Carburized spur gears were tested to determine bending fatigue life under variable loads. Fatigue life in random tests is
shorter than in block programmed tests. However, chronological order of loads in block programmed tests has
insignificant influence. Fatigue life in block programmed tests is decreased, if sequence length is shortened. Fatigue life
is similar to random load tests, if a few hundred sequences are repeated until tooth breakage. Loads below 50% of the
fatigue limit can be eliminated without changing test results. Fatigue life can be calculated by Miner’s Rule as modified
by Haibach using a damage ratio of D = 0,s. Single loads applied early decrease fatigue life up to 65%. The greater the
single load the greater the life reduction. Even single loads on reverse sides of teeth significantly reduce fatigue life.