Recently, Fisk et al.  have presented a theory that describes a number of features of the large‐scale coronal and heliospheric magnetic field. This theory predicts large‐scale transport of magnetic flux across the boundaries of the polar coronal holes, which leads to reconnection processes of open field lines with preliminary closed magnetic structures. Reconnection processes reveal themselves in solar wind composition data: Plasma released out of previously closed magnetic field structures exhibits hotter charge state distributions and has a tendency to be enriched by elements with low first ionization potentials. The idea of reconnection at the boundaries of coronal holes is not new. For example, Wang and Sheeley  and Luhmann et al.  found evidence for that mechanism by comparison of observations of the rotation and evolution of coronal holes with potential field models of the solar corona. We use Ulysses Solar Wind Ion Composition Spectrometer composition measurements and sophisticated numerical models [Linker et al., 1999; Riley et al., 1999] to accurately map these observations back to the solar surface. We then constrain the thickness of the stream interface at the Sun and compare the location of the source region with SOHO observations of the low corona. The results are discussed in the context of the global structure of the heliospheric magnetic field.