For social network analysis, the most common method used to generate social network data is the method of Name Generator. Nonetheless, the number of nomination has been all unsolved mystery for social networkers. Namely, for each respondents, to name how many people are good enough to generate a stable network, which is able to represent the truly association structure among these respondents, still, is all empirical research question for researchers. This study devoted to explore this question and to provide a preliminary answer. A set of social network data was collected from a sample of Taipei metropolitan junior high schools, including 44 classes. In each class the Students were asked to nominate tell best friends in the intimate order. It Was supposed that in each class has tell sociometric data for different nomination, and the total amount of sociometric data was 440. The software UCINET6.0 was applied to analyze the social network variables, and NEGOPY4.30 to define the network position. Comparing the betweenness, constraint, and efficiency, this study found that two names will generate more diverse network position with unstable Structure, three names are the minimum to get more stable network structure, four or five names are needed to observe the links between boys and girls, but more than five names seem to be redundancy.