The Board seeks enforcement of a cease and desist order against Bogart, and the union1 seeks review of the order to the extent it did not include all the relief which the union desired. The Board2 adopted the findings of the Trial Examiner ("TX") concerning Bogart's violations of Sec. 8(a)(1) by employee interrogations and threat of plant closure and of Sec. 8(a)(5) and (1) by refusing to bargain with the union, which was certified as representative of nonsupervisory production and maintenance employees at Bogart's Ft. Worth plant. We find substantial support in the record for these conclusions of the Board. A majority of the Board,3 however, rejected the findings of the TX of Sec. 8(a)(3) and (1) violations arising from the discharge of ten employees, and the union seeks review of this aspect of the order. The company opposes both enforcement of the order and any broadening of the order.

We withhold enforcement of the order as issued and remand to the Board for modification of the order to encompass the Sec. 8(a)(3) and (1) violations the TX found in the discharge of two employees.

I. The 8(a)(1) and 8(a)(5) violations.

The Board adopted the findings and conclusions of the TX that Bogart had violated Sec. 8(a)(1) by coercive interrogation of Vola Barrington on March 14 and 16, 1970, by a threat on May 13, 1970 to Joyce Elkins that the plant would close if the union succeeded, and by a supervisor's refusal on May 1, 1970 to assist Elkins because of the union's activity and of her display of a union badge. There is substantial evidence in the record as a whole to support these findings.

Bogart's challenge to the Sec. 8(a)(5) refusal to bargain findings is predicated solely on the argument that the bargaining unit was inappropriate because it did not include the nonsupervisory production and maintenance employees of Bogart's other two plants in nearby towns. The Board's discretion in making unit determinations is very broad and our standard of review narrow. NLRB v. Wood Mfg. Co., 466 F.2d 201, 202 (C.A.5, 1972). The fact that a multiplant unit might also be appropriate is not a basis for overturning the Board's determination. NLRB v. Fidelity Maintenance & Construction Co., 424 F.2d 707, 709 (C.A.5, 1970). The Board was required to choose only an appropriate unit, not the most appropriate unit. State Farm Mutual Ins. Co. v. NLRB, 411 F.2d 356 (C.A.7, 1969); NLRB v. Wood Mfg. Co., supra, at 202 of 466 F. 2d. The record fully supports the appropriateness of the bargaining unit chosen by the Board here.

Bogart manufactures sportswear. The ten alleged discriminatees whose discharges we review here were employed at Bogart's Ft. Worth plant as nonsupervisory production employees. In the late winter and spring of 1970 Bogart was engaged in a protracted defensive campaign against the union's efforts to organize the Ft. Worth plant. The organizational campaign had opened the year before with the union conducting its first meeting among Ft. Worth employees on May 23, 1969.4 The union's tactics were the usual-professional organizers were assisted by employees who served in an organizing committee and talked up the union among other employees and solicited authorization cards.

Bogart's responses to the union's organizational campaign were swift, direct and marked by activities whose unlawfulness has already been adjudicated by the Board and this court. Some of these unlawful acts had as their immediate targets some of the alleged discriminatees whose subsequent discharges we review here. In the spring and early summer of 1969 Bogart announced and granted improved vacation and insurance benefits to discourage the employees' union activity. And, through its Production Vice President Grodin and a supervisor, Bogart coercively interrogated and then discriminatorily discharged a Ft. Worth employee. The Board found that these acts were unfair labor practices, Bogart Sportswear Mfg. Co., 186 N.L.R.B. No. 90 (1970) and we enforced the Board's cease and desist order. NLRB v. Bogart Sportswear Mfg. Co., 461 F.2d 847 (C.A.5, 1972).

On February 12, 1970, the union filed a representation petition, and on March 5, 1970 the Board held a hearing to determine whether an election should be conducted in the proposed unit. On March 14 and 16 a supervisor unlawfully interrogated alleged discriminatee Barrington concerning union activities. See Part I supra. In April the Board ordered the election for May 19. On May 13 alleged discriminatee Elkins was threatened with closing of the plant if the union succeeded. See Part I supra. Seven of the discharges in question occurred between March 20 and 27 and three on May 14. The union won the May 19 election.

With the exception of two or three workers, it is not disputed that the company had pre-discharge knowledge of union adherence of the ten alleged discriminatees. Because of our disposition of the case we need not adjudicate the dispute as to these two or three.

A business-justified general reduction in force, unrelated to union activity, is not condemned merely because it coincides with anti-union sentiment. NLRB v. Materials Transportation Co., 412 F.2d 1074, 1078 (C.A.5, 1969). Nor does the mere existence of a just cause for discharge negate the possibility of an 8(a)(3) violation. See NLRB v. Whitfield Pickle Co., 374 F.2d 576, 582 (C.A.5, 1967), and Frosty Morn Meats, Inc. v. NLRB, 296 F.2d 617, 621 (C.A.5, 1961). Discrimination may exist not only in discharge of a single worker but also in selection of employees for a business-justified reduction. NLRB v. Materials Transportation Co., supra, at 1079 of 412 F.2d. If an employee is inefficient his engagement in union activities does not alone destroy just cause for discharge. NLRB v. Birmingham Publishing Co., 262 F.2d 2, 9 (C.A.5, 1959).

Bogart's version of the ten discharges is that they were incident to a reduction in force necessitated by economic reasons. There is substantial evidence of business justification for a reduction in force, and the TX assumed arguendo that such justification existed. The Board made no specific finding on the justification point, although it referred to adverse conditions faced by the company. The union, although it opposes the Board on the discharge issue, does not squarely contest the existence of business reasons for reducing the number of production employees. It takes some minor tangential shots from the flanks, but its basic objection is to the manner in which selection was made of the employees to be terminated. We turn then to that issue, which concerns the validity of a facially neutral computer study that in a mechanical sense was the basis on which those to be discharged were chosen.

Nine of the ten alleged discriminattees were piece rate workers. They worked at rates designed to produce (at the level of efficiency theoretically expected by Bogart) $1.80 to $1.90 per hour, based on time studies of various operations and differing materials and styles and types of garments. Bogart was, of course, required to pay each employee the minimum wage of $1.60 per hour irrespective of her piece rate production. The extent to which a piece rate employee's production at piece rates produced earnings of less than $1.60 per hour was referred to as "makeup pay" for "production not made."

Company Vice President Seals testified that management decided the most equitable way, and the most profitable for the company, to make the necessary reduction in force, was to terminate the employees who are least productive at the piece rate and therefore were requiring of the company the most "makeup pay." Bogart conducted a computerized study of its piece rate employees for a test period January 1-March 14, 1970. Its computer was "asked" to print out in ascending order, from lowest to highest earner, the names and average hourly piece rate earnings of each employee by production department for the test period. Seals testified that management then simply drew a line on each department's roster to reflect the number of employees still needed, retained all employees above the line and discharged all employees, irrespective of their support or nonsupport of the union, who had earned less than the last person to be retained.5 The discharges were not made all at once but over a period of time as production was "winding down."

Despite the facial neutrality of the computer study, the TX found that Bogart had discriminatorily discharged the ten employees in question. He made no specific findings as to Seals' credibility but implicitly assumed the mechanical correctness of the computer data while, at the same time, finding that the study was, in effect, tainted in a manner which we will explain below. In reaching his ultimate conclusion of discrimination he considered: Bogart's history of anti-union animus; the evidence of specific anti-union activities around the time of the discharges, especially the 8(a)(1) statements; the timing of the terminations; the fact that there had been no layoffs at other times of reduced production; and his conclusion that during the test period (or prior to and extending into the test period) and, after learning of their union adherence, the company had given the nine piece workers assignments less productive at peice rates or had impeded their job performance.

We have already summarized the history of anti-union animus and have reviewed the 8(a)(1) statements in Part I. The TX particularly noted a "spoken threat to close the plant rather than deal with the union." Other specific matters referred to by the TX included the following. Several of the affected employees had been employed by the company for many years without prior difficulty-one for nearly nine years, others two to six years. Despite the fact that terminations were alleged made by reason of economic conditions, no indications were given concerning possible recall,6 and no advance warnings of termination were given. Although operations of the plant were coordinated with operations of the two other nearby Bogart plants,7 there had been no substantial reduction in force at the other plants.

As to timing, the organizational campaign had been in progress for approximately a year. The representation petition had been filed and a hearing conducted March 5. As we have pointed out, the discharges in question occurred between March 20 and 27 and on May 14, and the election was held May 19.

Despite the existence of these background factors the validity of the computer data becomes central to the case for nine of the ten discharged employees. The business necessity for reducing the labor force is assumed by the TX, assumed (if not found) by the Board, not questioned by the dissenter, and not seriously contested by the union, and there is more than the required substantial evidence to support it. No one questions that the computer data correctly reflects the actual performance of all employees for the test period. Similarly no one claims that the test period was invidiously selected, or that there was anti-union animus in choosing the number of persons to be discharged from any one or more departments, or that an excessive number of employees were discharged, or that the computer study did not, in a mechanical sense, accurately select those to be terminated by the "above the line"-"below the line" method described by Seals. And there is no substantial evidence which would support any such contentions. Seals' credibility is thus material only to the extent of his conclusory statements that there was no anti-union bias in the selection process. To this extent the TX implicitly found him not credible.

We turn then to the inquiry into whether by reason of assignments to less productive work, or impediments to their job performance, the productivity figures for the nine piece workers had been forced down or held down. The majority of the Board rejected the TX's ultimate conclusion that the non-piece worker, Barrington, and all nine of the piece rate workers were discriminatorily discharged, and found that with respect to all ten the General Counsel had failed to carry his burden of proof.8 Member Jenkins dissented as to Barrington and five piece workers. We have considered the discharges in detail, one by one and overall, and the differing positions of TX, Board majority and dissenter. We think that the Board's conclusions are supported by substantial evidence as to all persons except piece workers Phelps and Evans.

As to the issue of discriminatory discharges, considered overall, the Board pointed to the following, inter alia. Twenty-one or twenty-two employees had been terminated, but at the unfair practice hearing the General Counsel claimed discriminatory discharge of only twelve and dropped two of these during the hearing. There was no pattern among the departments showing that the last employee terminated in the respective department was a union adherent or that the first employee retained was nonunion. We know that a total of approximately eighty union adherents were identified in a series of telegrams sent by the union to the company during the campaign. Using only the names of those identified in this manner, and looking at the various production departments, we find the following. In Department 1, seven of thirty-eight employees were terminated. Four of those terminated had been identified by the telegrams. The General Counsel sought to show that three of these were discriminatees. In Department 3 there were eight identified unionists among thirty-nine employees. Three persons were terminated, only one of these was alleged to be a discriminatee. In Department 4 there were twenty-four employees, fourteen identified unionists, three persons terminated, two alleged discriminatees. Department 5 had twelve employees, four identified unionists, one terminated and that one allegedly a discriminatee. In Department 17 there were thirty-six employees, five identified unionists, one person terminated, no discriminatees. In Department 19 there were twenty-seven employees, eight identified unionists, six persons terminated, but the General Counsel only claimed that three were discriminatees. Department 20 had twenty-one employees, two identified unionists, one person terminated. The General Counsel did not allege that she was a discriminatee.

The claims of several persons can be disposed of initially. Barrington, the non-piecework employee, was an assistant floorlady paid on a salary basis. She did not testify to having been given discriminatory work assignments. Rather, in effect, she claimed that she would have been transferred and not terminated but for her known union sympathy. The Board rejected the TX's conclusion that her discharge was discriminatory and found that she was terminated because the general reduction in her department made an assistant floorlady unnecessary, and that transfer to other departments was not practical when more experienced employees were already being terminated in those departments. These conclusions are supported by substantial evidence.

Two piecework employees, Kincannon and Washington, did not testify to discriminatory work assignments prior to the test period and extending into the test period, or made during the test period. As to them, the Board rejected the TX's findings of discriminatory discharge, the dissenter did not disagree, and we think the Board's finding has adequate evidentiary support.9

The majority found that Turpin's work assignment claimed to have been unfair was not made until the last two weeks of the test period and thus was not the controlling factor in her test results.10 Member Jenkins did not disagree.

We turn then to the other piece rate workers. As to each of them the TX had considered in detail the computer test data and had concluded that, although not claimed to be mathematically incorrect it was in fact "loaded", indeed was part of a pattern of discrimination because either prior to or during the test period the piece rate workers had been assigned to tasks compensated at lower piece rates for the time required, tasks requiring more time for the piece rate compensation paid, more difficult work, work requiring additional operations, work for which the employee was not experienced, and the like. The TX specifically credited the testimony of six of the nine pieceworkers, each describing how after the company knew of her union adherence she had been given assignments in which she was less productive because of one or more of the above factors. The TX rejected testimony of Bogart witnesses to the contrary.

The Board was not bound by the TX's credibility determinations, although where the TX has made specific credibility findings we must examine the evidence with greater care. Compare NLRB v. Buddies Supermarkets, Inc., 481 F.2d 714 (C.A.5, 1973), with NLRB v. Dal-Tex Optical Co., 325 F.2d 78 (C.A.5, 1963) and Ward v. NLRB, 462 F.2d 8, 12, 13 (C.A.5, 1972). See also Universal Camera Corp. v. NLRB, 340 U.S. 474, 496, 71 S. Ct. 456, 468, 95 L. Ed. 456, 472 (1951); NLRB v. Walton Mfg. Co., 369 U.S. 404, 408, 82 S. Ct. 853, 855, 7 L. Ed. 2d 829, 832 (1962), and Russell-Newman Mfg. Co. v. NLRB, 407 F.2d 247, 249 (C.A.5, 1969). The Board, however, disagreed with the TX's credibility findings in only small degree. Rather it drew upon other evidence that tended to destroy the effect of the findings concerning assignments to less productive work and interference with the alleged discriminatees' productivity. As to pieceworker Elkins the Board noted that the TX had accepted her oral testimony that prior to transfers in work assignments she had made or exceeded her "quotas" of work [i. e., sufficient work, calculated at piece rates, to produce $1.60 per hour]. But, the Board noted, payroll records for Elkins showed that she had not, before the transfers, made her quota all or a substantial part of the time. As to pieceworker Bruce, the TX found that she had no difficulty making production levels and was moved from job to job. But the Board noted that Bruce's testimony was that she had made her quota "a few times," and that there was uncontroverted evidence that employees always had been moved around because of production requirements. Elkins' payroll records and Bruce's own testimony were evidence that the test results for Elkins and Bruce were consistent with their prior performance rather than the consequence of any adverse changes in work assignments. The dissenter argued that selective discharge of workers allegedly for poor performance when their work has been poor all along was a classic example of discrimination. But the Board was entitled to infer that the matter of the company's discharging its poorer producers had never been a viable question until the necessity for reduction in force arose.

As to employee Gamez, the central point was the timing of her transfer to an assignment which the TX found she was not able to perform as efficiently. The majority considered determinative that the switch was in March or April, 1969, before she joined the union in May, 1969. The dissenter urges that she could have been known earlier to the company as a union adherent because organizational work already was under way. We cannot say that the Board's inference was wrong and that the General Counsel carried his burden of proof as to Gamez.

As to Morgan, the TX had found that she encountered little or no difficulty making production goals until her union activity started in May, 1969 and she was assigned additional duties which slowed her production and which were not required of nonunion employees. The Board, however, found no substantial support in the record, for these findings. All the record reveals about Morgan's productive efficiency, other than the adverse computer data and evidence that she was warned in April, 1969 about her low production, is Morgan's own testimony that she made the quota "off and on" after returning from an illness in August, 1969. This testimony relates to events after Morgan was identified with the union and thus is not probative of her productive efficiency before she became identified. Moreover, the only evidence relevant to whether nonunion employees in Morgan's department were not required to perform the additional duties Morgan claimed she was given was Morgan's testimony that "the Cuban girl" was not given such duties and that this girl had never worn a union badge to work. The Board reasoned that such testimony, even if credited, was not substantial support for the TX's finding that nonunion employees were not required to perform the additional duties because the Cuban girl's nonwearing of a union badge did not make her a nonunion employee. As the Board pointed out, one alleged discriminatee who was an active union campaign worker never wore her badge to work. Given the absence of substantial supporting evidence for Morgan's testimony, we cannot say that the Board was wrong in refusing to give weight to the TX's credibility-based finding that Morgan had been discriminated against in work assignments, see NLRB v. Buddies Supermarkets, supra, at 481 of 722 and in concluding that the General Counsel had failed also to carry his burden as to Morgan.

As to Phelps and Evans, the TX concluded that they were discriminatorily discharged by relying upon their credited testimony and circumstantial evidence supporting it. He recounted the following testimony by Phelps and Evans. Phelps testified that after she joined the organizing committee she began to receive constantly changing assignments, smaller bundles of work and even bundles lacking the accompanying card on which her piece work must be noted in order that she receive credit for pay purposes. She stated that after receiving a warning slip in January, 1970 because of her low production, she complained to Vice President Grodin about her work assignments and that he told her not to pay attention to the warning slip. According to Phelps, Grodin characterized the work she was getting as "garbage," and told her to let him know if she received any more such work. Phelps attended the March 5 representation petition hearing in response to a subpoena (apparently issued by the union) which she had shown to Grodin. Phelps stated that shortly after the hearing garments which she had sewn began to be returned to her with repair tape on them11 and she was ordered to correct "crooked stitches." Phelps testified that she showed the returned garments to a quality control inspector who told her that they were perfect.

Evans testified that before she was identified on one of the telegrams to the company, she was assigned to what she classified as "good work" because it was easier to make her quota with such work. After her name appeared on the telegram, Evans stated, she began receiving harder work-"scrapwork," "little old bitty stuff," bundles with different colors, smaller bundles-that made it more difficult for her to produce the quota. Evans testified that after she received a warning slip about her low production, she complained to her immediate supervisor that she could not make the production quota with the type of work she was being assigned. Her supervisor, Evans testified, admitted that she knew this.

The TX expressly found circumstantial support for the testimony of these employees and for his ultimate conclusion of their discriminatory discharges in the form of Bogart's background of unfair labor practices, the current Sec. 8(a)(1) violations and the timing of the discharges before the election.

The Board expressly accepted the TX's findings as to the credibility of Phelps and Evans and stated that the evidence as a whole came very close to supporting the TX's ultimate findings as to them. Nevertheless, the Board decided that the record did not support the TX's conclusion that Phelps and Evans were discharged because of their union activities. The Board posited that the TX's conclusion as to Phelps and Evans must have been based on Barrington's testimony that she was ordered to distribute small bundles to certain workers and large bundles to others. But Barrington's testimony, the Board reasoned, did not support the TX's conclusion as to Phelps and Evans because Barrington testified about work assignments made after the test period had ended. Furthermore, the Board pointed out, the TX's conclusion as to Phelps and Evans was also based on "a similar pattern" of discrimination which the TX had found among the ten discharges but for which the Board found no support in the record. Therefore, the Board concluded, although "the circumstances" raised "a suspicion" that Phelps and Evans were terminated because of union activities, the General Counsel had failed to carry his burden as to them.

The Board erred as to Phelps and Evans. Even if the TX's conclusion as to them was reached by his relying in part upon Barrington's testimony and a finding of a pattern of discrimination12 and even assuming the Board was correct in subtracting these factors from the evidentiary calculus,13 the General Counsel more than carried his burden with the testimony by Phelps and Evans which the Board itself accepted and the other circumstantial evidence upon which the TX expressly relied and with which the Board has not disagreed. The General Counsel amply established with such evidence that Phelps and Evans were treated differently by Bogart because of their union adherence14 in a manner which materially resulted in their selection to shoulder the burden of the necessary reduction in force. No more was required of the General Counsel under the Act. See Frosty Morn Meats v. NLRB, 296 F.2d 617, 621 (C. A.5, 1961) and NLRB v. Materials Transportation Co., supra, at 1078-1081 of 412 F.2d.

Under these circumstances, we hold that the Board incorrectly concluded that the General Counsel failed to carry his burden as to Phelps and Evans and also that the substantial evidence in the record concerning them which the Board has either accepted or not controverted supports no other conclusion than that they were discriminatorily discharged. See Ward v. NLRB, supra, at 12-14, of 462 F.2d; General Truck Drivers, etc. v. NLRB, 410 F.2d 1344, 1347 (C.A.5, 1969), and NLRB v. Dal-Tex Optical Co., supra.

Enforcement of the Board's order is withheld pending remand to the Board for modification of the order to encompass the Sec. 8(a)(3) and (1) violations found by the TX in the discharge of employees Phelps and Evans. When the Board issues a modified order to encompass these Sec. 8(a)(3) and (1) violations, the order shall be made part of the pending record for our final action and decree.15

Remanded.