Page:Graphic methods for presenting facts (1914).djvu/302

 *crease due to a seasonal period of slackness. The manager may think, therefore, that his business is not affected by seasons, when in reality the seasonal changes are very great. The extent of these might be very evident if, for instance, a panic year should come along and suddenly stop the upward tendencies in the curve resulting from the rapid increase in the size of the business. This is an important point for any small business to watch, for it may involve bankruptcy to assume that the particular business is not affected by slack seasons of the year such as affect most businesses. Fig. 211 gives an example of a curve in which a large amount of seasonal fluctuation might be easily seen if the rapid increase in the business did not make the upward trend from increased sales greater than the possible downward trend due to seasonal changes.

It is often necessary to compare curves for entirely different things. For instance, it is desirable to compare expenses with sales. When sales increase expenses per unit should decrease, and vice versa. Tremendous saving can be made in most large manufacturing plants by carefully watching the curves for the unit expenses and the output. Something is usually wrong in any department in which expenses per unit increase at the same time that the output curve goes upward. When the amount of work done is fluctuating greatly, as during periods of business depression, the executive may often get many vital hints for the operation of his plant if he will simply make periodic examinations to see whether the overhead-expense curves for each department react in the manner which would be indicated by fluctuations in the curves for direct labor or for quantity of output.

In making comparisons between separate curves there is a great advantage in having a standard arrangement of the five-year cards so that there may be no danger of comparing two curves for different years when it is intended to compare for the same year. It is desirable, as seen in Fig. 215, that all five-year cards should have the years arranged by half-decades. One arrangement includes those years ending in one to five inclusive, and the other arrangement takes those years ending in six to ten inclusive. The person reading the curves can then pick up any two cards relating to the same half-decade and compare the curves instantly and correctly, simply by placing the cards so that the edges coincide. If this half-decade arrangement were not followed there would be five different positions in making curve comparisons, and there would be grave possibility of frequent