+---------+-----------+-------+---------------+ | Season. | Littoral. | Tell. | High plateau. | +---------+-----------+-------+---------------+ | Winter | 41 | 36 | 45 | | Spring | 27 | 32 | 46 | | Summer | 4 | 7 | 11 | +---------+-----------+-------+---------------+ ] [Footnote 4: The most recent records available, 1897-1908, do not give a satisfactory account of the precipitation on the desert. For instance, meteorological records covering eight years for El Golea do not report on the rainfall. Records of five years at In Salah take the rainfall into account on one year only, and on that year no precipitation occurred.] [Footnote 5: Unless otherwise stated, the climatological statistics given in this paper were taken or compiled from Observations Météorologiques du Riseau Africain, 1907-1908. The evaporation data are based on readings of the Piche evaporimeter. The amount of evaporation given in the text can be reduced to the evaporation from a free-water surface by multiplying by 0.737 (Meteorological Notes, J. I. Craig. Cairo Scientific Journal, vol. vi, May 1912).] [Footnote 6: At Touggourt, however, no rain was reported during the summer season of 1908.] [Footnote 7: Kearney and Means, _loc. cit._] [Footnote 8: The Root Habits of Desert Plants. W. A. Cannon. Carnegie Institution of Washington Publication No. 131. 1911. This paper gives a descriptive classification of the main root-types in the deserts of the southwest, in which such a root as found in _Acanthyllis_ is called _generalized_, in distinction to roots like those of most of the cacti, or _Zizyphus_, the former having a system wholly superficial and the latter a system wholly deeply placed, as _specialized_. The specialized root-systems appear to be so fixed in character as to be not easily changed, while the generalized type is flexible. It will be self-evident that the type of root-systems may be of great importance in determining the local distribution of a species.] [Footnote 9: Compare the root-system of _H. scoparium_ at Biskra, p. 64.] [Footnote 10: The most striking change in the general character of the vegetation which the traveler notices as he goes from the less arid to the more arid portion of southern Algeria is its decrease in amount. This occurs through dwarfing effects of whatever cause and through a decrease in the number of individuals. Within certain limits the results observed are to be attributed mainly to the first of these, since there is often a surprisingly large number of perennials on any given area. But in other and more intensely arid regions (as portions of the Arabian-Egyptian desert, and indeed a limited area on the hamada between Ghardaia and Ouargla) plants are wholly wanting. Whether such is generally the case on the reg or the hamada farther south in southern Algeria is not known.]