This study was based on the assumption that it is not psychologically useful to research all criminals whether detected or not, as one group. It is argued that the group usually referred to as delinquent, is made up of a relatively homogeneous collection of law breakers, and further, that the essential characteristic of this group is a high degree of hostility. Previous work has indicated that prison staff tend to be highly authoritarian, and the theory concerning authoritarian people suggests that they are highly hostile, although this has not been emperically verified. It was therefore decided to investigate the relationship between these factors in the present study. Because it is considered that hostility is central to delinquency, it is likely that any significant effect of Borstal Training would show itself in a change of hostility levels. Both the frustration and modelling based theories of hostility, seem to suggest that the Borstal is unlikely to reduce hostility, particularly if, as is argued , most staff are of the authoritarian personality type. The study used four measures, one verbal and one non-verbal measure of authoritarianism, and one verbal and one mainly non-verbal measure of hostility. As expected the non-verbal tests gave better results with the delinquent samples. The results of the analysis of authoritarianism and its relationship with hostility were at best inconclusive and no significant change in authoritarian attitudes occurred as the result of Borstal Training. The hypothesis that delinquency is related to hostility was supported by an analysis of hostility scores and their relationship to four indices of delinquency. It was found that Borstal Training was associated with a very significant increase in fantasy hostility. The verbal measure of hostility gave few significant results, but neither did it suggest any conclusions contrary to those suggested by the non-verbal measure. Overall the two tests correlated with one another significantly. The non-verbal measure, the Rosenzweig Picture-Frustration Study scored according to the method devised by Chorost, proved a very sensitive instrument showing good correlations with the indices of delinquency and a highly significant increase after Borstal Training. The staff sample was found to be highly authoritarian as expected, and there was no evidence that the level of authoritarianism was decreasing. It was concluded that hostility is a basic factor in delinquency and that Borstal Training as it now operates increases hostility. It was also concluded that the employment of less authoritarian staff will not be sufficient to bring a change, because such staff are unlikely to remain long in the job as presently defined.