The system of license for nuclear installations has been administered within the technological framework of thinking. In administrative law cases, the defendant, i. e. the authority, always explains that she should only examine only the “basic design” of the reactor at the time of the procedure for license. The authority considers “basic design” for a technological concept. But I will not support this view. It is the legal concept of “hindrance to prevention against a disaster” (the former Atomic Law §24(1) Nr. 4) that determines what should be examined in the procedure. In the “Ikata-decision” of 29th April 1992, the Supreme Court followed the popular principle of the civil procedure law that it is plaintiff who should bear the burden of proof, albeit it was an administrative law case. On the other hand, the Court required that the authority should first “prove” that her conclusions were not unreasonable. This means that the Court will accept the consequence of technological thinking of the authority, if she succeeded in “proving” that her conclusions were not unreasonable. But the technological way of thinking is not enough for a good administrative decision. I will state in this essay that the authority should aim at mixing of the related sciences and quest for the best mix of knowledge.

　This paper presents a review of the literatures on “science in a court” focusing on the following points: 1) How should the judicial decision be made, once the scientific knowledge, which is the basis of an act reviewed in the court, has changed? 2) If the scientific knowledge is lacking in the judges, they may seek a written statement of the scientists or the expert witnesses. They, however, might not be able to understand what is written or testified in turn: the dilemma of the scientific trial. How can they overcome that? 3) Firstly, administrative law scholars argue that the judges should not try to review whether the administrative judgment on safety of a nuclear reactor is right or not. However, they ought to concentrate only on a review of the rationality of the administrative process. 4) Secondly, in any case, it would be necessary to improve scientific literacy of the judges. 5) Thirdly, when scientific opinions are divided, judges should not rush to conclude on the basis of one-sided opinion. In many cases the judgments can be given on legal issues before deciding the right scientific opinion. 6) Finally, it is often appropriate to discuss the treatment of new technologies and those potential risks not at the court but at the social schemes such as ADR or Consensus Conference.