The paper draws together a number of findings from the PhD Examination Project conducted by the authors as part of the SORTI (Centre for the Study of Research Training and Impact) program conducted at The University of Newcastle. Its particular focus is on an analysis of the roles of examiner and supervisor, and the interface between these, as seen through the lens provided by Habermas's 'Ways of Knowing' thesis. Early scripts appraised of examiner comment directed to the candidate seemed to reveal a preponderance of text that conformed with Habermas's 'empirical-analytic' way of knowing, displaying a fairly technical approach to the task and positioning the examiner in the role of 'expert'. At the same time, there was little evidence of 'self-reflective' knowing that might betray a more sophisticated task being undertaken and a role of some asymmetry between the examiner and candidate. Since the latter way of knowing would seem to fit better with a regime dealing with original thought and new contributions to knowledge, it has been postulated that the dominant text in PhD examination may work to constrain the generation of new knowledge rather than encourage it. A recent paper (Lovat & Morrison, 2003) explored this postulation with special reference to those aspects of examination script that made explicit mention of the role of the supervisor, finding essentially the same phenomenon, but with a slightly different balance in favour of 'self-reflective' text. This paper will draw strands out of SORTI's more comprehensive work that might inform this particular analysis, expand on the analysis itself and indicate ways in which the analysis could inform the practicalities of research training and especially the role of the supervisor.