Crick and Watson began feverishly devising models.

The "very pretty model" of which Watson had written Delbrück was one attempt, but a mistaken one, as Jerry Donohue pointed out. Donohue's input turned out to be critical. A magna cum laude graduate of Dartmouth who had worked and studied with Pauling at Caltech since the early 1940s, Donohue knew structural chemistry inside and out. Hydrogen bonding had been a specialty of his, and he saw that Crick and Watson, chemical novices that they were, had been playing with the wrong structures for guanine and thymine. He set them right, switching the hydrogen atoms essential for cross-bonding into their correct positions, destroying Watson's pretty model and pushing them toward the correct solution.

With Donohue's corrections, Crick and Watson could now see hydrogen bonds forming naturally between specific pairs of purines and pyrimidines: adenine to thymine and guanine to cytosine. Bases in the middle, phosphates outside. Two strands. Matching a large with a small base not only smoothed the structure's outline but provided a simple explanation for Chargaff's findings.

The resulting structure, a sort of ladder with base pairs as the steps and the sugar-phosphate backbone as the runners, formed easily into a helix that matched the x-ray data. More than beautiful, the structure had meaning. Each strand was a complementary mirror image of the other; if separated, each could act as a mold for forming a new double helix identical with the original. This immediately provided ideas about replication that Pauling's model, with its bases facing out and unrelated to each other, could not.