In the classical equations for the traction on concentric cylinders by a viscous substance, it is assumed that traction on the top and bottom of the inner cylinder (bob) is negligible in comparison with that on the curved surface. In viscometers of practical dimensions, however, it is necessary to compensate for the end‐traction or end‐effect by adopting one of several expedients in design or by determining the magnitude of the effect and accounting for it in terms of increased length of bob.In the experiments reported in this paper, the method of multiple bobs has been used, and the variation of end‐effect with the following factors studied: (1) radius of bob; (2) clearance between cylindrical surfaces; (3) separation between bottom of bob and cup; and (4) viscosity.The magnitude of the end‐effect increases with radius and with clearance. For separations at the bottom greater than 1 cm and for viscosities above 1 poise the end‐effect is nearly constant, but must be determined for each cup and bob combination. At lower viscosities the correction must be determined either by calibration with a standard liquid of about the same viscosity as the unknown, or by the multiple‐bob method.It is shown that trapping a layer of air beneath the bob is not effective in making the traction on the bottom negligible. The end‐effect is almost as large for a bob with an open bottom as for a closed one.When the method of multiple bobs is used, data with an accuracy of ±2 percent can be obtained without calibration. When the instrument constant is determined with standard liquids or computed from a value for the end‐effect previously found, data of somewhat lower accuracy result, but the uncertainty should be within ±5 percent.