The time elapsed till fracture under a constant load and the breaking strength of wood are widely scattered, even if many specimens which are so-called “non-defective” and are selected from one timber are tested under the same condition. This dispersion results from the difference of properties-specific gravity, annual ring breadth and so on-between each specimen and from the inhomogeneity in a specimen. Especially, the latter plays an important part.
For the dispersion resulted mainly from the latter alone, the treatment based on the theory of probability would be necessary. In the present paper, the authors have carried out the bending test of HINOKI (Chamaecyparis obtusa Sieb. et Zucc.) and BUNA (Fagus crenata Blume) under several kinds of constant stress, moisture content and temperature with a large number of samples, of which specific gravity and average annual ring breadth are made relatively uniform, and discussed the time-dependency of the probability of occurrence of fracture in unit time, that we pointed out previously, further in detail from the standpoint of stochastic process and the relation between this probability and moisture content, applied stress and temperature.
As a result, the following conclusions have been obtained:
(1) The probability of occurrence of fracture of wood in unit time can be considered to be constant, that is, independent of time in the fixed time region after the application of load.
(2) The relation between this constant probability m0 and applied stress S, absolute temperature T, and moisture content u can be given approximately by the following equation:
m0=Cexp{(-F0+αS+βu)/kT}
where F0 is the activation energy, k is Boltzmann constant and C, α, β are constants.
Then, fracture of wood under a constant load may be understood as a kind of rate process.
(3) After the lapse of the fixed time, the probability of occurrence of fracture of wood in unit time becomes to decrease in inverse proportion to 1/2-2/3 power of time.