probably most of the rootlets formed in any season perish before the close of the succeeding dry season.
As regards the second class of roots referred to, there is quite a different story to tell. These roots are apparently quite as deciduous as the leaves of many plants. It should be noted, however, that nothing has been done in an experimental way to test the longevity of these roots, and it is reasonable to suppose that under some favoring conditions they might endure, possibly becoming converted into large laterals, even if under conditions, which are the usual ones, their life is limited.
It may be well to describe the roots referred to. If we examine a root of such a desert shrub as Franseria, we shall find, along such of the roots as extend in a more or less horizontal direction from the stem of the plant, groups of filamentous rootlets. These occur at about 1 cm. intervals, in varying numbers usually about one half dozen together. They are from two to four cm. long and probably not more than one half millimeter in thickness. The rootlets appear promptly with the coming of the summer rains, and they cease their activity when the soil attains to an unbearably dry condition, as perhaps in adobe soil, 10 per cent, moisture, more or less.
The deciduous rootlets greatly increase the absorption surface without, at the same time, necessitating invasion of new root areas, or of causing a long transfer of water from the place of absorption to the stem. So far as is known, the deciduous rootlets are formed only when there is an abundance of water, and when the temperature is high. These rootlets have been seen on most of the desert shrubs, on all in the vicinity of the Desert Laboratory, and have been observed on a few of the shrubs in southern Algeria. Whether a similar kind of rootlets occurs on perennials in the more humid regions is not known to me.
The deciduous rootlets are thus of great importance to such desert plants as bear them. They appear adventitiously always, and apparently in the same place on the root year after year. In certain species it has been observed that the adventitious roots are formed precociously, but in other forms this is not the case. And again, where such rootlets are not to be found, it appears that they can not be induced.
The extension of the roots of the desert shrubs is various, perhaps in no case exceeding three or four meters. The position in the ground is also not uniform. In most instances the position occupied by the roots is characteristic for the species, but it is likely that the extension of the root-systems varies mainly with the age of the individual.
There are three main types of root-systems to be found in the shrubs of the desert plants of the southwest. (1) Root-systems which extend horizontally from the main plant axis and lie, for their whole course, near the surface of the ground. (2) Root-systems which are characterized by a strongly developed tap root going directly down to a depth