Is a spectrograph of hidden variables possible?

Résumé : A new definition of "Realism" is proposed: it is that a gedanken "spectrograph" of hidden variables behaves as an actual (say, wavelength) spectrograph. The question is: does this definition allow, by itself, the derivation of Bell's inequalities? If it were, then such a spectrograph would be impossible, for Bell's inequalities are observed to be violated. In this short paper it is reported that, on the contrary, such spectrograph is compatible with the violation of Bell's inequalities. This result puts some new light on the controversy about the hypotheses necessary to derive Bell's inequalities. In particular, "Spectrograph's Realism", and "Locality", are proven to be different, and both necessary, hypotheses to derive Bell's inequalities.