A little consideration regarding the practical operation of trains will show that it is impracticable for an engine to drop off or pick up cars in accordance with the grades which may be encountered along the line. The nearest approach to this is to divide a long road into divisions. At the termini of each division are located freight-yards at which the cars are assorted, and, if necessary, the train-load is increased or diminished, according to the capacity of the engines which are to haul the trains over the succeeding division. The locating engineer cannot determine the proper rate of ruling grade for a line until after the belt of country through which the road is to pass has been thoroughly surveyed, and he can study the project as a whole. There are usually a dozen or more points within a distance of 100 miles through which a railroad is almost compelled to pass, and which must be considered as governing points, unless there is very urgent reason why the road should not go through them. The selection of the rate of grade then becomes a problem of determining the best route between each pair of consecutive governing points. If the natural grade between two consecutive points is very high, and much higher than the grades between other consecutive ruling points, it may be advisable to immediately decide to operate the especially steep grade with pusher-engines, and thereby make it possible to use a much lower rate for the ruling grade. The endeavor should be made to cut down all grades which would naturally be somewhat higher than the other grades until a considerable number of grades have been determined, all of which are approximately equal and which cannot be materially reduced without a large expenditure. On the one hand, it will not pay to spend any amount of money to reduce a grade below this maximum which has been selected, although, on the other hand, it will pay to spend considerable to cut down the rate of grade on one or two grades which are higher than the general maximum. In this way may be determined, after considerable study, some limit of grade which can be used at all points without an extravagant expenditure of money, and yet which could not be reduced, on account of the large number and length of the grades which would be involved, without an extravagant expenditure of money. The engineer is frequently confronted with a definite problem substantially as follows: The rate of ruling grade at one or two places on the line is naturally at some definite figure, say, 1.4%. By spending an easily computable sum of money the line may be modified so that the ruling grade is reduced to perhaps 1.0%. The engineer must then consider the traffic to be run over the road, and must compute the effect on the operating expenses of reducing the rate of grade. If the annual saving in operating expenses,when capitalized materially exceeds the cost of obtaining the lower grades, it will probably be justifiable to construct them.