The opinion
LUTTIG, Circuit Judge,
dissenting in part:
One cannot possibly read the record and the administrative law judge’s decision at issue in this case without being left with the impression that, whether through a series of innocent legal errors or, as Fieldcrest Cannon contends, through manifest pro-Union bias, the ALJ never gave Fieldcrest a chance to have its defense to the Union’s claims fairly considered. Indeed, the opinion upon which the Company’s substantial liability is now predicated appears as nothing more than a series of rationalizations for a predetermined finding of liability against the Company. For example, after conducting some 36 days of hearings, which consume more than 7800 pages of transcript, the ALJ ended up crediting the testimony of nearly every single one of the 100 witnesses who appeared on behalf of the Union and discrediting the testimony of nearly every single one of the 150 witnesses presented by the Company, finding approximately 130 violations of the National Labor Relations Act from among the 130-plus violations alleged by the Union. As the majority notes, the comparative numbers of witnesses credited by a factfinder cannot, in itself, serve as a measure of the correctness of a decision. At the same time, however, we cannot be naive to the real possibility in this instance that, contrary to the majority’s uncritical assumption, the ALJ’s wholesale crediting of the Union’s witnesses and corresponding discrediting of the Company’s witnesses was more than simply the happenstance result of a dispassionate consideration of the substantive testimony of the 250 witnesses in question. Regardless of whether it was merely a composite of innocent errors or outright bias against the Company that resulted in the decision we review herein, the ALJ’s conclusions, and consequently the judgment of the Board, are wholly insupportable in law. Accordingly, although I join in the judgment rejecting the Board’s order of a retroactive salary increase for the Company’s union employees, I dissent from the majority’s affirmance of the Board’s liability determinations.
The ALJ’s errors are various and sundry, but three broad categories of errors combine to confirm the palpable unfairness of the proceeding below. First, in direct contravention of this court’s precedent, the ALJ uniformly credited those employee-witnesses who testified on the Union’s behalf,.merely because they were employees of Fieldcrest Cannon; second, the ALJ found violations of the National Labor Relations Act in many incidents that are not, as a matter of law, even colorable violations of the Act; and third, the ALJ based many of his findings of violation on no evidence whatsoever, merely inferring certain violations of the Act from what he erroneously found to be management animus toward the Union.
The first and foremost error of the ALJ was that, in direct contravention of this court’s precedent, he weighted heavily the testimony of Fieldcrest employees who testified against the Company, merely because they were employed by the Company. In L.S. Ayres & Co. v. National Labor Relations Bd., 551 F.2d 586 (4th Cir.1977), we held that, “the fact that a witness is an employee at the time that he testifies adversely to his employer is at most only a factor bearing upon his credibility and one entitled to little weight unless it is established that he does not know that he is protected in testifying.” Id. at 588 (emphases added). In direct disregard of Ayres, the ALJ expressly and repeatedly held that the testimony of current employees was entitled *79 to “considerable weight” because, he reasoned, it was “unlikely that a current employee will testify falsely against his employer.” J.A. at 61 & n. 12. Representative are the following: that, “[t]he testimony ■ of [Angela] Coleman, a current employee, [was] entitled to considerable weight,” JA. at 61; that Eric Strickland’s “status as a company employee at the time of his testimony ma[de] it unlikely that he testified against his Employer’s interest,” J.A. at 61-62; that Sandra Greene “was a current employee at the time of her testimony, and for that reason, it is unlikely that her testimony was false,” J.A at 62; that Benny McIntyre’s “testimony, as a current employee, [was] entitled to considerable weight,” J.A. at 64; that “it[was] unlikely that the[ ] testimony [of employees Térros Brown and Johnny High] against the interests of their employer was false,” J.A. at 66; that Drenia Smith’s status as a current employee “enhance[d] her credibility,” J.A at 67; that, “[a]s a current employee, it [was] unlikely that [Brenda Lyles] testified falsely against her Employer’s interest,” J.A. at 68; and that Oreida Clarke’s “trustworthiness [was] augmented by her status as a current employee,” J.A. at 69. In fact, the ALJ expressly credited employee witnesses due solely or partially to their employment status some SO times (and, likely, implicitly in many others), clearly a sufficiently large number of instances such that it is impossible to affirm the ALJ’s conclusions as the majority does. See J.A. at 61 (Angela Coleman), 61 (Eric Strickland), 62 (Sandra Greene), 64 (Benny McIntyre), 66 (Johnny High & Térros Brown), 67 (Drenia Smith), 68 (Brenda Lyles), 69 (Oreida Clarke), 76 (Sandra Greene), 77 (Tony Bumgarner), 77 (Sharon Davis), 78 (Joanne Diggs & Sylvia Crawford), 78 (Drenia Smith, Pat Boger, & Sheila Deal), 78 (Vicki Fink), 79 (Angela Coleman), 83 (Terry Smothers), 83 (Sherry Anthony), 84 (Sylvia Crawford), 87 (Sharon Davis), 88 (Benny McIntyre), 88 (Peggy Jordan), 89 (Euretha Lee, Sherry Anthony, & Paula Brice), 89 (Sharon Davis), 90 (Wade Story), 92 (Sharon Davis), 93 (Eric Strickland), 94 (Vickie Fink), 94 (Kemberly Watts), 95 (Brenda Harrell & Cynthia Hanes), 95 (Patricia Boone), 102 (Jenny Vires & Daniel Vires), 103 (Oscar Clark), 103 (Patsy Turner & Barbara Cook), 103 (Patsy Turner), 104 (George Cochran), 104 (Wilbert Williams & John Rossner), 105 (Cynthia Hanes), 105 (Sharon Davis), 105 (Drenia Smith), 105 (Clafter Jackson), 115 (Diana Hamilton), 143 (Norma Chapman).
The majority acknowledges that the ALJ erred by crediting witnesses based solely on their status as employees, but concludes that there are “other indicia of credibility” which render the error harmless. There are no such “other indicia of credibility” evident in the record before us. Tellingly, even the majority is able to identify, from the scores of incidents at issue, only a meager seven or eight examples which supposedly confirm that there were alternative reasons for presuming the truthfulness of the pro-Union witnesses’ testimony and presumptively discrediting the Company witnesses’ testimony. And, in fact, not one of even these noticeably few examples supports the majority’s conclusion.
The majority cites two examples of the “absence of direct denials by company witnesses” in support of its claim that there were other indicia of reliability for the ALJ’s impermissible crediting of pro-Union witnesses. Neither example supports its conclusion. In the first of these, the incident between James Allman and Graciela Whitley, ante at 70, although the ALJ did note that supervisor “Allman did not testify,” the ALJ did not even rely on that fact in crediting Whitley, nor did he mention that fact in his “factual and legal discussion.” J.A at 66. In any event, this was not an incident in which the ALJ explicitly credited a witness based on his employee status. In the second incident, Oreida Clarke alleged that the Company posted a Spanish notice threatening workers who signed union cards, ante at 71, but, as the ALJ himself noted, “all of the Company’s witnesses denied posting any Spanish notice, much less one of the nature described by Clarke.” J.A at 69. Of course, contrary to the majority’s conclusion, the fact that not all of the Company’s supervisors testified does not render Clarke’s testimony unrebutted. Compare ante at 71 (“[T]he company’s witnesses who denied posting such a notice did not ‘include all supervisors who *80 may have done so.’” (quoting J.A. at 69) (emphasis added)).
The majority cites but a single example of the “comparative vagueness” of the Union’s witnesses vis-a-vis the Company’s witnesses, ante at 71, and, likewise, this example does not support the majority’s assertion that there were alternative reasons for crediting the witnesses improperly credited by the ALJ. Whether or not supervisor Howard Hall “could hardly remember” the details of the conversation between himself and Brenda Lyles, ante at 71, Hall, as even the ALJ conceded, “denied saying anything about plant closure and added that supervisors had been warned against making any such statements.” J.A at 68.
Similarly, it is impossible to discern in the majority’s lone example of “internal inconsistency” of Company witness testimony any inconsistency at all, much less one that would justify the crediting of Union witness testimony over Company witness testimony. In that example, supervisor Perry Harkey simply denied that he spent additional time on the floor in order to fulfill his added responsibilities to “be available” to answer employee questions during the election. There is nothing even arguably inconsistent in this testimony; indeed, the ALJ’s finding (without explanation) that Harkey’s testimony was inconsistent because he could not have fulfilled his added responsibility to “be available” for questions without increasing his time on the floor, J.A. at 77, is nothing short of silly.
The “substantial corroborating testimony” which the majority would have one believe supports the ALJ’s credibility determinations, ante at 72, is not corroborating testimony at all. For example, when a supervisor and an employee agreed that a conversation took place or that an incident occurred, but disagreed as to what was said or what occurred, the ALJ, consistent with the manner in which he conducted the balance of the hearing, found that the agreement partially corroborated the employee’s testimony. See, e.g., J.A. at 64, 103.
“To a lesser extent,” the majority relies upon witness demeanor evidence assertedly relied upon by the ALJ in making credibility determinations, and it cites four examples of the ALJ’s supposed reliance on such demean- or evidence. Ante at 71-72. Even the majority must concede, however, that, in two of these instances, the ALJ himself did not even cite demeanor as a reason for his crediting of the Union’s witnesses. Of course, even when the ALJ did rely upon demeanor, it cannot possibly be known whether the ALJ would have credited the pro-Union witnesses’ testimony on the strength of demeanor alone, had the ALJ not impermissibly accorded those witnesses’ testimony considerable weight at the outset.
Finally, the majority holds that the “atmosphere of reprisal and recrimination against union supporters” justified the ALJ’s presumption of truthfulness with regard to Union employees, ante at 72. Obviously, even if such an atmosphere existed, this alone would be no ground for crediting, as a matter of law, one party’s witnesses over another’s. To uphold the ALJ’s findings on this ground is to relieve the Union altogether of its burden to prove violations of the Act.
In sum, contrary to the majority’s assertion, there is simply no basis in the record for sustaining the ALJ’s findings, in the face of his conceded error in presuming the truthfulness of all pro-Union employee witnesses and the contrary of all Company witnesses.
Wholly apart from the ALJ’s erroneous crediting of the Union’s employee witnesses over the Company’s witnesses, the ALJ found egregious violations of the NLRA where, as a matter of law, there were not even colorable violations. Illustrative is the first “violation” described by the ALJ. There, supervisor Diane Hartis handed out pro-company t-shirts to employees who asked for them. J.A. at 58. Employee Sylvia Crawford asked Hartis for a shirt, and Hartis said “[f]ine” and gave her one. Id. When Hartis learned that Crawford intended to deface the shirt, she told Crawford, “[i]f you search in your heart, will you please just give the shirt back.” Id. Hartis also said, “I don’t want to be made a fool of.” Id. Crawford admitted that some of the employees were defacing the shirts and told Hartis that she would give the shirt back to Hartis if she *81 changed the slogan. Hartis replied, “Please give it back,” and Crawford later returned the shirt. Id. On the basis of these exchanges alone, the ALJ concluded that Har-tis “coercively interrogated Crawford about her union sentiments.” Id. Of course, Har-tis’ request that Crawford return the shirt rather than deface it was not even arguably coercion or interrogation in violation of section 8(a)(1).
The ALJ even found that the Company violated the Act when it remedied or promised to remedy employee grievances, J.A. at 95-98, and when it held grievance meetings for employees. The utter ridiculousness of the ALJ’s findings is perhaps best summed up in the following passage, in which the ALJ described M.D. Ford’s “violation” of the NLRA:
Ford’s own testimony establishes that his meetings with employees after the advent of the union campaign increased markedly over the number of asserted meetings pri- or to that time. There is no question, as Ford admitted, that he asked employees about their concerns and problems.
J.A. at 96 (emphasis added). And, unfortunately, similar examples of the ALJ’s errors abound throughout the record.
Finally, as if to add insult to injury, the ALJ inferred countless violations from the Company’s “extraordinary animus” and “widespread violations,” both of which he found to exist only because he improperly credited all of the Union’s witnesses and otherwise misread section 8(a). Most notably, the ALJ relied on these unwarranted presumptions to establish prima facie cases that protected conduct was a motivating factor in the Company’s discipline or discharge of pro-union employees. J.A. at 110, 112, 114, 116, 117, 119. In fact, the ALJ sometimes relied exclusively on anti-union animus to establish such prima facie cases. J.A. at 110,112.
Thus, at bottom, the ALJ’s errors in this case are so egregious and pervasive that it cannot possibly be said that the record offers substantial evidence for his findings. In holding to the contrary, the majority has, for purposes of this case at least, redefined our role in the review of such ALJ findings into an empty act of proforma ratification.
I respectfully dissent.