%  Paper der interessanten IR Linien (vgl. Paper Solanki, Biemont, Muerset)
\advance\vsize by 3 truecm
\nopagenumbers
\magnification=\magstep0
\font\mini=cmr7 scaled \magstep0
\font\klein=cmr9 scaled \magstep0
\font\mittel=cmbx10 scaled \magstep1
\font\gross=cmb10 scaled \magstep2
\font\grossi=cmmib10 scaled \magstep2
\vsize24.5cm
\vglue1.5cm
\leftline{\gross Interesting lines in the infrared solar spectrum. 
 II: Unblended lines between}
\leftline{\gross {\grossi \char021} 1.0 and {\grossi \char021} 1.8 
{\grossi \char022}m}
\vskip1cm
\hskip0.7cm{J. Ramsauer($^1$), S.K. Solanki($^2$) and 
 E. Bi\'emont($^3$)}
\vskip0.1cm 
\hskip0.5cm{\klein\obeylines 
\qquad($^1$)Institute of Astronomy, University of Graz, A-8010 Graz, Austria
\qquad($^2$)Institute of Astronomy, ETH-Zentrum, CH-8092 Z\"urich, Switzerland
\qquad($^3$)Institute of Astrophysics, University of Li\`ege% 
 , 5, Avenue de Cointe, B-4000 Li\`ege, Belgium}
\bigskip
\leftskip=1.4cm\rightskip=0.45cm{\noindent\klein{\bf Abstract.}
 --- We list 603 spectral lines between 1.0 and 1.8 $\mu$m that are judged 
to be relatively unblended from a visual inspection of spectra of the 
quiet sun. Much of the atomic data of relevance to studies of solar and 
stellar magnetism, convection and atmospheric structure are also provided.
Particular attention is paid to blending by telluric lines. We determine 
the level of blending both in the presence and the absence of telluric 
lines. We also describe how telluric blends may be removed from spectra 
with high spectral resolution.}
\medskip
\leftskip=0cm\rightskip=0cm
\hskip0.7cm{\klein{\bf Keywords:} the sun: spectrum --- lines: 
unblended --- lines: identification --- atomic data --- infrared.}
\medskip
\hskip0.7cm{\klein{\bf Thesaurus:} 
 02.01.3, 02.12.2, 02.12.3, 06.09.2}
\vglue0.9cm
\leftline{\bf 1. Introduction}
\bigskip
\noindent
The present paper forms the second of a series that explores the
infrared solar spectrum by identifying spectral lines of interest. The 
aim of this series is to produce a compilation of spectral lines which 
can be used to determine abundances and thermal properties of the 
atmosphere of the sun and of solar-type stars, or to investigate their 
magnetic and velocity fields. Such investigations profit greatly
from the use of unblended lines and, as a consequence, numerous lists 
of unblended optical spectral lines exist (e.g. Holweger 1967; Stenflo 
\& Lindegren 1977; Rutten \& Van der Zalm 1984). The situation in
the infrared is less satisfactory. In the present paper we list the 
unblended lines, as identified from a visual inspection of 
Fourier-transform-spectrometer (FTS) data, in the infrared J and H-bands,
i.e. in the wavelength ranges 1.00 - 1.34 $\mu$m and 1.49 - 1.80 $\mu$m. 
The present paper may be considered to be an extension of the work by 
Solanki et al. (1990, henceforth called Paper I) in different respects.
Firstly, it is an extension towards shorter wavelengths (we did not
consider the J-band at all in Paper I). Secondly, we distinguish between
telluric and solar blends. This is topical, since the work of Livingston
\& Wallace (1991), Wallace \& Livingston (1991) and Wallace et al. (1993)
has demonstrated the feasibility of removing telluric blends from infrared
solar spectra observed from the ground. This extension has led to a 
doubling of the listed H-band lines.
\smallskip    
There are a number of reasons for investigating the spectral range 1.00 - 
1.34 $\mu$m. The continuum opacity is similar to that in the visible (the
J band lies between the opacity maximum near 0.8 $\mu$m and the opacity 
minimum at 1.6 $\mu$m), but the temperature sensitivity of the Planck function
is lower, while the Zeeman sensitivity of the spectral lines is larger 
(a 1.2 $\mu$m line is roughly twice as Zeeman sensitive as a line of equal
strength and Land\'e factor at 0.6 $\mu$m). Although the Zeeman sensitivity
of lines in the H-band (1.5 - 1.8 $\mu$m) is even larger, the spectral 
region 1.00 - 1.34 $\mu$m is spectrally richer in the sense that the spectral
lines in the J-band cover a much wider range of properties than lines in
the H-band. For example, compare the excitation potentials of the lines
listed in Table I  with those listed in Table III. Obviously, the lines in 
the J-band span a much broader interval of temperature sensitivities. They 
are also formed over a larger range of heights. Finally, the J-band is not
so dominated by Fe I lines as the H-band. 
\smallskip
Unfortunately, the J-band is less clean, as far as telluric absorption is
concerned, than the optical or even the H-band. Since this is one of the
major disadvantages of the investigated wavelength range, we discuss the
implications of telluric blends and simple ways of compensating for them 
in Sect. 3.
\eject
\vglue-0.3cm
\leftline{\bf 2. Unblended lines}
\bigskip
\noindent
Table I lists the 308 lines between 1.00 and 1.34 $\mu$m that appear to 
be unblended upon visual inspection of three spectra of the quiet sun.
One of the spectral atlases we use was compiled by Delbouille et al. 
(1981) based on observations obtained at an heliocentric angle of almost 
zero with the McMath telescope on Kitt Peak and the Fourier transform 
spectrometer (FTS) on 5 May 1979 and on 26 Oct. 1979. The water vapour 
column was 5.7 - 7.7 mm for the employed spectra, which is not untypical 
for Kitt Peak. Observations made at drier sites (e.g. Jungfraujoch, space, 
or even Kitt Peak at a drier period) may well reveal additional unblended 
lines. The other spectral atlases we consider are those of Livingston \& 
Wallace (1991), covering roughly 1.1 - 5.4 $\mu$m and of Wallace et al.
(1993), covering roughly 0.7 - 1.1 $\mu$m. By comparing 2 spectra 
obtained at different atmospheric column masses, these authors were able
to extrapolate to the solar spectrum at zero atmospheric column mass. 
These spectra, therefore, provide us with a source for determining 
unblended lines without any telluric contamination (except for the 
wavelengths at which the telluric absorption is complete and no 
information on the solar spectrum reaches the earth). 
\smallskip
In spite of the problems posed by telluric absorption, the Delbouille et al.
(1981) atlas is well suited for the work in hand. The wavelengths have
been carefully calibrated, the spectral resolution is very high ($\lambda
/\it\Delta\lambda\rm = 1.9-2.5 \times {10}^6$ for the J-band) 
and the signal-to-noise ratio is excellent. This last point is 
particularly remarkable, since the spectral region around 1.1 $\mu$m 
lies in an instrumental no-man's-land: both CCDs, whose peak sensitivity 
lies at shorter wavelengths, and InSb detectors, whose sensitivity 
peaks at longer wavelengths, are relatively insensitive around 1.1 $\mu$m.
\smallskip
The considerable advantage of the quiet sun spectra published by Livingston 
\& Wallace (1991) and Wallace et al. (1993) is that they are 
unaffected by telluric blends. This advantage is somewhat 
offset by the lower S/N ratio and minor distortions to the line profiles 
caused by wavelength shifts between the two spectra used to remove the 
telluric contamination. The spectral resolving power of these spectra, which 
were obtained with the FTS at the McMath telescope on 18 Dec. 1990 and
on 26 and 27 June 1983, respectively, is 600\phantom{.}000 for 
$\lambda \ge$ 1.22 $\mu$m, 150\phantom{.}000 for 1.22 $> \lambda >$
1.1 $\mu$m (Livingston \& Wallace 1991) and 660\phantom{.}000 for shorter 
wavelengths (Wallace et al. 1993)\footnote{$^1$}{The resolving power at 
the shorter wavelengths of the Livingston \& Wallace atlas was reduced by
Doppler filtering in order to ensure a high S/N value.}. Since a given 
spectral line may be sufficiently pure (i.e. unblended) for 
a certain type of analysis, but not for a more sensitive one, we have
also listed lines that are not totally free of blends. Thus, Table I 
should contain almost all lines which are relatively unblended in the 
J-band under normal atmospheric conditions, as well as the relatively
unblended lines in the absence of terrestrial absorption. For some 
analyses a more rigorous selection will have to be made from among these 
lines. Therefore, following Paper I, we have added a ``blending index" to 
Table I. This index mirrors our subjective impression of the amount of 
blending (see below). The structure of Table I is identical to
that of Table I of Paper I, except that we now list two blending
indices, one derived from the Delbouille et al. (1981) atlas and one
from the atlases of Livingston \& Wallace (1991) and Wallace et al. (1993). 
The latter is an estimate of the blending in the absence of telluric lines. 
We have also included lines into Table I that are reasonably unblended 
only in the Livingston \& Wallace and the Wallace et al. atlases. 
\smallskip  
Column 1 gives the solar wavelength (\AA , in air), 
$\lambda_{\odot}$, of the lines. Whenever possible $\lambda_{\odot}$ has
been determined from the Delbouille et al. (1981) atlas. If the line is 
too strongly blended in their spectra, then we have obtained the wavelength 
from the Livingston \& Wallace or Wallace et al. spectra\footnote{$^2$}{All 
$\lambda_{\odot}$ values were only corrected for Doppler shifts. 
As in Paper I, the gravitational redshift was not removed.}.
Column 2 lists the corresponding vacuum wavenumbers, $\sigma_{\odot}$
(in ${cm}^{-1}$). The conversion from wavenumber to wavelength 
is made using Edl\'en's (1953) formula. The blending index is tabulated in 
column 3. It is calibrated by comparing with the lines listed in Paper I 
and those listed by Solanki et al. (1986): 0 = no detectable blend, 
1 = minor blending, 2 = significant blending, 3 = complete blending. 
Subcolumn I contains the blending index derived from the Delbouille et al. 
(1981) spectral atlas and subcolumn II the blending index determined from 
the Livingston \& Wallace (1991) and Wallace et al. (1993) atlases. 
\smallskip
Columns 4 and 5 show the relevant ions and diatomic molecules that we
identify as the sources of the solar spectral lines, together with
their laboratory wavelengths, $\lambda_{\rm lab}$ (taken from e.g. 
Bi\'emont et al. 1985a,b, 1986; Kurucz 1991c). The identification rests 
to a large extent on the intensities and on the correspondence between 
$\lambda_{\odot}$ and $\lambda_{\rm lab}$, which is often 
(but not always) better than 0.03 \AA. A sizeable fraction of this 
difference is probably of solar origin (cf. Nadeau 1988). In some cases
no direct laboratory measurement of the wavelength is available and 
wavelength values calculated from the difference between measured or
calculated energy levels have been listed under $\lambda_{\rm lab}$.
All lines found in the literature with $\lambda_{\rm lab}$ sufficiently
close to $\lambda_{\odot}$ are listed, except those fulfilling one of the
following conditions. Lines found to be too weak when calculated using $gf$ 
values from Kurucz \& Peytremann (1975) or Kurucz (1991c) are listed 
separately in Table II, while unlikely candidates for which not much more 
than their laboratory wavelength is known 
(e.g. As I $\lambda$ 10575.02 \AA), are removed completely from the lists. 
Whenever multiple identifications of a given 
solar line are possible we have listed them in what we think is the order
of decreasing probability, based on their calculated equivalent widths 
(wherever possible) and proximity of wavelengths. In a few cases an 
unlikely identification, if it is the only one known, has been retained
in Table I. Such identifications are marked by an $^\alpha$ just following
the ion and must be considered with caution.  
\smallskip
Column 6 lists the atomic and molecular transitions, whenever known, and 
column 7 indicates their sources. The notation of the terms in the identified
transitions follows the National Bureau of Standards (NBS) compilations of 
atomic energy levels, Kurucz (1991c) and Nave et al. (1994). The 
final digits of 
the wavelengths corresponding to the listed transitions (calculated on the 
basis of available energy levels) are presented in column 14 of Table I,
marked $\lambda_{\rm calc}$. A small $|\lambda_{\rm calc} - \lambda_{\rm lab}|$
is a necessary (but not sufficient) condition for the identification of a 
particular transition with an observed spectral line. Only transitions with 
$|\lambda_{\rm calc} - \lambda_{\rm lab}| \le 0.5$ \AA\ have been retained.
The much lower precision with which the energy of some atomic levels, 
especially of highly exited ones is known explains the often larger 
discrepancy between $\lambda_{\rm calc}$ and $\lambda_{\odot}$ than between 
$\lambda_{\rm lab}$ and $\lambda_{\odot}$. All the transitions due to
iron group elements have been checked with the newest atomic energy levels
available. This has led to the rejection of some
previous identifications, either due to the fact that one of the levels 
is no longer present in the NBS tables, or that 
$|\lambda_{\rm lab} - \lambda_{\rm calc}| > 0.5$ \AA. 
Column 8 contains the excitation potential of the lower level of the 
transition, $\chi_e$, in eV. Most of the tabulated values are from Kurucz 
(1991c). The effective Land\'e factor calculated from the listed transitions,
$g_{\rm eff}^{\rm calc}$, is presented in column 9, while the effective
Land\'e factor derived empirically from laboratory measurements,
$g_{\rm eff}^{\rm emp}$, is listed in column 10 for those lines for which
such data are available. The importance of empirical Land\'e factors has
been pointed out by, e.g., Landi Degl'Innocenti (1982), Solanki \& Stenflo
(1985), Solanki (1987) and Mathys (1989). 
The $g_{\rm eff}$ values have been calculated using the equation
\smallskip


$$g_{\rm eff} = {1 \over 2}(g_l + g_u) + 
  {1 \over 4}(g_l -g_u)(J_l(J_l + 1) - J_u(J_u + 1))\eqno(1)$$


\medskip
\noindent (Shenstone \& Blair 1929), which is independent of the coupling
scheme. Here $J_u$ and $J_l$ are the total angular momentum quantum numbers 
of the upper and lower levels involved in the transition, respectively,
while $g_u$ and $g_l$ are the corresponding Land\'e factors. The $g_u$ and 
$g_l$ resulting from the standard LS-coupling formula are used for 
calculating $g_{\rm eff}^{\rm calc}$, except those taken from Johansson \& 
Learner (1990). For $g_{\rm eff}^{\rm emp}$, $g_u$ and $g_l$ are mostly 
taken from the measured values listed by Sugar \& Corliss (1985).  
If only either $g_l$ or $g_u$ is available from laboratory 
measurements, the missing atomic level Land\'e factor is assumed to be 
represented by its calculated value. A question mark has been placed behind 
such $g_{\rm eff}^{\rm emp}$ values. We expect that even in such situations
$g_{\rm eff}^{\rm emp}$ is to be preferred to the $g_{\rm eff}^{\rm calc}$
values. 
\smallskip
Columns 11, 12 and 13 of Table I contain $X_\pi$, $X_\sigma$
and $Y_\sigma$, the second and third order coefficients of the expansion
of a spectral line according to its Zeeman moments (cf. Mathys \& Stenflo
1987). They are defined as follows:
\removelastskip
\baselineskip=16pt
\vglue-0.7cm
{$$\eqalignno{X_\pi &= (g_l - g_u)^2(3s - d^2 - 2)/10, & \cr
              X_\sigma &= (g_l - g_u)^2(8s - d^2 - 12)/80, &(2) \cr
              Y_\sigma \hskip1.5pt &= (g_l - g_u)^3d(4 - d^2)/160, & \cr}$$}
\vglue-0.7cm\noindent 
where
\vglue-0.7cm
\baselineskip=16pt
{$$\eqalign{\hskip169pt s &= J_l(J_l + 1) + J_u(J_u +1), \cr
            \hskip169pt d &= J_l(J_l + 1) - J_u(J_u +1). \cr}
            \qquad
            \hskip142.4pt (3)$$} 
\vglue-0.4cm
\vskip10pt
\baselineskip=12pt
\noindent The third order moment of the $\pi$-component is identically 
zero. The $Y_\sigma$ value refers to the red $\sigma$-component. The   
corresponding value for the blue $\sigma$-component is $-Y_\sigma$.
The $X_\pi$ and $X_\sigma$ coefficients represent, respectively,
the scatter of the $\pi$- and $\sigma$-components about their centres
of gravity (Landi Degl'Innocenti 1982, 1985). 
Still higher order moments may be obtained from the tables published
by Mathys \& Stenflo (1987). We have used the $g_{\rm eff}^{\rm calc}$ 
values to calculate $X_\pi$, $X_\sigma$ and $Y_\sigma$.
\smallskip
Finally, column 15 lists the logarithm of the statistically weighted 
oscillator strengths log$(gf)$ of the corresponding atomic lines. Most of 
them are taken from Kurucz (1991c). Values obtained from other sources
are marked.
\smallskip
Some of the identifications made in Table I may 
be due to the chance coincidence of $\lambda_{\rm calc}$ and 
$\lambda_{\rm lab}$. For most of the identifications it was possible 
to calculate the equivalent width of the lines by means of a solar 
atmospheric model (Holweger \& M\"uller 1974) and compare the results 
with the observed equivalent widths (cf. Bi\'emont 1976). This procedure has 
allowed us to rule out some transitions giving a negligible contribution to 
the observed line, i.e. transitions whose equivalent width is smaller than 
1m\AA. These rejected identifications are listed in Table II. It should be
noted that for other temperatures and chemical abundance ratios (on other 
stars) it cannot be ruled out that the rejected transitions give a more 
significant contribution to the spectra.
\smallskip
Table III has exactly the same structure as Table I. It lists the 295 
unblended lines in the H-band between 1.49 $\mu$m and 1.80 $\mu$m. It 
differs from Table I of Paper I mainly in that it distinguishes between
telluric and solar blends, also lists lines that are relatively clean
only after the removal of the telluric contamination and provides log$(gf)$
values whenever available. The importance of telluric blending even in
the relatively clean H-band is indicated by the fact that this table 
contains more than twice as many lines as Table I of Paper I. Practically
all of the new entries are lines that are unblended only after removal
of the telluric spectrum.
\smallskip
Finally, Table IV is the H-band counterpart of Table II.  
\vglue0.7cm
\leftline{\bf 3. Discussion}
\bigskip
\noindent 
In the present paper we have compiled, identified and provided
atomic data for the unblended lines between 1.00 and 1.80 $\mu$m, thus 
extending the spectral region investigated in Paper I to shorter wavelengths. 
Although our current knowledge of the atomic structure relevant for line 
identifications is still incomplete, recent investigations (e.g. Litz\'en 
\& Verg\'es 1976; Litz\'en 1976; Bi\'emont 1976; Bi\'emont et al. 
1985a,b 1986; Bi\'emont \& Brault 1987a,b; Johansson \& Learner 1990; 
Kurucz 1991a,b,c; Nave et al. 1994) have substantially improved the picture. 
In particular, the extensive compilation by Kurucz (1991a,b,c) of relevant 
atomic data of more then 400 000 individual atomic lines from the extreme UV 
to the far IR have led to a considerable improvement. Thus almost all the 
lines in Table I could be, at least tentatively, identified with particular 
atomic or molecular transitions.
Nevertheless, improvements in term structure of the high-lying states of 
Fe I and other iron group elements should be continued. It is likely that,
given further atomic data, a number of the identifications will have to
be revised.   
\smallskip
Since blending by telluric lines is relatively important in the J and K-bands, 
we now discuss it in greater detail. In addition to the deep and broad 
absorption separating the analyzed wavelength ranges there 
are also numerous atmospheric absorption features inside each spectral 
region (e.g., near 1.12 and 1.32 $\mu$m). Consider first the effect of 
telluric line blending on the polarized Stokes parameters. As already 
pointed out in Paper I, Stokes $Q$, $U$ and $V$ are affected by telluric 
blends: In the extreme case when all the light at a particular wavelength is 
absorbed in the earth's atmosphere no Stokes $Q$, $U$ or $V$ signal is left.
Nevertheless Stokes $Q$, $U$ and $V$ are affected quite differently from 
Stokes $I$. Whereas the $I$ profile of a solar spectral line is seemingly 
strengthened by a telluric blend (i.e. the total absorption is increased), 
the $Q$, $U$ and $V$ profiles can only be weakened by it, or, for that 
matter, by any unpolarized blend: an unpolarized blend can only destroy 
polarization, it cannot create any. 
\smallskip
In addition, the quantitative influence of a telluric blend is often much
larger on the Stokes $I$ profile than on $Q$, $U$ or $V$, particularly if 
both lines are not very deep. The relative reduction in Stokes $Q$, $U$ or 
$V$ at a given wavelength equals the fraction of the solar intensity (at 
that wavelength) which is absorbed in the earth's atmosphere. Consider, 
e.g., a visible or near infrared solar 
line with a central depth corresponding to 10\% of the continuum value,
which is blended with a telluric line whose depth is also approximately
10\% of the continuum intensity. Now, although the $I$ profile shape is
completely distorted (its depth may change by a factor of up to 2) the $Q$, 
$U$, $V$ signals are reduced by a relative amount of the order of 10\%. Thus,
in this example, there is an order of magnitude difference in the effects
on the $I$ profile, on the one hand, and the $Q$, $U$, $V$ profiles on 
the other. Consequently, for many purposes the $Q$, $U$, $V$ profiles of 
numerous spectral lines which are considerably blended by telluric lines 
can still be used with sufficient accuracy (cf. R\"uedi et al. 1995,
Paper III of the present series). Recall that this is only true as long 
as the telluric absorption is not too strong. 
\smallskip
The second point we wish to stress is that it is possible to remove the
influence of telluric blends, as long as these are not too strong, by
combining two spectra which differ exclusively in the amount of telluric
absorption. This technique has been used to produce clean spectral atlases
of the quiet sun (Livingston \& Wallace 1991; Wallace et al. 1993) and 
of a umbra (Wallace \& Livingston 1991). See Hall (1974), Saar \& Linsky 
(1985) and Paper III for other techniques to tackle this problem. 
Unfortunately, these techniques work perfectly only for solar features
that do not, at the spatial resolution of the observations, change over
a period of a few hours. Some very special spectral lines, such as the 
Ti I multiplet at 2.2~$\mu$m, constitute exceptions since they disappear
outside sunspots, so that nearly simultaneous observations of the quiet
sun provide clean templates of the telluric absorptions to be removed. Thus 
we are left with the problem of removing telluric blends when observing, e.g.,
with high spatial resolution (since small-scale features tend to evolve 
relatively rapidly).
\smallskip
We propose the following solution, which works if the spectral resolution
is sufficiently large. In order to remove a telluric blend care must be 
taken to observe an unblended telluric line of the same molecular species 
as the blend as nearly contemporaneously as possible. Let the observed 
spectrum be $I_{\lambda}^{\rm obs}(x)$, the purely solar spectrum
$I_{\lambda}^{\odot}$, the blending telluric line profile 
$I_{\lambda}^{\rm tb}(x)$ and the profile of the unblended telluric
line $I_{\lambda}^{\rm tu}(x)$. The strength of the telluric absorption
is parameterized by $x$ and is related to the zenith distance angle $z$
by $x \sim \sec z$. Then we can determine $I_{\lambda}^{\odot}$ from
$I_{\lambda}^{\rm obs}(x)$ if $I_{\lambda}^{\rm tb}(x)$ is known for
the $x$ at the time of the observations:
 
$$\hskip-16pt I_{\lambda}^{\odot} = {{I_{\lambda}^{\rm obs}(x)}\over
                                     {I_{\lambda}^{\rm tb}(x)}}.\eqno(4)$$
\medskip
\noindent In order to determine $I_{\lambda}^{\rm tb}(x)$, first 
$I_{\lambda}^{\rm tb}(x_{0})$ and $I_{\lambda}^{\rm tu}(x_{0})$ must be
formed, with $x_{0}$ being some reference value of $x$ (e.g. 
$x_{0} = 1$). This can be done by applying standard procedures (e.g.
B\"ohm{\phantom{.}}-Vitense 1989) to 2 or more spectra of the quiet sun 
observed at different $x$ values. For $\lambda$ between 0.7 and 5.4 $\mu$m 
Livingston \& Wallace (1991) and Wallace et al. (1993) have produced such 
a purely telluric spectrum. Now, we can determine $x$ of the observed 
spectrum by comparing $I_{\lambda}^{\rm tu}(x)$ with 
$I_{\lambda}^{\rm tu}(x_{0})$:
\medskip


$${x\over x_{0}} = {{\ln(I_{\lambda}^{\rm tu}(x))}\over
                   {\ln(I_{\lambda}^{\rm tu}(x_{0}))}}.\eqno(5)$$


\medskip
\noindent Using the obtained $x/x_{0}$ ratio $I_{\lambda}^{\rm tb}(x)$ 
may then be determined:
\medskip


$$\hskip-7pt I_{\lambda}^{\rm tb}(x) = 
                    {(I_{\lambda}^{\rm tb}(x_{0}))}^{x/x_{0}}.\eqno(6)$$


\medskip
\noindent Inserting this into Eq.{\phantom{.}}(4) then gives the 
tellurically unblended profile of the observed solar feature. 
This technique requires that the blending telluric line lies at the
same wavelength in both spectra. This can be achieved by calibrating
the wavelength scale of the reference telluric spectrum such that the
unblended telluric line has the same wavelength in both spectra.
The reliability of the proposed technique is 
expected to improve if more than one unblended telluric line
$I_{\lambda}^{\rm tu}(x)$ is measured, and possibly if reference spectra
at different $x_{0}$ are used. Ideally, the reference spectra should
be obtained with the same instrument as the solar spectrum of interest,
so that they have the same spectral resolution and spectral stray light. 
It is also possible to use reference spectra having higher spectral 
resolution by degrading them first using the appropriate instrumental profile. 
\smallskip
If Stokes $Q$, $U$, $V$ are measured, then it is best to carry out the 
correction directly on the originally measured $I \pm Q$, $I \pm U$, 
$I \pm V$ profiles and to form $Q$, $U$, $V$ only from the corrected 
spectra (see Paper III for a successful application to $I$ and $V$). 
The above procedure can, of course, also be applied to the
removal of telluric blends from stellar spectra, although most of these
will not have the necessary spectral resolution to resolve the often 
very narrow telluric features. 
\smallskip
Another relatively general but complex way of removing telluric blends
has been described by Lallement et al. (1993).
\vglue0.7cm
\leftline{\bf Acknowledgements} 
\bigskip
\noindent 
We thank Dr. R.L. Kurucz for providing us with his newest compilation of 
atomic data of the iron group elements, Prof. G. Elste for providing us with
a copy of Mohler\'\negthinspace s line list and Prof. H.F. Haupt for his 
continuing encouragement and support.
\vfill
\eject
\leftline{\bf References}
\bigskip
\parindent=0pt
{\klein\obeylines B{\mini I\'EMONT} E.: 1976, {\sl Astron. Astrophys.%
\ Suppl. Ser.} {\bf 26}, 89.     
B{\mini I\'EMONT} E., B{\mini RAULT} J.W.: 1987a, {\sl Physica Scr.}%
{\ \bf 34}, 751.
B{\mini I\'EMONT} E., B{\mini RAULT} J.W.: 1987b, {\sl Physica Scr.}%
{\ \bf 35}, 286.
B{\mini I\'EMONT} E., G{\mini REVESSE} N.: 1973, {\sl Atomic Data Nuclear%
\ Data Tables} {\bf 12}, 217.
B{\mini I\'EMONT} E., B{\mini RAULT} J.W., D{\mini ELBOUILLE} L., R{\mini%
OLAND} G.: 1985a, {\sl Astron. Astrophys. Suppl. Ser.} {\bf 61}, 107.     
B{\mini I\'EMONT} E., B{\mini RAULT} J.W., D{\mini ELBOUILLE} L., R{\mini%
OLAND} G.: 1985b, {\sl Astron. Astrophys. Suppl. Ser.} {\bf 61}, 185.     
B{\mini I\'EMONT} E., B{\mini RAULT} J.W., D{\mini ELBOUILLE} L., R{\mini%
OLAND} G.: 1986, {\sl Astron. Astrophys. Suppl. Ser.} {\bf 65}, 21.     
B{\mini I\'EMONT} E., M{\mini ARTIN} F., Q{\mini UINET} P., Z{\mini%
EIPPEN} C.J.: 1994, {\sl Astron. Astrophys.} {\bf 283}, 339.     
B{\mini \"OHM -}V{\mini ITENSE} E.: 1989, Introduction to Stellar%
\ Astrophysics. Vol. I: Basic Stellar Observations and Data.
\ \ \ Cambridge University Press, pp. 16-20.
C{\mini ORLISS} C., S{\mini UGAR} J.: 1981, {\sl J. Phys. Chem. Ref. Data}%
{\ \bf 10}, 197.
D{\mini ELBOUILLE} L., R{\mini OLAND} G., B{\mini RAULT} J.W., T{\mini %
ESTERMAN} L.: 1981, Photometric Atlas of the Solar Spectrum 
\ \ \ from 1850 to 10 000 $cm^{-1}$, NOAO, Tucson, Az.
E{\mini DL\'EN} B.: 1953, {\sl J. Opt. Soc. America} {\bf 43}, 339.
H{\mini ALL} D.N.B.: 1974, An Atlas of Infrared Spectra of the Solar% 
\ Photosphere and of Sunspot Umbrae, Kitt Peak 
\ \ \ National Observatory, Contribution No. 556, Tucson, Az.
H{\mini OLWEGER} H.: 1967, {\sl Z. Astrophys.} {\bf 65}, 365. 
H{\mini OLWEGER} H., M{\mini \"ULLER} E.A.: 1974, {\sl Solar Phys.}%
{\ \bf 39}, 19.
J{\mini OHANSSON} S., L{\mini EARNER} R.C.M.: 1990, {\sl Astrophys. J.}%
{\ \bf 354}, 755.
K{\mini URUCZ} R.L.: 1991a, in Stellar Atmospheres: Beyond Classical%
\ Models, L. Crivellari, I. Hubeny (Eds.), Kluwer,
\ \ \ Dordrecht, p. 441.
K{\mini URUCZ} R.L.: 1991b, in Precision Photometry: Astrophysics of the%
\ Galaxy, A.G. Davis Philip et al. (Eds.),
\ \ \ L. Davis Press, Schenectady, N.Y.
K{\mini URUCZ} R.L.: 1991c, Magnetic tape with atomic data.
K{\mini URUCZ} R.L., P{\mini EYTREMANN} E.: 1975, Smiths. Astrophys. Obs.%
\ Spec. Report 362. 
L{\mini ALLEMENT} R., B{\mini ERTIN} P., C{\mini HASSEFI\`ERE} E.,%
\ S{\mini COTT} N.: 1993, {\ \sl Astron. Astrophys.} {\bf 271}, 734.
L{\mini ANDI} D{\mini EGL'}I{\mini NNOCENTI} E.: 1982, {\sl Solar Phys.}%  
{\ \bf 77}, 285.
L{\mini ANDI} D{\mini EGL'}I{\mini NNOCENTI} E.: 1985, {\sl Solar Phys.}%  
{\ \bf 99}, 1.
L{\mini ITZ\'EN} U.: 1976, {\sl Physica Scr.} {\bf 14}, 165.
L{\mini ITZ\'EN} U., V{\mini ERG\`ES} J.: 1976, {\sl Physica Scr.}%
{\ \bf 13}, 240.
L{\mini IVINGSTON} W., W{\mini ALLACE} L.: 1991, An Atlas of the Solar% 
\ Spectrum in the Infrared (1.1 to 5.4 {$\mu$}m), Techn.
\ \ \ Rep. \# 91-001, National Solar Obs., Tucson, Az. 
M{\mini ARTIN} W.C., Z{\mini ALUBAS} R.: 1979, {\sl J. Phys. Chem. Ref. Data}%
{\ \bf 8}, 817.
M{\mini ARTIN} W.C., Z{\mini ALUBAS} R.: 1983, {\sl J. Phys. Chem. Ref. Data}%
{\ \bf 12}, 323.
M{\mini ATHYS} G.: 1989, {\sl Fundam. Cosmic Phys.} Vol. 13, No. 3, 143.
M{\mini ATHYS} G., S{\mini TENFLO} J.O.: 1987, {\sl Astron. Astrophys.%
\ Suppl. Ser.} {\bf 67}, 557.
M{\mini OHLER} O.C.: 1955, A Table of Solar Spectrum Wave Lengths:%
\ 11984 \AA\ to 25578 \AA , The University of
\ \ \ Michigan Press, Ann Arbor.
M{\mini UGLACH} K., S{\mini OLANKI} S.K.: 1992,%
{\ \sl Astron. Astrophys.} {\bf 263}, 301.
N{\mini ADEAU} D.: 1988, {\sl Astrophys. J.} {\bf 325}, 480.      
N{\mini AVE} G., J{\mini OHANSSON} S.: 1993,%
{\ \sl Astron. Astrophys. Suppl. Ser.} {\bf 102}, 269.
N{\mini AVE} G. et al.: 1994,%
{\ \sl Astrophys. J. Suppl. Ser.} {\bf 94}, 221.
R{\mini EADER} J., C{\mini ORLISS} C., W{\mini IESE} W.L., M{\mini ARTIN} W.%
C.: 1980, NSRDS-NBS$^*$ 68: Wavelengths and transition 
\ \ \ probabilities for atoms and atomic ions.
R{\mini \"UEDI} I., S{\mini OLANKI} S.K.,  L{\mini IVINGSTON} W.,%
\ H{\mini ARVEY} J.W.: 1995, {\sl Astron. Astrophys. Suppl. Ser.} % 
\ in press
\ \ \ (Paper III).
R{\mini UTTEN} R.J., V{\mini AN DER} Z{\mini ALM} E.B.J.: 1984,%
{\ \sl Astron. Astrophys. Suppl. Ser.} {\bf 55}, 143.
S{\mini AAR} S.H., L{\mini INSKY} J.L.: 1985, {\sl Astrophys. J.} {\bf 299},%
\ {\sl L} 47.
S{\mini HENSTONE} A.G., B{\mini LAIR} H.A.: 1929, {\sl Phil. Mag.}%
{\ \bf 8}, 765.
S{\mini OLANKI} S.K.: 1987, Ph.D. Thesis, No. 8309; E.T.H., Z\"urich.
S{\mini OLANKI} S.K., B{\mini I\'EMONT} E., M{\mini \"URSET} U.:% 
\ 1990, {\sl Astron. Astrophys. Suppl. Ser.} {\bf 83}, 307 (Paper I).
S{\mini OLANKI} S.K., P{\mini ATELLINI} F.G.E., S{\mini TENFLO} J.O.:%
\ 1986, {\sl Solar Phys.} {\bf 107}, 57.
S{\mini OLANKI} S.K., S{\mini TENFLO} J.O.: 1985, {\sl Astron. Astrophys.}%
{\ \bf 148}, 123.
S{\mini TENFLO} J.O., L{\mini INDEGREN} L.: 1977, {\sl Astron. Astrophys.}%
{\ \bf 59}, 367.
S{\mini TRIGANOV} A.R., S{\mini VENTITSKII} N.S.: 1968, Tables of spectral%
\ lines of neutral and ionized atoms, IFI/Plenum. 
S{\mini UGAR} J., C{\mini ORLISS} C.: 1985, {\sl J. Phys. Chem. Ref. Data}% 
{\ \bf 14}, Supplement No. 2.
S{\mini WENSSON} J.W., B{\mini ENEDICT} W.S., D{\mini ELBOUILLE} L.,%
\ R{\mini OLAND} G.: 1970, The Solar Spectrum from {$\lambda$} 7498 to 
\ \ \ {$\lambda$} 12016. M\'emoires de la Soci\'et\'e Royale%
\ des Sciences de Li\`ege, Special Volume No. 5.
W{\mini ALLACE} L., H{\mini INKLE} K., L{\mini IVINGSTON} W.: 1993, An Atlas% 
\ of the Photospheric Spectrum from 8900 to
\ \ \ 13600 {$cm^{-1}$} (7350 to 11230 \AA), National Solar Obs., Techn.%
\ Rep. \# 93-001, Tucson, Az. 
W{\mini ALLACE} L., L{\mini IVINGSTON} W.: 1991, An Atlas of a Dark Sunspot% 
\ Umbral Spectrum from 1970 to 8640 {$cm^{-1}$} 
\ \ \ (1.16 to 5.1 {$\mu$}m), Techn. Rep. \# 92-001, National Solar Obs.,%
\ Tucson, Az. 
Z{\mini AIDEL'} A.N., P{\mini ROKOF'EV} V.K., R{\mini AISKII} S.M., S{\mini% 
LAVNYI} V.A., S{\mini HREIDER} E.Ya.: 1970, Tables of spectral
\ \ \ lines, IFI/Plenum.
\bigskip
$^*$NSRDS-NBS: National Standard Reference Data Series - National Bureau of%
\ Standards.}
\vfill
\bye