This petition for review and cross-application for enforcement of a National Labor Relations Board (Board) order is brought pursuant to 10(e) and (f) of the National Labor Relations Act (Act), 29 U.S.C. 160(e) and (f). Torrington Company (Company) was found by the Board to have violated 8(a)(1) of the Act by its anti-union coercion of employees, and 8(a)(3) by unlawfully discharging employee Charles Crumley for union activity. We believe substantial evidence supports the finding of anti-union coercion but, because the uncontradicted evidence supports the Company's contention that Crumley's discharge was due to economic reasons, we decline to enforce that portion of the Board's order.

The record shows that employee Crumley had worked for the Company, which manufactured bearing products, on two different occasions between 1966 and 1971, for a total of over two years, and on both occasions had left his employment voluntarily. Crumley again applied for a job in mid-August, 1971 and was told by their Quality Control Manager (General Foreman as of October, 1971), Art Nord, who had been Crumley's supervisor on both previous occasions when he worked for the Company, that the Company was tentatively planning to lauch an experimental program to reduce the cost of scrap materials. Nord advised Crumley that, if he was interested in the new experimental program, Nord would let him know something in a few days. Shortly thereafter, Nord instructed plant Personnel Director Mobley Jeter to contact Crumley for rehire, which was done. Crumley reported to work about August 23 and was interviewed by Jeter, as was customary for newly-hired employees. Jeter told Crumley that the plant was non-union, that he had information that Crumley was involved in the 1969 union campaign, and that the union was not good for Union County or the plant, and they hoped Crumley would not be involved in the future. Jeter further testified that he normally reads a statement on unionism to newly-hired employees, but told Crumley that he did not think it necessary to read it in its entirety since Crumley had worked there before.

Crumley was assigned as a probationary employee to the inspection department, which was supervised by Nord. He and Stanley Moody, an employee of some years' standing who was transferred from another department to work with Crumley in the experiment to reduce scrap costs, were assigned as floor inspectors in the Acme Screw Machine Department (company Department No. 14). Crumley and Moody were to work the two night shifts in Department 14, which had not previously been manned with floor inspectors. Department 14 had consistently been one of the biggest problem areas in the company insofar as scrap expense was concerned.

Crumley became actively involved in the union campaign which was underway when he began work. On one occasion, shortly after he began work, Crumley was talking to another employee, Heatherly, about what happened during the previous union campaign in 1969. A long-time friend of both parties, Bryce Castle, who had just been promoted to assistant foreman on that same day, walked by the two as they were talking. Crumley remembered that Castle had been one of the more outspoken of the union adherents during the prior campaign and apparently sought to ask him what had happened to the union in the other campaign. Castle responded to Crumley's question by saying that his job would require him to report Crumley if he found that Crumley was trying to help the union. Crumley freely testified Castle told him: 'I can't be talking about the union. If I got caught talking to you about the union, I might lose my job.' Castle's testimony was to the effect that his remarks were make jokingly, although both Crumley and Heatherly stated that the remarks were made in a 'matter of fact' fashion. In any event, Crumley testified that he did not think Castle would report his talking about the union, and neither Crumley nor Castle made any reference to Crumley's pro-union activities during the present or past campaigns. The Board found this conversation was not an unfair labor practice.

During the first few weeks after Crumley and Moody began their new jobs, the daily scrap tickets maintained by the Company indicated that not only was there no reduction in scrap from Department 14, but if the trend continued, the month of September would have been the high for the year of 1971 to date. Around September 10, Nord called Crumley into his office, where the two of them discussed the scrap report and what the Company was endeavoring to do to reduce the increasing trend of scrap material, '. . . the scrap material and the costs, the amount of costs from the scrap.' Crumley admitted having this interview with Nord and Nord's telling him that the reason he had been hired was to reduce scrap.

About a week after Nord's conversation with Crumley on September 16, Nord met with plant Superintendent Donald Klinger to discuss the rising trend in scrap, and they mutually agreed that the experiment of placing floor inspectors on the night shifts had not worked. Both men further agreed that the experimental program should be terminated immediately. Nord then advised Personnel Director Jeter that Crumley was to be taken off the job, and Moody, who was not a probationary employee, was asked to report to a different shift. Probationary employees are those who are in training for the particular job they are hired for, and have no recall rights. Thus, since Moody had job seniority in his old job, he was transferred back, while Crumley, the job he was hired for having been discontinued, was terminated. Both Klinger and Nord stated they knew nothing of Crumley's union activities at that time.

That afternoon, Nord called Crumley into Mobley Jeter's office, whereupon Jeter informed Crumley that his job 'had not panned out,' and that the experiment had proved wrong. Crumley was then terminated after Jeter's exit interview.

The Board relies heavily on an incident that occurred the day before Crumley's discharge, in finding union activity as a cause for the discharge. Crumley was engaged in conversation with another employee, while at work, when Crumley remarked that the particular problem they were discussing would be straightened out once the union comes in. Immediately following the remark, the two noticed that a supervisor named Nicholson (who was not one of Crumley's supervisors) was standing closeby. Nicholson apparently made no indication of having heard the conversation, and in fact denied having overheard it. He stated there was considerable noise from machinery in operation in the work area. Crumley testified that Nicholson was 'within reaching distance' when he first saw him.

Crumley also stated that Nicholson was in Personnel Director Jeter's office the next day when he was discharged and that Nicholson left when Crumley walked in. Nicholson denied being in Jeter's office that day.

Other evidence pertaining to the exit interview indicates that Jeter told Crumley of the company's unwritten policy of not rehiring someone again after that person has worked for the company on three different occasions. Since this was Crumley's third severance of employment, Jeter indicated on the exit interview form that he would not be for rehiring Crumley.

Crumley returned to the Company in January, 1972 because he heard that the job he was on had been reopened. He discussed the situation with Jeter, and told Jeter that he felt like he'd been fired because of his union activities. Jeter told him that he would arrange an interview with Mr. Cook, the plant manager, and the interview was set for two days later. Cook and Crumley had never met prior to this interview, and Cook was not plant manager and did not work at that plant at the time Crumley was terminated. Crumley repeated his feeling that he was fired for union activity, and Cook told him that he had been let go because of the elimination of his job, but he would thoroughly investigate the matter to see if there was any substance to that charge. Cook also told Crumley that the three times and out rule was not mandatory but was for him to decide in each case. Crumley was told by Cook that if he submitted an application for employment, Cook would see if he couldn't employ Crumley in some other department of the plant. Crumley inexplicably refused to fill out a new application, telling Cook that he had only filled in one application but had worked for the Company on three previous occasions. Crumley then stated that Cook told him he was doing himself 'a great injustice by not filling one out.' Crumley therefore was not considered for further employment. Cook testified that he did subsequently investigate Crumley's allegation of being fired for union activities, and found, as he told Crumley during the interview, that Crumley's dismissal occurred only because his experimental job did not work out.

The record further shows that no one worked as a floor inspector on second or third shift in Department 14 either before or since the experiment involving Crumley and Moody. One floor inspector, Heatherly, worked first shift in Department 14 throughout the entire period of time involved here. There were three inspectors hired between the time Crumley was terminated, in September, 1971 and January, 1972; two of the newly-hired inspectors worked as spherical inspectors and one as a roving inspector. The difference between floor inspectors and spherical inspectors is that floor inspectors are generally responsible for checking the first part produced by a machine when it is placed into operation on any given day (called 'first article' inspection), while spherical inspectors make final inspections of all parts produced; the record does not indicate what a roving inspector does. Ferguson, one of the newly-hired spherical inspectors, does spend approximately 25% Of his time doing first article inspection, which was formerly done by Crumley as a part of his job, but the bulk of floor inspection in Department 14 was done by Heatherly. The record also indicates that James White became foreman of the Inspection Department, assuming duties as to that department formerly done by Nord, in October, 1971; and the new inspectors were hired by White. White testified that he knew Crumley, had heard rumors that he might be interested in the new jobs that came up after his termination, but did not consider hiring Crumley because the new openings for spherical inspectors and the one roving inspector were lower paying jobs than the floor inspector's job, and Crumley had not seen him to indicate he wanted a job.

On these general facts, the Board found that Crumley's discharge was in substantial part motivated by his union activity. The Administrative Law Judge discounted the Company's production machining scrap summary for 1971, which showed a substantial increase in scrap costs from September over August in Department 14 ($1,881 in September as opposed to $1,412 in August), apparently because the total figures for September were not available when Crumley was discharged. He was not impressed by Plant Superintendent Klinger's testimony that the daily scrap tickets were relied on in reaching his conclusion that the experiment was not working, because the daily tickets were not introduced into evidence.

The Administrative Judge found two factors of paramount importance in his analysis: (1) the timing and abruptness of the discharge and (2) the hiring of inexperienced people after the discharge. The judge pointed out, as the Company freely admits, that Crumley was a capable employee, and found wonder in the assertion that the Company would discharge such an employee without warning and through no fault of his own. The judge felt this was especially doubtful in light of the fact that a Company supervisor, Nicholson, overheard Crumley make a pro-union statement to a fellow employee on the day before he was terminated. He found equally implausible the subsequent hiring of three untrained individuals for the inspection department when the company knew that Crumley, an experienced, competent person, was interested in being rehired.

Crumley's refusal to file another application following his January, 1970 interview with Plant Manager Cook, the judge found to be of no relevance since the discriminatory discharge had already taken place. He concluded by stating that while he did not disbelieve the Company's evidence with respect to the experimental jobs, he did believe that Crumley's union activity played a substantial part in bringing about his termination. The Company was ordered to reinstate Crumley with back pay.

The Administrative Law Judge issued a supplemental decision some nine months later when it was discovered that the part of the transcript containing supervisor Nicholson's testimony had been inadvertently omitted by the court reporter when he initially delivered the transcript to the judge and the parties. He made an additional finding of fact, stating he disbelieved Nicholson's testimony that he did not hear the pro-union conversation between Crumley and another employee because: 'It seems to me that anyone standing within touching distance of two people talking has to be able to hear what is said as well as the two principals can.'

Generally, of course, an employer may dismiss an employee for good cause, no cause, or for any reason he chooses except the employee's union activity. See NLRB v. M & B Headwear Co., 349 F.2d 170, 173 (4th Cir. 1965). Business reasons may not be used as a pretext to get rid of active union employees, e.g., NLRB v. Overnite Transportation Co., 308 F.2d 284, 288 (4th Cir. 1962), and if an anti-union motive, or a desire to stifle protected activity, is a factor in the employer's decision, the discharge is discriminatory. Winchester Spinning Corp. v. NLRB, 402 F.2d 299, 304 (4th Cir. 1968). But we have stated in NLRB v. Hanes Hoisery Division, 413 F.2d 457, 458 (4th Cir. 1969), 'undoubtedly, the fact that a worker takes part in protected activity does not insulate him from discharge for legitimate business reasons.'

In determining employer motivation for an alleged unlawful discharge, all the circumstances of each case must be weighed; this is the function of the Board, and conflicts in testimony, or inferences to be drawn therefrom, must be resolved in the Board's favor. J. P. Stevens & Co. v. NLRB, 406 F.2d 1017, 1020 (4th Cir. 1968); Filler Products, Inc. v. NLRB, 376 F.2d 369, 378 (4th Cir. 1967). This court has held, however, that an inference of discrimination resulting from mere participation in union organization and activity, followed by a discharge, disappears when a reasonable explanation is presented to show that the discharge was not for union membership. NLRB v. United Brass Works, Inc., 287 F.2d 689, 693 (4th Cir. 1961).

The crux of the matter is that the Board, after discounting all explanations offered by an employer for a discharge of employees, must find unlawful motivation through substantial direct or indirect evidence. Substantial evidence means 'evidence furnishing a substantial basis of fact from which the fact in issue can reasonably be inferred; and the test is not satisfied by evidence which merely creates a suspicion or which amounts to no more than a scintilla or which gives equal support to inconsistent inferences.' Dubin-Haskell Lining Corp. v. NLRB, 375 F.2d 568, 573 (4th Cir. 1967); Riggs Distler & Co. v. NLRB, 327 F.2d 575, 580 (4th Cir. 1963); Appalachian Electric Power Co. v. NLRB, 93 F.2d 985, 989 (4th Cir. 1938).

The uncontradicted facts in this case show that the Company had a legitimate business reason for discharging Crumley. Indeed, this is found by the Board. Crumley was hired on an experimental basis as a probationary employee. He was told that the experimental program, involving placing floor inspectors on shifts not theretofore manned, had as its aim the reduction of scrap costs in one of the most troublesome Company departments. Daily scrap reports showed that not only did the scrap problem not improve, it got worse, and this in addition to the increased labor costs due to the two new inspectors in the department. The Quality Control Manager, Nord, who hired Crumley, was sufficiently concerned about the increase in scrap expense that he had a conference with Crumley to discuss the problem. One week later, and a little more than three weeks after Crumley started work, Plant Superintendent Klinger and Nord mutually agreed that the experiment had not worked and should be immediately abandoned. Crumley, because he was a probationary employee, was terminated that same afternoon, and the other floor inspector involved in the experiment, Moody, was transferred back to his old job. No floor inspectors were hired, and the jobs done by Crumley and Moody remained vacant, except for part-time coverage by inspectors from other departments, during the four month period after Crumley's discharge and before he returned to the company to seek another job. The record shows that scrap expense from Department 14 was $1,881 for September, 1971 as opposed to $1,412 for August, and that the September total, in fact, was the third highest monthly total for scrap expense from that department during the year of 1971. These uncontroverted facts show clearly that, whatever inference the Board drew or might have drawn from the evidence, Crumley would have been discharged whether there was a union campaign or not.

We find no acceptable basis for the Board's discredit of the uncontradicted testimony of Company officers that the daily scrap tickets showed a substantial increase in scrap costs. The rearly scrap report was in evidence which supports their testimony, and it is obvious that some form of daily tally must be kept, if for no other reason than to compile the monthly and yearly reports. Sworn testimony cannot be arbitrarily disregarded on the mere suspicion or assumption that Nord and Klinger were lying. NLRB v. Atlanta Coca-Cola Bottling Co., 293 F.2d 300, 306 (5th Cir. 1961); see Universal Camera Corp. v. NLRB, 340 U.S. 474, 484 and n. 17, 71 S. Ct. 456, 95 L. Ed. 456 (1951).

Likewise untenable was the Board's analysis that the abruptness of the discharge and subsequent hiring of inexperienced people indicated anti-union bias against Crumley. Of course any discharge during a union campaign could be said to be ill-timed because suspicions on both sides are then at their peak. But here, Crumley was admittedly hired only for the purpose of trying an experimental job; he was told a week before the decision was made to discharge him that the scrap trend had worsened, and the evidence shows beyond doubt that the experiment was a failure. The Company, as was its practice with probationary employees, then dismissed Crumley because the job he was hired for no longer existed. We see nothing abrupt in this sequence of events.

The hiring of inexperienced people has no relevance here. No one was hired to do the job formerly done by Crumley; there were no recall rights existing within the Company when a lay-off occurs because of elimination of the employee's job; Crumley did not reapply for a job until after the new employees were hired, and, in any event, the Company's general policy of hiring no individual on more than three separate occasions would make it more doubtful that he would be hired for a fourth time. Even so, Crumley was offered an opportunity to submit a new employment application in January, 1972, and he refused. If Crumley had truly been interested in being rehired, it would seem that he would have made application before four months had elapsed after his discharge. Also, once Crumley decided to reapply and Plant Manager Cook told him that he would be considered for a job despite the 'three times and out' rule, Crumley's refusal to fill out a new application is inexplicable in almost any context except that the Company was acting in good faith.

Added to this, or of even greater importance, is the finding of fact of the Administrative Law Judge as to the knowledge of the Company of Crumley's union activities. Such is required to sustain the holding of the Board. NLRB v. Lexington Chair Co., 361 F.2d 283, 291 (4th Cir. 1966). The original opinion relied almost solely on the conversation (revealing his union activity) Crumley had with Louis Jeter and during which Nicholson, a foreman, was nearby. Only Crumley and Jeter testified to the conversation with inconsequential differences. The opinion then must be based on such testimony which it describes as it finds: 'Nicholson did not testify.' Nicholson had in fact testified, and had denied that he had overheard the conversation. Thue transcript of Nicholson's testimony which was not with that part of the transcript initially filed had inadvertently not been brought to the attention of the Administrative Law Judge. Upon the discovery of this error, the Board remanded for reconsideration in the light of Nicholson's testimony, with leave to reopen the record to take additional evidence. The record was not reopened, however, and after review of the written transcript, the Administrative Law Judge, who apparently still did not remember Nicholson testifying, recited: 'In my (original) decision I found that Crumley, on the basis of his undenied and credited testimony, had been overheard by his foreman, Jack Nicholson, talking in favor of a union to a fellow employee.' The opinion then goes on to discredit Nicholson by inferring that Nicholson overheard the conversation because 'It seems to me that anyone standing within touching distance of two people talking has to be able to hear what is said as well as the two principals can.' And this finding was made despite the uncontradicted and unexplained evidence that it was noisy, a machine shop being thirty feet to the rear; a threshing machine with a 50 horsepower motor being there; a turret lathe running just behind him; and a machine operating on compressed air being in the area. Also, the finding was made without reopening the record and without even a recital of any recollection of Nicholson testifying. Had the supplemental opinion had its finding on the credibility of witnesses, it may be that we would be required to accept it, although the evidence is essentially negative; see Jones v. Chesapeake and Ohio Ry. Co., 371 F.2d 545 (4th Cir. 1966); II Wigmore on Evidence, Third Ed., 664; but since the ultimate fact (knowledge) is based on an inference which is based on a supposition ('it seems to me'), we are of opinion that even the most liberal construction of the substantial evidence rule should not allow a critical fact to be so proved.

In sum, we hold that there is not substantial evidence to support the Board's finding of a discriminatory discharge. A finding of discrimination may not reasonably be inferred, for this record, at the very least, gives equal support to inconsistent inferences. Dubin-Haskell Lining Corp., supra, 375 F.2d at 573.

We do think, however, that substantial evidence supports the Board's finding of coercive employee interrogation in violation of 8(a)(1) of the Act. The record shows that three different Company officers had, at different times during the campaign, called employees into their respective offices to discuss the union. The conversations, in some instances, did involve rather extensive questioning of the employees concerning their union attitudes, activities, and support. Of course, the employer is entitled to take part in free and frank discussions over unionization, but not where such discussions take the form of interrogations which have a tendency to inhibit the free exercise of rights protected by the Act. As we have noted, employer interrogation is generally a dangerous practice, not helped in this case by the fact that supervisors' offices were frequently used to conduct it. See Winchester Spinning Corp. v. NLRB, 402 F.2d 299, 302-303 and n.1 (4th Cir. 1968).

Enforced in part and denied in part.