These are appeals by the Secretary of Agriculture and a group of dairy farmers from a judgment of the District Court declaring illegal and enjoining further enforcement of the "farm location differential" provision1 in the 1964 federal milk marketing order for the Massachusetts-Rhode Island area. We recently invalidated a like provision of the 1957 New York-New Jersey milk marketing order on grounds of its incompatibility with the Agricultural Marketing Agreement Act of 1937.2 Blair v. Freeman, 125 U.S.App.D.C. 207, 370 F.2d 229 (1966). Since we find that the farm location differential of the Massachusetts-Rhode Island order is no less irreconcilable with the statute, we affirm the judgment and order of the District Court.

* The rationale and mechanics of federal regulation of the marketing of milk and other dairy products have been fully described in Blair and elsewhere.3 Briefly, the marketing order is designed to eliminate the destructive competition which grew out of the surplus conditions which characterized the milk industry before federal regulation. The device chosen to this end is a uniform, "blended" price to be paid by handlers to each producer in the marketing area, regardless of whether his milk is ultimately disposed of by the handler at the high Class I price for fluid use, or at the lower Class II price paid for milk to be used in the manufacture of various milk products (butter, cheese, ice cream, and so on). An equalization pool, called the "producer settlement fund," is administered by the agent of the Secretary of Agriculture to maintain the blended price; and handlers make payments to or withdrawals from the pool in the amounts by which the use value of milk which they handle exceeds or is less than the blended price.

The Act specifies several limited exceptions which the marketing order may make to the requirement that each producer receive a uniform price. Thus, Section 8c(5) permits the inclusion in the order of provisions

(B) Providing:

* * * * * *

(ii) for the payment to all producers and associations of producers delivering milk to all handlers of uniform prices for all milk so delivered, irrespective of the uses made of such milk by the individual handler to whom it is delivered; subject * * * only to adjustments for (a) volume, market, and production differentials customarily applied by the handlers subject to such order, (b) the grade or quality of the milk delivered, (c) the locations at which delivery of such milk is made * * *.

Purportedly acting pursuant to the authority of this section, the Secretary included in the 1964 Massachusetts-Rhode Island order several differentials which operate to vary the payments to producers. These are (1) a "butterfat differential," based on the quality of the farmer's milk;4 (2) "zone differentials," which vary inversely with the distance from Boston of the handler's plant to which the producer delivers his milk;5 and (3) the "farm location differentials," which are paid to producers whose farms are located near the principal consumption centers. This last form of differential, commonly known as the "nearby differential," is forty-six cents per hundredweight for farmers located in certain designated towns and twenty-three cents per hundredweight for those in more distant towns.

The plaintiffs, 168 dairy farmers who are ineligible for any farm location differential, brought this action against the Secretary of Agriculture seeking a halt to these allegedly discriminatory and unauthorized payments.6 They are aggrieved in that the differential is paid out of the equalization pool, rather than the handler's own pocket, and consequently reduces the blended price received by all producers. The District Court granted the plaintiffs' motion for a preliminary injunction which provided for the escrowing of the challenged differential, and thereafter, relying wholly on our decision in Blair v. Freeman, granted their motion for summary judgment. The escrowing, though later amended to release amounts attributable to nearby producers, was continued pending a decision by this court. When it appeared that the Secretary might not appeal,7 we granted Zuber and seven other nearby farmers, whose motion to intervene had been denied by the District Court, leave to intervene for the sole purpose of protecting their rights by appeal. The Secretary subsequently perfected an appeal, which we heard together with that of Zuber.

Appellants have pressed upon us several distinct positions. They argue, first, that the farm location differential is valid because it is founded not upon the factor of fluid milk utilization condemned in Blair but on other considerations which are authorized by the Act. Secondly, they contend that, whatever might be the fate of another farm location differential under Section 8c(5), this particular provision was explicitly ratified and validated both by Congress in 1937 and by the United States Court of Appeals for the First Circuit in 1939. Thirdly, appellants assert that, if the legality of the provision is not clear, neither is its illegality; and that the case should be remanded to the Secretary to afford him an opportunity to elucidate the legal and factual premises of the nearby differential, or to determine whether, if the differential must fall, the marketing order should be continued, with or without other changes. Finally, it is argued that, even if the District Court is affirmed, considerations of equity dictate that its judgment be wholly prospective from the date of our decision, and that the escrowed monies should be distributed to the nearby farmers who have relied on the farm location differential for so many years. We deal with these points in order.

[T]he Act forbids consideration of the use to which the milk of a particular producer or class of producers is put, historically or potentially, in adjusting the uniform minimum price to be paid to such producers.8

In adopting the New York-New Jersey order in 1957, the Secretary was explicit in his reliance upon this outlawed consideration.9 He is certainly less so in his justification of the 1964 Massachusetts-Rhode Island order:

Such farm location differentials have been in effect under the several New England orders since the inception of the orders. The differentials were adopted to reflect in the pricing structure of the orders historical price relationships by location which prevailed in these markets. It was found that customarily somewhat higher values, above those which normally reflected transportation costs, attached to milk produced near the principal consumption centers as compared to the market value of milk produced in the more distant areas of the milkshed.

* * * * * *

* * * [W]e believe there are adequate reasons on this record for continuing the farm location differentials at their present rates.10

But we do not think it can be said, as appellants now do, that this explanation does not reflect any reliance whatsoever upon the prohibited factor of fluid use patterns. To say, of the order under review, that milk produced in locations close to principal consumption centers has enjoyed "customarily somewhat higher values" above transportation costs, and that the differential is provided to reflect "these historical price relationships by location," may be only a less direct way of saying, as in the case of the New York-New Jersey order, that the differential was designed "to reflect the fact that under competitive conditions" nearby producers got a higher price because they had a bigger share of the fluid use market. If examination is made of the "record" to which the Secretary adverted in his 1964 statement — the various New England marketing orders which preceded the 1964 order and the administrative records on which they were based — it is revealed inexorably that, as in the case of New York-New Jersey, those "historical price relationships by location" were the fluid use advantages which the Act expressly condemns.

The 1964 order consolidated the former federal orders regulating the handling of milk in four separate marketing areas: Greater Boston, Worcester, Springfield, and Southeastern New England. The oldest and most prominent of these was the Greater Boston marketing order. Under federal "licenses" pursuant to the 1933 Agricultural Adjustment Act,11 milk in the Boston area was marketed, as it had sometimes been by cooperative associations before federal regulation, through base-rating plans. In 1934 the market administrator

allotted preferential bases to nearby producers to give effect to the historically higher prices they had received in the market. A study was made of the eight years 1926 to 1934 to obtain the proper differential between nearby and distant producers' returns from the sale of milk. It was determined that all producers where milk was delivered directly to plants located within 35 miles of the State House in Boston should have bases equal to 85 percent. * * * [as compared with 61 percent for all other producers].12

Because many nearby producers refused to comply with the license, and "as a concession to win the cooperation of the Massachusetts Milk Control Board in enforcing, or trying to enforce, the terms of the federal license,"13 in 1935 cooperative associations agreed to allow producers whose farms were within forty miles of the State House bases of 100 per cent. This forty mile radius later became official under Greater Boston Order No. 4.

This Order No. 4, promulgated in 1936 after the 1935 amendments to the Act, retained the base-rating system but abandoned the preferential bases for nearby producers. In their place it was provided that producers whose farms were located within forty miles of the State House and who delivered to plants within the same area be paid the full Class I price for all milk not in excess of their bases. Other farmers received only the blended price for their bases. Distant producers could, however, receive an additional twenty-one cents by delivering to a handler within the forty mile radius — a "country station" differential which (unlike the situation today) was not also available to the nearby producer. The latter differential was found by the Secretary to be a "reasonable allowance for the extra handling costs incurred by handlers operating country station plants." As to the former, the Secretary stated that

the payment of class I prices for the bases of producers whose farms are within forty miles of Boston is a proper recognition of the economic position of such producers, due to the greater availability of their milk for class I use. * * * (Emphasis added.)

The order did not expand upon this ground for the preferred treatment of nearby producers. It is a finding which appears to be derived directly from the testimony at the hearing of the representative of the U. S. Department of Agriculture. In explaining the proposed order, he said this about the differential in question:

Now, I presume that you know some reason for that special provision located in that zone. I am not sure that I made such a statement of that as we here in this brief, and so I think I might make it at this time, and then perhaps there will be further questions.

With respect to this particular provision, it should be noted that a producer whose farm is located near the market is in a position to supply Class I milk for use at all times, at less cost and distances, and if necessary he can, after each twice-a-day milking, put his milk on his own truck and deliver it to a handler or to consumers. Forty miles, however, is approximately a limit of the use of such facility in handling milk.

Evenness of production was certainly not assigned as the reason, presumably because this quality was already compensated by base-rating itself. Higher local production costs had been mentioned at the hearings in support of the special treatment, but this ground was evidently rejected by the Secretary. Since nearby producers were not also to receive the "country station" differential, their preference must have been attributable in large part to transportation and handling advantages. This surmise is supported not only by the Secretary's use of the word "availability" but also by the following argument in the Department of Agriculture's Economic Brief in 1936:

It should be noted that a producer whose farm is located near the market is in a position to supply milk for Class I use to the market more easily at all times than those at a greater distance. Much less time is required for the movement of the milk and a greater variety of transportation facilities are available to such a producer. If necessary, he can, after each twice-a-day milking and within two hours deliver it to a handler or even to consumers. * * * Moreover, those producers whose farms are located beyond the forty-mile line who so desire may deliver their milk to handlers within the forty-mile line, in which event they are to be paid the blended price, plus [21] cents. * * *

The savings in transportation and handling expenses were not, of course, the whole story. The Agriculture Department's testimony and brief identify another element which was also heavily emphasized at the hearings, namely, the nearby producers' opportunities for direct access to the consumer. This fact was cited at the hearings as the explanation for traditionally higher prices to nearby producers ("he can sell it to the dealer or become a dealer himself, and go direct to the [consumer], and you cannot stop him") and as a potential "menace to the enforcement of the agreement."

In sum, the Secretary did not indicate that he had accorded nearby producers a preferred position because their ever-available milk supply was itself more valuable to the handlers. His action, rather, was based on the nearby producers' "economic position," a position which had enabled them to command higher prices from handlers before regulation and which enabled them to coerce this concession in the marketing order. The reason for this "economic position" was, as the Secretary recognized, "availability for Class I use." We are unable to distinguish this factor from the consideration of historical fluid use patterns condemned in Blair.

After passage of the 1937 Agricultural Marketing Agreement Act, Order No. 4 was amended to do away with base-rating. The nearby producer's preferred status was preserved, however, by the payment of farm and delivery location differentials. For the first time, the condition of nearby delivery was dropped from the farm location differential, and the farm location and delivery location differentials were paid cumulatively. The Secretary found only:

[t]hat in view of changes in economic conditions since the date of the original findings, the payment to all producers and associations of producers delivering milk to all handlers of uniform prices for milk so delivered, without the use of a base-rating plan, but with the inclusion of special location differentials, is a fair and reasonable method of distributing to producers the proceeds of sales of milk to handlers.14

The farm location differential was designed to reproduce as closely as possible the nearby farmer's preferred position under base-rating. Thus, the differential was (and still is) payable only to the extent that it did not increase the producer's payment above the Class I price.15 As the Agriculture Department argued in its 1937 Economic Brief:

It is thought that the proposed plan, which gives consideration to those producers located nearby the market who are predominantly even and who as a group have very uniform deliveries, and which returns a price higher than the average to the other even producers, accomplishes the purpose for which the rating plan in Order No. 4 was intended.

Evenness of production, which was awarded on an individual basis under base-rating, was an alleged advantage of nearby farmers frequently cited by proponents of the differential during the hearings as explaining their historically higher prices. The Market Administrator's explanation is typical:

The idea is, without regulation over a period of years producers whose farms are near the market, due to the fact that their milk is accessible to small dealers, and their milk is rather even throughout the year, have been able to obtain a price greater than the country station charges. So that, if we set up a plan which did not give them the same advantage which they would have without regulation it would be unfair to them. * * * It is the desire of the Administration to maintain the relative positions and to set that figure at a point which will do that.

This assertion about seasonality was somewhat undermined by the Department's admission that only two-thirds of the nearby producers could be classified as even, and that "even producers were scattered throughout the milkshed so that, for the most part, their evenness was completely offset by the irregularity of their neighbors." The principal relevance of evenness was not that it constituted an element of value in the sense of dependability to the handler, but that nearby producers, who as a group contributed less to the surplus in the market, should not share the burden of that surplus by receiving only the blended price. This resort to seasonability, then, was not so much in explanation of historically higher prices but rather in recognition of the fact that the nearby farmers' evenly produced milk had been sold primarily at Class I prices, i.e., for fluid use.16

Aside from references to evenness of production and the unfairness to the nearby farmer of disturbing his traditionally preferred market position, the evidence revolved mostly around what would happen if nearby farmers were not favored. The New England Milk Producers Association swore they would "oppose the license with every means in their power," and the witness referred to in footnote 16 made the pregnant observation that "there is always a row in the market unless [the nearby producers] are compensated in some way." One source of their power was also recognized: "We are all within retailing distance of the market and we are potential producer-dealers." This potentiality also underlay the observations in the Department's brief and elsewhere that "their milk is accessible to small dealers." This reference to availability was limited to small dealers as it had not been in the Secretary's finding in the 1936 Order, presumably because only large handlers could afford to bring in their milk from distant farms. That nearby farmers had a captive market, however, does not alone explain the higher price, without the additional fact that the farmers or handlers could dispose of the milk at Class I prices. Economist Ellsworth Bell's brief to the Secretary concisely described this economic advantage:

Being adjacent to markets the [nearby] dairymen command a natural sales advantage whereby it is taken first at the fluid price. * * * [Historical price difference] exists due to peculiar advantages existing for disposal and sale of milk at the higher market price.

In 1949 the Secretary first promulgated orders for the Springfield and Worcester markets, both of which contained nearby farm location differentials. Those in favor of the differentials had given testimony similar to that which preceded the Boston orders, and had argued for uniformity of treatment with the Boston market. In approving the differentials, the Secretary adverted to "the customary market practice of paying somewhat higher prices to producers located near the sales area." These historically higher prices were explained as follows:

This difference cannot be attributed to transportation cost alone. The many opportunities for dairymen to market their own milk directly influence the price which they demand for their product.17

Those direct marketing opportunities were, of course, for fluid use at Class I prices.

In 1951 the Secretary held hearings on a proposal to eliminate the nearby differential in the Boston order. Opponents of the differential argued that it constituted a trade barrier; that evenness of production, if compensable at all, should be recognized on an individual basis; that the validity of the differential should be reexamined in the light of contemporary economic conditions; and that if nearby milk has any extra intrinsic value it should be paid for by the handlers rather than by the pool. Proponents of the differential brought out all the old arguments, with somewhat greater emphasis this time on the disruptive market effects of eliminating the differential, and on the unfairness to nearby farmers for whom the nearby differential had become capitalized into greater land values. A major pillar of their argument was the unabashed assertion that in terms of "historical price relationships * * * producers located near the market were able to sell a larger proportion of their milk at Class I." The representative of the Massachusetts Milk Control Board, for example, testified that the differential was not discriminatory but rather had been

adopted by way of partial compensation to nearby producers for the losses which they suffered when the equalization provisions deprived them of those natural advantages which prior to federal regulation always enabled them to sell at [sic] a relatively higher percentage of their production of Class I milk.

The Secretary concluded that "no change should be made at this time" because the opponents had "presented no material facts different from those considered in establishing specified location differentials as a part of the order provisions."18

In the 1958 southeastern New England order, one of those now consolidated into the 1964 Massachusetts-Rhode Island order, the Secretary for the first time gave a detailed economic justification for the nearby differential, which he described as "identical with that employed in adjacent Federal order markets," a characterization which includes the New York-New Jersey order invalidated by us in Blair:

Historically, dairy farmers in the States of Massachusetts, Rhode Island, and Connecticut, because of their location with reference to the large population concentrations of New England have disposed of a substantially larger percentage of their production for fluid uses than have dairy farmers in the upcountry area. Hence nearby producers have been able to realize a price higher in relation to more distant producers than can be accounted for by the advantage in the cost of transportation to market. Under the marketwide pooling system herein proposed and without some adjustment mechanism the nearby producer, notwithstanding, would be paid on the basis of the average utilization of all milk on the milkshed rather than according to the utilization of his milk. Under a regulated market, however, he obtains the benefit of an established Class I price which may be higher than in the absence of regulation, and has assurance that his milk will not be displaced by cheap, unregulated milk from more distant sources. The distant producer also benefits from an established Class I price, and by the fact that he gains a larger share of the fluid market than is likely without marketwide equalization.

The use of a nearby differential will return to nearby producers a somewhat greater share of the Class I price than they would otherwise obtain through marketwide pooling. Moreover such a differential will tend to provide returns to both nearby and upcountry producers more nearly in line with their prospective returns without marketwide equalization.19

This rationale — which is clearly reflected in the 1964 order's language about "historical price relationships by location" — is indistinguishable from that advanced the year before in support of the 1957 New York-New Jersey order condemned by this court in Blair v. Freeman.20

The testimony at the hearings preceding the 1964 order served mainly to reinforce the Secretary's reliance on historical fluid use patterns as the primary justification for the nearby differential. Most of the witnesses in favor of retaining the differential either ignored, or rejected as irrelevant, the question of whether it had ever been authorized under the Act. Their testimony dwelt upon the absence of adverse economic effects of the differential, and the dire consequences which would flow from its elimination. When their attention was directed to the issue of the differential's original justification, some reiterated the testimony of 1937 to the effect that the differentials recognized the evenness of production of nearby milk and its accessibility to small dealers; that the nearby farmers enjoyed a favorable bargaining position because of their ability to become producer-dealers; and that the differential was a necessary compromise to win the support of the nearby farmers and Massachusetts Milk Control Board for the federal order.

Almost to a man, however, the witnesses at the hearing testified — and no one denied — that the principal reason for the differential was to preserve for nearby producers their traditionally high percentage of the fluid use market. One witness even argued as follows for an increase in the nearby differential:

We don't feel that these market pools were set up to make it possible for distant milk to take the Class I sales that rightfully belong to nearby farmers away from them. Again I think you have to consider the use value of the milk in determining the prices to the producers.

And a committee report specially prepared for the Department of Agriculture in 1962 was read into the record. While nearby milk was claimed to have various advantages, the report concluded that:

the major purpose of nearby differentials apparently has been to compensate nearby producers for sharing market Class I sales with more distant producers in a marketwide pool.

In view of the concurrence of all concerned — including on occasion the Secretary himself — that the differential was a recognition of historical use values, we find it impossible to conclude that the Secretary did not rely on this illegal consideration in approving the 1964 order as well as all of its predecessors. Although in patent conflict with the blended price regulatory scheme propounded by Congress, it was apparently thought that the resulting distortion was a price necessary to be paid in order to get sufficient agreement to support a marketing order, and that it was better to pay that price than to have no order at all.

Section 4 of the 1937 Agricultural Adjustment Act provided:

Nothing in this Act shall be construed as invalidating any marketing agreement, license, or order, or any regulation relating to, or any provision of, or any act of the Secretary of Agriculture in connection with, any such agreement, license, or order which has been executed, issued, approved, or done under the Agricultural Adjustment Act, or any amendment thereof, but such marketing agreements, licenses, orders, regulations, provisions, and acts are hereby expressly ratified, legalized, and confirmed.21

Appellants argue that, since nearby differentials appeared in the Greater Boston order before 1937, Section 4 immunizes the farm location differential from invalidation under another section of the Act.22 We find this contention unpersuasive for several reasons.

First, it does not appear that a farm location differential of the kind now under attack ever appeared in a greater Boston license or order before 1937. There were differentials based on delivery location, and for a time farm location was one condition along with delivery location, but none seems to have been based on farm location alone. Moreover, the differential was never one payable over and above the transportation differential available to distant producers who delivered to a nearby zone.23

Second, Section 4 itself ratifies only those provisions and acts which were "done under the Agricultural Adjustment Act." This qualification negates any inference that Congress intended to sanction orders which were illegal under other sections of the Act. And this construction is consistent with what was clearly the sole legislative purpose underlying Section 4, namely, to prevent technical abatements which arguably might result from reenactment of the 1933 Act. In United States v. Butler, 297 U.S. 1, 56 S. Ct. 312, 80 L. Ed. 477 (1936), the Supreme Court had invalidated certain of the production control provisions of the 1933 Act; and several lower courts, construing these provisions as not severable, thought that the marketing agreement and order sections must also fall. Section 1 of the 1937 Act specifically reenacted the latter provisions of the Act in order to avoid that result.24 Congress feared, however, that such reenactment might itself be construed as suggesting that, pending its accomplishment, the Act, and all actions taken by the Secretary under its aegis, had been illegal. Hence Section 4 of the 1937 Act.25

Appellants further assert that the legality of the farm location differential was established in 1939 by the United States Court of Appeals for the First Circuit in Green Valley Creamery, Inc. v. United States, 108 F.2d 342 (1st Cir. 1939). That case began as an action by the United States and the Secretary of Agriculture seeking a marketing injunction requiring the defendants, a group of milk handlers, to comply with Order No. 4 regulating milk marketing in the greater Boston area. One of the defenses was that the farm location differential was not authorized by Section 8c(5). After holding that handlers had no standing to raise this defense, the court volunteered the view that the differential could be justified under the Act either as a delivery location differential or a "market differential customarily applied."26

We took note of the Green Valley opinion in Blair, where it was also pressed upon as dispositive; and at that time we expressed our disagreement with the First Circuit's analysis.27 In any event, the opinion in that case could not control the outcome here. In the first place, only one of the orders which was consolidated into the Massachusetts-Rhode Island order in 1964 was under attack in Green Valley. Moreover, the defendants in that case were all handlers, and there was no representative of the class of non-nearby producers before the court. Indeed, as we said in Blair, the approval of the differential was a dictum, since "the court's holding was that the handlers lacked standing to challenge the disposition of the producer settlement fund."28 We must conclude, therefore, that neither Congress nor the courts have validated the farm location differential of the Massachusetts-Rhode Island order.

Appellants insist that, even if the Secretary had in the past relied upon historical fluid use patterns, there are other legitimate legal and factual grounds for the farm location differential. Therefore, they urge, the court should, before striking the provision from the order, in effect remand the case to the Secretary for an opportunity to offer a more explicit justification for the differential. We are not impressed with this suggestion since we do not think that a court should, in the face of what it finds to be a clear and continuingly burdensome statutory violation, albeit one of long standing, defer a declaration of its illegality. We are especially not disposed to do so where there appears to be no serious possibility of finding warrant in the statute for a preference based solely on farm location.

We need not speculate as to what the Secretary would do upon such a remand, for he has already, since our decision in Blair and the judgment of the District Court in this case, issued a recommended decision on the farm location differential in the Massachusetts-Rhode Island order. He there concluded after further hearings on the subject that the nearby differential "should be continued." He noted that nearby farmers had received higher prices "prior to the advent of federal regulation," that this advantage had been preserved in the base-rating plans of the federal licenses and 1936 Boston order, and that it had been carried over into the nearby differential of the 1937 order. The Secretary quoted from the 1937 record to the effect that the historically higher price was:

due to the fact that their production is relatively even, and that their milk is accessible to small dealers. * * * If we should take away from these producers a differential that they have been able to get without Government regulation, they would have a legitimate objection that we are placing them at a comparative disadvantage. * * *29

The fact of historically higher prices is, of course, not alone sufficient. The reason for that difference must not, under the Act, have been related to sales advantages for fluid use. To avoid this pitfall, the Secretary assigned two other explanations for the high prices: availability and evenness of production. The former is not a legitimate factor. To the extent that it reflects the nearby producers' ability to become distributors directly to the consumer, it is no more than an aspect of the high fluid utilization advantage. To the extent that availability means dependability and ease of access, it is, as we held in Blair,30 and as the Secretary himself recognized in the New York-New Jersey order, "related to the extra value of such milk to the handler who buys it rather than to producers generally and [is], therefore, * * * payable by handlers rather than out of the pool."31 Nor would evenness of production appear to support a differential based on farm location. While nearby production may as a general matter be more even, there have been wide individual variations in seasonality throughout the milkshed. Moreover, as the Secretary stated in his New York-New Jersey decision,32 uniform supply should be compensated on an individual basis. The Act makes provision for base-rating,33 and it appears to be the only sanctioned method by which to stimulate and compensate even production by individual farmers.

None of the differentials permitted by the Act contemplate in terms preferential treatment on the basis of farm location.34 A dual legal justification for the nearby differential, was, however, offered in the 1967 recommended decision:

Clearly * * * these greater returns * * * were, in the language of the statute "market differentials customarily applied by the handlers." * * * Inasmuch as these greater returns for milk were directly related to geographical source, it was considered that these differentials were also adjustments based on "location" * * * within the meaning and context of [the Act].35

But the only "location" differential permitted by the Act is one for "the location at which delivery" of the milk is made to the handler. Plainly, this exception was designed to recognize differences in transportation costs only, and does not encompass a preference based solely on farm location.36

Nor is farm location comprehended within the provision for "market * * * differentials customarily applied by the handlers." The Committee Reports in respect of the 1935 Act are explicit that "the market differential is a differential which is given to the producer to compensate him for delivering his milk to a city market instead of to a country plant."37 This adjustment, of which the "country station" differential of the 1936 greater Boston order was an example,38 recognized the savings in handling costs when milk in fluid form is delivered to the city rather than first to a country plant. Provision is now made for this differential in the Massachusetts-Rhode Island order as a fixed portion of the zone differential.39 As we held in Blair,

it is plain that this type of allowance turns not on the location of the producer's farms but on the nature of the market to which they deliver their milk. The "market differential" rationale is plainly not a tenable basis for the nearby differential.40

On August 2, 1968, the United States District Court for the Northern District of New York elected to treat a nearby differential of the kind here in issue as a "market differential." Cranston v. Freeman, 290 F. Supp. 785. It recognized that the result it reached was directly in conflict with our decision in Blair, but it went on to say that while it "concurs with the Court of Appeals for the District of Columbia that the uniform price requirement was designed to eliminate ruinous competition for the fluid milk market, it cannot concur in [that Court's] conclusion that the requirement prohibits recognition of the historic fact that farmers located in particular areas near the market were accorded a greater share of that market." In other words, although Congress authorized a price pooling arrangement in order to eliminate revenue disparities among producers due to varying access to the fluid milk market, those who enjoyed that advantage before pooling must continue to have it because a larger percentage of their production customarily goes into fluid milk use. This is, of course, precisely what the Secretary gave as his primary reason for including the nearby differential in the Connecticut marketing order involved in Cranston:

Provision should be made also for the payment of a market differential to producers in the nearby supply area. The monies for the payment of these differentials will be obtained by a deduction from the total value of all milk before completing the computation of the market-wide uniform price.

* * * * * *

Milk received from the nearby area has long represented the major part of the total Connecticut supply. The differentials adopted recognize the higher percentage of the milk near to the market than of milk in the more distant zones which customarily has been used for fluid purposes. Nearby producers have been able to obtain a price higher, in relation to more distant producers, than can be accounted for by the advantage in the cost of transportation to market. This historical pattern of pricing has been typical in the markets of this region, and market differentials of this type have been adopted in the nearby regulated market. The differentials adopted will reflect more representative values with respect to the milk of both nearby and distant producers since the nearby producer will share in the higher-priced Class I market to an extent somewhat greater than otherwise would be the case under market-wide pooling. (Emphasis added.) 24 Fed. Reg. 1049, 1063 (1959).

The Cranston court does not, therefore, sustain the nearby differential on some reason other than that relied upon by the Secretary in propounding it. It, rather, embraces fully the Secretary's persistent assumption that Congress did not really mean what it said about end-use not being allowable as a factor in fixing the amount received for milk by the producer.^40.1 The Cranston court's acceptance of the "market differential" rationale thus flies in the face of the congressional purpose explicit in the statute, as well as the flat language of the Senate and House Committee Reports defining "market differential." See supra p. 673.

Even if the farm location differentials cannot be saved, appellants' argument continues, the case should be remanded to the Secretary so that he can determine whether other amendmants are necessary and whether the order, without a nearby differential, effectuates the declared policy of the Act.41 Appellants rely in this regard upon cases such as Addison v. Holly Hill Fruit Products, 322 U.S. 607, 64 S. Ct. 1215, 88 L. Ed. 1488 (1944), and FPC v. Idaho Power Co., 344 U.S. 17, 73 S. Ct. 85, 97 L. Ed. 15 (1952).

Those decisions do enunciate the principle that a court cannot always assume that an agency, had it been aware of the limits of its authority, would have promulgated an order differing only in the absence of the provision judicially declared unlawful. That principle, however, does not obtain here. The 1964 order contains an explicit severability clause, clearly indicating that all parties contemplated that judicial invalidation of some provisions would not upset the remainder. Moreover, unlike the provisions involved in Addison and Idaho Power, the nearby differential is discrete. There is nothing to suggest that elimination of it irretrievably impairs the Secretary's power and purpose to administer the significant regulatory scheme embodied in the remaining provisions of the order. He has simply been told to discontinue illegal payments under the order. Since any discrimination attributable to farm location would appear to be invalid, there is no reason to afford him a chance to disburse them under a different name.42

Although we affirm the judgment of the District Court, we must deal with the distant producers' claim of equitable entitlement in respect of the escrow fund.43 The complaint was filed on December 27, 1966, and the escrowing commenced with the preliminary injunction of January 16, 1967. Final judgment was entered by the District Court on June 15, 1967. In the exercise of our duty, as a court of equity, to "tailor * * * relief with a critical and balanced view of the ramifications of [the] decision,"44 we have decided that the monies escrowed before June 15, 1967 should be disbursed to the nearby farmers and that the distant producers are entitled to amounts thereafter collected.

In support of their claim, the nearby producers point out that the nearby differential was in operation for almost 30 years before it was challenged in court and that the nearby farmers have relied upon it over that period in making economic decisions and investments.45 In particular, they allege reliance upon the First Circuit's decision in Green Valley.46 Because it was, in language at least, an explicit judicial approval of the farm location differential, they suggest that the payments should not be terminated until the differential has finally been determined invalid by a court of at least equal rank in the federal judiciary.

We recognized in Blair that the plaintiffs' long acquiescence in the differential and their years of enjoyment of a blended price and stable marketing conditions in the milkshed were significant factors bearing upon their entitlement to equitable relief. In view of these factors, the nearby farmers could legitimately have expected to receive, and the distant farmers to pay, the differential until it has been held invalid, after full opportunity for the presentation of views on both sides, by a court. We think this is a fair view of the equities; and, since a preliminary injunction does not constitute a final judgment on the merits, appellant producers should receive the amounts escrowed between January 16 and June 15, 1967.

We are not persuaded that the endorsement of the nearby differential in Green Valley further alters the equities between the parties. The nearby producers' reliance on that opinion is amply recognized by preserving their preference until it has been finally invalidated by a responsible court having jurisdiction to do so. Their expectations cannot seriously be said to encompass a decision to that effect by a Court of Appeals but not by a District Court.

The judgment appealed from is affirmed and the case remanded to the District Court for further proceedings consistent herewith.

It is so ordered.