We discuss the statistics of the chains of Ag atoms self-assembled on the steps of a vicinal Pt surface as established experimentally and calculated within a lattice gas model by Gambardella et al. [Phys. Rev. B rong>73rong> (2006) 245425]. We suggest that the discrepancy between the theory and experiment may be due to additional interatomic interactions inside the clusters unaccounted for in the model. Our consideration is based on an exactly solvable one-dimensional equilibrium model of self-assembly proposed by us recently. We argue that the model provides an adequate approximate description of the Ag/Pt system and show that the chain length distribution in the model can be fitted to the experimental data with high accuracy.