply a scaled-up version of solar activity.

Some systems have been observed at different orbital phases and different epochs, covering all the orbital cycle and it was possible to study the variability of the chromospheric emission. The excess H $\alpha$ and Ca II IRT emission of BF Lyn in the McD98 observing run shows anti-correlated variations with the orbital phase (see Fig. 8). This anti-correlation could indicate that the chromospheric active regions are concentrated on faced hemispheres of both components but at about 0.4 and 0.9 orbital phases for the hot and cool component, respectively. Evidence of an active longitude area (0.29-0.44 orbital phase interval) has been found in HU Vir when we compare with the higher level of activity in this phase interval also reported by Strassmeier (1994a), in 1991, and Hatzes (1998), in 1995.

$\begin{figure} \par\includegraphics[height=10cm,width=8.7cm,clip]{ds1878f25.ps}\end{figure}$

Figure 25: Spectra of OU Gem, BF Lyn and HU Vir in the Li I $\lambda$ 6707.8 line region. The expected position of the Li I line for each star is marked with a short vertical line. We have also plotted a K1III reference star with some photospheric lines identified

5 Discussion and conclusions