When the plate thickness is less than 10 mm, testing with standard sized Ch
arpy-V notch specimens is impossible. In such cases the testing must be bas
ed on sub-sized specimens. Sometimes, standard size specimens are impractic
al also for thicker sections. The use of sub-sized specimens requires extra
polation of the result from the sub-sized specimen to correspond to the res
ult from a standard sized specimen. In this work, literature data combined
with newly generated data were collected and examined in order to find the
correct normalisation procedure for sub-sized specimen, upper shelf impact
energy data. The resulting normalisation procedure, combined with a new Cha
rpy-V-J-R curve correlation, was shown to produce realistic estimates of th
e materials upper shelf fracture toughness. In the case of materials showin
g pronounced splitting in the Charpy-V test, the estimates may be over-cons
ervative. (C) 2001 Elsevier Science Ltd. All rights reserved.