A new test coke oven of 7 cu. ft. capacity is described. The oven was designed to simulate conditions in a central section of a battery oven and a special feature is the achievement of unidirectional heating from theflues on either side of the charge.The method of charging permits close control of charge conditions such as bulk density and moisture content.These conditions, as well as coking conditions (e.g. flue temperatures and coking time) must be varied to simulate conditions in different batteries. The effects of several variables on coke quality are discussed.Using test oven conditions corresponding to those of Newcastle steelworks batteries, good agreement was obtained between strength indices of test oven and battery cokes. For 17 tests, the average ASTM stability and hardness indices for the test oven coke were 38 5 per cent and 697 per cent, while those for the battery coke were 37 4 per cent and 676 per cent.Observed relationships between coke yield and productivity, and volatile matter and charge moisture are given.The use of a test oven to classify test coals into prime-, medium-, poor-, and non-coking coals is discussed.This may be achieved by observing the effects of the test coal in blends with other coals. The results of a series of blending tests are presented.The dangers of attempting to predict coke quality from a single coal property are pointed out, since this can be done only for very restricted ranges of coals.A diagram is given showing the established trends in the effects of blending various types of Australian coals on coke strength.