The paramagnetic shift resulting from f orbital covalency for lanthanide fluorides has been successfully treated theoretically in earlier studies. This theoretical method has been modified to calculate the paramagnetic shift for 13C and 1H in uranocenes. It is shown that the very large anisotropy in the magnetic moment makes the treatments in the literature incorrect. The so-called "contact shift" in 13C is in reality three terms: (1) a dipolar contribution from spin in the carbon pz orbital, (2) a dipolar contribution from spin in the nearest neighbor carbon pz orbitals, and (3) an induced Fermi contact term. Contribution (1) is the largest. For 1H there are contributions from the direct dipolar interaction of spin in the nearest carbon pz orbital and the induced Fermi contact term. Both terms are nearly equal in magnitude. The theory, however, predicts a much smaller magnitude and the incorrect sign for the shift as measured experimentally. The measured shift must, therefore, result from a polarization mechanism in which exchange interactions between the f electrons and electrons in filled orbitals cause spin delocalization into the ligand σ orbitals. It is argued that this polarization mechanism operates primarily through the overlap of the outer 6s and 6p orbitals of the uranium atom with the σ orbitals of the ligand making the observed nmr shifts insensitive to the nature of the bonding between the ligand and the uranium atom. We conclude, therefore, that the question of how much f orbitals are involved in covalent bonds between uranium and the ligands in uranocene cannot be answered by measuring the paramagnetic shift in 13C or 1H.
