Most 6-axis force and torque transducers (FTT) have a common feature: they are statically indeterminate. This fact induces some highly undesirable effects on the performance: the coupling of forces and moments, and the axial and lateral sensitivity differences. These drawbacks seriously impair the performance of the force control loop. A possible remedy is to choose determinate structures with the definitive payoff of better controllable performance indices. There is not much freedom to select such a design. It would have to be like a Stewart platform (SP). The only problem is the realizability. The central problem for the mechanical SP design is the choice of the bearings of the supporting structure. Due to these bearings we have only linear stress in the supporting limbs. Simple ball and socket joints are inapplicable. The backlash and stick slip caused by dry friction lead to intolerable nonlinearities as found in Gaillet and Reboulet . They equipped the SP with a delicate bearingless isostatic supporting structure. This design is hard to miniaturize. Our proposal has SP geometry and is equipped with elastic joints, see Fig. 1. This monolithic design provides of two necks on the supporting Stewart limbs. The compliant behavior of the necks provides the partial torsional (rotary) and bending (cardanic) compensation of the limbs. The result of this compensation should be an approximate linear stress in the supporting framework. Our aim was to develop a linear elastic model with concentrated parameters so that we could examine the influence of the elastic joints. To cope with this task we have to solve a statically (or kinematically) indeterminate problem. The solution delivers the computed stiffness matrix, and thus can be compared element by element with that of a SP based FTT with ideal frictionless joints.