dissenting in part:
I agree with the conclusions of the panel opinion, with one important exception. I do not think that the Board seriously or adequately confronted the fact that when Marion Hospital withdrew recognition on October 28, 1998, it had in hand petitions which, after we exclude signatures “tainted” by the September 11 refusal to bargain and signatures of persons not in the unit, amounted to 44% of the unit employees. For me this raises a vexing issue under Allentown Mack Sales & Serv. v. NLRB, 522 U.S. 359, 118 S.Ct. 818, 139 L.Ed.2d 797 (1998), which perhaps the Board could answer with reasoned deci-sionmaking, but has not. And because the Board’s justification for the affirmative bargaining order evidently depended in part on finding the October 28 withdrawal of recognition unlawful, I would remand the case for it to consider the matter.
On October 20, 1998 Marion received petitions purporting to carry the signatures of 82 of the 157 bargaining unit employees. The Board excluded seven on the ground that they were not members of the unit. And it excluded another six on the ground that their signatures were dated after September 11 and thus presumptively tainted by the refusal to bargain of that date. See Lee Lumber & Bldg. Mate*1189rial Corp. v. NLRB, 117 F.3d 1454 (D.C.Cir.1997). After the Board’s deductions, the number of bargaining unit employees expressly disavowing the union was 69, or 44% of the total.
As the panel opinion notes, Allentown Mack authoritatively construed the applicable Board standard: “good-faith reasonable doubt” or “genuine, reasonable uncertainty,” based on objective considerations and “supported by evidence external to the employer’s own (subjective) impressions.” Allentown Mack, 522 U.S. at 367 & n. 2, 118 S.Ct. at 824 & n. 2. Explaining the type of uncertainty or doubt required, the Court said:
“Doubt” is precisely that sort of “disbelief’ (failure to believe) which consists of an uncertainty rather than a belief in the opposite. If the subject at issue were the existence of God, for example, “doubt” would be the disbelief of the agnostic, not of the atheist. A doubt is an uncertain, tentative, or provisional disbelief.
Id. at 367, 118 S.Ct. at 823. Although Allentown Mack involved a combination of several express disavowals of the union and other evidence (including statements from a union shop steward), the Court said that some number of express disavowals alone sufficed to create a good-faith reasonable uncertainty:
The Board did not specify how many express disavowals would have been enough to establish reasonable doubt, but the number must be less than 16 (half of the bargaining unit) since that would establish reasonable certainty.
Id. at 368-69, 118 S.Ct. at 824 (emphasis in original). The Court seems to be saying at a minimum that disavowals from 15 in the bargaining unit of 32 (i.e., the largest possible number under 50%) would have been enough to establish the requisite “uncertainty.” 15/32 would, of course, have been 46.8%. On its face, then, one would expect the Board to undertake some subtle explanation of why 46.8% sufficed to create uncertainty in Allentown Mack, but 44% does not do so here. Instead, it simply declared that the 44% were “insufficient to establish a good-faith reasonable uncertainty.” See 335 N.L.R.B. No. 80, 2001 WL 1126579, at *5.
Although the Allentown Mack Court nowhere qualified the language just quoted, it may have been colored by context. As noted above, there was softer (i.e., less quantitative) evidence supporting “uncertainty,” including the statement of a member of the union’s bargaining committee and shop steward that it was his feeling that the employees did not want a union, and that “ ‘with a new company [transfer of the business was in fact imminent], if a vote was taken, the Union would lose.’ ” Id. at 362, 118 S.Ct. at 821. Further, it may be that statements of individual workers adding up to less than a majority generate more reasonable doubt than less-than-a-majority names on a petition, as the latter suggests an organized effort, and the lack of a majority may suggest the circulators’ inability to attract additional signatures. But we have no discussion of any of this from the Board — no indication that the Board applied its “fund of knowledge and expertise” about labor relations to what appears to be a close question. NLRB v. Gissel Packing Co., 395 U.S. 575, 612-13 n. 32, 89 S.Ct. 1918, 1939-40 n. 32, 23 L.Ed.2d 547 (1969).
Although the Board is giving up even nominal allegiance to the Allentown Mack test in the context of withdrawal of recognition, see Levitz, 333 N.L.R.B. No. 105, 2001 WL 314139, the Allentoivn Mack test has significance for the future. The Board explained in Levitz that it was retaining the test as the basis on which employers can obtain a Board-certified election. Id. *1190at *2 (“We have also decided to allow employers to obtain RM elections by demonstrating good-faith uncertainty (rather than disbelief) as to unions’ continuing majority status.”). Time spent by the Board articulating the test’s meaning will by no means be wasted, quite apart from fairness to the litigants here.
Does the Board’s failure of analysis require a remand? The court appears to take the position that the October 28 withdrawal of recognition was completely tainted regardless of the number of signatures obtained before September 11, for want of a “cure” of the September 11 refusal to bargain. Maj. Op. at 1181, 1187. On this view the subtleties of Allentown Mack are irrelevant. But the Board disclaimed any idea of total taint; overruling the ALJ, it held that the taint extended only to signatures obtained after the refusal to bargain. See 335 N.L.R.B. No. 80, 2001 WL 1126579, at *4. (“Therefore, under Lee Lumber, only employee signatures dated after September 11 are presumptively tainted by the unlawful refusal to bargain. No such presumption of unlawful taint attaches to those employee signatures dated before September 11.”) (emphasis in original).
Alternatively, we might disregard problems with the Board’s view of the October 28 withdrawal of recognition on the theory that it would have issued the bargaining order anyway. Perhaps so. But the Board suggested nothing of the sort, and there is no special reason to think so. Indeed, the bulk of its opinion is devoted to October 28, and its justification for the bargaining order speaks consistently of the employer’s “withdrawal of recognition” (the Board’s precise finding for the October 28 conduct), as opposed to the “refusal to bargain” (its precise label for September 11). Our oft-repeated ruling that a bargaining order is an “extreme remedy that must be justified by a reasoned analysis,” see, e.g., Vincent Indus. Plastics v. NLRB, 209 F.3d 727, 738 (D.C.Cir.2000), would seem to call for explicit consideration by the Board when one of two supports for an order (seemingly the larger one) falls for want of reasoned decision-making.