1: \begin{abstract}

2:   Spectral enhancement -- which aims to undo spectral broadening -- leads to 

3: integral equations which are ill-posed and require special regularisation

4: techniques for their solution. Even when an optimal regularisation technique is

5: used, however, the errors in the solution -- which originate in data

6: approximation errors -- can be substantial and it is important to have good

7: bounds for these errors in order to select appropriate enhancement methods. A

8: discussion of the causes and nature of broadening provides regularity or source

9: conditions which are required to obtain bounds for the regularised solution of

10: the spectral enhancement problem. The source conditions do only in special cases

11: satisfy the requirements of the standard convergence theory for ill-posed

12: problems. Instead we have to use variable Hilbert scales and their interpolation

13: inequalities to get error bounds. The error bounds in this case turn out to be

14: of the form $O(\epsilon^{1-\eta(\epsilon)})$ where $\epsilon$ is the data error

15: and $\eta(\epsilon)$ is a function which tends to zero when $\epsilon$ tends to

16: zero. The approach is demonstrated with the Eddington correction formula and

17: applied to a new spectral reconstruction technique for Voigt spectra. In this

18: case $\eta(\epsilon)=O(1/\sqrt{|\log\epsilon|})$ is found.

19: \end{abstract}

20: