The Double Tension Test or the ESSO Test with temperature gradient has been carried out to estimate the brittle fracture propagation arrest characteristics of the steel used in the welded structures. However the effects of specimen size and loading condition on these tests have been considered vaguely. In this report, the authors have investigated these effects by the statical elastic analysis using the simple spring model and by the systematical experiments. In fact they have been carried out for the 5 test series changing the parameter of the loading condition and the aspect ratio of the test specimen, the distance of pin-to-pin to the specimen breadth. In the test results these effects have been found as was expected and explained by the above-mentioned analysis satisfactorily. Then the one of the experimental methods to avoid these effects in the test of the brittle fracture propagation arrest are proposed.