In solution, the correlation time of the overall protein tumbling, tau(R), plays a role of a natural dynamics cutoff-internal motions with correlation times on the order of tau ( R ) or longer cannot be reliably identified on the basis of spin relaxation data. It has been proposed some time ago that the 'observation window' of solution experiments can be expanded by changing the viscosity of solvent to raise the value of tau(R). To further explore this concept, we prepared a series of samples of alpha-spectrin SH3 domain in solvent with increasing concentration of glycerol. In addition to the conventional (15)N labeling, the protein was labeled in the Val, Leu methyl positions ((13)CHD(2) on a deuterated background). The collected relaxation data were used in asymmetric fashion: backbone (15)N relaxation rates were used to determine tau(R) across the series of samples, while methyl (13)C data were used to probe local dynamics (side-chain motions). In interpreting the results, it has been initially suggested that addition of glycerol leads only to increases in tau(R), whereas local motional parameters remain unchanged. Thus the data from multiple samples can be analyzed jointly, with tau(R) playing the role of experimentally controlled variable. Based on this concept, the extended model-free model was constructed with the intent to capture the effect of ns time-scale rotameric jumps in valine and leucine side chains. Using this model, we made a positive identification of nanosecond dynamics in Val-23 where ns motions were already observed earlier. In several other cases, however, only tentative identification was possible. The lack of definitive results was due to the approximate character of the model-contrary to what has been assumed, addition of glycerol led to a gradual 'stiffening' of the protein. This and other observations also shed light on the interaction of the protein with glycerol, which is one of the naturally occurring osmoprotectants. In particular, it has been found that the overall protein tumbling is controlled by the bulk solvent, and not by a thin solvation layer which contains a higher proportion of water.