When using a recently developed method of Doppler-Zeeman mapping (Vasilchenko et al., 1996) for analysis of a real star and real observational data, we are confronted with limitations due to the model simplifications and unavoidable errors in observed spectra. We discuss the errors introduced by probable inaccurracies of the mathematical model: analytical fit of the local Stokes parameters, influence of magneto-optical effect, ignorance of the true atmosphere model to compute local Stokes profiles, non-uniform surface brightness. The magnetic field configuration is found in the form of arbitrarily shifted dipole and sum of dipole and quadrupole, along with the distribution of Si, Ti, Cr and Fe over the surface of the star. Lines of different elements lead to the same magnetic field configuration, which is reliably determined for the part of the stellar surface which faces the observer. This allows to compare the magnetic field and chemical maps of the surface of HD 215441. A large-scale ring structure with the magnetic pole at its center is clearly seen on the abundance maps. Si, Cr and Ti are highly deficient where the magnetic field lines are vertical (near the magnetic pole) while Fe is highly overabundant there.