William C. Humphreys, Jr., with whom Todd R. David and Alston & Bird, Atlanta, Ga., were on brief for appellant.

Louis N. Massery with whom Cooley, Manion, Moore & Jones, P.C., Boston, Mass., was on brief for appellees Milliken, Sinclair Paint Co., Western Bench Craft, Inc., Barber Coleman, Diversitech, Ross & Roberts.

Before CAMPBELL, Chief Judge, and TORRUELLA and SELYA, Circuit Judges.

SELYA, Circuit Judge.

There are two consolidated matters before us at this juncture. They are an odd couple: an interlocutory appeal and a petition for writ of mandamus.1 Because we find that the challenged orders (1) are not "final" within the purview of 28 U.S.C. Sec . 1291 (1982), (2) do not come within the encincture of our jurisdiction under any recognized exception to the finality principle, and (3) are not suitable grist for the rarely-used mandamus mill, we pretermit the present proceedings shy of the merits.

I. BACKGROUND

On New Year's Eve 1986, a conflagration engulfed the San Juan Dupont Plaza Hotel. The blaze resulted in ninety-six deaths, numerous personal injuries, and extensive property damage. Upward of two thousand persons sued. The suits brought in federal fora were consolidated for discovery purposes in the United States District Court for the District of Puerto Rico. Petitioner-appellant Recticel Foam Corporation (RFC) is one of roughly two hundred defendants in this massive multi-district litigation. It is an especially reluctant respondent, because it insists that the district court lacks jurisdiction over its corporate person. RFC filed a dismissal motion, Fed.R.Civ.P. 12(b)(1), but as of the date of oral argument herein, the district court had not acted upon the motion. The jurisdictional issue is not now before us, and we take no view it.

Confronted with a litigatory monster, the district court took several innovative steps to handle the obvious complexities of pretrial discovery. These maneuvers included entry of an elaborate case management order (CMO); appointment of liaison counsel for plaintiffs and defendants, respectively; establishment of a document depository; and formation of a joint discovery committee (JDC). The JDC, drawn from the ranks of the lawyers on both sides, was apparently designed to promote quick discussion and resolution of discovery disputes and to explore novel methods of discovery that would ensure expeditious progress of the litigation.

In January 1988, with the blessing of the JDC, a codefendant (not RFC) moved to compel production of various videotapes and photographs. The holder of the depictions, the San Juan Dupont Plaza Hotel Corporation, despite conceding materiality and relevancy, claimed a work product immunity. Rather than fight this claim to the bitter end, the JDC and the Hotel worked out an agreement; the asserted immunity was waived in exchange for reimbursement of one-half the cost of generating the film. The district court ratified the agreement, ordering all served defendants to share in the expense.2 RFC objected. It unsuccessfully sought reconsideration of the cost-sharing order and appealed the ensuing denial.

But there is more. On February 11, 1988, the district court entered an order ancillary to the earlier CMO. Recticel--which contended inter alia that the district court, having no personal jurisdiction over the corporation, could not require it to share in ongoing discovery expenses--objected, but to no avail. The February 11 order allocated the continuing costs of the document depository and the defense liaison equally among the served defendants and third-party defendants. Displeased with this turn of events, RFC sought mandamus. On April 4, a duty panel of this court, in an unpublished order, rejected portions of Recticel's petition out of hand. Insofar as the prayer that the cost-sharing order be quashed (or levies thereunder postponed until resolution of RFC's motion to dismiss), we consolidated that segment of the petition with the appeal from the order directing shared payment of production costs. Thereafter, we granted petitioner's motion to supplement the record in the consolidated proceedings to allow concurrent consideration of yet another cost-sharing order of the same genre, to wit, an order entered April 18, 1988 which covered essentially the same ground as the February 11 order, but concerned expenses subsequently incurred for the document depository and for liaison counsel.

Against this backdrop, we consider petitioner-appellant's eclectic offerings. We look first at the appeal, focusing on its jurisdictional underpinnings (or, more aptly put, on the lack of any), and then turn to the question of mandamus.

II. THE APPEAL

It is too elementary to warrant citation of authority that a court has an obligation to inquire sua sponte into its subject matter jurisdiction, and to proceed no further if such jurisdiction is wanting. As we recently stated: "No matter how tantalizing a problem may be, a federal appellate court cannot scratch intellectual itches unless it has jurisdiction to reach them." Director, OWCP v. Bath Iron Works Corp., 853 F.2d 11, 13 (1st Cir.1988). In this instance, we are persuaded that jurisdictional constraints preclude us from inquiring into the merits of RFC's appeal.

A. The Finality Principle. Our jurisdiction over appeals stems primarily from 28 U.S.C. Sec . 1291, which provides that the courts of appeals may review "final decisions of the district courts of the United States." An order is usually considered "final" only when it "resolv[es] the contested matter, leaving nothing to be done except execution of the judgment." United States v. Metropolitan Dist. Comm'n, 847 F.2d 12, 14 (1st Cir.1988). See also Firestone Tire and Rubber Co. v. Risjord, 449 U.S. 368, 373, 101 S.Ct. 669, 673, 66 L.Ed.2d 571 (1981); Coopers & Lybrand v. Livesay, 437 U.S. 463, 467, 98 S.Ct. 2454, 2457, 57 L.Ed.2d 351 (1978); Catlin v. United States, 324 U.S. 229, 233, 65 S.Ct. 631, 633, 89 L.Ed. 911 (1945).

Discovery orders, in general, are not final. See, e.g., Appeal of Licht & Semonoff, 796 F.2d 564, 568 (1st Cir.1986); Boreri v. Fiat S.P.A., 763 F.2d 17, 21 (1st Cir.1985); City of Las Vegas v. Foley, 747 F.2d 1294, 1297 (9th Cir.1984); cf. Bridge C.A.T. Scan Associates v. Technicare Corp., 710 F.2d 940, 943 (2d Cir.1983) (discovery orders not immediately appealable, generally, under 28 U.S.C. Sec . 1292(a)(1)). Short of electing not to comply and being adjudged in criminal contempt, a party may ordinarily obtain review of such an order only after judgment has entered. See Firestone Tire, 449 U.S. at 377, 101 S.Ct. at 675; Doyle v. London Guarantee & Accident Co., 204 U.S. 599, 607-08, 27 S.Ct. 313, 315-16, 51 L.Ed. 641 (1907); Licht & Semonoff, 796 F.2d at 568. Although pretrial cost-sharing orders can validly be characterized as case management orders rather than as "pure" discovery orders, the characterization is not dispositive. Federal jurisdiction cannot be dictated by the simple expedient of artful labelling. See Royal v. Leading Edge Products, Inc., 833 F.2d 1, 5 (1st Cir.1987).

The similarity between discovery orders and case management orders is, we suggest, striking. In both instances, the main litigation continues to pend in the district court, as to all parties. In both instances, the orders deal with preliminary matters, leading up to--but not comprising--the main event. In both instances, the activities to which the orders relate tend to be frequent, repetitive, fragmentary, and bound up in the progress of the suit as a whole. Whether or not case management orders are discovery orders in the conventional sense, they are sufficiently akin to discovery orders to warrant precisely the same treatment under the finality principle.

Having determined that, in the usual case, orders issued incident to a CMO are on a par with discovery orders vis-a-vis finality concerns, we now discuss those considerations which, in this instance, weigh against appellate jurisdiction. In the first place, the cost-sharing orders do not seem "final." Not only does the pot continue to boil furiously below, but the rights of the parties are not settled by the disputed orders in any relevant sense. Although the orders require some temporary shifting of funds, they do not purport to resolve any party's rights in any definitive way. Given the district court's continued exercise of jurisdiction, the orders remain subject to modification. See Fed.R.Civ.P. 26(f) (orders relating to "proper management of discovery.... may be altered or amended"). The district court can reassess their content, and make adjustments as it thinks best. The orders work less than an immutable rejection of Recticel's position because no irreversible legal consequences flow from them, as they presently stand.

Nor are the orders "final" in a monetary sense. The costs in question continue to mount, and further orders will doubtless be forthcoming. The fact that we are being asked to review what are admittedly the first flurry of a potential blizzard of similar orders argues strongly against their finality. Here, as with discovery orders generally, interruption of the ongoing process carries with it a potentially high price in terms of diminished efficiency. Moreover, to the extent that an inquiry into finality requires "some evaluation of ... competing considerations," Eisen v. Carlisle & Jacquelin, 417 U.S. 156, 171, 94 S.Ct. 2140, 2149, 40 L.Ed.2d 732 (1974), we believe that the dangers inherent in piecemeal review of cost-sharing orders3 far overbalance any realistic possibility of denying justice by a delay in appellate oversight. As we said in Bath Iron Works,

... [M]erely saying that it would be "practical" to hear an appeal at an earlier time does not make it so. The kind of practical consideration which would warrant a departure from so well settled a principle [as finality] requires, at the least, some specialized showing of out-of-the-ordinary circumstances. These might include, say, a demonstration that what remained to be done was self-executing, or that cognizable harm of an unusual sort would result from delay, or that blind adherence to the letter of the finality principle would work some great injustice or grave systemic diseconomy. A petitioner, we think, has the burden of making the specialized showing needed to complete the arduous climb over the jurisdictional threshold.

Director, OWCP v. Bath Iron Works Corp., at 14. Having stated the task, it is readily evident that RFC's jeremiad is entirely inadequate to it. The cost-sharing orders are nonfinal, hence nonappealable, unless appellate jurisdiction attaches in some other fashion.

B. The Collateral Order Exception. The only detour around the finality principle which could conceivably make ends meet in this instance is the so-called "collateral order" exception. See Cohen v. Beneficial Industrial Loan Corp., 337 U.S. 541, 545-47, 69 S.Ct. 1221, 1225-26, 93 L.Ed. 1528 (1949). Collaterality, in the Cohen sense, demands conformity to certain hard-and-fast essentials:

The order must involve: (1) an issue essentially unrelated to the merits of the main dispute, capable of review without disrupting the main trial; (2) a complete resolution of the issue, not one that is "unfinished" or "inconclusive"; (3) a right incapable of vindication on appeal from final judgment; and (4) an important and unsettled question of controlling law, not merely a question of the proper exercise of the trial court's discretion.

United States v. Sorren, 605 F.2d 1211, 1213 (1st Cir.1979); accord Licht & Semonoff, 796 F.2d at 570-71; Boreri, 763 F.2d at 21. We have styled the four requisites as comprising "separability, finality, urgency and importance." In re Continental Investment Corp., 637 F.2d 1, 4 (1st Cir.1980). Not surprisingly, discovery orders rarely satisfy all four of these criteria. See Boreri, 763 F.2d at 25-26; International Union, Etc. v. National Caucus of Labor Committees, 525 F.2d 323, 324 (2d Cir.1975). By analogy, the same can be said of case management orders and CMO derivatives.

It is at least arguable that the particular cost-sharing orders of which Recticel complains meet none of these four requirements. In multi-party, multi-case litigation, the district court's success is largely dependent upon its ability to uncomplicate matters. In that endeavor, the case management order is one of the court's trustiest tools. It is central to, and in a sense inextricably intertwined with, the main case. Flinging sand in the finely-tuned case management gears has the unhappy potential for bringing the entire litigation to a grinding halt. Then, too, the issue of how to defray aggregate costs is not finished business: the costs are ongoing, the district court retains jurisdiction, and further orders can (and doubtless will) be forthcoming. Lastly, although we express no firm opinion on the subject, we are skeptical that a district court's authority to impose reasonable cost-sharing orders in multi-district litigation is much in doubt. See, e.g., Fed.R.Civ.P. 26(f) (court's order following discovery conference may "includ[e] the allocation of expenses" anent discovery). The interposition of a jurisdictional defense seems to us flimsy armor against the force of that authority.

We need not dredge these waters too deeply, for this circuit has long held to the view that "urgency"--the likelihood of irreparable harm from a delay in obtaining review of the dispute--is the "central focus," if not the "dispositive criterion" in the Cohen equation. Boreri, 763 F.2d at 21 (quoting In re San Juan Star Co., 662 F.2d 108, 112 (1st Cir.1981)). Because RFC cannot swim this channel, it fails the Cohen test. Detailed discussion of the remaining criteria becomes, therefore, supererogatory.

We think this appeal not "urgent" within the usage of that term in our jurisprudence for several reasons. First, if the district court eventually denies RFC's motion to dismiss, petitioner-appellant's position with respect to the cost-sharing orders is then virtually indistinguishable from that of a swarm of other parties. In that event, the power of the district court to issue those orders, the extent of its discretion to utilize various allocation methods, and a myriad of like issues, will be appealable after final judgment. Nothing will be lost: money wrongfully paid by one party to another can be restored by judicial decree. Conversely, much will be gained: series of inchmeal attacks on successive cost-sharing orders will be avoided, and an appeal at the end of the litigation will permit resolution of the issue with the participation of all interested parties--not merely Recticel alone. This appears to us a fairer, more expedient, and more efficient procedure than RFC's scattershot approach.

Placing the shoe on the other foot does not alter the fit. Should the district court eventually grant the dismissal motion, petitioner-appellant's rights pertaining to these orders would not be irretrievably lost. A district court which has the power to allocate litigation costs retains continuing control over the allocation of those costs during the entire pretrial period (and beyond). Wholly apart from the adjudication of Recticel's dismissal motion, we believe that the district court can reshape and refashion its cost-sharing orders as new information comes to light, or as information already known takes on added significance. See Fed.R.Civ.P. 26(f) (discovery order "may be altered or amended whenever justice so requires" ). That being so, it is certain beyond peradventure that the district court can--and, in materially changed circumstances, should--entertain motions for the reallocation of expenses. If RFC prevails on jurisdictional grounds, such a motion will be available to it. If the motion to reallocate is granted, the problem disappears. If denied, any error could presumably be rectified on appeal from the eventual final judgment in the case, or RFC could apply for the entry of a judgment under Fed.R.Civ.P. 54(b), so as to pave the way for an earlier appeal. In any event, the right to efficacious review would not be lost.

Because petitioner-appellant's claim is fully capable of vindication on appeal from final judgment in the usual course, and Recticel would suffer no irreparable harm from waiting its turn, the urgency prong of Cohen has not been met. Thus, we have no jurisdiction over this appeal under the collateral order doctrine. See Van Cauwenberghe v. Biard, --- U.S. ----, 108 S.Ct. 1945, 1950, 100 L.Ed.2d 517 (1988) (if claim "is effectively reviewable on appeal from final judgment, [it] is not an immediately appealable collateral order under Cohen" ).

III. THE MANDAMUS PETITION

We need not dawdle over RFC's request that we invoke mandamus to address its grievances anent the district court's cost-sharing orders. To be sure, the mandamus power is not some vestigial remnant of a bygone era, to be wrapped in cellophane and left untouched by human hands. Though ancient in its origins and potent in its effect, mandamus may be a permissible--sometimes necessary--vehicle for obtaining immediate judicial review of nonfinal orders that would otherwise escape timely scrutiny. But, "the remedy of mandamus is a drastic one, to be invoked only in extraordinary situations." Allied Chemical Corp. v. Daiflon, Inc., 449 U.S. 33, 34, 101 S.Ct. 188, 189, 66 L.Ed.2d 193 (1980) (per curiam); accord Will v. United States, 389 U.S. 90, 107, 88 S.Ct. 269, 280, 19 L.Ed.2d 305 (1967); Boreri, 763 F.2d at 26; Acton Corp. v. Borden, Inc., 670 F.2d 377, 382 (1st Cir.1982); In re Arthur Andersen & Co., 621 F.2d 37, 38-39 (1st Cir.1980). A valuable asset when appropriately employed, mandamus has the potential, if promiscuously used, to spawn piecemeal litigation and disrupt the orderly processes of the justice system. Therefore, we have guarded against the too-free availability of the writ: its "currency is not profligately to be spent." Boreri, 763 F.2d at 26. Mandamus should be dispensed sparingly and only in pursuance of the most carefully written prescription, not made available over the counter, on casual demand. It is not a substitute for interlocutory appeal. See Sorren, 605 F.2d at 1215.

Mandamus entreaties are generally subject to a pair of prophylactic rules, which together require that a petitioner show (a) some special risk of irreparable harm, and (b) clear entitlement to the relief requested.4 Illustrating the first of these rules, the Court has written that a party seeking mandamus must "have no other adequate means to attain the relief he desires...." Allied Chemical Corp., 449 U.S. at 35, 101 S.Ct. at 190; accord Department of the Navy v. FLRA, 835 F.2d 921, 923 (1st Cir.1987); Boreri, 763 F.2d at 26; In re Justices of the Supreme Court of Puerto Rico, 695 F.2d 17, 20 (1st Cir.1982). Because interlocutory orders are reviewable on appeal from final judgment, a mandamus petitioner who attacks such an order must ordinarily demonstrate that something about the order, or its circumstances, would make an end-of-case appeal ineffectual or leave legitimate interests unduly at risk. Compare Boreri, 763 F.2d at 26 (mandamus inappropriate where petitioner "not in unavoidable danger of forfeiting its perceived rights ... [and] ... not unable to appeal the issues under other (more propitious) circumstances") with United States v. Christian, 660 F.2d 892, 897 (3d Cir.1981) (mandamus appropriate where, otherwise, district court's failure to act would be forever shielded from appellate review). In the case at bar, petitioner-appellant has advanced no thesis which would warrant a departure from the usual practice. There is no irremediable harm,5 and RFC's rights vis-a-vis the cost-sharing orders (such as they may prove to be) are susceptible of full vindication during an appeal taken in the ordinary course, see supra Part II.

The second rule, too, comes into play here. As a general matter, mandamus will not issue to control exercises of discretion. Rather, a petitioner's right to the writ must be "clear and indisputable." Bankers Life & Casualty Co. v. Holland, 346 U.S. 379, 384, 74 S.Ct. 145, 148, 98 L.Ed. 106 (1953). Where an action is committed to judicial discretion, it cannot often be said that a party's claim of entitlement to a particular result meets this benchmark. Allied Chemical Corp., 449 U.S. at 36, 101 S.Ct. at 190. For that reason, mandamus has customarily been granted only when the lower court was "clearly without jurisdiction," Sorren, 605 F.2d at 1215, or exceeded its discretion "to such a degree that its actions amount[ ] to a 'usurpation of power.' " In re Puerto Rico Elec. Power Auth., 687 F.2d 501, 503 (1st Cir.1982) (quoting DeBeers Consolidated Mines, Ltd. v. United States, 325 U.S. 212, 217, 65 S.Ct. 1130, 1132, 89 L.Ed. 1566 (1945)). Interlocutory procedural orders, particularly case management orders of the type here in issue, rarely will satisfy this precondition for mandamus relief. Trial courts enjoy a broad measure of discretion in managing pretrial affairs, including the conduct of discovery. Decisions regarding the scope of discovery, the allocation of expenses, the most appropriate areas for enforced economy, and the protections to be afforded parties in the discovery process, are ordinarily left to the informed judgment of the district judge, who is in a unique position to gauge and balance the potentially conflicting interests at stake. That, we think, is the case in this situation.

Further discussion of the point would serve no useful purpose. Recticel--as the party seeking issuance of the writ--had the burden of showing us that its right to extraordinary relief was "clear and indisputable," Acton Corp., 670 F.2d at 382, and that other available means of redress were inadequate. It has failed to do so. Thus, without reaching the merits of the district court's allocative orders, we decline petitioner's invitation to review the matter here and now, in the context of an application for mandamus.

IV. CONCLUSION

We prize strict adherence to the rules which define and delimit our jurisdiction. The finality principle, in particular, should be accorded great deference. The Court has termed it "crucial to the efficient administration of justice." Flanagan v. United States, 465 U.S. 259, 264, 104 S.Ct. 1051, 1054, 79 L.Ed.2d 288 (1984). Moreover, the importance of the finality requirement is at its zenith in large cases such as this. The finality rule, after all, was designed to conserve judicial energy and eliminate the "delay, harassment and cost" that would result from a barrage of interlocutory appeals. Manual for Complex Litigation 2d Sec. 25.1 (1985). The disruptive potential of successive interlocutory appeals is magnified in complex cases such as the one before us, where the district judge has made--and must yet make--numerous pretrial rulings, all of which are interrelated and tied into an integrated case management scheme. It follows ineluctably that "[w]e should not rush to deviate from [the final judgment rule], nor should we do so lightly." Director, OWCP v. Bath Iron Works Corp., at 16.

In the circumstances of this case, we abjure premature intrusion into precincts which are, for the present, reserved to the district court. It would disserve the proper relationship between trial and appellate courts in the federal system, and wreak havoc with the taxing demands of modern-day case management, were the court of appeals gratuitously to inject itself as a super-navigator of sorts, second-guessing the district court from turn to turn as that tribunal wended its way through the thickets and brambles of complex litigation. To do so, we suggest, would be to concentrate on the trees at the expense of a balanced vision of the forest.

We need go no further. The cost-sharing orders are not final within the intendment of 28 U.S.C. Sec . 1291; the Cohen exception to the finality principle does not apply; and there is no other hook upon which appellate jurisdiction may suitably be hung. Similarly, the paired petition fails to portray circumstances so extraordinary as to energize our mandamus powers. The merits of the cost-sharing controversy are not properly before us in this proceeding.6

In No. 88-1298, the appeal is dismissed without prejudice for want of appellate jurisdiction.

In No. 88-1204, the petition for writ of mandamus is denied, as improvidently brought.

1 The petition, No. 88-1204, is brought under the All Writs Act, 28 U.S.C. Sec . 1651(a) (1982), which empowers the courts of appeals to "issue all writs necessary or appropriate in aid of their respective jurisdictions and agreeable to the usages and principles of law." See Boreri v. Fiat S.P.A., 763 F.2d 17, 26 (1st Cir.1985). For ease in reference, we call the mode of relief sought in this action "mandamus." Technically, however, the writ of mandamus lies to compel action by an inferior court, whereas the writ of prohibition is used to prevent such action. The instant petition sounds more in the nature of prohibition. Nevertheless, we need not split fine hairs; the two are counterpart remedies for which the same grounds must be established, and they are frequently used in concert. See, e.g., Ex parte Oklahoma, 220 U.S. 191 , 209, 31 S.Ct. 426, 431, 55 L.Ed. 431 (1911); Morrow v. District of Columbia, 417 F.2d 728, 733 (D.C.Cir.1969)

2 Assembling the materials was estimated to have cost in the neighborhood of $600,000. Under the cost-sharing arrangement, therefore, the defense moiety was approximately $300,000. The moiety was to be prorated among the various served defendants

3 We need not recite chapter and verse as to these dangers. We have warned, time and again, of the perils inherent in leaving "the way clear for the four horsemen of too easily available piecemeal appellate review: congestion, duplication, delay, and added expense." Spiegel v. Trustees of Tufts College, 843 F.2d 38, 46 (1st Cir.1988)

4 We recognize that, within limits not yet precisely defined, mandamus may also be appropriate--albeit rarely--to resolve issues which are both novel and of great public importance. See In re Justices of Supreme Court of Puerto Rico, 695 F.2d 17, 25 (1st Cir.1982); Sorren, 605 F.2d at 1215-16; In re Ellsberg, 446 F.2d 954, 956-57 (1st Cir.1971); Morrow v. District of Columbia, 417 F.2d at 736; see generally Note, Supervisory and Advisory Mandamus Under the All Writs Act, 86 Harv.L.Rev. 595, 613-19 (1973). We need not plumb these murky depths, however, for it is abundantly clear that here, where interstitial matters of case administration are involved, the exceptional circumstances necessary to trigger what has been termed "advisory mandamus" are utterly lacking. As Sorren teaches: "Invocation of our advisory mandamus power is not to be used as a bootstrap device to circumvent the limits on our jurisdiction to review discretionary interlocutory rulings of district judges." 605 F.2d at 1216

5 We have earlier "rejected the general burdensomeness of litigation as a basis for assuming mandamus jurisdiction." In re Justices, 695 F.2d at 20. We unabashedly reaffirm that view today

6 On the same day that we heard arguments in these matters, we also heard argument in a related case, In re San Juan Dupont Plaza Hotel, No. 88-1436. There the issue--which involved a claim of work product privilege unrelated to the cost-sharing orders--had been certified for interlocutory appeal under 28 U.S.C. Sec . 1292(b). See generally Spiegel v. Trustees of Tufts College, 843 F.2d at 46 n. 7 (discussing Sec. 1292(b) certification). Although we do not now address the propriety of that certification, it plainly affords no toehold for accelerated review of the cost-sharing orders with which we are presently concerned