The five principles that make up the hypothesis of columnar organization for somatosensory cerebral cortex are considered in relation to their experimental foundations. Several important implications that flow from these principles are examined, and the requirements for their experimental evaluation are elaborated. A number of specific measurements are found to be lacking in necessary precision: the total amount of somatosensory cerebral tissue in each area, the sizes, shapes and degree of overlap of aggregate excitatory receptive fields, the number of distinct modalities and the modality equivalences in our different linguistic categories, the sizes and shapes of the columns and, finally, the number of columns for each modality and the total number of columns. Some possible columnar arrangements are set up, and their experimental detectability is assessed, using optimal conditions and values. It is shown that if a columnar structure exists, it defies clear detection by current neurophysiological techniques and experimental approaches. The concept of an ''elementary functional unit'' is found to need clear definition; taken at face value, it is shown to yield some rather unusual predictions. The conditions under which the hypothesis of columnar organization can be distinguished from its rival, the hypothesis of topographic organization, are also reviewed.