fell to the lot of the last-formed or uppermost leaves of the set. As need arose for better protection of the infant shoots, the simplest way of meeting it would be to increase the efficiency of the parts already in use by widening them as far as might be necessary. As this was going on, the same fate which overtook the lateral leaflets of the original three would now extend to the terminal one of each of these upper leaves; for with the shortening of the stem they would be brought to lie so closely above the others as to shade them injuriously if not reduced in size. Moreover, as being the latest to develop, they would get but a small share of the reserve food provided for the rosette. Still, their relation to the supply of nutriment as well as their uselessness or power for harm in the rosette would, after all, be more a matter of degree than in the case of the lateral leaflets, since these latter would have to lie practically in one plane and so must interfere not only with the terminal leaflets but with each other. This may help us to explain why, although the lateral leaflets have so entirely
Fig. 11.—Berberis vulgaris. Series of spiny leaves passing into spines.
disappeared, we still find on some of the lower bud scales traces of a blade which thus afford connecting links in our morphological chain (Fig. 13).
This evolution of the bud scales must, of course, have been closely connected with that shortening of the petiole which we have already noticed in the typical rosette leaves as having culminated in the production of persistent overlapping scales forming an outer bark for the secondary branches; and it would seem most probable that the development of bud scales and bark scales proceeded side by side. Finally, as accounting further for their similarity of form, it may be remarked that in both, the protective function, at first merely incidental to that of mechanical support, comes at length to be the sole use for which they are retained: in one case it is a matter of years, in the other of generations.
