AbstractThe fidelity achieved in first derivative profiles of DNA thermal denaturation is shown to depend on a number of factors including the thermal increment of data gathering, the precision of absorbance readings, and the manner in which data are smoothed prior to calculating the derivative of hyperchromicity. The closeness with which thermal denaturation data can be fitted by a cubic polynomial is carefully considered, and a derivation is presented for the estimated error in calculated values of the derivative of hyperchromicity with respect to temperature. After reviewing both theoretical and experimental evidence for the expected minimum width of a thermal transition in DNA, we conclude that thermal increments of 0.05°C or less are required for an adequate representation of transitions in naturally occurring DNA's.Data gathered under conditions meeting the requirements suggested here for quantitative recording of thermal denaturation profiles (Vizard and Ansevin, submitted for publication) show that virtually all of the high‐resolution thermal denaturation profile of a simple, naturally occuring DNA may consist of small subtransitions, which we call thermalites. The finding of substransitions is consistent with current theories of DNA melting. A particularly well‐resolved thermalite of λ bacteriophage DNA has a breadth of only 0.30°C (2σ width), and thus is narrower than previously reported thermal transitions for DNA. For this thermalite, the combination of width, shape, and position in the profile suggests that the substransitions observed in accurately recorded DNA thermal denaturation profiles are not described satisfactorily by existing theories.Knowledge of the requirements for the quantitative recording of thermal denaturation profiles should greatly favor the usefulness of denaturation experiments for physical genomic a