OPINION OF THE COURT

ADAMS, Circuit Judge.

The present appeal, marking the second time this litigation comes before this Court, is taken by the Secretary of Labor and the United Steelworkers of America from an order of the district court on remand from the judgment of the Supreme Court in Dunlop v. Bachowski. The appellants contend that the district court transgressed the boundaries established by the Supreme Court for permissible review of a decision by the Secretary not to bring suit under 29 U.S.C. § 482(b) to overturn a union election. As a threshold matter, however, it is necessary to determine whether the order of the district court constitutes a final judgment, appealable under 28 U.S.C. § 1291.

I.

Walter Bachowski was one of three candidates for the office of Director of District 20 of the United Steelworkers of America in an election held on February 13, 1973. He was defeated by the incumbent, Kay Kluz, by a margin of 907 votes out of approximately 24,000 votes cast. After exhausting his internal union remedies, Mr. Bachowski filed a complaint with the Secretary of Labor, alleging that there had been violations of Title IV of the Labor-Management Reporting and Disclosure Act which affected the outcome of the election. The Secretary declined to bring suit, and did not set forth the reasons for his decision.

Mr. Bachowski then initiated an action against the Secretary and the Steelworkers in the district court, seeking to compel the Secretary to file a complaint to upset the election. The district court dismissed the action for lack of subject matter jurisdiction, concluding that it was without authority to review the Secretary's determination.

In Bachowski v. Brennan, a panel of this Court reversed the district court's order of dismissal, holding that the decision of the Secretary in refusing to file suit was subject to judicial review. The panel also held that the scope of review encompassed a challenge to the factual basis for the Secretary's conclusion, an inquiry that might necessitate a trial-type hearing. Finally, the Secretary was ordered to provide Mr. Bachowski with a detailed statement of reasons for not bringing suit.

The Supreme Court granted certiorari, and in Dunlop v. Bachowski, reversed the judgment of this Court. It agreed with the holding that the Secretary's determination not to bring suit under Title IV is judicially reviewable and that he must submit a statement of reasons to serve as a basis for judicial review. However, the Court found the conclusions regarding the proper scope of this judicial review to be unacceptably broad. The Supreme Court noted that one of Congress' chief concerns in shaping the mechanisms of Title IV was "`to settle as quickly as practicable the cloud on the incumbents' titles to office'", and that the procedures envisioned by this Court could serve only to prolong Title IV disputes. Thus, the Supreme Court held that review should ordinarily "be confined to examination of the reasons statement, and the determination whether the statement, without more, evinces that the Secretary's decision is so irrational as to constitute the decision arbitrary and capricious."

On remand, Mr. Bachowski filed a motion with the district court for further proceedings, alleging that he was entitled to a trial-type hearing under the limited circumstances established by the Supreme Court and, in the alternative, that the Secretary's statement was arbitrary and capricious on its face.

In an opinion filed on December 19, 1975, the district judge, the Honorable Edward Dumbauld, rejected Mr. Bachowski's request for a trial-type hearing. However, he ordered the Secretary to submit a supplemental statement of reasons, ruling that the initial statement did not adequately explain the basis for not bringing suit. In particular, Judge Dumbauld expressed dissatisfaction with certain aspects of the methodology employed by the Secretary to determine the possible effect of violations on the outcome of the election.

After receiving the supplemental statement, the district court ruled that the standard utilized by the Secretary in deciding whether to bring suit was "irrational," and remanded the cause to him for a "recount" to be conducted pursuant to guidelines established in the district court's opinion. The district judge, however, did not direct the Secretary to file suit. Both the Secretary and the Steelworkers appealed from the second order that was entered by Judge Dumbauld.

On June 17, 1976, Mr. Bachowski filed a motion to dismiss the appeals for want of an appealable order. This motion was denied without prejudice on July 2, 1976. Mr. Bachowski's contention that this Court lacks statutory jurisdiction has been raised anew before this panel. We have concluded that the order of the district court is not a final order, appealable under Section 1291, and that the appeals, accordingly, must be dismissed.

II.

Section 1291 is the sole basis relied upon by the Secretary and the Steelworkers to establish our jurisdiction. The cases are clear in their command that only final decisions of the district court are appealable under this statute. Courts have recognized that the precepts forming the bulwark of this historic doctrine are the realization that piecemeal appeals are costly and inefficient and do not serve the cause of a just and speedy resolution of controversies. Yet, the cases also admonish us to give section 1291 a "practical rather than a technical construction."

Mr. Bachowski maintains that the district court's order is not final because it does not bring about a final resolution of all the claims and defenses in this matter. He notes that the ultimate relief requested in his complaint was a decree ordering the secretary to file suit. Judge Dumbauld did not address himself to the demand that the Secretary file suit; instead, he merely remanded the cause to the Secretary for further proceedings.

If we were to affirm the district court and remand to the Secretary for the mandated recount, Mr. Bachowski argues, the Secretary might continue in his refusal to bring suit. He claims that such a course would necessitate a further round of proceedings in the district court, this Court, and perhaps the Supreme Court, dealing with the difficult constitutional issue whether a court may order the Secretary to bring suit. The consequence of such a splitting of appeals, Mr. Bachowski insists, would result in a significant delay in the final resolution of the litigation, the paradigmatic evil that the final order doctrine seeks to avoid.

The Secretary and the Steelworkers have countered with two lines of argument. They claim that the district court's judgment is final because it conclusively resolves what they characterize as the ultimate issue in this lawsuit — namely, whether the statement of reasons provided by the Secretary is arbitrary and capricious.

Alternatively, they contend that the judgment below is final in its effect on the Secretary. If the Secretary cannot appeal the judgment of Judge Dumbauld at this point, they maintain that the Secretary may never be able to do so, since the question of the propriety of Judge Dumbauld's decision would be overtaken by events. For example, if the Secretary would decide to bring suit after undertaking the mandated recount, the issue whether Judge Dumbauld deviated from the teachings of Dunlop would clearly be moot. Thus, the Secretary and the Steelworkers advocate that we entertain the present appeal lest the Secretary be burdened with an allegedly incorrect and disruptive precedent in this rapidly-evolving area of law.

In support of their position, the Secretary and the Steelworkers have directed our attention to a number of opinions that have found appealable orders by district courts that remanded cases to administrative agencies for further consideration. Chief among these authorities is Cohen v. Perales. In Perales, the district court had ruled that physicians' affidavits did not constitute the substantial evidence required to support a denial of social security disability benefits, and remanded the action for a new hearing. The Fifth Circuit held such an order appealable, invoking the Supreme Court's statement in Cohen v. Beneficial Industrial Loan Co. that the concept of finality must be interpreted in a practical manner. "Unless the Secretary is allowed to appeal from this order," the Fifth Circuit reasoned, "he will never be able to reach the questions involved, because on the next appeal, if there is one, the sole question may be the substantiality of the evidence, and not its admissibility."

III.

The requirement that only final judgments are reviewable in the federal courts dates back to the beginnings of the federal judicial system. The Judiciary Act of 1789 provided for review only of final judgments and decrees, whether at law or in equity. And indeed, this limitation can trace its roots to the practice of the English common law courts that entertained appeals solely from the final disposition of a completed controversy.

Until the last decade of the nineteenth century, the final judgment statutes provided the sole authorization for review within the federal judicial system. Congress began to realize, however, that rigid application of the final judgment rule in all cases might inflict irreparable harm upon litigants in certain instances, and might actually have the effect of unnecessarily prolonging the litigation. Its response, however, was not an abrogation of the general rule of finality, but rather a careful delineation of narrow categories of interlocutory orders that would be appealable. These initial enactments, now codified in 28 U.S.C. § 1292(a), encompassed limited types of interlocutory orders that would be appealable, the most familiar of which are orders granting, continuing or denying a preliminary injunction.

Although these statutes ameliorated, to some degree, the hardships caused by the final judgment rule, there were other instances where the underlying concerns of serious harm and wasted effort, which section 1292(a) spoke to, were also present. At the instance of the Judicial Conference of the United States, Congress enacted 28 U.S.C. § 1292(b). Unlike earlier statutes permitting certain substantive classes of interlocutory appeals, section 1292(b) established a procedure for judicial identification of cases where the grant of an interlocutory appeal might aid in promptly resolving litigation. This would be done by a process whereby the district judge and the court of appeals would each certify a case as comporting with specified criteria. The certification procedure is not mandatory; indeed, permission to appeal is wholly within the discretion of the courts, even if the criteria are present.

Thus, the reaction of the Congress to the rigors occasionally imposed by the final judgment rule has been very cautious. Congress has retained the finality doctrine as the general rule of appealability, in recognition of the rule's general contribution to the maintenance of an orderly judicial system. But it has also provided limited exceptions to the rule of finality so as to relieve any hardship that remorseless application of the rule might create.

IV.

Although the venerable lineage of the final judgment rule and the reticence of Congress to deviate too far from its strictures are certainly significant factors in our inquiry, they do not conclude it. As commentators have noted, the draftsmen of the Judiciary Act of 1789 did not articulate any functional justification for their adoption of the rule of finality. And the Supreme Court opinions explicating the formula for the finality rule by no means speak with harmonious voices. It is therefore necessary for us to examine in some detail the purposes of the final judgment rule, the gloss placed upon the provision by the Supreme Court, and the opinions of this Court concerning finality.

The hostility towards piecemeal appeals expressed by the final judgment rule has a strong basis in logic and practicality. Forbidding appeals from all interlocutory judgments of the district courts achieves significant savings in time and resources on the part of litigants and courts. This is so since if litigation proceeds, the intermediate ruling may lose its significance for a number of reasons. For example, a settlement might be reached, or the party aggrieved by the allegedly erroneous ruling might ultimately win the suit.

An immediate appeal from certain interlocutory orders, such as denial of a motion for summary judgment, may, admittedly, serve to expedite the termination of lawsuit. But this will not necessarily be the outcome. Indeed, as Professor Wright has indicated, the majority of trial court rulings are affirmed. Consequently, such an appeal could very well result in prolonging, rather than promptly terminating an action. Moreover, by interdicting most interlocutory appeals, we deprive litigants of a tool that could possibly be used to harass parties with lesser resources, and to drive them into submission.

The rule of finality, however, is not without costs. As we have previously noted, certain interlocutory rulings can inflict irreparable interim harm upon aggrieved parties. And denying review of some interlocutory orders could result in a net waste of effort. We have already described Congress' response to such tension. This problem has elicited two responses from the courts. First, the opinions have varied in their emphases, some hinting that the concept of finality be given a flexible interpretation, while others stress the importance of safeguarding the integrity of the final judgment rule. And, second, the courts have crafted new exceptions to the final judgment rule that have increased the number of interlocutory decisions that can receive immediate review.

V.

The finality opinions of the Supreme Court, as one might expect, reflect this duality to a considerable degree. Several decisions express the most rigid conceivable formulations: a final judgment is one which ends the litigation on the merits and leaves nothing for the court to do but execute the judgment. Other cases, while eschewing this formulation, have highlighted the values served by the final judgment doctrine and emphasized the dangers of deviating from it. In Cobbledick v. United States, one of the leading cases, the Supreme Court held that an order denying a motion to quash a subpoena duces tecum served upon persons who were to appear before a grand jury was not a final order. In arriving at this conclusion, Mr. Justice Frankfurter stressed the purposes behind the final judgment rule and its importance in maintaining the smooth functioning of the federal judicial system:

"Congress from the very beginning has, by forbidding piecemeal disposition on appeal of what for practical purposes is a single controversy, set itself against enfeebling judicial administration. 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. To be effective, judicial administration must not be leaden-footed. Its momentum would be arrested by permitting separate reviews of the component elements in a unified cause. * * * For purposes of appellate procedure, finality * * * is not a technical concept of temporal or physical termination. It is the means for achieving a healthy legal system. * * *"

In contrast to these stern admonitions, the Court has also spoken in terms which suggest a more lenient, balancing, approach to finality. Thus, in Dickinson v. Petroleum Conversion Corp., the Supreme Court noted that in resolving problems of finality, courts "sometimes choos[e] one and sometimes another of the considerations that always compete in the question of appealability, the most important of which are the inconvenience and the costs of piecemeal review on the one hand and the danger of denying justice by delay on the other."

The most prominent of the Supreme Court's cases prescribing a pragmatic stance to finality, however, is Gillespie v. United States Steel Corporation. The Court, in ruling that certain district court orders were appealable, made several sweeping observations about the nature of the final order requirement. Urging that finality be given a practical interpretation, the Court suggested that close questions should be determined in favor of appealability. And the opinion repeated the suggestions of Dickinson that questions of finality be resolved by a particularized balancing of the costs of accepting or declining review in each case. Moreover, the Court noted that the traditional touchstone of finality, whether the district court's order was the last possible one, is a less than infallible guidepost.

Although many commentators initially anticipated that Gillespie would profoundly alter the final judgment doctrine, these prophecies have not been borne out. Instead, Gillespie has been restrictively interpreted and sometimes ignored altogether. This has led one commentator who is sympathetic to a more pragmatic approach to note that the "spectacular possibilities of the Gillespie opinion" have descended into a "dismal limbo."

The other response of the Supreme Court to the hardships engendered by the final judgment rule, the creation of new exceptions, is best illustrated by the now-familiar collateral order doctrine announced in Cohen v. Beneficial Industrial Loan Corporation. Under this tenet, an appellate court may review a determination of a trial court, even though the judgment lacks the traditional indicia of finality, if the substance of the appeal is separate from the merits of the underlying suit, and if effective review of the matter would be doubtful on an appeal from a conventional final order.

In short, the opinions of the Supreme Court on the subject of the final judgment rule do not contain clear guidance for the manner in which we are to confront matters of appealability. As Justice Powell has noted, "No verbal formula yet devised can explain prior finality decisions with unerring accuracy or provide an utterly reliable guide for the future." Although there are some cases supporting a generous attitude to appealability, they have had limited influence and coexist uneasily with precedents urging a stricter interpretation of finality.

Given the lack of definitive signals from the Supreme Court, it is necessary to plow further and to consult the cases in our circuit that discuss the finality concept.

VI.

The opinions of this Court present a somewhat more consistent pattern than those of the Supreme Court. This Court has adhered to a less pliable definition of finality, and has frowned upon expansive judge-made exceptions to the final judgment rule.

Several examples will suffice to establish this position. We have consistently noted that the collateral order doctrine should be narrowly construed, lest all parties aggrieved by interlocutory orders bombard the courts of appeals with allegations that they too fit within the exception. A felicitous expression of our reticence to label orders as "collateral" is the opinion by Judge Aldisert in Borden Co. v. Sylk, a case refusing to allow an appeal by a non-party witness from an order directing him to answer deposition questions:

Every interlocutory order involves to some degree, a potential loss. That risk however, must be balanced against the need for efficient federal judicial administration as evidenced by the Congressional prohibition of piecemeal appellate litigation. To accept the appellant's view is to invite the inundation of appellate dockets with what have heretofore been regarded as non-appealable matters . . . To accept the appellant's view is also to invite a geometrical increase in the already unacceptable delay between the date of filing and trial in the metropolitan district courts. The present case, filed over three years ago and now held in abeyance pending the outcome of this appeal is a splendid example of the Homeric proportions that such litigation can assume. Our over-burdened courts have little time or appetite for such protractions.

The concerns set forth in Borden have been voiced by other members of this Court in not acceding to requests to expand the doctrine of Cohen.

A further instance of this Court's reluctance to inflate the boundaries of section 1291 can be found in the rejection of the "death knell" doctrine in Hackett v. General Host Corp. Although Judge Gibbons' majority opinion recognized that there was merit in permitting appeals from refusals to certify cases as class actions, he stated that this "must be balanced against the competing policies which have historically protected the federal appellate courts from being overwhelmed by review." Judge Gibbons also noted that any arguments in favor of interlocutory appellate review as of right must be scrutinized in light of the availability of discretionary channels of review, particularly section 1292(b) and FRCP 54(b). Section 1292(b) and Rule 54(b), he concluded, were the preferable routes since they permit immediate review of genuinely pressing questions while affording the courts of appeals greater control of their already bulging dockets.

The opinion of this court sitting en banc in United States v. Estate of Pearce is additional evidence of our narrow reading of the final judgment doctrine. In that case, we held that a decision of a district court denying a motion to quash a sequestration order was not appealable under section 1291. In addition to rejecting a request to find the case reviewable under the collateral order doctrine, Judge Garth, speaking for the Court, specifically refused an invitation to ignore the non-appealability, under traditional criteria, of the district court's order on account of the greater hospitality to appeals suggested by Gillespie. We indicated that whatever benefit might redound to the litigants under such an approach would be more than counterbalanced by the harm that piecemeal litigation would impose on parties in general and upon this Court.

With these principles in mind, we now turn to the precise issue before us, namely whether the remand by Judge Dumbauld, following his rejection of the Secretary's order, is appealable under section 1291.

VII.

Under the simplest test of finality — whether the order in question is the last one in the case — we could not entertain the present appeal. A declaration that the analytical techniques employed by the Secretary are irrational is not the total relief that Mr. Bachowski demanded in his complaint, and indeed still seeks. He asked and continues to ask, that the district court order the Secretary to file suit. The order now in question, although an important way-way station for Mr. Bachowski, does not address this ultimate request.

Gillespie might be read to indicate that this inquiry may not be dispositive of the question of finality. A more searching examination of the particular order brought to the court is therefore required before an appeal is disallowed.

The Secretary and the Steelworkers thus argue that the district court's order is final as a practical matter. The crux of their contentions is that a decision that jurisdiction is now lacking may very well deprive the Secretary of any opportunity to test the correctness of the scope of review employed by Judge Dumbauld. They maintain that the Perales line of cases provide clear authority for the entertainment of this appeal. We do not agree.

This case does not present the same type of situation that the Fifth Circuit encountered in Perales. In that case, a failure to review the district court's guidelines as to the admissibility of and the weight to be given to hearsay evidence, almost certainly would have foreclosed any possibility of review of that matter. For if the agency followed the district court's standards on remand, and once again held that the complainant was not entitled to benefits, the issue on a second appeal would be the substantiality of the non-hearsay evidence adduced at the second hearing. The propriety of the district court's prior action that rejected the use of affidavits would have become moot.

Such is not the case in the present appeal. While we recognize that there are circumstances in which the Secretary would not be able to obtain later review of Judge Dumbauld's determination, it is not inexorably so, as was the case in Perales. By way of illustration, if the Secretary, after remand, would continue in his refusal to bring suit, and the district court ordered him to do so, the viability of the mode of review employed by Judge Dumbauld would be before us on review. Indeed, it might well be the duty of a court to consider the question regarding the scope of review used by the district court before examining the difficult constitutional issue that would be posed by the attempt of a judge to compel an executive official to file suit.

To follow the path charted for us by the Secretary and the Steelworkers would thus require an extension of the Perales doctrine. We are not inclined to do this, in view of our circuit's disinclination to expand the class of appealable final orders. Our conclusion is reinforced by this Court's preference, expressed in Hackett and other cases, to allow for appeals from pressing interlocutory orders by the mechanism of certification under section 1292(b). This was a course that was open to the Secretary or the Steelworkers in this case. That they did not choose to make use of it can in no way affect our analysis of whether the order of the district court constitutes a final judgment.

VIII.

This case provides ample illustration of the proposition that application of the final order doctrine is often made with considerable diffidence. The substance of the dispute here is highly significant, and immediate resolution would clarify an important aspect of federal law as well as possibly terminating a lengthy controversy. However, the wisdom of the final judgment rule lies in its insistence that we focus on systemic, as well as particularistic impacts. The appellate system has become increasingly overburdened and the future would appear to promise no relief from the continuous increase in case loads. Accordingly, it would seem to us to be a disservice to the Court, to litigants in general and to the idea of speedy justice if we were to succumb to enticing suggestions to abandon the deeply-held distaste for piecemeal litigation in every instance of temptation. Moreover, to find appealability in those close cases where the merits of the dispute may attract the deep interest of the court would lead, eventually, to a lack of principled adjudication or perhaps the ultimate devitalization of the finality rule as enacted by Congress.

IX.

The appeals will be dismissed.