A reasonable J-R curve determination procedure is suggested by correcting the amount of overestimation in J-value whose effect becomes significant when crack grows. J-value estimated according to this procedure is compared with those obtained from the other conventional methods proposed by Rice, Merkle & Corten, and Ernst & Paris.
From experiments using compact tension specimens of a steel (SCM4), the conditions of a specimen for consistent J-R curves are determined. Under these conditions the J-R curve of the material seems to be independent of crack length.