Fitting the quark and lepton masses in string theories

The capability of string theories to reproduce at low energy the observed pattern of quark and lepton masses and mixing angles is examined, focusing the attention on orbifold constructions, where the magnitude of Yukawa couplings depends on the values of the deformation parameters which describe the size and shape of the compactified space. A systematic exploration shows that for $Z_3$, $Z_4$, $Z_6$--I and possibly $Z_7$ orbifolds a correct fit of the physical fermion masses is feasible. In this way the experimental masses, which are low--energy quantities, select a particular size and shape of the compactified space, which turns out to be very reasonable (in particular the modulus $T$ defining the former is $T=O(1)$). The rest of the $Z_N$ orbifolds are rather hopeless and should be discarded on the assumption of a minimal $SU(3)\times SU(2)\times U(1)_Y$ scenario. On the other hand, due to stringy selection rules, there is no possibility of fitting the Kobayashi--Maskawa parameters at the renormalizable level, although it is remarked that this job might well be done by non--renormalizable couplings.