Acidic proteins found in mineralized tissues act as nature's crystal engineers, where they play a key role in promoting or inhibiting the growth of minerals such as hydroxyapatite (HAP), Ca10(PO4)6(OH)2, the main mineral component of bone and teeth. Key to understanding the structural basis of protein-crystal recognition and protein control of hard tissue growth is the nature of interactions between the protein side chains and the crystal surface. In an earlier work we have measured the proximity of the lysine (K6) side chain in an SN-15 peptide fragment of the salivary protein statherin adsorbed to the Phosphorus-rich surface of HAP using solid-state NMR recoupling experiments. 15N{31P} rotational echo double resonance (REDOR) NMR data on the side-chain nitrogen in K6 gave rise to three different models of protein-surface interaction to explain the experimental data acquired. In this work we extend the analysis of the REDOR data by examining the contribution of interactions between surface phosphorus atoms to the observed 15N REDOR decay. We performed 31P-31P recoupling experiments in HAP and (NH4)2HPO4 (DHP) to explore the nature of dipolar coupled 31P spin networks. These studies indicate that extensive networks of dipolar coupled 31P spins can be represented as stronger effective dipolar couplings, the existence of which must be included in the analysis of REDOR data. We carried out 15N{31P} REDOR in the case of DHP to determine how the size of the dephasing spin network influences the interpretation of the REDOR data. Although use of an extended 31P coupled spin network simulates the REDOR data well, a simplified 31P dephasing system composed of two spins with a larger dipolar coupling also simulates the REDOR data and only perturbs the heteronuclear couplings very slightly. The 31P-31P dipolar couplings between phosphorus nuclei in HAP can be replaced by an effective dipolar interaction of 600 Hz between two 31P spins. We incorporated this coupling and applied the above approach to reanalyze the 15N{31P} REDOR of the lysine side chain approaching the HAP surface and have refined the binding models proposed earlier. We obtain 15N-31P distances between 3.3 and 5 A from these models that are indicative of the possibility of a lysine-phosphate hydrogen bond.