4 Results

4.1 Description of tabular results

Individual velocity measurements are given in Table 7 (available in electronic form only; see footnote to title page). The columns are: HIP and HD numbers, heliocentric Julian Day for the middle of the integration time JJ, radial velocity v_i and its error $\sigma_{i0}$, $T_{\rm eff}$ and rotation parameters of used synthetic spectrum and difference $dv_{\rm gp}$ between Gaussian and parabolic fitting.

When there are several observations for the same star, the average of the radial velocities $\overline{v}$ and its rms $\sigma_{\rm e}$ are computed.

$$\sigma_{\rm e}=\sqrt{1/(n-1)\,\sum{(v_{i}-\overline{v})^{2}}}.$$

Table8 (available in electronic form only; see footnote to title page) gives in order, the HIP and HD numbers, ICRS coordinates (consistent with J2000) from the Hipparcos catalogue, V (Johnson) magnitude and its source from the Hipparcos catalogue ("G'': ground-based photometry, "H'': from $H_{\rm p}$ and a colour index, "T'': from Tycho photometry), the spectral type from the CDS (Centre de Données Astronomiques de Strasbourg), the mean radial velocity $\overline{v}$, its rms $\sigma_{\rm e}$, its error $\epsilon_{\rm e} = \sigma_{\rm e}$$\sqrt{n}$,the number of observations n, the mean adopted internal error $\sigma_{i}$, the ratio E/I of external to internal error, the indication of radial velocity variability (Chi-square test), a number function of the shape of the peak obtained by correlation with minimum rotation and described in Table4, "R'' if a remark function of the object is given in Table6 and finally Hipparcos flags of possible duplicity as described in Sect. 4.2.3.

 

Table 4: Meaning of the flag (Cols. 106-107 of Table 8 and 58-59 of Table 9)

 

Table 5: Stars with no radial velocity. "R" refers to Table 6

 

Table 6: Notes about peculiar and multiple stars, these last stars belonging to the MSC (catalogue of physical multiple stars of Tokovinin 1997). $\lambda$Boo-type stars are taken in Paunzen et al. 1997 (1), Gray 1997 (2) and Abt & Morell 1995 (3)

4.2 Binary and multiple stars

An important number of suspected binaries are found in our sample by the shape of the peak. A lot of these binaries are physical, but there are optical systems also. So it is important to compare with two other criteria: the variability test upon the radial velocity and the data of Hipparcos mission. It is obvious that the conditions of detection are not the same for the three criteria. In fact they are complementary in terms of separation, difference of magnitude and orbital period. When the shape of the correlation peak is the only investigation mean, systematic simulations help in explaining the system of components.

4.2.1 Variable radial velocities

The width of the slit is $1.5\,\hbox{$^{\prime\prime}$}$ on the sky. So when the separation $\rho$ of components is higher than $2\,\hbox{$^{\prime\prime}$}$, the star is always seen as single. Using the Chi-Square test, the probability of variability with a confidence level of $99\%$ was computed. In these cases, this information is given in Cols. 102-104 by "var'' in Table8. This test of variability depends on the dispersion of radial velocities around the mean value and the internal error, i.e. $T_{\rm eff}$, vsini and possible peculiarity of the star.

4.2.2 Non Gaussian peaks

The interpretation of the shape of the peak is carried out by simulations as seen in Sect. 2.5. Detection of a binary depends on difference of radial velocities between the two components, difference of parameters $T_{\rm eff}$ and vsini and ratio of their intrinsic luminosities. Figure 1 gives an example of the simulation of a binary peak: $T_{\rm eff}$ are 8500 and 6500K and rotation 100 and 10kms^-1 respectively, their ratio of intensities is 22.2 and their difference of radial velocities is 32.07kms^-1. A faint late-type component is detectable in an early-type peak until a difference of magnitude of 4.5. So with propitious conditions, a component G0V can be detected on the peak of a A0-type star with rotation 200kms^-1. First a little tip is seen on the top of the peak, then it is replaced by an asymmetry at the bottom if the ratio of their luminosities increases. If the two components have close parameters, detection depends on their rotation and radial velocities difference essentially. First a deformation of the top of peak is obvious, then it is replaced by a bulging of sides if their rotation increases.

  

Figure 1: Simulation (in dashed line) of the double correlation peak of HD 6668 (in full line)

If the faint secondary peak was well defined, its radial velocity was measured by correlation with a cool synthetic spectrum ($T_{\rm eff}\,=$6000K and low rotation) and is quoted "b'' in Table7 after the HD number. It has been verified that the lunar light was not responsable. When peaks are well separated, the two are measured and results are given in Table7, the most important peak being quoted the component "a''. The average radial velocity is not computed and the HIP and HD numbers, coordinates, V magnitude, the flag for the peak shape and the flags of Hipparcos catalogue are given in Table9, as in Table8.

4.2.3 Detected binaries by Hipparcos

Hipparcos Catalogue (ESA, 1997) gives useful indications about binaries. The satellite measured more 24000 solved or suspected double (or multiple) systems 3000 of which being new with a separation $\rho < 1\hbox{$^{\prime\prime}$}$ generally. Lindegren (1997) gives the conditions of detection which are very different from the spectroscopic ones. A binary system could be resolved if $\rho$ is between $0\hbox{$.\!\!^{\prime\prime}$}1$ and $30\hbox{$^{\prime\prime}$}$, $\Delta H_{\rm p} < 4\,$mag, $H_{\rm p}$ being the Hipparcos magnitude and the period P < 30years with some difficulties if $\rho\,<\,0\hbox{$.\!\!^{\prime\prime}$}4$ or $\Delta H_{\rm p}\,\gt\,3.5\,$mag. The detection could result from photometric or astrometric data, generally both. If the photocenter deviated from the uniform motion expected for a single star, it is a suspected binary with an invisible companion. These indications are given in Tables 8 (and 9) respectively in the columns (bytes):
111(and 63): D indicates a probable binary detected by the Hipparcos photometry.
114 (and 66): the number of astrometric components.
116 (and 68): a double and multiple system flag about astrometry.
- C: resolved system ($P\,\gt\,3.3\,$years).
- O: orbital solution.
- G: solution with acceleration term (unresolved system, probable $P\,\gt\,10\,$years).
- V: variability of one of the components.
- X: stochastic solution (probable binary of $P\,<\,3\,$years).
118 (and 70): solution quality flag (A: good, B: fair, C: poor, D: uncertain, S: suspected non-single).
120-1 (and 72-3): the concerned components.
123-129 (and 75-81): separation $\rho\hbox{$^{\prime\prime}$}$.
131-135 (and 83-87): difference of magnitude $\Delta H_{\rm p}$.
136 (and 88): an asterix indicates that $\rho\hbox{$^{\prime\prime}$}$ and magnitude difference are from CCDM (Dommanget & Nys 1994).
More details are given in Hipparcos Catalogue Vol.1, Chap.1.4.

These three criteria can be complementary. Taking into account only stars with variable radial velocities and detected as double by Hipparcos, the minimum binaries percentage is 60%. If the criterion given by the shape of peak is added, this percentage increases to 74% with probable doubles or multiples (codes 4 and 6 of Table 4) and reaches 88% with suspected doubles (code 5 of Table 4), noticing these last percentages include optical systems.

Two spectroscopic binaries (HIP 33077 and 110838) have been found by McAlister et al. (1993) and are not binaries in Hipparcos Catalogue.