Proper identification of the transition temperature and the shape of the lower-bound (KIc) fracture toughness curve in the transition range has been a long-term objective of work at Oak Ridge National Laboratory. A past practice has been to test a large number of specimens of varying sizes, from 1/2 to 8T compacts, in expectation that size effects and statistical variability of KJc could be resolved empirically. Recently, statistical and constraint-based models have been developed that purport to explain much of what has been seen. Weakest-link theory has been successfully used to predict specimen size effects for the lower part of the transition curve. Constraint-based models of βc - βIc and Jssy (small-scale yield) also can model size effects, but these tend to conflict among themselves with regard to the prediction of full constraint KJc. All lack potential for defining the absolute lower bound of fracture toughness. Statistically based models have the benefit of quantifying data scatter characteristics and provide a basis for making lower-bound toughness estimates with assigned error estimates. The appropriate characterization of transition temperature is of value in industrial problems and is of particular importance to the nuclear industry where safety issues are involved. Here the KJc data are obtained from small specimens, the size of which is dictated by volume limitations of surveillance capsule size. A basis has been explored for establishing a lower-envelope curve from such data.