In nature, repeated mutations must be of far greater significance than isolated ones. How [568] great is the chance for a single individual to be destroyed in the struggle for life? Hundreds of thousands of seeds are produced by _lamarckiana_ annually in the field, and only some slow increase of the number of specimens can be observed. Many seeds do not find the proper circumstances for germination, or the young seedlings are destroyed by lack of water, of air, or of space. Thousands of them are so crowded when becoming rosettes that only a few succeed in producing stems. Any weakness would have destroyed them. As a matter of fact they are much oftener produced in the seed than seen in the field with the usual unfavorable conditions; the careful sowing of collected seeds has given proof of this fact many times. The experimental proof of this frequency in the origin of new types, seems to overcome many difficulties offered by the current theories on the probable origin of species at large. VI. The relation between mutability and fluctuating variability has always been one of the chief difficulties of the followers of Darwin. The majority assumed that species arise by the slow accumulation of slight fluctuating deviations, and the mutations were only to be considered as extreme fluctuations, obtained, in the main, by a continuous selection of small differences in a constant direction. [569] My cultures show that quite the opposite is to be regarded as fact. All organs and all qualities of _lamarckiana_ fluctuate and vary in a more or less evident manner, and those which I had the opportunity of examining more closely were found to comply with the general laws of fluctuation. But such oscillating changes have nothing in common with the mutations. Their essential character is the heaping up of slight deviations around a mean, and the occurrence of continuous lines of increasing deviations, linking the extremes with this group. Nothing of the kind is observed in the case of mutations. There is no mean for them to be grouped around and the extreme only is to be seen, and it is wholly unconnected with the original type. It might be supposed that on closer inspection each mutation might be brought into connection with some feature of the fluctuating variability. But this is not the case. The dwarfs are not at all the extreme variants of structure, as the fluctuation of the height of the _lamarckiana_ never decreases or even approaches that of the dwarfs. There is always a gap. The smallest specimens of the tall type are commonly the weakest, according to the general rule of the relationship between nourishment and variation, but the tallest dwarfs are of course the most robust specimens of their group. [570] Fluctuating variability, as a rule, is subject to reversion. The seeds of the extremes do not produce an offspring which fluctuates around their parents as a center, but around some point on