These lines would not appear this way if a piece of paper had been torn out of a large-size chart. The curve lines should have continued up to the break in the drawing more nearly on the slope seen in the left-hand portion of the chart. The lines for curves "G", "F" and "E" would, if correctly made, show much less slope also at the right of the break where they lead down to the lower corner of the chart. This error is mentioned here simply that the reader may have some guidance if he finds it necessary to make charts including a break, similar to that shown in Fig. 158.

It is customary in cost-keeping work to make costs which show only an average of the cost per unit, on an average for all orders completed during a period of time, say one month or one year. Though this method of averaging all orders together without respect to the size of the order is sufficiently accurate for many purposes, there are times when such a method may gravely mislead an executive. It is almost invariably true that small orders cost more per unit of output than orders of large size. A man who makes a selling price for his work on the average cost of small orders and large orders combined may be

FREQUENCY CURVES 191

losing money unnecessarily, because he does not realize the true cost of work when it is done in only small quantities on different orders.