2.1. The sample
We have chosen our sample from the list of Mathieu (1994). In Paper I, Monin et al. already presented some spectroscopic measurements on five objects in this list, with separations ranging between and . In this paper we present complementary observations of closer binaries from the same list. This new sample (see Table 1) now includes all the binaries new sample (see Table 1) now includes all the binaries in this list with separations ranging between and , to the exception of HBC 411 (CoKu Tau/3) and HBC 389 (Haro 6-10).
Table 1. Complete list of spectroscopically observed binaries (Paper I and this paper). Listed are the Herbig & Bell (1988) catalogue numbers (hereafter: HBC) of the primary and secondary when available, the binary separation and the previous classification of the whole system as CTTS or WTTS (from HBC unless explicitly quoted).
2.2. New spectroscopic observations
The observations were conducted on 1996 November 5 and 6, and December 1, at the Canada-France-Hawaii Telescope on Mauna Kea. We used the STIS2 detector with a /pixel scale. Using SIS (Subarcsecond Imaging Spectrograph) providing tip-tilt correction, we obtained an angular resolution of about to . Differential VRI imaging photometry was also performed during the first two nights for some targets. For each system, the primary has been defined as the brightest star in the band.
Long-slit spectra were obtained using a 1" slit and a grism. The usefull range of the spectra is 4000 to Å, yielding a Å/pixel scale. However, the actual resulting spectral resolution is Å, except for HBC 356-357 where it is Å. Spectra of calibration lamps and of a spectrophotometric standard (Feige 110) were obtained every night. All spectra have been wavelength calibrated, cosmic-ray cleaned, flat fielded, sky emission subtracted and flux calibrated. All data reduction steps were performed with standard IRAF 1 routines. The two stellar spectra of each binary were deblended and extracted using a task fitting two gaussians with the same FWHM profile. This reduction procedure is accurate as long as the separation remains larger than the seeing, which was the case for all our sources except FX Tau and UY Aur, the closest systems of our sample (see Sect. 3.1 for details).
Our estimates of the spectral types are based on the strength of TiO bands for M stars, and on relative strengths of CaI 6122,62, NaI , CaH 6350,80 and CaH 6750-7050 for K stars. We used the standard grids from Allen & Strom (1995) and Kirkpatrick et al. (1991), and we also observed a series of spectral type standards during the same nights as the binary targets. From these standard stars measurements, we find that our estimates are accurate to within one subclass for the whole sample. However, we are unable to determine spectral types later than M5, because most of the spectral features we use do not change anymore with effective temperature for such late type stars. Spectra at longer wavelengths are needed for the classification of the reddest objects.
Uncertainties on emission line equivalent widths (hereafter EWs) were estimated by using the maximum and minimum acceptable continuum values next to the lines. They are typically smaller than 5%, except for the weakest lines, where they are of the order of 0.1-0.2 Å. In the blue part of the spectrum, for the faintest stars, uncertainties can reach 10-15%.
We evaluated differential photometry for 8 of our sources in the VRI bands. Uncertainties are usually smaller than 0.02 mag and never exceed 0.03 mag.
© European Southern Observatory (ESO) 1999
Online publication: November 16, 1999