A hypothesis is described on the nature of gene splicing. It is based
largely on the notion of multiplicity of codes carried by the nucleoti
de sequences. That is, not only the classical triplet code is harbored
by the sequences, but at least several other codes of different natur
e-like chromatin code and framing code. The codes all overlap, so that
one and the same letter frequently, if not always, partakes in severa
l messages of different types. Such overlapping, however, compromises
each of the necessarily degenerate messages involved, so that the best
functional performance (closeness of the corresponding sequence patte
rn to an ideal one) of each one of them can only be achieved if the me
ssages (codes) are spatially separated, not superimposed anymore. This
is suggested as a possible reason for the original insertion of numer
ous intervening sequences into eukaryotic genes. Due to this fragmenta
tion of the protein-coding sequences the exons, presumably, are primar
ily loaded by this coding function, while intervening sequences are pr
imarily responsible for formation of sequence-specific chromatin struc
ture. Computer experiments with the sequences are described, strongly
supporting this view on the nature of gene splicing in eukaryotes.