The resistant stress was evaluated by Petersson's model  as shown in Figure 4. The region where this stress was acting could be called as the fracture process zone, and the zone was actually observed by Ohtsuka using X-ray technique . In order to obtain the appropriate magnitude of the tensile resistant stress in the fracture process zone, the computation was repeated until the proper coincidence was obtained between the crack width and the cor responding stresses. In detail, it was necessary to determine the principal direction of stresses in the fracture process zone. Since the nominal direction of the principal stress in the elastic range was com puted as about 60 degree to the assumed failure plane at the maximum load, the stresses in the fracture process zone were, then, supposed to act in the same manner (Figure 3). Finally, the stress distribution at the maximum load of 6.3 tonf was obtained as shown in Figure 5. PROPOSED FORMULA FOR PULL-OUT CAPACITY In order to extend the analytical results, the stress distribution in Figure 5 should be examined with relation to the embedment length. Since the shape of ruptured cone was identical irrespective of the embedment length, the stress acting area in the elastic range should increase proportionally to the second power of the embedment length (h). It is because both the band width and the periphery length of the elastic stress zone increase in proportion to (h). On the contrary, the band width of stress acting area in the cracked zone could be constant because it depends upon the absolute crack width. The periphery length is only proportional to the embedment length (h). Comparing the analytical results with the test results, the proposed equation for the maximum capacity is, then, expressed as follows: Pmax= 18-°-ft (0-9*h + 0.1•h2) (D where, f j^tensile strength of concrete ( = 0.58-(f )^ 2/3 , f^ ;compressive strength of concrete, kgf/cm ) and h;embedment length (cm). P max is expressed in (kgf). The applicability of Eq.(l) was examined by the test data of this program and the results are shown in Figure 6 in which the equation proposed by Eligehausen  was compared. His equation was as follows: max c
Presently a number of tests have been performed on fastening elements consisting of one anchor that is embedded in uncracked or cracked concrete . Tests in cracked concrete are usually arranged in a way that the crack plane is passing through the anchor axis. Crack initiation as well as the crack width is controlled by reinforcement that is perpendicular to the crack plane. Comparison between concrete cone failure loads obtained for uncracked and cracked specimens indicate a decrease of the failure load in the case of cracked concrete. Measurements show that by increasing the crack width to w > 0.3 mm the concrete cone failure load of headed or undercut anchor decreases to about 50 - 80 % of the failure load obtained for uncracked concrete . However, in spite of the number of experiments the failure mechanism is not yet quite well explained and understood.