The author suggests, arguing from the work of Antibi (2003) that there is an inherent bias in assessment, the macabre constant. It is argued that there is a tendency for markers to create categories of assessment and balance the numbers between these irrespective of pupils / students’ actual performance. Teachers know that if all their students have good grades on a regular basis they will be considered too ‘nice’; conversely, if they consistently give marks below average, they will be considered too “strict”. A ‘good’ assessment thus divides the class into three groups: the ‘good’, the ‘average’ and the ‘bad’. Some students will certainly move to an adjacent group, but the ternary structure will remain. About a third of students are thus condemned to failure regardless of educational conditions: whatever the level of the class, the quality of teaching, the subjects taught, failure will happen. No proposal has so far succeeded in really improving the situation: failure remains. The reason is that all evaluations are finally caught up by the macabre constant. For a significant change to really take place, it is argued that the process of assessment must take account of the central role of the macabre constant.