In this case we review a National Labor Relations Board finding that Concord Furniture Industries violated the National Labor Relations Act §§ 8(a)(1) and 8(a)(3) in the course of a unionization effort by Teamsters Local 82. This effort led to an election among Concord’s warehouse employees on January 12, 1978, at which two former employees, Douglas Goranson and John Williams, voted under challenge. Following the election the Union filed charges with the Board. The Board found that statements made to employees by Concord’s president, Anthony Ferreira, its secretary-treasurer, Henry Greenberg, and by a “warehouse supervisor,” Charlie Lupis, were coercive and therefore violated § 8(a)(1).1 It also found that Concord’s motivation was union animus when it selected Goranson and Williams for discharge, thus violating both § 8(a)(1) and § 8(aX3).2 The Board then counted the challenged ballots, changing the election results from 5-4 *428against, to 6-5 in favor of the Union, and certified the Union as bargaining representative. Concord has refused to bargain with the Union as a means of securing review of the Board’s orders.

Since Concord does not challenge the Ferreira/Greenberg § 8(a)(1) violations, this case presents only two significant issues. 1) Is there substantial evidence that Charlie Lupis was a “supervisor” and that his statements can therefore be attributed to Concord? 2) Is there substantial evidence that an anti-union motive was a necessary condition for the discharge of Goranson and Williams? We believe the answer to both questions is yes.

“Substantial evidence” here, of course, means “such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Universal Camera Corp. v. NLRB, 340 U.S. 474, 477, 71 S.Ct. 456, 459, 95 L.Ed. 456 (1951). We must take contradictory evidence in the record into account, id. at 487-88, 71 S.Ct. at 463-464, but “the possibility of drawing two inconsistent conclusions from the evidence does not prevent an administrative agency’s finding from being supported by substantial evidence.” Consolo v. Federal Maritime Commission, 383 U.S. 607, 620, 86 S.Ct. 1018, 1026, 16 L.Ed.2d 131 (1966). “Even if reasonable minds could also go the other way, we must uphold the Board if its ultimate finding is supported by substantial evidence on the record as a whole.” NLRB v. J.K. Electronics, 592 F.2d 5, 7 (1st Cir. 1979).

These admonitions to give weight to the Board’s expertise and judgment are particularly apt when we review its determination that Lupis was a “supervisor.” A supervisor is defined in the Act broadly: he is one who 1) has authority 2) to use independent judgment 3) in performing a supervisory function 4) in the interest of management. See NLRB v. Security Guard Service, Inc., 384 F.2d 143 (5th Cir. 1967). A “supervisory function” includes hiring, transferring, suspending, laying off, recalling, promoting, discharging, assigning, rewarding, disciplining, responsibly deciding or adjusting the grievances of other employees (or effectively recommending such action). 29 U.S.C. § 152(H).3 The definition of supervisory function is broad and general; that breadth is limited by the requirement that a supervisor’s judgments and directions must be significant. Stop & Shop Companies, Inc. v. NLRB, 548 F.2d 17, 19 (1977); NLRB v. Swift & Co., 292 F.2d 561, 563 n.2 (1st Cir. 1961). Significance, however, is not always easy to judge, given the “infinite and subtle” “gradations of authority ‘responsibly to direct’ the work of others from that of . . . top executive to ‘straw boss.’ ” Id. at 563. At the same time, that subtlety, along with the likely firm-specific, and fact-specific nature of a judgment, warrant the exercise of “a large measure of informed discretion” by the Board. NLRB v. Swift & Co., supra; NLRB v. Security Guard Service, 384 F.2d at 146 (“the tensile strength of [the Board’s] . . . findings increases as they are applied to the minutiae of management-employee relations.”). See NLRB v. Erie Resistor Corp., 373 U.S. 221, 236, 83 S.Ct. 1139, 1149, 10 L.Ed.2d 308 (1963). Moreover, we are particularly reluctant to overturn the Board’s exercise of discretion here, for it depends in large part upon determinations of credibility. Universal Camera Corp. v. NLRB, 340 U.S. at 496-97, 71 S.Ct. at 468-469.

Indeed, if one accepts Lupis’ own testimony at face value, there is no question but that he was a supervisor. He came to Concord Furniture from American Homestead, Inc., when Concord bought that company in July 1977. He had been the warehouse supervisor at American Homestead *429with major responsibilities. According to his testimony, he was introduced to Concord Furniture employees “as the warehouse manager and as their boss;” he was told by Concord’s executives that he was “to run the warehouse and perform [his] .. . duties as a warehouse manager;” and he hired, fired, disciplined, rewarded, and managed other warehouse workers — drivers and furniture finishers. The problem is that Lupis’ testimony is contradicted by Ferreira and Greenberg. And, much in the record suggests that Lupis had an exaggerated idea of his own position in the firm. When he sought to fire one employee, for example, Ferreira and Greenberg would not back him up — a fact that apparently made him so angry that he quit. In other instances, there is conflicting testimony about whether he or Ferreira actually hired particular workers. According to Ferreira, the warehouse ran itself; Lupis was the “lead” man; but this task involved only routine duties.

Nonetheless, according to credible testimony of Lupis and others, at a minimum Lupis gave other employees permission to leave early and to take time off, he organized overtime, and he assigned work to other employees, making changes in delivery routes. The record contains testimony by several witnesses to the effect that Ferreira was in the warehouse actively supervising much more after Lupis quit than before he left. It suggests that while Lupis was there Ferreira was often absent from the warehouse — suggesting that he relied upon Lupis, if anyone, to supervise. At the same time, the claim that the warehouse staff— truck drivers, finishers and helpers — ran itself is not very credible. See NLRB v. Supreme Dyeing & Finishing Corp., 340 F.2d 493 (1st Cir. 1965). All this pro-Lupis testimony, even when weighed against the anti-Lupis testimony, is sufficient support for the Board’s finding.

In fact, the conflicting testimony in the record indicates that this is a classical case for Board discretion. The issue is not even whether one side or the other is lying; the question really is one of comparative exaggeration (or minimization) in describing Lupis’ position. While Lupis voted in the union election, we do not believe that fact determinative given the possibly “borderline” nature of his duties and the fact that the Company and the Union may each have thought he would vote for them— hence no one challenged his vote. Since reasonable minds could come to different conclusions on the basis of the record, the Board’s determination that Lupis is a “supervisor” is lawful. If Lupis is a supervisor, his statements that Williams might be discharged because of his union activity and that Goranson and Williams would be reinstated “as soon as this union thing goes through” are in context sufficiently coercive to warrant findings of § 8(a)(1) violations.4

We also believe the Board’s finding that the Company violated § 8(a)(3) in discharging Goranson and Williams is supported by substantial evidence. The issue is not whether the Company had a right to discharge two employees. In our view, it did.5 The Company’s financial records, showing a decline in sales after its American Homestead acquisition and its stipulation with the Board about the financial need for discharges establish that right. *430The issue is whether Goranson and Williams were selected, instead of other warehouse employees, because of their known union support. There is no doubt but that the Company knew of that support. But it is uncertain whether this knowledge was in fact the basis for selection or whether instead the selection was made on the basis of seniority as the Company asserts. Goranson was plainly low man on the totem pole with least seniority; but as to Williams, his seniority depends on the method of calculation used. Should his seniority have been calculated from the time he began to work at Concord several months earlier or should his worktime at American Homestead be counted, too? If the latter, should he be credited with time worked prior to certain breaks in service or not? These questions had to be answered and Williams’ dates of work examined before Concord could know whether Williams was also low man, or higher up by one. The record reveals an absence of clear seniority policy on these points or at least such confusion that the fact that Concord’s executives never sought to calculate Williams’ seniority or to answer the questions strongly suggests they were not truly interested in seniority as to Williams, which in turn suggests the same as to Goranson. This inference is bolstered by the fact that Charlie Lupis stated on two occasions that Goranson and Williams had been discharged because of their union activity.

Whether this court would have found a violation on this evidence is not the point. The Board did so. Having read the record, we find the Board’s reasoning and its findings sufficiently supported and its orders are therefore

Enforced.