The first attempt to get statistical demand functions was made by H. L. Moore. H. Schultz, taking after Moore's research, worked out his bulky volume, "The Theory and Measurement of demand, 1938". Since then, quite a few demand functions of various goods and of various forms have been calculated by many investigators. Though the history of statistical demand function can be traced back to comparatively old times, we can not frequently expect good results when we make an attempt to get a demand function of producer's or intermediate goods. In this field of empirical demand function, many problems have still been left untouched. Firms demand intermediate goods because they are useful to produce other goods, while consumer's goods are demanded to keep consumer's lives alive and comfortable. Accordingly, when we try to get statistical demand functions of intermediate goods, we must take into account this fact. We will not be able to expect to get a good demand function of intermediate goods if we apply the same method adopted to get a demand function of consumer's goods. The magnitude of demand for intermediate goods depend upon that of final demand. A firm that produces consumer's goods demands enough intermediate goods to supply a certain amount of consumer's goods which will counterbalance the anticipated demand for the goods. This kind of demand for intermediate goods is apparently caused by the final demand. It will in turn cause, another demand for intermediate goods. Thus final demand will stimulate demand for intermediate goods from industry to industry. In this paper, putting an importance to the causality from the final demand to the intermediate goods, I tried to present a method of short-range forecasting, applying it topaper pulp and pulp wood industries.