Ill two recent papers (~':) the most accurate exper imenta l verif icat ions of special r e l a t iv i ty have been summarized. There are m a n y accurate a tomic (or part icle) experiments but only three macroscopic exper iments . Two of then1 are null exper iments : ~he Miehelson-Morley (2) in terference (which excludes ether drift, effects) and the t iming of pulses f rom binary stars (~) (which excludes a ve lec i ty dependence of l ight on the source velocity) . Only the th i rd exper iment gives a mt~eroseopic posi t ive resul ts : the proof of the so-called (~ twin parodox ~) by moving clocks (a). Unfor tmla te ly the accuracy of this exper iment is v ( r y poor (300o as given in the original paper (a) and not 3~o as repor ted in ref. (1)). But there is ano ther more importan~ fact. All the exper iments so far perfor lned bring no evide~ce against the dragged, compressible c ther of Stokes and P lanck (a). i ndeed the Miehelson-Morley exper iment and the stellar aber ra t ion are correct ly explained by the S tokes-Phmck theory (~). Also the mass dependence on the speed and all the electronm~'netie phenomena are all t aken into account if one assumes the Maxwell equMions as w~lid only in the reference frame at rest wi th the dragged ether. A Michelson-Morley exper iment performed on a satell i te could disprove such a theory. We propose here a , o t h e r feasible exper iment for d iscr iminat ing be tween the StokesPlal lck theory a~ld speeial re la t iv i ty . I t would be a new macroscopic test which should give a posi t ive (non null), easily observable result , at the same t ime being local, since i t does ~of require different clocks brought around the Ear th . Moreover i t would be the first exper iment nwasur ing the length contract ion. The conceptual exper inw, l ta l set-up is ve ry s imple: it consists of two al ternators , one wi th a ro ta t ing a rmature and the o ther wi th a ro ta t ing inductor . The difference between thei r highest e lec t romot ive forces (e.m.f.) is the desidered re la t iv is t ic effect. Precisely (see fig. 1), the set-up should consist of a s t ra ight wire ?A." 1 ro ta t ing between the opposite poles of a single (or of nn)re) nmgnet . Another equal nlagnet is fixed to the same ro ta t ing shaft w]fich drives wi, and induces an e.m.f, in another wire w 2 equal to wl but at rest witfi the laboratory. The (e.m.f.) 1 induced in the ro ta t ing wire w 1 is