Ajter a brief discussion of work on experimenter-bias, it is suggested that the possibility of less well documented biases should be considered. A number of such biases are suggested. They are presented in terms of their source, namely the political ideologies, cultural backgrounds, biographical characteristics and personal characteristics of scientists. In considering implications of biases, three reactions are discussed: ignoring, controlling and understanding them. In particular, it is proposed that understanding the operation of bias might be furthered by working towards a taxonomy of biases, organized in terms of the sources of biases and the points in the research process at which they intrude.