The wood in standing trees undergoes internal stress during the entire life of the tree. This stress, commonly named growth stress, originates in maturation strains and is impeded by the mass of the entire trunk. Released strain at the stem surface (ie strain on a small piece of wood isolated from the stem by cutting grooves, drilling holes etc) measures maturation strain. Our research program on 'Tree architecture, anatomy and mechanics' aims at i) understanding the biological control of maturation strain with regard to tree morphogenesis (branching patterns, crown form, stem and shoot positions in relation to vertical direction); ii) qualifying correlations between maturation strain and the anatomical features of the wood and other wood characteristics (stiffness, shrinkage, hygrothermal recovery); iii) modelizing and qualifying the strew in the entire tree that results from the cumulative maturation of successive layers during cambial growth, in order to understand cracks and strains when the wood is processed. This paper focuses on results concerning longitudinal maturation strain at the stem periphery measured on different species. Two methods have been used. i) the single hole method with a special sensor designed in the Centre Technique Forestier Tropical; and ii) measurements of strain due to 2 grooves sawn above and below a classical electric sensor. Comparisons between the 2 methods show quite good, but not perfect, agreements in beech and eucalyptus but not in chesnut. These results are discussed from the sensor dimensions, the principles of the method and the anatomical and mechanical properties of wood. Measurements on one tree show important variations with height and angular position, which are correlated to tree morphogenesis (proximity of branches and righting movements of stems). High strain values are never homogeneous in the tree but concentrated in small angular sectors. This angular asymmetry of maturation strain is obviously related to stem bending movement as one side of the stem 'pulls' or 'pushes' the other. Furthermore, histograms of values measured on Pinus pinaster, clones of Eucalyptus (PF1 1.45, UAIC-CTF7, Congo) and poplar (Populus euramericana cv 1214), Castanea saliva, Fagus sylvatica and Eperua falcata show that the distribution of strains is not Gaussian. The long tail of tensile (in hardwoods) or compressive (in softwood) values correlates with the formation of reaction wood (compression or tension). The main difference between populations is not the mean value (out of the tail, in normal standard wood) but the width and maximum of the tail. Hence, to study variability of growth stress in a population of trees, we must study the frequency and the magnitude of peaks of high maturation strains' within trees, rather than mean strains. Therefore, growth stress should be analyzed in correlation with the regulation of the form of the tree and particularly in correlation with the kinetics of stem movement (changes of curvature and lean). Finally, uncommon patterns of release strain with 2 opposite angular peaks have been observed in some tropical trees as Dichostemma sp, Saccoglotis gabonensis, Eperua falcata and Castanea sativa. These patterns can be related to the tree architecture. In sympodial trees (ie trees in which the trunk is built by stacks of branches as stems formed from axillary buds take over from the former leader), peaks of high maturation strains seem to be induced by the different axes (the present leader and the former) and thus, in a cross-section, 2 flows of highly strained wood can be observed. A functional explanation of such patterns is not obvious.
