dissenting:
Although the substantive issue presented by this case seems relatively unimportant, the court’s decision dramatically broadens the scope of judicial review of National Mediation Board decisions. Despite extensive recitation of and homage to precedent establishing a highly deferential standard, Maj. Op. at 136-37, the court in practice uses an approach no more deferential (and perhaps even less so) than ordinary review under Chevron U.S.A., Inc. v. NRDC, 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). I dissent.
Judicial review under § 2 Ninth of the Railway Labor Act, 45 U.S.C. § 152 Ninth, started with a bang — Switchmen’s Union v. National Mediation Board, 320 U.S. 297, 64 S.Ct. 95, 88 L.Ed. 61 (1943).; Like this case, it involved a question of statutory construction — namely, whether the Act permitted the Board to divide a “craft or class” of employees of a single carrier into smaller units for collective bargaining purposes. Plaintiff union attacked a Board decision that was based solely on the Board’s notion that it had no such power, a view without visible support in the statutory language. The Court found the decision absolutely unreviewable, on the principle (as the Court later put it) “that it was for the Board, not the courts, finally to resolve such questions.” Railway Clerks v. Employees Ass’n, 380 U.S. 650, 659, 85 S.Ct. 1192, 1197, 14 L.Ed.2d 133 (1965).
In Leedom v. Kyne, 358 U.S. 184, 79 S.Ct. 180, 3 L.Ed.2d 210 (1958), the Court somewhat tempered the nonreviewability rule of Switchmen’s, but not much. At issue was an NLRB order in certification proceedings under § 9 of the NLRA, and the Court appeared to assume that the Switchmen’s principles applied. Id., 358 U.S. at 187-90, 79 S.Ct. at 183-85. The statute had explicitly barred inclusion of professional employees with nonprofessional ones without the former’s consent. The NLRB conceded that it had done just that, id. at 187, 79 S.Ct. at 183, and the Court found judicial relief available. Evidently not regarding conceded legal error as grounds enough to justify judicial interference, the Court emphasized that the Board order was “contrary to a specific prohibition in the Act.” Id. at 188, 79 S.Ct. at 184 (emphasis added). It stressed how patent a violation the Board had committed, describing the statutory provision at issue as “clear,” “mandatory,” and “definite”. Id. at 188, 189, 79 S.Ct. at 184.
*143In applying the Switchmen’s/Leedom mandate, this court has stated that the level of judicial review permitted by the decisions is “one of the narrowest known to the law.” International Ass’n of Machinists v. TWA, 839 F.2d 809, 811 (D.C.Cir.1988). Specifically, a reviewing court may only take a “peek at the merits” and may interfere only if the peek reveals an error “as obvious on the face of the papers as the violation of specific statutory language”. International Bhd. of Teamsters, Chauffeurs, Warehousemen & Helpers of Am. v. Brotherhood of Ry., Airline & S.S. Clerks, 402 F.2d 196, 205 (D.C.Cir.1968).
Taking the prescribed peek, one finds no such blazing error. The Board’s Merger Procedures provide that railroad carriers that have “merged” (as defined in the Board’s regulations) “may invoke the Board’s services for a determination of the post-merger status of any NMB certifications on the applicable properties.” See NMB Merger Procedures, 17 N.M.B. 44, 54 (1989). The Procedures also allow the Board to act on its own initiative. See id. at 50-51. Investigation and certification— sorting out which union represents which employees — are indisputably what the Board is supposed to do. “[S]o long as the Board is acting with the purpose of ‘findpng] the fact’ as to who is the employees’ representative, the courts are deprived of jurisdiction to review Board decisions.” America West Airlines v. NMB, 969 F.2d 777, 781 (9th Cir.1992). The Board here does not propose to go off settling union/carrier disputes or engage in any other activity plainly outside of the Board’s mission.
Appellants’ only claim is that the Board may not pursue its investigation and certification mandate at a carrier’s instigation or on its own hook. But the statute says nothing to bar its doing so in either circumstance. The pertinent provision, § 2 Ninth, reads:
If any dispute shall arise among a carrier’s employees as to who are the representatives of such employees designated and authorized in accordance with the requirements of this Chapter, it shall be the duty of the Mediation Board, upon request of either party to the dispute, to investigate such dispute and to certify to both parties, in writing, within thirty days after the receipt of the invocation of its services, the name or names of the individuals or organizations that have been designated and authorized to represent the employees involved in the dispute, and certify the same to the carrier.
45 U.S.C. § 152 Ninth.
The language requires the Board to engage in investigation and certification at the behest of a “party” to the dispute. The Board does not argue, that carriers are parties entitled to “request” such investigation or certification; the Supreme Court rejected a carrier claim to a similar entitlement — participation—in Railway Clerks, 380 U.S. at 666-68, 85 S.Ct. at 1200-02. The Board’s idea is simply that, as the statute does not preclude investigation and certification activities under other circumstances, it leaves the Board free to act in the circumstances that it has identified in the Merger Procedures. No statutory words suggest the contrary.
One might claim a conflict with the statutory language on the theory that if neither a union nor an employee brings the representation issue to the Board, there is no “dispute” to be resolved, so that the statutory .predicate — a dispute — is absent. The theory is valid only in the most pedantic and meaningless sense. If investigation procedures are initiated at the behest of a carrier and it proves that the pertinent unions and employees are in harmony as to who represents whom, clearly the proceeding must come to a prompt ending; the Board is not to go out into the world fomenting trouble. But plaintiffs here offer no suggestion that the Board has so acted, or even that the Merger Procedures would authorize such conduct.
Of course the Board could fall into “gross” or “clear” error even without the sort of direct flouting of the statute that was conceded in Leedom. Judge Leven-thal’s phrase in International Bhd. — an “error ... as obvious as the violation of *144specific statutory language” — was well chosen. For example, some powerful linguistic norm might come into play, such as the proposition that when a legislature uses a string of words whose meanings all bear a strong family resemblance, plus a more general word with some meanings that stretch much farther, one reads the general word only in the limited sense common to the others. In Latin, noscitur a sociis. Plaintiffs here identify no such linguistic norm that the Board's decision might violate, and none comes to mind.
Alternatively, a strong substantive background norm might control. See generally Cass R. Sunstein, Interpreting Statutes in the Regulatory State, 103 Harv.L.Rev. 405, 469-508 (1989). The majority implicitly invokes this idea in its citation of Adams Fruit Co. v. Barrett, 494 U.S. 638, 649, 110 S.Ct. 1384, 1390, 108 L.Ed.2d 585 (1990), for the proposition that a statute need not express the “truism that States may not preempt federal law” to be read as embracing that principle. Maj. Op. at 138. I agree. Again, the difficulty is that no such substantive norm appears pertinent to this problem. At least neither the plaintiffs nor the majority has identified one.
Possibly one might find such a norm in the statute’s legislative history. Plaintiffs and the majority have dredged that history and come up with very little. There is evidence of concern that unions should “be free from employer influence and control”, Maj. Op. at 140, but it is not clear why Board sorting out of representation disputes on its own initiative or employer impetus will tend to subject the unions to such control. There is similar evidence of legislative insistence on the Board’s neutral status as between carriers and unions, see Maj. Op. at 140, but again there is no evident link to our problem; it is unclear why giving unions and employees the sole initiative is necessary to that neutrality or even conducive to it.
Finally, one might find a controlling background norm in prior judicial authority (though it, in turn, would presumably draw on some independent source such as those discussed above). The cases that the majority invokes, Maj. Op. at 141, are plainly inapposite. Court decisions denying any Board obligation to investigate absent a request by employees or their representatives simply do not cut one way or the other in the present case. In fact, the only authoritative judicial pronouncement on the subject points to the Board’s freedom of maneuver on the subject: in the course of rejecting a carrier’s claim that the statute entitled it to be a party to § 2 Ninth proceedings, the Court said firmly that the scope of the carrier’s role (if any) was up to the Board. Railway Clerks, 380 U.S. at 666-68, 85 S.Ct. at 1200-02. “Whether and to what extent carriers will be permitted to present their views on craft or class questions is a matter that the Act leaves solely in the discretion of the Board.” Id. at 666-67, 85 S.Ct. at 1201.
Because my peek at the merits reveals no obvious error, I would uphold the Board.
The majority’s approach is radically different. At its core the decision rests primarily (though not by name) on the expres-sio unius canon: to express one thing is to exclude or to compel the opposite result for all alternatives not expressly stated. See Maj. Op. at 138-39. The majority reasons that if Congress mandated certification activity “upon request of either party” and did not explicitly grant the Board authority to investigate on its own initiative or carrier request, then it must have forbidden investigation under those circumstances.
The use of this canon is troubling on many counts. First, the canon is implausible as a matter of linguistic practice. A statute’s silence on a matter is typically as easily explained by the legislature’s failure' to address the point as by an inference that it intended a negative. “Come home right after school” is not a directive to dawdle after Little League.
Second, the canon is peculiarly inapposite in the context of judicial review of administrative agencies under Chevron. The canon is triggered by legislative silence, precisely the condition that under Chevron is understood to create a gap to be filled by *145the agency. See 467 U.S. at 843-44, 104 S.Ct. at 2782-83. See also Cheney R.R. v. ICC, 902 F.2d 66, 68-69 (D.C.Cir.1990), and cases cited therein.
Where the scope of judicial review is, as here, even more limited, use of expressio unius is exceptionally unjustifiable. Not surprisingly, in Switchmen’s/Leedom review in the past we have rejected such negative inferences. In Professional Cabin Crew Ass’n v. NMB, 872 F.2d 456 (D.C.Cir.1989), we considered the Railway Labor Act’s definition of “employee” — a "person in the service of a carrier (subject to its continuing authority to supervise and direct the manner of rendition of his service) who performs any work,” 45 U.S.C. § 151 Fifth — and refused to infer from its reference to current activity any actual requirement of current work; rather we upheld the Board’s decision that former strikers not presently working could vote. “[Njeither section purports to speak exclusively.” Id. at 460. We also refused to draw negative implications from the contrast in wording between the RLA and the NLRA. See id. at 461. The majority dismisses PCCA, apparently on the theory that there the Board could “pin” its interpretation on statutory language. See Maj. Op. at 138 n. 7. But in both this case and PCCA there is statutory language that states something affirmative; in both the issue is whether that affirmative implies a negative. Structurally, the issues are just the same.
Despite the recital of the many judicial expressions of exceptional deference, the majority’s actual practice is no “peek at the merits”. Instead, the court explores the “[cjoncerns underlying the [Railway Labor] Act,” Maj. Op. at 140 and “proposed statutory language Congress rejected,” id. at 140 (emphasis in original). Thus it moves beyond any patent error and is reduced to the sort of contentions that Railway Clerks forbids us to consider — “arguing in terms of policy and broad generalities as to what the Railway Labor Act should provide”. 380 U.S. at 671, 85 S.Ct. at 1203.
Indeed, the court’s review appears less deferential even than garden-variety Chevron analysis. This is of course apparent in its strong reliance on expressio unius. In addition, while the Court in Chevron expressly discounted prior agency positions on the issue, see 467 U.S. at 862, 104 S.Ct. at 2791 (rejecting the proposition that an agency interpretation “is not entitled to deference because it represents a sharp break with prior interpretations”), here the majority treats a Board observation from 1935 as helpful in locking in today’s Board. See Maj. Op. at 140.
Although the majority never explicitly repudiates the standards of Switchmen’s and Leedom, it appears to rely on a view that the issue here is “essentially jurisdictional”. Maj. Op. at 141. But the level of deference owed an agency’s statutory interpretation — and Switchmen’s, Leedom, and Railway Clerks all involved statutory interpretations — does not necessarily shift just because the issue may be characterized as jurisdictional. See Commodity Futures Trading Comm’n v. Schor, 478 U.S. 833, 844-45, 106 S.Ct. 3245, 3253-54, 92 L.Ed.2d 675 (1986) (addressing scope of the Commission’s jurisdiction over counterclaims); NLRB v. City Disposal Systems, Inc., 465 U.S. 822, 830 n. 7, 104 S.Ct. 1505, 1510 n. 7, 79 L.Ed.2d 839 (1984) (assuming jurisdictional character of “concerted activities” provision of § 7 of the NLRA and declining to apply less deferential standard to the NLRB’s interpretation). If our non-deference here is to depend on any such characterization, we must both distinguish the Court’s refusal to create such an exception in other contexts, and find some definition of - “jurisdictional”, a notoriously elusive word. See Mississippi Power v. Miss. ex rel. Moore, 487 U.S. 354, 381, 108 S.Ct. 2428, 2444, 101 L.Ed.2d 322 (1988) (Scalia, J., concurring) (“[T]here is no discernible line between an agency’s exceeding its [jurisdictional] authority and an agency’s exceeding authorized application of its authority. To exceed authorized application is to exceed authority.”). See also Louis L. Jaffe, Judicial Control of Administrative Action 356-57 (1965). The majority, understandably, attempts neither task.
Equally unhelpful is the majority’s suggestion that the issue has a “one-time only *146character”. Maj. Op. at 141. This is so only in the sense that the issue is one of statutory interpretation — and is equally true of all statutory interpretations. Restricting courts to “one-time only” issues, in that sense, places off limits only review for factual correctness and for application of law to facts. In the past, the Switch-men’s/Leedom line of cases has been understood to confine judicial intervention far more strictly.
I would uphold the Board.