The wavelength coverage of the spectra taken during this observing run is about three times larger than that of the spectra acquired during the October 1994 run. Despite this fact, the number of arc lines available for wavelength calibration is still small and atmospheric lines had to be used. Furthermore, the zero point of the wavelength calibration is seen to change during the night. Such changes are not detected for the data obtained in the UT9410 run.
This basically renders the obtained arc spectra useless for wavelength calibration purposes.
At Br Gamma, the eight lines used to wavelength calibrate were: three OH
airglow emission lines (2.15376 
, 2.17111 
and 2.18022 
(from the CGS4DR
list of OH lines)), four telluric water lines (2.15910 
, 2.16347 
, 2.16869 
and 2.17452 
) and one
telluric methane absorption line ( 2.17259 
). The telluric lines were identified with the help of
the HITRAN database [Rothman et
al. 1992]. The spectra were classified into groups, each
with the same wavelength calibration. For each group the calibration was
obtained by fitting a straight line to the pair
(pixel number,wavelength). Higher degree polynomials were also tried but
the fits obtained were never as good as the ones performed with a
straight line.
At Pa Beta
the wavelength calibration was firstly performed in the same way
as described above for the Br Gamma spectra. The eight lines used to
wavelength calibrate were: four OH airglow lines (1.27636 
, 1.27831 
, 1.28065 
and 1.28248 
[Maihara et al. 1993]) and four
telluric oxygen lines (1.27519 
, 1.27599 
,
1.27743 
and 1.28043 
(HITRAN database)). The wavelength
range covered with the Pa Beta spectra is
roughly 
. From the lines indicated above, the one
with the longest wavelength is the OH line at 1.28248 
, i.e. all the lines used to wavelength
calibrate lie roughly on the first half of the range covered by the
spectra. Absorption lines with photospheric origin were identified in
the spectra of quite a few T Tauri stars (using the Near Infrared Solar
Atlas of Livingston & Wallace 1991). However, for lines on the
second half of the range covered by the spectra, the larger the
wavelength of a line the larger the offset relative to the expected
wavelengths were. Photospheric lines on the first half of the wavelength
range were at the expected positions. This led one to conclude that
while the wavelength calibration was good up to roughly 1.28248 
its
quality degraded towards longer wavelengths. In view of this fact, the
wavelength calibration was re-done making use of the information
provided by the photospheric lines. Obviously, only stars with
photospheric lines could be re-calibrated in this way.
The problem that arises when using photospheric lines together with the telluric airglow and absorption lines to wavelength calibrate, is the fact that they are effectively in a different reference frame. The telluric lines are obviously stationary relative to the observer while the photospheric lines are formed in an object moving with a radial velocity relative to the observer. The way around this problem, is to consider that the origin of the function from which the wavelength calibration is derived is different for atmospheric and photospheric lines.
The only difference between a calibration given by the photospheric lines and by the telluric lines should be an offset in the origin. Given this fact, the function chosen to wavelength calibrate the Pa Beta spectra where photospheric lines were present was:
where, 
and 
contains atmospheric lines
only and 
and 
contains photospheric lines only.
This function was fitted to the pairs (pixel number,wavelength) by using a general least squares procedure. A routine implementing such a procedure was written in IDL. The calibration relative to the telluric lines is given by the top branch of the the function above.
As one can see, the two branches of the function differ only in the constant term of the second degree polynomial. The difference a-d is the offset of the wavelength of photospheric lines relative to the atmospheric lines and it is due to the non zero radial velocity of the star relative to the observer. The offset value is easily translated into the heliocentric radial velocity of the star (see Section 3.5). This procedure for the wavelength calibration gives as a bonus an estimate for the heliocentric radial velocity of the star. These estimates are shown in Table 3.3 for the T Tauri stars for which photospheric lines were identified in their Pa Beta spectra. Radial velocity values found in the literature (see Table 2.1) are also shown here for comparison. As can be seen, the results obtained here agree (within the uncertainties) with the values found in the literature. Exceptions are DI Cep and V807 Tau.
For Pa Beta spectra where no photospheric lines are available the wavelength calibration is that obtained with the telluric lines only. This calibration is good enough throughout the wavelength range where the Pa Beta line is seen.
The uncertainty in the wavelength calibration is, both for Br Gamma and Pa Beta, given by the rms of the fit residuals. At Br Gamma the uncertainty (one-sigma) is 14-17 km/s whereas at Pa Beta the uncertainty (one-sigma) is 6-8 km/s.
Comparing the typical uncertainties in the wavelength calibration of this run with those of UT9410 one sees that while the Pa Beta spectra have very similar uncertainties in both runs the Br Gamma uncertainties are considerably larger (by at least a factor of 2) in spectra taken in the UT9512 run.
