retrogress unless that tendency is checked by selection; for, as regards any trait, an individual may vary from his parent primarily in two ways. He may vary from him either towards the ancestry or away from it, i.e. he may either undergo retrogression or evolution; and, so far as we know, the chances are equal of his doing either the one or the other. But if he vary away from the ancestral type, it does not necessarily follow that the variation will constitute an extension of the previous evolution. It may constitute a reversal of it, or a divergence in an altogether new direction, and, therefore, in the absence of selection, the variation of the offspring from the parent must tend on the whole to bring about retrogression—a tendency which is checked and reversed in an evolving species only by a sufficiently severe process of selection.
Though each multicellular organism has its starting-point in a unicellular organism (the germ cell), yet nevertheless each germ cell, counting from the time of the first evolution of the multicellular organism, when all the cells were more or less germ cells, through succeeding generations, during which ever-increasing differentiation in structure and specialization in function took place, must have become in one sense a more and more complex entity, differing more and more from its ancestor, the unicellular organism, in that it was the starting-point of a more and more complex and heterogeneous multicellular organism. In other words, each successive conjugating germ cell differed more and more from its ancestor, the conjugating unicellular organism, not only in that its non-conjugating descendants, however remote, remained adherent in one mass, but also in two other particulars.
First, in that definite lines of its cell-descendants multiplied at increasingly different rates, whereby were produced differentiations in the shape of the whole
