The coding parts of DNA sequences are regarded as clusters of connected sit
es of a random Canter-like set, while the non-coding parts are regarded as
the empty regions of the same set. Under this representation, we find that
higher eucaryotes are mapped on random Canter sets with fractal dimension a
round 0.85, while lower organisms are mapped on Canter sets with fractal di
mension 1. This result indicates that the coding/non-coding partition in th
e DNA sequences of lower organisms is homogeneous-like, while in the higher
eucaryotes the partition is fractal. This result agrees with the power law
distribution observed in the non-coding parts of higher eucaryotes.