For a polystyrene melt with a zero shear rate viscosity of 44.5 kPa s at 190 degreesC, shear stress and first normal stress difference, in step shear rate experiments with cone-standard plate and cone-partitioned plate tools are compared. The cone angle alpha is 8.5degrees throughout. With the partitioned plate and a central stem of radius 4 mm, steady stateviscosities can be obtained up to shear rates of 100 s(-1), about a factor of 3 higher than for the standard plate tool. The strain at the maximum viscosity grows from 23 below gamma = 3 s(-1) to 6.5 at gamma = 100 s(-1). This increase can hardly be seen with the standard plate tool. The ratio of the maximum to the steady state viscosity shows a tendency to saturate beyond gamma = 30 s(-1) and reaches 1.75 at gamma = 100 s(-1). This finding is again beyond the scope of the standard plate tool. The ratio of the maximum to the steady state first normal stress difference only shows a minor increase to 1.03 at gamma = 30 s(-1). N-1 beyond this strain rate cannot be determined because the maximum thrust of 20 N narrows the range of sample radii too much. It is found that experiments at high shear rates are not limited by edge fracture but by insufficient normal force transducer capacities and by viscous dissipation heating. Calculations show that small cone angles (below 6degrees) should be used beyond gamma = 30 s(-1) to deal with this. Small cone angles necessarily require very stiff rheometers for a good short time resolution. (C) 2003 The Society of Rheology.