Two new methods for finding the second normal stress difference are proposed; one depends on an analysis of edge effects in a cone‐plate device, the other depends on the shape of the free surface in an open‐channel flow. Additionally, the effect of the edge conditions on the total thrust in a cone‐plate device is considered carefully. Some possible errors in the interpretation of these measurements is found. For two solutions (polyisobutylene/cetane, polyethylene‐oxide/water) the viscometric functions are presented. In both cases the second normal stress difference was found to benegative, in contrast to some earlier conclusions.