The stability of a DNA double helix of any particular sequence is conventionally estimated as the average of the stabilities of the 10 different nearest-neighbor (NN) base pair doublets that it contains. Therefore, much effort has been devoted to the experimental characterization and tabulation of the enthalpy, entropy, and free energy of melting for each of the NN doublets. Although data from different research groups generally agree for the NN free energies and melting temperatures, there are major disagreements for the enthalpies and entropies. The largest differences are between the parameters obtained on oligomeric relative to polymeric DNA. This disagreement interferes with the practical application of NN thermodynamic parameters. It also raises doubts regarding several fundamental assumptions about DNA melting, such as the absence of longer range interactions, the length dependence of DNA melting parameters per base pair, the applicability of polyelectrolyte theory to the description of salt effects on oligomers, and the purely enthalpic difference between NN doublets. Here we show that if one takes into account the significant heat capacity increase associated with DNA melting, all of the above assumptions are self-consistently reconciled with experiment.