If among the subdivisions of a species, no single one can be pointed out as playing a primary part, and the others can not be traced back to it, the relation between these lesser units is of course of another character. They are to be considered of equal importance. They are distinguished from each other by more than [36] one character, often by slight differences in nearly all their organs and qualities. Such forms have come to be designated as "elementary species." They are only varieties in a broad and vague systematic significance of the word, not in the sense accorded to this term in horticultural usage, nor in a sharper and more scientific conception. Genera and species are, at the present time, for a large part artificial, or stated more correctly, conventional groups. Every systematist is free to delimit them in a wider or in a narrower sense, according to his judgment. The greater authorities have as a rule preferred larger genera, others of late have elevated innumerable subgenera to the rank of genera. This would work no real harm, if unfortunately, the names of the plants had not to be changed each time, according to current ideas concerning genera. Quite the same inconstancy is observed with species. In the Handbook of the British Flora, Bentham and Hooker describe the forms of brambles under 5 species, while Babington in his Manual of British Botany makes 45 species out of the same material. So also in other cases. For instance, the willows which have 13 species in one and 31 species in the other of these manuals, and the hawkweeds for which the figures are 7 and 32 [37] respectively. Other authors have made still greater numbers of species in the same groups. It is very difficult to estimate systematic differences on the ground of comparative studies alone. All sorts of variability occur, and no individual or small group of specimens can really be considered as a reliable representative of the supposed type. Many original diagnoses of new species have been founded on divergent specimens and of course, the type can afterwards neither be derived from this individual, nor from the diagnosis given.