Exponential expression for temperature dependence of the effective surface energy of brittle fracture was early proved its validity by Y. Akita and K. Ikeda. G. R. Irwin's stress intensity factor at instability point K_C was lately applied by M. Yoshiki, T. Kanazawa and S. Machida to the crack-arrest phenomenon of double tension test. Combining these two approaches, the authors propose a new expression for double tension test data, i. e. the formula (7) in conjunction with the formula (4) and Fig. 1. If the material constants K ₀ and T ₀ are given, the conventional σ- T curve of double tension test can be drawn by calculation only. Examples of the calculated curve are given in Fig. 10, where the curves for C/B =0.2 and C/B =0.6 are corresponding to the transition curves for the iso-thermal test and for the temperature-gradient type test respectively. So far as the ferrite-pearlite plates are concerned, K ₀ (kg/mm²·√mm) for the double tension test can be predicted from the pressed-notch Charpy fracture transition temperature _PT_C (° K ) through the formula (13) and Fig. 6. In the light of the experiments illustrated in Fig. 8 and Table 1, the effects of the parabolic crack-front and the temperature-gradient on the double tension test result are discussed. Figs. 7, 9 and 10 are also associated with the discussion, and it is concluded that the arrest-transition characteristic of steel plate in service condition may be represented by the double tension σ- T curve for C/B =0.2. This agrees in principle with the idea previously suggested by one of the authors that the service performance of plate can be predicted by the iso-thermal double tension σ- T curve.