We conducted a theoretical and experimental reexamination of the solution transfer experiment proposed by Sprunt and Nur (1977). Limestone parallelepipedic samples with cylindrical holes were submitted to uniaxial load in a fluid which was hydrochloric or acetic acid. A theoretical treatment similar to that of Griffith's crack is proposed, which allows the calculation of the equilibrium shape of the hole by assuming that it deforms into an elliptical cylinder with its major axis perpendicular to the applied stress. Experiments conducted to check the model showed that, when all basic theoretical assumptions are fulfilled, the kinetics of hole shape evolution is too slow to make any hole deformation observable after a two months run. The disagreement between our results and Sprunt and Nur's observations is explained by experimental procedure differences. The model proposed should give a new insight into the solution transfer phenomenon. ¿ American Geophysical Union 1988