An important point about these mutations is that their production is a constant feature of _Lamarckiana_. Whenever large numbers of the seeds of this plant are grown, a certain proportion of the plants developed present these _same_ mutations; not always all of them--some may be absent in one culture, present in another, but four of them are fairly common and of constant occurrence. The total proportion of mutant plants compared with the normal was 1.55 per cent. in one family, 5.8 per cent. in another. It would appear therefore, supposing that mutations arose subsequently in the same determinate way from previous mutations, that evolution, though in a number of divergent directions from one ancestral form, would proceed along definite lines, and that there would be nothing accidental about it. We should thus arrive at a demonstration of what Eimer called orthogenesis, or evolution in definite directions. The mutation _lata_ cannot be said to breed true, as the pollen is almost entirely sterile. It has therefore been propagated by crossing with _Lamarckiana_ pollen, with the result that both forms are obtained with _lata_ varying in proportion from 4 per cent. to 45 per cent. _Rubrinervis_ is a mutation from _Lamarckiana_, chiefly distinguished by red midribs in the leaves and red stripes on the sepals. When propagated from self-fertilised seed it produced about 95 per cent. of offspring with the same characters, and the remaining 5 per cent. mutants, one of which was _laevifolia_ which had been found by De Vries among plants growing wild at Hilversum. Gates obtained a single plant among offspring of _rubrinervis_ in which the sepals were red throughout, and to this he gave the name _rubricalyx_. When selfed this plant gave rise to both _rubricalyx_ and _rubrinervis_, and in the second generation when the _rubricalyx_ was selfed again the numbers of the two were approximately 3 to 1. _Rubricalyx_ is therefore a dominant heterozygote, and this fact was further confirmed in the third generation when a selfed plant gave 200 offspring all _rubricalyx_, the mother plant having evidently been homozygous for the red character. In this case, therefore, we have what Bateson was seeking, the origin of a new dominant character under observation, the original mutation having arisen in a single gamete of the zygote which gave rise to the plant. It is claimed by mutationists that mutations are not new combinations or separations of Mendelian unit characters already present, but are themselves new characters, though not always necessarily, as in the case of _rubricalyx_, new unit characters in the Mendelian sense.