What test did the double helix suggest?
In 1953, Watson and Crick proposed the double-helical structure of deoxyribonucleic acid, or DNA, and discussed its genetic implications. Each complementary strand could serve as a template for a new strand, leaving each replicated molecule with one old strand and one new one: the idea of semiconservative replication. A convincing structural model, however, did not mean that replication inside cells had already been observed experimentally.
Other possibilities included conservative replication, in which the old molecule remained intact while an entirely new molecule formed, and dispersive replication, in which old material was distributed throughout daughter molecules. Distinguishing these explanations required tracking old material and its distribution across successive rounds. Matthew Meselson and Franklin Stahl at the California Institute of Technology turned the problem into measurable density differences and published their study of Escherichia coli in 1958.
Giving old and new material distinguishable weights
They labelled material with stable nitrogen isotopes. Bacteria used nutrients containing the heavier isotope to synthesize cellular material, including DNA. After transfer to an ordinary, lighter nitrogen source, newly synthesized material bore a different label. Instead of counting radioactive decay, the method used density differences caused by the two isotopes, making old and new material traceable even though their appearance alone could not distinguish them.
Measurement depended on caesium chloride density-gradient ultracentrifugation. Within the gradient formed during centrifugation, DNA tended to concentrate where the medium matched its buoyant density, forming bands recorded optically. Meselson, Stahl and Jerome Vinograd had published a paper on equilibrium sedimentation in 1957. The instrument's ability to resolve very small density differences supplied the sensitivity needed to test the biological question in 1958.
A single intermediate band in the first generation
The researchers first labelled bacterial DNA with heavy nitrogen, then continued growing the culture in light-nitrogen conditions, taking samples at different times. Initial DNA formed a heavier band. After approximately one replication generation, the principal result was a single band between the heavy and light positions. Separately prepared heavy- and light-labelled material supplied reference positions, making 'intermediate' a comparison with known samples rather than a visual impression.
This contradicted a simple conservative model: if the original double-stranded molecules remained intact, the first generation should contain both old heavy molecules and wholly new light molecules, rather than all concentrating at an intermediate position. One intermediate band alone, however, could not distinguish semiconservative replication from certain dispersive explanations, since both could produce first-generation molecules containing old and new material. Observing the next generation further separated the models' predictions.
The second generation and molecular subunits
After approximately two generations, the sample showed intermediate and light bands in roughly equal amounts. Under the semiconservative model, every first-generation molecule contained one old and one new component. When those separated for the next replication, some descendants retained old label while others contained only light label, producing two density classes. A simple dispersive model expected continued mixing of old and new material in every molecule, with the entire band gradually becoming lighter. It did not readily explain separation into two populations.
The paper also reported density changes after heat denaturation. Treated first-generation material of intermediate density formed two classes of smaller subunits, corresponding to the results from separately treated heavy and light material. The researchers used this to consider continuity of the retained components. Density bands were not direct photographs of two helical strands. The conclusion arose from the combined constraints of label conservation, intergenerational proportions and treatment results, consistent with the double-helix interpretation.
What the experiment answered and what remained
The 1958 paper directly concerned E. coli DNA under the conditions studied. It strongly supported the separate passage of the old molecule's two components into daughter molecules after replication: a semiconservative distribution. It did not simultaneously explain replication origins, enzyme action, the direction of strand extension or every organism's replication details. Those required other experiments. The theoretical proposal in 1953, methodological development in 1957 and central publication in 1958 should therefore be distinguished.
The experiment later became celebrated for elegance and simplicity, but Frederic Lawrence Holmes's history, reconstructed from research records, correspondence and interviews, reveals methodological exploration and repeated adjustments behind the result. The lesson is not that one density-band image resolves a major question. It is to turn competing models into different predictions, design labels preserving temporal order and continue observing when evidence remains insufficient. The final conclusion gained strength because each comparison restricted the explanations that could still hold.