In these cases the Interstate Commerce Commission undertook to prescribe just, reasonable, and equitable divisions of joint rates pursuant to § 15(6) of the Interstate Commerce Act, 24 Stat. 384, as amended.1 The Commission found that existing divisions violated § 15(6) because they allocated to Northern lines a lesser share of the revenues from the joint rates than would be warranted by their share of the expenditures made in providing the joint service. 325 I.C.C. 1, 50.

The Southern lines brought suit before a three-judge District Court to enjoin and to set aside the Commission's order. The District Court set aside the Commission's order and remanded the case for further proceedings. 270 F.Supp. 695. We noted probable jurisdiction. 390 U.S. 940, 88 S.Ct. 1031, 19 L.Ed.2d 1128.

Both Northern and Southern lines used Rail Form A as their basic formula, that form being a rail freight formula for determining freight service costs which utilizes the expenses and statistics for a given year as reported to the Commission by the carriers and supplemented by special studies of the carriers.

The Southern lines proposed 12 adjustments, five of which the Commission accepted and seven of which it rejected. The year 1956 was the one both Southern and Northern lines used in the final cost analysis. The cost level for that year, said the Commission, was higher in the North than in the South for like services; and it concluded that that situation would most likely continue in the immediate future. In that year the Northern lines received 44.64% of the revenues while incurring 46.35318% of the fully distributed costs. Accordingly, the Commission prescribed new divisions based on the fully distributed costs and divided the revenues in the same proportion to those costs. The shift in revenues resulting from the new divisions was approximately $8,000,000 a year, giving the Northern lines an overall increase in revenues from the traffic involved of 3.5% and reducing the revenues of Southern lines by about 3%.

When the Southern lines sued to set aside the new divisions, the Northern lines intervened as defendants. The District Court held that the Commission's order was not supported by substantial evidence and reasoned findings within the meaning of §§ 8(b)2 and 10(e)3 of the Administrative Procedure Act and, as noted, remanded the case for further proceedings.

The present problem of divisions deals only with North-South traffic which represents 6% of the total traffic of the North and 21.4% of the total traffic of the South. The costs of that North-South traffic are not isolated in the findings. The average costs used relate to all Northern traffic and to all Southern traffic. Nearly 80% of the total Northern traffic is intra-territorial and handled entirely in the North, and it is therefore argued that that traffic has the dominant influence on the Northern average. As the District Court said, it is difficult to maintain that these intraterritorial Northern costs are the same or approximately the same as Northern costs in handling traffic between North and South. In another divisions case, the Commission ruled that territorial average costs are entitled to little weight in determining the costs of handling particular movements. Increased Freight Rates, 1967, 332 I.C.C. 280, 303. The use of 'unsifted averages' of costs does not necessarily establish greater costs either in rate cases (ICC v. Mechling, 330 U.S. 567, 583, 67 S.Ct. 894, 902, 91 L.Ed. 1102) or in divisions cases. The ruling of the District Court was, not that territorial average costs were irrelevant or that Rail Form A was not a usable and useful tool for cost determination, but that territorial average costs could not be used consistently with the statutory requirements for precise and relevant findings without any evidence relating the territorial average costs to North-South traffic. While Southern lines had offered evidence showing the costs of handling North-South traffic in the South, there was not always any such Northern offer; nor did the Commission always exercise its undoubted authority to gather it on its own.

On the question whether territorial average costs represent costs of the North-South traffic, the Commission only replies that where particular traffic uses physical facilities and employees' services in common with other traffic 'and has been shown to have no distinguishing characteristics,' the application of Rail Form A costs is proper. Yet the Commission stated 'its exclusive standard' for resolving this divisions question to be 'the relevant cost of handling the specific freight traffic to which the divisions apply.' 270 F.Supp., at 710.

We agree with the District Court that there is no substantial evidence that territorial average costs are necessarily the same as the comparative costs incurred in handling North-South freight traffic. If we were to reverse the District Court, we would in effect be saying that the expertise of the Commission is so great that when it says that average territorial costs fairly represent the costs of North-South traffic, the controversy is at an end, even though the record does not reveal what the nature of that North-South traffic is. The requirement for administrative decisions based on substantial evidence and reasoned findings which alone make effective judicial review possible—would become lost in the haze of so-called expertise. Administrative expertise would then be on its way to becoming "a monster which rules with no practical limits on its discretion." Burlington Truck Lines v. United States, 371 U.S. 156, 167, 83 S.Ct. 239, 245, 9 L.Ed.2d 207. That is impermissible under the Administrative Procedure Act. If indeed that lax procedure were sanctioned in a North-South divisions case, whose solution turns solely on costs, the class rate discrimination in favor of the North and against the South which we condemned in State of New York v. United States, 331 U.S. 284, 67 S.Ct. 1207, 91 L.Ed. 1492, could well flourish in another form.

Rail Form A was used in Chicago & N.W.R. Co. v. Atchison, T. & 1585, 18 L.Ed.2d 803, and we approved 1585 18 L.Ed.2d 803, and we approved its use. Moreover, ever since the New England Divisions Case, 261 U.S. 184, at 196—197, 43 S.Ct. 270, at 275, 67 L.Ed. 605, it has been held that mathematical exactness in dividing each rate of each carrier is not necessary, because practical necessities demand otherwise. In addition we repeat what we said in Chicago & N.W.R. Co. v. Atchison, T. & S.F.R. Co., supra, 387 U.S. at 358, 87 S.Ct. at 1603, that there are no 'mechanical restrictions on the range of remedies from which the Commission may choose' in solving a divisions case or making its expert judgment as to what scale of costs should be used in making the allocation. Precision and exactitude in the mathematical sense are not possible. Yet the nature and volume of the traffic in question must be known and exposed, if the costs of other traffic are to govern a division of rates. Moreover, where Rail Form A costs are shown to be a distortion when applied to the particular traffic over which the divisions dispute arises, some effort must be exerted to make an adjustment which fairly reflects the difference in the costs or to make clear that there is in fact no basic, material difference. The Commission states to us that it cannot be expected to know whether peculiar characteristics may exist respecting the traffic involved in the divisions dispute or whether special studies may be needed. Yet if that is true, the Commission's expertise is not equal to the task and the opposed carriers must be directed to expose the various versions of the conflict so that the Commission may make its informed decision. That was done on aspects of the present cases (325 I.C.C., at 25) and no reason is aparent why it cannot be done on other aspects of the controversy.

The Commission in its argument before us said that Rail Form A territorial average costs were 'adjusted' to reflect the costs attributable to the North-South traffic issue, which is true as respects five4 of the 12 adjustments proposed by the Southern lines.

On remand of the cases to the Commission we think specific findings must be made on that several items of so-called 'adjustment' of average territorial costs to which we now turn.

One is the question of commuter deficits, which swell the average territorial costs in the North while they are less important in the South that does not yet have substantial commuter operations. Passenger deficits generally are considered as part of the costs of providing freight service, since the common facilities that support each must be maintained for both types of service. There is, however, evidence that in some territories as much as one-half of the track facilities are maintained solely because of the company's suburban service and even a larger proportion of other facilities such as stations, terminals, coach yards, and repair shops is maintained exclusively for commuter service.

The Commission, however, ruled that costs of commuter service include 'common costs which must be incurred to provide freight service or intercity passenger service' and that the deficit from suburban operations was properly included in 'the constant costs.' The Commission on the other hand found that 'many individual items of suburban service can be considered solely related * * * to suburban service.' 325 I.C.C., at 78. How these two findings can be reconciled is not apparent. The Commission in its argument before us rests primarily on revenue needs—'Such losses must be recovered from railroad freight operations if railroads are to remain solvent.' Section 15(6) makes plain that revenue needs come into focus in divisions cases. Revenue problems under § 15(6) at times have resulted in putting a part of one area's transportation costs upon other sections of the country. See New England Divisions Case, 261 U.S. 184, 191—195, 43 S.Ct. 270, 273—275, 67 L.Ed. 605. But that issue is not presented in these cases. The issue in the present cases was costs, not revenue.5 The allocation either to the North or to the South of costs peculiar to its territorial traffic is a task with which the Commission is familiar. Thus, in these very cases it excluded certain platform deficits incurred by the Northern lines because they were not related to North-South freight traffic. 325 I.C.C., at 56. There is no apparent reason why costs related solely to commuter service in the North cannot be determined.

As to the costs of interchanging cars in North-South traffic at territorial border points, there is evidence in the record that the interchange operations performed by Northern lines are no more costly than those performed by Southern lines. Yet the Commission allowed the Northern lines a border interchange cost that is 58% higher than the one allowed the Southern lines. That apparently was done solely because Rail Form A showed higher interchange costs when all territorial interchanges were considered. We cannot bridge the gap by blind reliance on expertise which in this instance would be a mere assertion that no difference means a substantial difference.

The empty freight car return ratios is another example of deficient findings. There is evidence that higher costs of Northern lines result from the Commission's use of higher Northern territorial average empty return ratios. There was no attempt made to show that the latter were at all applicable to North-South traffic. The problem arises in the North by reason of boxcars on shuttle from Detroit to automobile plants, most of which are in the North. These shuttle boxcars return empty to Detroit. We know from the record that this is a major cost item as 800,000 carloads of automobile parts move out of Detroit each year. The record does not show the extent to which these empty returns swell the territorial average costs in the North, though it does show that Northern use of these shuttle boxcars is substantially higher than the Southern proportion. The District Court concluded the territorial average boxcar empty return ratios could not be said, absent specific findings to reflect the costs of the North-South freight traffic relevant to this problem of divisions.

There are other proposed adjustments6 on which we think the Commission's findings are adequate.

The judgment of the District Court is modified and as modified it is

Affirmed.

Affirmed as modified.