Nature and six journals on analytical chemistry have been studied scientometrically. A mathematical model describing the evolution of the number of articles with time is proposed. This model approximates satisfactorily the publication dynamics but cannot be used for prognosis because of bifurcation.