To validate the use of the curve-of-growth technique with such a small set of lines, the derived abundances were compared to abundances obtained via synthetic-profile fits. Using UCLSYN, we synthesized spectral windows around the Mn lines in Table 2 in several of the single sample stars, selected to form a representative subsample over the abundance range. Fits of the synthetic spectra to the observations were made by eye. Table 3 shows the differences between the abundances derived using the two methods, and Fig. 1 shows representative synthetic fits to the Mn lines in HR 7361.
The Mn I lines at 
4030 and
4034 and Mn II lines at
4478, 4365, 4363 and 3917 were all well fitted by the
synthetic profiles, and contain known blends contributing no more than
2-3 percent of the total 
over
our entire range of 
and composition. Neither Mn II 
4206 nor 
4326 could be
fitted with single line profiles with parameters similar to the other Mn
lines. In most cases no fit could be obtained where the wings and core of
the line fitted simultaneously. This effect is similar to that seen in
Ga II, which recent work by Dworetsky et al. [1998]
demonstrated is due to hyperfine structure. These two lines were thus
excluded from the abundance analysis and we shall return to them later.
Table 3: Differences between abundances derived using
spectrum synthesis
and exact curve of growth
For our final selection of six Mn lines, there would appear to be no significant systematic difference between the abundances derived by the profile and curve-of-growth techniques; the mean difference is only -0.04dex, somewhat smaller than the mean standard deviation on the abundances derived from our curve-of-growth analysis on the six Mn lines (0.09dex). We conclude that the curve-of-growth technique is sufficiently accurate, when using a carefully selected set of lines, to reproduce results from the more time-consuming spectrum-synthesis method.
