The employer ("Newpark") filed a petition in this court to review an order of the Benefits Review Board with regard to the claim of its disabled former employee Roundtree for benefits under the Longshoremen's and Harbor Workers' Compensation Act. 33 U.S.C. Secs. 901 et seq. The Board's order determined the method by which the claimant Roundtree's periodic compensation rate was to be calculated, but it also remanded the case to the administrative law judge for further findings and proceedings. A motion to dismiss the appeal as being from a non-final administrative determination was denied by a divided panel, 698 F.2d 743 (5th Cir.1983), which proceeded to determine a substantive issue on its merits and likewise remanded for further proceedings. We granted en banc rehearing, 706 F.2d 502 (1983), primarily to determine whether the Board's determination was a reviewable "final order", a statutory prerequisite under the Act for judicial review of the Board's rulings. Sec. 921(c).1

Contrary to the panel, we dismiss the present petition for review. Applying the well-settled general rule that a judgment or order is not final unless it ends the litigation on the merits and leaves nothing for the trier to do but execute the judgment, we hold that the present Board order--which determined a central issue of liability, but which nevertheless also remanded the administrative proceedings to the administrative law judge for further findings--is not a "final order", Sec. 921(c), so as to be statutorily subject to judicial review at this time.

We will detail the facts somewhat more fully below. For present purposes, we note only that the Board's order, which determined the method of computing compensation and remanded the proceedings for further findings, was concededly not final in any "technical" sense. The panel majority, however, found it to be reviewable at this time on a concept of pragmatic finality, since upon the court panel deciding a central issue (and reversing the Board as to it), the factual record as to this issue was now complete, although some unresolved collateral issues also remained for decision by the administrative proceedings upon remand. 698 F.2d at 747 & n. 2, 748 & n. 3.

The issue thus drawn is whether the Board's order, under such circumstances, should be deemed a "final order" and thus subject to judicial review under Sec. 921(c). The required finality for reviewability of an order of the Board follows, for the same reasons of policy, the contours of the finality-requirement under 28 U.S.C. Sec. 1291 for appealability of decisions of the district courts. Simms v. Valley Line Company, 709 F.2d 409, 413 (5th Cir.1983); Director, Office of Workers' Compensation Programs v. Brodka, 643 F.2d 159, 161 (3d Cir.1981); National Steel and Shipbuilding Company v. Director, Office of Workers' Compensation Programs, 626 F.2d 106, 107-08 (9th Cir.1980); Newport News Shipbuilding and Dry Dock Company v. Director, Office of Workers' Compensation Programs, 590 F.2d 1267, 1268 (4th Cir.1978) (all four decisions dismissing petitions of review because of non-finality of the Board's order).

In Firestone Tire & Rubber Company v. Risjord, 449 U.S. 368, 373-74, 101 S. Ct. 669, 673, 66 L. Ed. 2d 571 (1981), the Supreme Court has recently restated the general test for finality as being a decision "that 'ends the litigation on the merits and leaves nothing for the court to do but execute the judgment.' " Coopers & Lybrand v. Livesay, 437 U.S. 463, 467, 98 S. Ct. 2454, 2457, 57 L. Ed. 2d 351 (1978), quoting Catlin v. United States, 324 U.S. 229, 233, 65 S. Ct. 631, 633, 89 L. Ed. 911 (1945).

This finality rule is designed to avoid piecemeal trial and appellate litigation and the delays and costs of multiple appeals upon both parties and courts, as well as to provide a clear test so that needless precautionary appeals need not be taken lest substantive rights be lost. "Restricting appellate review to 'final decisions' prevents the debilitating effect on judicial administration caused by piecemeal appellate disposition of what is, in practical effect, a single controversy." Eisen v. Carlisle & Jacquelin, 417 U.S. 156, 170, 94 S. Ct. 2140, 2149, 40 L. Ed. 2d 732 (1974). "This insistence on finality and prohibition of piecemeal review discourage undue litigiousness and leaden-footed administration of justice * * *." DiBella v. United States, 369 U.S. 121, 124, 82 S. Ct. 654, 656, 7 L. Ed. 2d 614 (1962). "Thereby is avoided the obstruction to just claims that would come from permitting the harassment and cost of a succession of separate appeals from the various rulings to which a litigation may give rise, from its initiation to entry of judgment." Cobbledick v. United States, 309 U.S. 323, 325, 60 S. Ct. 540, 541, 84 L. Ed. 783 (1940), quoted with approval in Firestone, supra, 449 U.S. at 174, 101 S. Ct. at 673. See also 15 Wright, Miller, and Cooper, Federal Practice and Procedure, Secs. 3907, 3909 (1976).

Nevertheless, the Court has recognized a small number of narrow exceptions to this general rule that appealable finality is accorded only to a judgment or order that completely terminates the litigation below.2 Under these exceptions, the requirement of finality is "given a 'practical rather than a technical construction' ". Eisen, supra, 417 U.S. at 171, 94 S. Ct. at 2149. The exceptions arise from considering "the competing considerations underlying all questions of finality--'the inconvenience and costs of piecemeal review on the one hand and the danger of denying justice by delay on the other.' " Id. See 15 Wright, Miller, and Cooper, supra, Sec. 3907. Both the petitioner Newpark and the panel majority, in asserting the reviewable finality of the present Board order, essentially rely upon an exception of pragmatic finality that was recognized by Gillespie v. United States Steel Corporation, 379 U.S. 148, 85 S. Ct. 308, 13 L. Ed. 2d 199 (1964).

We will not here detail the facts nor attempt to analyze the several strands of reasoning by which Gillespie, because of the competing considerations, held to be appealable a technically non-final order, save to observe that none of Gillespie's exceptional reasons are apparent here. As recently summarized by the Court,

In Gillespie, the Court upheld an exercise of appellate jurisdiction of what it considered a marginally final order that disposed of an unsettled issue of national significance because review of that issue unquestionably "implemented the same policy Congress sought to promote in Sec. 1292(b) [permitting discretionary reviewing of interlocutory rulings]," id., at 154, 85 S.Ct. at 312, and the arguable finality issue had not been presented to this Court [after certiorari was granted] until argument on the merits, thereby ensuring that none of the policies of judicial economy served by the finality requirement would be achieved were the case sent back with the important issue undecided.... If Gillespie were extended beyond the unique facts of that case, Sec. 1291 [the finality requirement] would be stripped of all significance.

Cooper & Lybrand, supra, 437 U.S. at 477 n. 30, 98 S. Ct. at 2462.

Newpark contends, however, that despite this limited construction of Gillespie recently accorded by the Court itself, that decision's continued viability for application to the present Board order is demonstrated by American Export Lines, Inc. v. Alvez, 444 U.S. 274, 100 S. Ct. 1673, 64 L. Ed. 2d 284 (1980). There, citing Gillespie, the Court did not reject review of an interlocutory amendment ruling in a state court action as non-final. (The amendment in the state court action, affirmed by the state's high court, had permitted a harborworker's wife to assert a claim under general maritime law for loss of her injured husband's society.) The issue before the Court in Alvez, however, is distinguishable from that before us now.

The issue before us concerns the concept of finality under 33 U.S.C. Sec. 921 and 28 U.S.C. Sec. 1291, which afford an aggrieved party review in the court of appeal as of right. In Alvez, the issue arose under 28 U.S.C. Sec. 1257(3), where a "final" judgment of a state's highest court is reviewable by discretionary grant of certiorari where a right is claimed under federal law.

In Alvez, after the court had granted certiorari, it sua sponte noted the jurisdictional issue, observing that "were the case in the posture it stood when petition for certiorari was granted, we might well determine that the judgment lacked sufficient characteristics of finality to warrant an assertion of our appellate jurisdiction." 446 U.S. at 277, 100 S.Ct. at 1675. However, after certiorari had been granted, the case was tried in the state trial court and the plaintiffs prevailed, with "no federal question, except that which we are now asked to resolve," remaining in the litigation. 446 U.S. at 277-78, 100 S.Ct. at 1675-76. The Court noted that " 'the federal issue, finally decided by the highest court in the State, will survive and require decision regardless of the outcome of future state-court proceedings.' " 446 U.S. at 279, 100 S.Ct. at 1676.

Alvez is not apposite to the present issue. While the concepts of finality for Supreme Court review of state court judgments under 28 U.S.C. Sec. 1257 and for court of appeal review of district court judgments under 28 U.S.C. Sec. 1291 are often parallel, they also diverge in some aspects because of the Supreme Court's different and additional function in Sec. 1257 review of state court judgments to protect federal rights. 15 Wright, Miller, and Cooper, supra, Sec. 3909;3 16 Wright, Miller, Cooper, and Gressman, Federal Practice and Procedure, Sec. 4010 (1977). Thus, one of the exceptions in Sec. 1257 review according reviewable pragmatic finality to a technically interlocutory state ruling is where the federal issue has been finally determined in the state court proceeding. Cox Broadcasting Corporation v. Cohn, 420 U.S. 469, 482-83, 95 S. Ct. 1029, 1042, 43 L. Ed. 2d 328 (1975).4

Alvez accorded pragmatic finality to the state court ruling on the federal issue partly because of the functional purposes of review of state court judgments under Sec. 1257--different and distinguishable in this respect from those according "finality" to district court judgments under Sec. 1291 or to Board decisions under 33 U.S.C. Sec. 921(c)--and also because of the exceptional posture of the proceeding before it: after certiorari was (improvidently) granted, the state court judgment became (the court found) final for Sec. 1257 purposes. It was in that context that Alvez relied upon the Gillespie exception to strict finality, quoting from that decision: " '[N]ow that the case is before us ... the eventual costs, as all the parties recognize, will certainly be less if we now pass on the questions presented here rather than send the case back with those issues undecided.' " 444 U.S. at 279, 100 S. Ct. at 1676.

Alvez thus cannot be regarded as blanket authority for a court of appeal--to which review is sought of right from a "final" decision--to disregard the Congressionally mandated finality limitation upon its review powers and to ignore the powerful reasons of policy underlying the finality requirement, see I supra, in order to accord reviewable pragmatic finality to the trier's decision, whenever the reviewing court on a case-by-case basis determines that it would be more efficient to review immediately a central issue raised in review sought from a non-final decision.5

The issue before us arises in the following factual and procedural context:

Roundtree claims disability benefits under the Longshoremen's and Harbor Workers' Compensation Act. Roundtree, a welder, was injured on his first day of work as an employee of Newpark. Prior to this employment, he had worked for twenty years as an independent contractor. By the time the proceedings came before the Board, the principal (but not only) issue concerned the determination of his weekly wages, 33 U.S.C. Sec. 910, for purposes of computing the periodic compensation payments he claimed to be due to him.

Newpark contended that the weekly wage basis should be determined on the basis of Sec. 910(b) (the prior year's wages of co-workers). The Board affirmed the administrative law judge's decision that, instead, the determination of the wage basis should be based upon Sec. 910(c) (an alternative section applicable to "arriving at the average annual earnings of the injured employee," if the other statutory methods, Sec. 910(a) and Sec. 910(b) "cannot reasonably and fairly be applied", which may take into consideration other factors, "including the reasonable value of the services if engaged in self-employment" during the preceding year). However, the Board disagreed with the method used by the administrative law judge to determine the annual earnings of the preceding year, vacated the award of compensation, and remanded for further proceedings, including the determination of unresolved issues, such as attorney's fees.6

The uniform jurisprudence of this and every other circuit that has considered the issue is that a Board order, which determines liability but which remands to the administrative law judge to determine the award, is not a reviewable "final order" of the Board under Sec. 921(c). Director, Office of Workers' Compensation Programs v. Brodka, 643 F.2d 159, 161 (3d Cir.1981); National Steel and Shipbuilding Company v. Director, Office of Workers' Compensation Programs, 626 F.2d 106, 107-08 (9th Cir.1980); Newport News Shipbuilding and Dry Dock Company v. Director, Office of Workers' Compensation Programs, 590 F.2d 1267, 1268 (4th Cir.1978); United Fruit Company v. Director, Office of Workers' Compensation Programs, 546 F.2d 1224 (5th Cir.1977); and reported and unreported decisions therein cited. As stated in the seminal reported decision of this line of cases, "[i]t is a well-established rule of appellate jurisdiction, however, that where liability has been decided but the extent of damage remains undecided, there is no final order." Sun Shipbuilding & Dry Dock Co. v. Benefits Review Board, 535 F.2d 758, 760 (3d Cir.1976). Therefore, in all of these decisions, petitions for review of a Board remand order under such circumstances have been dismissed for lack of reviewable finality under Sec. 921(c).

The panel opinion recognized that this principle ordinarily applied, but it felt that, weighing the competing values, the danger of denying justice by delay outweighed the value incorporated in the finality rule of the inconvenience and costs of piecemeal litigation. 698 F.2d at 746-48. It felt that, by deciding the issues presented on the appeal, "the remaining issues will not be questions of law but merely the largely ministerial job of applying the law to the pre-existing record," 698 F.2d at 747, and it distinguished the above precedents as, unlike the present case, having "unresolved factual determinations that appellate review could not settle." Id. The panel recognized that, however, "some unresolved collateral issues may remain in the instant case, as presented in this appeal", such as determining the amount of attorney's fees (see also note 6 supra ). 698 F.2d at 748.

The panel then reached the merits of Newpark's petition for review and, reversing the Board's determination that Sec. 910(c) applied, held that Sec. 910(b) provided the appropriate statutory methodology by which to determine Roundtree's average weekly wage. We note, however, that if instead--after deciding the merits--the panel had affirmed the Board's determination that Sec. 910(c) provided the appropriate methodology, then (as the Board held) the record was not factually complete. The panel's determination that, on its deciding the legal principle of law, the remand would be a largely ministerial function of applying that principle to the preceding record thus depended on its first deciding the merits of the appellate proceeding, i.e., whether Sec. 910(b) applied (in which event, in the panel's view, the record was factually complete), instead of Sec. 910(c), as the Board had determined.7 In essence, under this methodology, neither the court nor the parties could know whether the record was factually complete for purposes of reviewable "finality" until the court had first determined the merit-issue and thereafter decided whether the administrative record was factually complete so as to justify review at this time.

The panel's case-by-case methodology of determining pragmatic finality for purposes of reviewability is thus in fundamental conflict with the values and purposes of the finality rule to avoid the delay and system-costs of piecemeal and multiple appeals, and to provide a relatively clear test of appealability so that needless precautionary appeals not be taken. See I supra. "Any such ad hoc decisions [ascribing reviewable finality to technically non-final decisions] disorganize practice by encouraging attempts to secure or oppose appeals with a consequent waste of time and money." Baltimore Contractors, Inc. v. Bodinger, 348 U.S. 176, 181, 75 S. Ct. 249, 253, 99 L. Ed. 233 (1955). As we stated in Freeman v. Califano, 574 F.2d 264, 267 (5th Cir.1978), where under circumstances somewhat similar to the present case we dismissed an appeal and rejected Gillespie pragmatic-finality arguments,

the potential evil of here establishing a precedent for other piecemeal appeals in other cases is real. Here the balance appears to be struck in favor of a finding of nonappealability. There is simply no significant countervailing interest here to outweigh the dominant and traditional policy of requiring a final decision to avoid piecemeal appeals.8

For these reasons, we therefore find, consistent with all prior jurisprudential determinations under such circumstances,9 that the present Board order, which determined the employer's compensation liability to the injured worker but which remanded to the administrative law judge for further proceedings, is not a final order within the contemplation of Sec. 921(c) and is, thus, not subject to judicial review at this time, since it did not end the litigation on the merits and leave nothing for the trier to do but execute the judgment.

The panel relied not only upon Gillespie, but also upon the decision of this circuit in Ingalls Shipbuilding Division, Litton Systems, Inc. v. White, 681 F.2d 275 (5th Cir.1982), which itself had relied upon Gillespie, in according reviewable pragmatic finality to a Board's remand order. The issue there involved an extremely important general question in the administration of the Longshoremen's Compensation Act whether an administrative law judge had the power to approve settlements by an injured worker (as the administrative law judge and the Board held), or instead whether this power was vested in the Deputy Commissioner and the Secretary (as the Director of the Office of Workmen's Compensation Programs urged). The Board, affirming the power of the administrative law judge to approve a settlement, nevertheless remanded the settlement to the judge for him to approve or disapprove it under Board-specified guidelines. The Director filed a petition for review in this court.10

The Ingalls panel accorded pragmatic finality to the Board's remand order and entertained review of it under Sec. 921(c), based upon Gillespie and Gillespie-type reasoning. The panel pointed out the delay inherent in remanding and the ease of deciding the issues then so as to expedite earlier termination of the proceedings--on reasoning similar to that of the panel in the present case, albeit here on issues applicable to benefits due a single claimant rather than, as in Ingalls, on central issues of the administration of compensation scheme. Ingalls is, however, the only judicial decision that has held that a Board remand order is reviewable under Sec. 921(c), insofar as we can ascertain, and in Ingalls the issue decided on that appeal could ultimately have been brought to the court for review after a final order of the Board (although admittedly the somewhat exceptional circumstances (both employer and employee were satisfied with the Board ruling) made that course somewhat unattractive and cumbersome).

A dismissal of the present petition for review for lack of jurisdiction because the remand order is not final, Sec. 921(c), could possibly rest on leaving Ingalls undisturbed, distinguishable and limited to its facts in its pragmatic-finality ruling because of its exceptional circumstances and the central importance of the issues requiring decision. We have decided, however, that Ingalls should be overruled, for its value in occasionally permitting reviewable pragmatic finality to Board remand orders is outweighed by its erosion of the values of the finality rule mandated by Congress as a prerequisite for our appellate jurisdiction, considering also the necessary uncertainty (as is illustrated by the present case) that must exist in the administration of a judicially created pragmatic-finality exception, where definable standards for allowing this extra-statutory appeal of right are virtually impossible to formulate.

Since the remand order of the Board is not a final order as required by 33 U.S.C. Sec. 921(c), we lack jurisdiction to review it, and the Newpark's petition for review is DISMISSED.

PETITION FOR REVIEW DISMISSED.

JERRE S. WILLIAMS, Circuit Judge, with whom CLARK, Chief Judge, BROWN and GARZA, Circuit Judges, join, dissenting:

In this case, the Court regrettably retreats from a justifiable flexibility in the law to a rigid rule which disserves the interests of justice in the judicial review of the actions of an administrative agency. We recognized the importance of such flexibility in Ingalls Shipbuilding Division, Litton Systems, Inc. v. White, 681 F.2d 275 (1982), and the panel opinion in this case followed Ingalls in turning its back on absolutism in such review. The Court overrules Ingalls and refuses to recognize that in the proper case the saving of time and cost to the litigants in judicial review should prevail over an unduly and unnecessarily stringent application of the finality rule.

In this dissenting opinion I do not undertake to repeat in full detail the reasoning which led the majority of the panel to hold that there was a "final order" in this case. See Newpark Shipbuilding & Repair, Inc. v. Roundtree, 698 F.2d 743, 746 (5th Cir.1983). Instead, I emphasize briefly the actual factual situation in this case and relate it to the admittedly controlling concept of "finality". Under Sec. 10 of the Longshoremen's and Harborworkers' Compensation Act, 33 U.S.C. Sec. 910 there is provision for alternative determination of the average weekly wages as a basis of calculating benefits for injured workers. Section 10(a) provides for calculation of benefits based upon the injured worker's own earnings if he has been employed in the same job for a year. Only if 10(a) cannot be applied is 10(b) applied. It provides for calculation of benefits based upon the wages of workers in comparable jobs for the past year. Only if 10(b) cannot be applied is 10(c) applied. Its method of calculation is based upon earnings potential of the employee rather than actual prior wages of the employee or of employees in comparable work.

In this case the Benefits Review Board found that 10(c) was applicable because it would not be "fair and equitable" for 10(b) to apply even though there were a number of welders in comparable employment whose average wages for the past year were readily available and were in evidence.

The sole ground which the Benefits Review Board gave for the abandonment of the progression from 10(a) to 10(b) to 10(c) was that the comparable employees had received a substantial wage increase during the preceeding year. The panel decision found on the merits that this was not an adequate justification in the law for moving from the 10(b) calculation to the 10(c) calculation simply because it is almost universal for employees to receive a wage increase during the year, and the progression contemplated by Congress would be totally destroyed by the holding of the Benefits Review Board.

Why this brief description of the merits of the dispute in view of the fact that the issue before us is whether the decision of the Benefits Review Board is appealable or not? The answer is to be found simply in the fact that without an understanding of the completeness of the decision by the panel of this Court on review from the Benefits Review Board can the justification for treating the decision of the Benefits Review Board as a final order be seen.

The critical point in this case is that the fundamental legal decision controlling the case has been made by the Benefits Review Board and is now on appeal to this Court. If we do not consider this issue as being presented to us in a "final order" the case will go back to the administrative law judge for further extensive proceedings based upon an interpretation of the statute which the panel of this Court found to be in clear contravention of the congressional policy. It is difficult to accept under these circumstances the "judicial economy" in the rigid rule which the majority here establishes.

The Supreme Court cases upon which the opinion for the Court relies most heavily are cases which are easily distinguishable and require the evaluation of wholly different considerations. They involve attempts to obtain judicial review of preliminary matters in administrative proceedings before the final decision on the merits is made. It is beyond cavil that allowing appeals in those kinds of cases promotes constant appeals to the Court and would fail completely a test of judicial economy. Thus, in Firestone Tire & Rubber Co. v. Risjord, 440 U.S. 368, 101 S. Ct. 669, 66 L. Ed. 2d 571 (1981), the Court was asked to review a ruling on a motion to disqualify plaintiff's counsel in advance of trial.

The case of Coopers & Lybrand v. Livesay, 437 U.S. 463, 98 S. Ct. 2454, 57 L. Ed. 2d 351 (1978) is cited as limiting the sweep of the seminal case of Gillespie v. United States Steel Corp., 379 U.S. 148, 85 S. Ct. 308, 13 L. Ed. 2d 199 (1964). We relied strongly upon Gillespie both in Ingalls and in the panel opinion in this case. But Coopers v. Lybrand involved another preliminary matter--whether a class action should or should not be certified before trial on the merits. Both of these cases involved attempts to apply the collateral order doctrine of Cohen v. Beneficial Industrial Loan Corp., 337 U.S. 541, 69 S. Ct. 1221, 93 L. Ed. 1528 (1949). That doctrine is not here involved. Further, both Firestone and Coopers & Lybrand are federal cases so they do not involve the distinction properly and painstakingly drawn by the majority of the Court in American Export Lines, Inc. v. Alvez, 444 U.S. 274, 100 S. Ct. 1673, 64 L. Ed. 2d 284 (1980) recognizing a broader concept of non-finality as nevertheless subject to federal judicial review in appeals from state courts. This issue, too, is not here involved.

The Supreme Court has not abandoned the key statement it made in Gillespie v. United States Steel Corporation, 379 U.S. 148, 152, 85 S. Ct. 308, 311, 13 L. Ed. 2d 199 (1964). There the Court said that the question of finality deserves a "practical rather than a technical construction" because it would be "impossible to devise a formula to resolve all marginal cases coming within what might well be called the 'twilight zone' of finality." While Gillespie itself involved a district court decision resolving an issue in advance of trial, the issue to be resolved was fundamental in that it sought to review the decision by the district court that the administratrix of decedent's estate could not bring suit on an unseaworthiness claim against a shipowner but could only bring suit under the Jones Act. The Supreme Court held that the issue was properly reviewable, as the decision was a "final order."

Contrary to the conclusion in the opinion of the Court, our earlier decision in Ingalls is distinguishable from the case sub judice only in that Ingalls went further in recognizing an exception to a strict finality rule. What Ingalls decided was an interpretation of the law as to what agency within the administrative processes of the Longshoremen's and Harborworkers' Compensation Act had the authority to approve a compromise settlement. That decision clearly was not final in the sense that after it was made it was then left to that agency to decide whether or not to approve the compromise settlement in the case, a discretionary decision which would clearly be subject to additional review. In the case before us, in contrast, the core legal issue is over once the decision is made. As the panel opinion pointed out, all that was left was the ministerial act of taking the figures already in the record and calculating the average weekly wages based upon them.

While the majority of the Court mentions the possible additional issue of attorneys' fees, it properly does not draw any specific conclusions with respect to that issue once it is mentioned. Attorneys' fees can be part of the core of the case, requiring that they be set before a judgment is final. Holmes v. J. Ray McDermott & Co., Inc., 682 F.2d 1143 (5th Cir.1982). The award of attorneys' fees, on the other hand, can be a collateral matter independent of the merits of the case and uniquely separable from a cause of action. White v. New Hampshire Dept. of Employment Security, 455 U.S. 445, 102 S. Ct. 1162, 1166, 1167 n. 14, 71 L. Ed. 2d 325 (1982); Obin v. Dist. No. 9, Int'l Ass'n of Machinists & Aerospace Workers, 651 F.2d 574, 584 (8th Cir.1981) ("[A] judgment on the merits of an action, otherwise final, is final for purposes of appeal notwithstanding that a claim for attorney's fees may remain to be decided.") Attorneys' fees clearly are collateral in the case before us.

Finally, it is not amiss to point out that the rules concerning finality and judicial review of administrative rulemaking make no fetish whatsoever of a principle that nothing more must remain to be done to have finality. FCC v. WNCN Listeners Guild, 450 U.S. 582, 593, 101 S. Ct. 1266, 1273, 67 L. Ed. 2d 521 (1981) (reviewing a "policy statement" of an agency without even discussing the issue of finality); Harrison v. PPG Industries, Inc., 446 U.S. 578, 100 S. Ct. 1889, 64 L. Ed. 2d 525 (1980); Abbott Laboratories v. Gardner, 387 U.S. 136, 149, 87 S. Ct. 1507, 1515, 18 L. Ed. 2d 681 (1967) ("the cases dealing with judicial review of administrative action have interpreted the 'finality' element in a pragmatic way.") Of course, I recognize the difference between administrative rulemaking and adjudication. But what is critical is that the rulemaking review cases accept a pragmatic approach to finality because the agency is declaring a fundamental governing legal rule. That is exactly the situation we have in the case which is before us, even though that principle or "rule" is being declared through adjudication. Cf. Securities & Exchange Comm. v. Chenery Corp., 332 U.S. 194, 67 S. Ct. 1575, 91 L. Ed. 1995 (1947).

This is not a case then in which a preliminary matter is being submitted to the Court for judicial review as in Firestone and Coopers & Lybrand. In this case a controlling legal principle is submitted to the Court for judicial review leaving nothing but a routine application of that principle to be carried out on a record which is complete, as the panel decided the case. Even if the decision on the legal question were the other way, the record may be complete, and in any event is virtually so. Certainly, even under the BRB decision, less remains to be resolved than was the circumstance in Ingalls Shipbuilding where a serious decision on the merits concerning the approval of an agreed settlement remained to be decided.

It is my view that Gillespie is the law. Ingalls Shipbuilding was a proper application of Gillespie as a recognized narrow exception to a requirement that there must remain absolutely nothing to be done after the order is issued in the review of orders of administrative agencies. Ingalls Shipbuilding should not be overruled. It follows then that in the case before us we have a proper situation for application of the exception. Instead, we are compelled by the Court in its decision to force the parties to go back through the administrative process in ignorance of what this Court will decide is the controlling law of the case. This case does not call for nor open the door to "piecemeal" judicial review of administrative action. It calls for straightforward controlling judicial review of orders which are for all reasonable purposes final orders within the meaning of 33 U.S.C. Sec. 921 and 28 U.S.C. Sec. 1291.

The opinion for the Court sacrifices control by this Court over the concept of finality. It will substantially increase the expenditure of administrative and judicial services needlessly in that narrow class of cases such as Gillespie, Ingalls, and this case in which the Court can readily perceive that its decision will resolve the fundamental legal issue about which the entire administrative appeal revolves. There is no reason to fear misuse of such expeditious resolution of core issues. We should not fear a "case by case" approach, recognized and accepted by the United States Supreme Court, when it is narrowly confined as here set out. We should not abdicate our judicial responsibility to dispose of pragmatically final orders in an expeditious fashion and with saving to all persons and agencies involved.

For these reasons I am compelled to dissent from the opinion of the Court.