The large-scale temperature, salinity, oxygen and nutrient distributions of the North Atlantic Ocean are explored with the aim of identifying the minimum data set necessary to describe the general circulation and to identify the dominant physical characteristics. A form of empirical onthogonal function (principal component analysis) based upon the singular value decomposition is used. Six modes are capable of describing 92% of the variance of the full hydrographic fields. A second singular value decomposition applied to the horizontal property modes shows that there are only two independent tracers among the six used. The modal reconstruction satisfactorily reproduces the large-scale transport properties of the original data. A clear physical interpretation of the modes has not been found, but they are obviously connected to the large-scale water mass characteristics of the ocean. Much remains to be learned about how to use such descriptions.