Our sample of 33 background stars was selected from the list of field
stars provided by Whittet et al. (1987). Linear polarization
measurements were carried out in 1992 March 13 to 17 at the ESO/MPI
2.2 m telescope equipped with the two-channel photo-polarimeter PISCO
(Stahl et al. 1986; Schwarz 1989). Measurements were obtained in the
Cousins UBVRI system using a diaphragm of 15 arcsec. The instrumental
polarization compensation mode, involving a rotation of a compensating
phase plate unit by 180 degrees, was systematically used. The data
acquisition was performed using the two-apertures mode for sky
polarization compensation (the two apertures being separated by 66'' in
the East-West direction). Calibration in polarization, P, and
position angle, 
, was done through observations of the
polarization standard star HD 147084 (o Sco) (Hsu & Breger 1982).
The uncertainty in the position angle in all cases was estimated to be less than 2 degrees, using polarized standard stars and the internal calibration of the polarimeter. The instrumental polarization was measured by observing the unpolarized standard stars HD 100623, HD 115617 (61 Vir) and HR 6060 (18 Sco) (Serkowski 1974; Walborn 1968). Measurements of some bright stars were also performed with a polaroid inserted in order to determine the instrumental efficiency. These yielded negligible corrections.
The reduction of the polarimetric data was performed using a revised
version of the MIDAS procedure PISCO. For each star the degree of
polarization was determined, with values ranging from zero up to about
8 percent. The observational errors are determined from the power in
the 5th and higher harmonics in the Fourier-transform, using an
algorithm which is part of the PISCO data reduction package (Schwarz
1989). Also, the correction for noise biasing, according to the
analytical formula 
has been
applied to the observed polarization, p', using the estimated
standard deviation of the polarization, 
, derived from the
PISCO reduction procedure (Clarke & Stewart 1986).
The list of observed stars is given in Table 1 (click here), together with other relevant information. The optical polarization data are given in Table 2 (click here). For reference, the first column of each table gives the list number from Whittet et al. (1987).
The background stars earlier than A0, and suitable for the spectroscopic study of the cloud were selected from the polarimetry sample. In addition, we included the stars HD 97300 and HD 97048, which are thought to be members of the Cha I association, to improve the completeness of our spectroscopic sample.
High resolution spectroscopy was done with the Coudé Auxiliary
Telescope (CAT) of the European Southern Observatory at Cerro La Silla
(Chile) during two observing runs in 1992 March and 1994 January. The
data were obtained using an RCA CCD (ESO # 9) and the CES short
camera in the blue path. This instrument configuration provided a
resolving power, 
at the studied
wavelengths. A few
additional spectra were obtained in 1994 July using the CAT with the CES
long camera, operated in remote control from Garching. The July data
have a higher resolving power 
= 130 000,
and were used to complete coverage of the region, and obtain more
detailed velocity profiles for selected stars.
Data reduction was performed using the IRAF system, in which the
standard CCD reduction packages were used to remove cosmic rays,
perform bias subtraction and flat-field the images. Spectral
extraction was done using the optimal extraction algorithm in the IRAF
task 
, in which pixel values across an order of a
spectrum are weighted according to the calculated signal to noise
ratio. The signal to noise was estimated using the readout noise and
gain of the CCD. Spectra were wavelength calibrated using a
Thorium/Argon calibration lamp, and a linear interpolation of matched
features.
For each spectrum a fit to the continuum was performed using a high
order spline, and a correction was applied to transform the spectra to
a uniform LSR velocity reference frame, by removing the velocity
contributed by terrestrial and solar motions. Column densities were
computed by using a Voigt profile model, based on a version of the
STARLINK curve of growth programme adapted for IRAF. For some of the
spectra, line profiles were modeled using the MIDAS image processing
package. Each line profile was modeled using the minimum number of
components necessary to reproduce the observed data with residual
values comparable to the noise in the data. Our spectral resolution
limited us to separating line components with velocity differences of
more than 1-2 km 
, and we cannot rule out the possibility of
very narrow blended components. However, since the turbulent
velocities in molecular clouds are expected to be 1-2 km s
, we
believe our instrumental resolution is able to resolve the components
present in the Chamaeleon cloud. This is also consistent with the
determinations based on CO observations in the Cha I cloud obtained by
Dubath et al. (1995).
For the Chamaeleon region we obtained the latest coadded plates for
all four bands of the ISSA survey. The pixel size within the plates is
1.5' which, in the case of the 100 
m band, oversamples the actual
instrumental resolution of the satellite, which is closer to 4'. We
removed zodiacal background residuals from each plate by using an
iterative background removal programme which selects background
points based on an analysis of the histogram of pixel values, and fits
a flat background to the selected points. IRAS colors were derived
using the programme skyview from IPAC, which averages pixel
fluxes in a 9 pixel region centered on the coordinates of our program
stars. The larger sampling region makes that the IRAS colours have a
higher signal to noise ratio, and samples a region 4.5''
square, with uncertainties in the IRAS flux, 
,
of 
0.08 mJy Sr
for 12 
m and
25 
m, and 
0.15 mJy Sr
for the
60 
m and 100 
m passbands.
