A careful analysis of the (001)-reflection profiles of different pyrolytic carbons and of the modulation of the (ll)-and (30)-rods based on the theory of paracrystals proves that the model of Ruland cannot explain all experimental results, because if contains no bimodal statistics. Franklin's degree of graphitisation I -p and Mering-Maire's parameter g are not sufficient to describe the paracrystalline structures, because they give no quantitative measure of the probability n for T-defects and other variances. It is shown that the mean distance of layers with T-defects is not A = 0.09 1%1 larger but for low ordered pyrolytic carbons and tar coke is larger than 0.2 A depending on the history and treatment of the carbons. High ordered pyrolytic carbon gives a value A = 0.04 A and a value a so small that Mering-Maire's model is practically applicable. The a-values for all other samples under consideration are only about l/3 of Franklin's p-values. In low ordered pyrocarbons two components with different &-and u,-values are found. The main result is that all pyrolytic carbons under consideration have bimodal distance statistics of the layers, which can be quantitatively described by two independent parameters w and A, different from p resp. I -g' and from Franklin's A = 0.09 A values. The integral widths of the reflections along the (1 l)and (30)-&s can be anaiysed qu~titatively with the help of the theory of paracrystals as lateral v~i~cesof (ri2 and a&. In the theory of mesophases their values can only be 0 or 1. As a consequence of a bimodal statistic we get information of & within the T-defect free layer stacks and 01~ between these single stacks.