[4] See later. Galton gave a simple diagrammatic representation of his law in _Nature_, 1898, vol. LVII. p. 293. But already many cases are known to which the rule in any simple form will not apply. Galton points out that it takes no account of individual prepotencies. There are, besides, numerous cases in which on crossing two varieties the character of one variety almost always appears in each member of the first cross-bred generation. Examples of these will be familiar to those who have experience in such matters. The offspring of the Polled Angus cow and the Shorthorn bull is almost invariably polled or with very small loose “scurs.” Seedlings raised by crossing _Atropa belladonna_ with the yellow-fruited variety have without exception the blackish-purple fruits of the type. In several hairy species when a cross with a glabrous variety is made, the first cross-bred generation is altogether hairy[5]. [5] These we now recognize as examples of Mendelian ‘dominance.’ Still more numerous are examples in which the characters of one variety very largely, though not exclusively, predominate in the offspring. These large classes of exceptions--to go no further--indicate that, as we might in any case expect, the principle is not of universal application, and will need various modifications if it is to be extended to more complex cases of inheritance of varietal characters. No more useful work can be imagined than a systematic determination of the precise “law of heredity” in numbers of particular cases. Until lately the work which Galton accomplished stood almost alone in this field, but quite recently remarkable additions to our knowledge of these questions have been made. In the year 1900 Professor de Vries published a brief account[6] of experiments which he has for several years been carrying on, giving results of the highest value. [6] _Comptes Rendus_, March 26, 1900, and _Ber. d. Deutsch. Bot. Ges._ xviii. 1900, p. 83. The description is very short, and there are several points as to which more precise information is necessary both as to details of procedure and as to statement of results. Nevertheless it is impossible to doubt that the work as a whole constitutes a marked step forward, and the full publication which is promised will be awaited with great interest. The work relates to the course of heredity in cases where definite varieties differing from each other in some _one_ definite character are crossed together. The cases are all examples of discontinuous variation: that is to say, cases in which actual intermediates between the parent forms are not usually produced on crossing[7]. It is shown that the subsequent posterity obtained by self-fertilising these cross-breds or hybrids, or by breeding them with each other, break up into the original parent forms according to fixed numerical rule. [7] This conception of discontinuity is of course pre-Mendelian.