A variety of simple measurements were made on Baobab trees (Adansonia digitata) near Kibwezi, Kenya, to gain some information on their water economy. Water loss from cut shoots was compared with that of eight other woody species growing in the same region. The Baobab was found to be among the most effective at controlling its water loss. Daily shrinkage of the trunks was measured, giving an estimate of c. 400 litres daily water deficit when the trees are in leaf. Seasonal shrinkage indicates a loss of up to 1500 litres of water during dry periods. An attempt was made to determine root distribution. Measurements on exposed roots show that they are relatively shallow (down to c. 1.8 m), but are spread out to a distance greater than the heights of the trees. Such an extensive shallow root system is probably the best adapted to exploiting the low annual rainfall, most of which falls in the form of infrequent heavy downpours. ALTHOUGH THE BAOBAB is one of the most familiar trees in the drier parts of Africa, very little work has been done on its ecology or physiology. The distribution of the species is fairly well documented; it occurs in the semi-arid areas south of the Sahara, where the rainfall is in the range 150 to 1500 mm (with the greater part of the distribution below 800 mm), and from sea level to 1250 m in altitude (G. E. Wickens, pers. comm.). Its range extends right across the continent south of the Sahel region, as well as into east, central, and southern Africa. It is of considerable ecological importance, providing food for a wide range of animals, and also acts as an important source of water for both man and beast during conditions of drought. The baobab is of local economic importance as a producer of edible fruits and leaves, and as a source of fiber. Because of their great size and bizarre shapes, baobabs are often the most prominent tree species where they occur. They are the subject of many local legends. Although numerous attempts have been made to determine the ages of individual trees (Hobley 1922, de Villiers 1951, Swart 1963), only one paper seems to have been published dealing specifically with the ecology of this species (Owen 1974). The present study examines the role of water relations in the baobab's physiological ecology. The methods used consist of very simple measurements made in the field over a period of one year. Although the results must be rather provisional at this stage, they do at least provide some quantitative data on the water economy of this species. THE SITE AND VEGETATION All the measurements were made on trees in the 1Present address: Biology Department, The University, Southampton S09 5NH, England. vicinity of Masalani Camp, 25 km east of Kibwezi, Kenya (38.10E, 2.30S), at an altitude of 670 m. The vegetation of the area is similar to the bushland described by Lind and Morrison (1974) as occurring in the Tsavo National Park which adjoins the study area. Pratt and Gwynne (1977) refer to this community as 'dry thorn-bushland' occurring in their 'Eco-climatic zone V.' Where the vegetation is undisturbed, the trees and shrubs form three strata: Up to 3 m there is a dense impenetrable thicket formed by Acacia mellifera, A. senegal, Boscia coriacea, Combretum exalatum, Commiphora boiviniana, Grewia bicolor, and G. villosa; between 3 and 8 m approximately there is a stratum of scattered trees with an open canopy. The main species present are Acacia tortilis, Commiphora africana, C. baluensis, C. campestris, Delonix elata, and Sterculia rhynchocarpa; and the third layer consists of occasional emergent Adansonia digitata trees which are mostly 20-25 m in height. Although they are very prominent, these trees are relatively infrequent, perhaps not more than 10 per km2. They are remarkably uniform in size, so most of them may be about the same age. Of the 21 specimens measured, 16 were standing in undisturbed vegetation. In the case of the other five, the surrounding natural woody vegetation had been cleared, and had been replaced by herbaceous species characteristic of disturbed habitats. With the exception of Acacia tortilis and Boscia coriacea, nearly all the woody species lose their leaves during the long dry season from May to October. The baobabs spend only four months of the year in leaf. In the area studied, the leaves were shed in midFebruary 1978 and did not re-appear until mid-Octo-