This is an action brought by 441 plaintiffs, black former employees of appellee ESB, Incorporated. The case arises under Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e et seq. The district court, following trial on the merits, dismissed the action and entered judgment for the defendant ESB. We affirm.

The action was brought initially under the Civil Rights Act of 1866, 42 U.S.C. § 1981. A second action was filed under both § 1981 and Title VII. The district court, without prejudice, consolidated the cases and dismissed the § 1981 claims and struck references to them from the complaint.

The plaintiffs, who allege racial discrimination by ESB, sought to bring this case as a class action representing all black applicants for employment with ESB and all present and former black employees of ESB. This class was limited to the Sumter facility of the defendant by stipulation and court order. The plaintiffs alleged discrimination by ESB in hiring, layoffs and discharges, and in pay, promotion and other employment practices.

Shortly after the first complaint was filed, ESB served a set of written interrogatories on each of the named individual plaintiffs and also sought to take depositions of such plaintiffs. Due to the failure of 16 of the plaintiffs to comply with the discovery requirements, the district court found them in default and dismissed them as parties, with prejudice. Such dismissal came only after the district court granted them a second chance to obey an order of the court that they comply with discovery procedures.

The plaintiffs proceeded with the class action under a conditional order of the court pursuant to Rule 23(c)(1), in which the court reserved the right to alter or amend its order after discovery had been taken and evidence introduced. The motion to dismiss the class action was renewed at the conclusion of all testimony and was granted by the court.

The district court stated that, while there had been testimony as to alleged racial discrimination in several areas, the main thrust of the action and the primary complaint of virtually every plaintiff was that the layoff by ESB of 30 whites and 53 blacks in July 1970 was discriminatory in nature, as was the failure of the defendant to rehire the laid-off black employees. Of the 44 original plaintiffs, 42 were laid off on July 24, 1970, one prior to that date, and one several months later.

The court held that the class consisted of the 53 blacks who were laid off, and ruled that the class was not so numerous that joinder of all its members was impracticable. The court also questioned the ability of the plaintiffs to fairly and adequately represent and protect the interest of the class.

The district court, as to the merits of the action, found that the plaintiffs failed to prove any racial discrimination against them by the defendant and accordingly dismissed the action.

* The factual setting of this case should be noted in some detail. ESB opened its Sumter plant in August 1965, but it was never a profitable operation. The principal product of the plant was the nickel-iron battery, an expensive, long-term energy cell produced for industrial consumption. Sales of these batteries were in large part to railroads, with the Penn Central Railroad being the largest of its customers. Sales dropped through the late 1960's, and in 1970 its sales to Penn Central declined sharply. By early 1970, the Sumter plant was losing approximately $120,000 per month, with future sales projected to decline even further. The Sumter plant had developed problems in quality control, lack of efficiency, high labor costs, and engineering and inventory problems.

In an effort to cure these problems and to make the Sumter operation profitable, ESB employed one Peter May, a trouble-shooter in manufacturing problems. May arrived at ESB in March 1970 and began his investigation. He found a 50% defective rate in the production of the tubes made by ESB for their nickel-iron batteries. ESB at that time had over one million defective tubes on hand. Changes in production techniques introduced by May reduced the sub-quality tubes to almost zero. May also decided that the labor force was too large to work efficiently and for the plant to operate profitably. The management of ESB, in June, 1970, approved his suggested reduction of the work force. May suggested what job areas in the plant needed to be reduced, but had nothing to do with the selection of the actual employees to be laid off.

On July 24, 1970, the layoff was effected and involved 83 workers, 53 of whom were black. Prior to that time, there were 365 hourly employees in the Sumter plant, of which 197 or 54% were non-white. Of those laid off, 63% were non-white. Following the layoff, the hourly work force was 53% non-white and at the time of the trial 57% were non-white. According to the 1970 census, Sumter County had a non-white population of 41% and the City of Sumter was 35% non-white.

The layoff was carried out in compliance with the policy set forth by defendant ESB in its employee handbook, which policy had been in use for a considerable period of time. Under that policy, the reduction in the work force was made by reducing the numbers in various departments, as required, on the basis of plantwide seniority within each job classification of each department. Employees in a higher job classification, who had been previously assigned to a lower job classification within the department, were given the opportunity to move back to the lower classification so long as it was consistent with the layoff policy. Permanent job classifications were used throughout, including situations in which an employee might be working at a temporary job at the time of the layoff.

Under ESB employment policy, employees with less than five years' service (all persons involved in the layoffs had less than five years' service) retained rights of recall to work for a period of six months. No new employees were hired within this six-month period. The employees were advised at that time that the layoff was permanent.

Once recall rights had expired, former employees were required to reapply for employment and were treated on reapplication as new applicants for employment. ESB had two other applicable rules that are pertinent here. The first was that job applications would be considered active for only 30 days. If no job became available during this 30-day period, the application was destroyed and a new application required. Notice of this rule was posted in a conspicuous place in the room where persons filled out job applications. The second rule was a policy that former employees would not be rehired to jobs of a lower classification than the job held at the time the employee was laid off or terminated. This was based on prior experience of ESB that former employees who were rehired at a lower grade usually became unhappy or discontent and did not make good employees. The district court found both of these rules to be of long standing and that they did not result in racial discrimination nor were they added due to any racial discrimination.

In March 1971, ESB began to hire again. Eleven^1a of the original 44 plaintiffs in this action came to the plant and filled out applications after that time. Of these, three were offered jobs and two accepted and were rehired. One was offered a job and refused it. The remaining eight, the district court found, were not rehired due to either an absence of available jobs or due to work problems encountered during prior employment, i. e., absenteeism and transportation difficulties.

The plaintiffs at trial focused on the period of March, April and May 1971, in which 29 employees were hired, only one of whom was black. The trial court found that while this might appear to be racially discriminatory, it actually was not. The court compared the hiring during this isolated period with overall plant figures, which showed that ESB had a far higher percentage of black employees than the percentage of blacks present in the community. The district court also found that 10 of the 28 whites hired during this isolated period were temporary employees, and that three blacks were reinstated during this period from temporary layoffs and maternity leave. Another three blacks were offered employment during this three-month period but refused it.

Appellants challenge the district court's dismissal of the class action aspect of this suit, arguing that there was sufficient statistical evidence of racial discrimination for the class action to be properly maintained.2

As noted already, the district court delayed ruling on the motion to dismiss the class action and allowed the trial on the merits to proceed on a conditional order pursuant to FRCP Rule 23(c)(1),3 reserving the right of the court to alter or amend its order with respect to the class action.

At trial, the court received the evidence of the plaintiffs in accord with its broad conditional class action order. We note that appellants do not contend that the district court refused to admit any evidence that would properly be admissible in a class action or that the district court refused to consider any such evidence. After considering all of the evidence presented, the district court determined there was not sufficient evidence of racial discrimination by ESB to allow the action to proceed as a class action.

In view of the disposition we make of this case, especially considering that the district court considered all the evidence in arriving at its decision, the individual plaintiffs have received all the benefit they could have from the broad class designation in force until after the evidence had been taken, so they have not been harmed by the later modification of the order describing the class. In this connection, we note that the authorities apparently are not uniform as to whether relief may be awarded, in a proper case, absent a class action, which extends beyond those who are plaintiffs in the suit. Cf. Brito v. Zia, 478 F.2d 1200 (10th Cir. 1973), and Sprogis v. United Airlines, 444 F.2d 1194 (7th Cir. 1971), cert. den. 404 U.S. 991, 92 S. Ct. 536, 30 L. Ed. 2d 543 (1971), with Danner v. Phillips Petroleum Company, 447 F.2d 159 (5th Cir. 1971). See 42 U.S.C. § 2000e-5(g); 7 Wright & Miller, Federal Practice and Procedure, Civil, § 1754. We mention this not to express an opinion on the subject, but to indicate the wide range of choices within the reach of the district judge. We will address ourselves to the remaining members of the broad class later in this opinion.

We affirm the district court's denial of class action status. After receiving all the evidence, it determined that the thrust of the plaintiffs' suit was the layoff. Since only 53 black employees were involved in that layoff, the court ruled that the class was not sufficiently numerous to make impracticable a joinder of those not already plaintiffs in the action. Since there were 44 named plaintiffs, this would have meant joining only 11 additional persons. It further questioned the ability of the plaintiffs to fairly and adequately protect the interest of the class they sought to represent.

In order to determine whether a class action is proper, the district court must determine whether a class exists and if so what it includes. Although not specifically mentioned in the rule, the definition of the class is an essential prerequisite to maintaining a class action. 3B Moore's Federal Practice, P 23.04, pp. 23-251, 2d ed. 1974; 7 Wright & Miller, Federal Practice and Procedure, Civil, § 1760. This determination is a question of fact that will be determined on the basis of the circumstances of each case. 7 Wright & Miller, Federal Practice and Procedure, Civil, § 1760, pp. 579-80.

The district court here heard all of the evidence and determined that the proper class was not all black applicants and present and former employees of ESB, but rather was those involved in the layoff. We do not think the district court abused its discretion in defining the class. A court has broad discretion in deciding whether to allow the maintenance of a class action. 7A Wright & Miller, Federal Practice and Procedure, Civil, § 1785, p. 134. And it has been said the determination usually should be predicated on more information than the complaint itself affords. 7A Wright & Miller, Federal Practice and Procedure, Civil, § 1785; Baxter v. Savannah Sugar Refining Corp., 46 F.R.D. 56, 60 (S.D.Ga.1968).

Once the court defined the class, the determination was then made that the action was not properly maintainable as a class action either on the basis of numerosity or on representation.

FRCP Rule 23(a) allows a class action only if: "(1) the class is so numerous that joinder of all members is impracticable . . . and (4) the representative parties will fairly and adequately protect the interest of the class."

The determination of a district court that an action does not meet the requirements of a class action will not be disturbed unless it is clearly erroneous. Brito, at p. 1204. And generally, unless abuse is shown, the trial court's decision on this issue is final. Cypress v. Newport News General & Nonsectarian Hospital Association, 375 F.2d 648, 653 (4th Cir. 1967). We do not find error in the determination of the class by the district court nor in the court's finding of failure to meet the requirements of FRCP Rule 23(a).

The numerosity ruling of the district judge is so apparently correct as to require no discussion. If only 11 persons were added, all in the class would have been parties. His alternative ruling as to inadequacy of representation is equally correct. From the inception of the suit, its main thrust had been the layoff. Although all the circumstances surrounding the layoff were tried in detail, little or no attention was paid by the plaintiffs to other aspects of the employment situation existing at ESB.4 Hardly a witness was called to testify in any detail except as to the layoff.5 More will be said of the inadequacy of representation later in the opinion.

Error is assigned to the district court's dismissal of the individual plaintiffs who failed to cooperate during discovery. We find none. The plaintiffs who were dismissed had failed to respond to interrogatories; failed to respond to an order entered by the district court requiring a response to the interrogatories; and additionally failed to respond upon specific request after the court had denied, without prejudice, a first motion to dismiss for such failure. Only when the motion to dismiss was renewed after another failure to respond following the denial of the first motion to dismiss did the court invoke sanction. The court acted within options available to it under FRCP 37(d), and we are of opinion there was no abuse of discretion. Moore v. Island Creek Coal Company, 375 F.2d 732 (4th Cir. 1967); 4A Moore's Federal Practice, P 37.05 2d Ed. 1974.

Plaintiffs contend that the district court erred in not finding discrimination on the part of ESB in light of statistical evidence offered by them.

The effect of statistical evidence has been considered in at least three opinions of this court, Brown v. Gaston County Dyeing and Machine Company, 457 F.2d 1377 (4th Cir. 1972); Barnett v. W. T. Grant Company, 518 F.2d 543 (4th Cir. 1975); and Patterson v. American Tobacco Company, 535 F.2d 257 (4th Cir. 1976).

In Brown, we held that we "need not decide whether the 1969 statistics, revealing as they are, should be regarded as conclusively showing violation of Title VII or whether they establish a prima facie case . . . (footnote omitted). It is sufficient to hold that they have not been rebutted by the company's efforts since 1965 to hire and promote black employees." p. 1382. We stated that, as well as specific acts of racial discrimination, courts must also examine statistics, patterns, practices, and general policies to ascertain whether racial discrimination exists. p. 1382. We found that "statistics indicate that race is the only identifiable factor explaining the disparity between the jobs held by white employees and those held by black employees. The proof discloses no objective standards based on education, experience, ability, length of service, reliability or aptitude to account for the preferential employment of whites." Brown, p. 1383. In Barnett, we stated that "statistics can in appropriate cases establish a prima facie case of discrimination without the necessity of showing specific instances of overt discrimination." p. 549. There, we were of opinion that the statistics were strengthened by the evidence of discriminatory practice in that there were no objective criteria for the hiring of supervisors and at least one qualified black employee had been twice passed over for supervisory status in favor of whites, one of whom had less seniority and was brought in from outside the company. In Brown, we held that the lack of objective guidelines for hiring and promotion and the failure to post notices of job vacancies were badges of discrimination that served to corroborate racial bias shown by the statistical pattern of the company's work force. In Patterson, however, we held that the district court should not have imposed racial and sex quotas for supervisors based on population percentages in the area from which the work force was drawn because the percentages of black people and women in the same area in a category which included supervisory personnel were smaller. We thus held that the relevant statistic was the availability of the affected people in the work force, not merely the overall percentage of discriminatees in the area. We said such "percentages furnish a more realistic measure of the company's conduct than the gross percentages of blacks and women in the whole work force, including unskilled labor." p. 275.

We have no quarrel with the principles above related from Brown and Barnett. But we do not believe that isolated bits of statistical information necessarily make a prima facie case when divorced from other and contrary statistics and from the statistical picture of all the employment at the plant. We also think the absence of other evidence of discrimination should be considered in determining whether a prima facie case is made, just as the presence of other evidence of discrimination should be considered in arriving at the same conclusion.

In this respect, it is worthy to note that the establishment of a prima facie case does not require a finding in favor of the party establishing it; but only permits that finding. Wright v. Rockefeller, 376 U.S. 52, 57, 84 S. Ct. 603, 11 L. Ed. 2d 512 (1963). A risk of non-persuasion is yet on the plaintiff, as also has been noted in at least one learned journal cited by us with approval in Barnett, Note: Employment Discrimination: Statistics and Preferences under Title VII, 59 Va.L.Rev. 463 (1973), just as is a risk of non-persuasion on the defendant if he does not rebut the prima facie case. See Duplan Corp. v. Deering Milliken, Inc., 540 F.2d 1215 (4th Cir. 1976). While the burden of going forward with the evidence may shift to the defendant if the plaintiff has made a prima facie case, risk of non-persuasion is yet on the plaintiff as well as the defendant, and the defendant is not prevented from having the court consider all of the evidence before it in determining whether a case has been made. So the court, in determining whether a party has successfully overcome the risk of non-persuasion, should consider all of the statistical information before it, as well as all the other evidence bearing on the presence or absence of discrimination in employment.

The district court received into evidence and considered the statistical data as well as the other evidence offered, including much oral testimony, and determined that the record before it did not show discrimination on the part of ESB. The district court especially noted that the office force and the supervisory and salaried positions might arouse suspicion and that on a different record a finding of discrimination might be justified. But the court properly decided the case on the record before it, not its suspicions. It took care to provide, as do we, that the class plaintiffs are not adversely affected by the failure of the individual plaintiffs to establish their claim for them.

If we consider that the statistical evidence called to our attention by the plaintiffs makes a prima facie case, we are of opinion that the whole statistical evidence does not, or, to put it another way, if the plaintiffs' prima facie statistical case has been made, it has been rebutted by other statistical evidence as well as the absence of other evidence of discrimination. We think the finding of the district judge in this respect is not clearly erroneous, and, with two exceptions, a more detailed discussion of the statistical evidence is made elsewhere in the opinion.

We do not believe either Brown or Barnett require a different holding for the reasons we have set forth above and the brief comparison which follows.

In Brown, black workers were employed in only 11 job classifications out of 45. Here, blacks were employed in 73 and whites in 83 of 115 job classifications. Blacks were employed in 24 of the 26 departments in September 1972. In Brown, slightly less than half of the black employees were employed near the bottom of the company's pay scale. Here, at the bottom of the pay scale, the division between whites and blacks is almost exactly 50%. And leaving out the craftsmen, about which more will be said later, black employees are spread throughout the operating force of the company and employed in 10 of the 11 pay grades. In Brown, there was a lower percentage of black employees than the percentage of black people in the general population; here, there are more. In Brown, see p. 1383, there was no objective standard for promotion based on "education, experience, ability, length of service, reliability or aptitude." Here, promotion is based on "proven ability, proper qualifications, sufficient experience, and length of service." Here, all other things being equal, length of service controls. While in Brown we held that failure to post notices of job vacancies was a badge of discrimination which served to corroborate statistical evidence, here the percentage of promotions received by black employees was slightly larger than their percentage in the work force. So, so far as the record before us shows, there is no statistical imbalance in promotions, and thus no adverse inference, for the absence of job postings to corroborate. We do not mean to imply that absence of job postings may not be evidence of discrimination, and we do not so hold.

In Barnett, black employees were in only one of 17 non supervisory job classifications; here, black employees worked in 73 and white employees in 83 of 115 job classifications. As noted, blacks were employed in 24 of the 26 departments in September 1972. In Barnett, there were no black supervisors, as here, but we must assume, for the opinion does not indicate it, in view of our holding in Patterson, above referred to, that their availability in the work force was not before the court in Barnett. More will be said of this later. In Barnett, the percentage of black employees in the plant was less than that in the general population; here, it is greater. In the job at issue in Barnett, over-the-road driver, there were no black employees out of 27. No such claim is made here. In Barnett, there was word of mouth hiring for over-the-road drivers as contrasted to open recruiting for other jobs. Such is not present in our case, and the defendant here is not shown to have any different hiring policies.

Thus, in the case before us, we do not find a statistical imbalance indicating "that race is the only identifiable factor explaining the disparity between the jobs held by white employees and those held by black employees." Brown, p. 1383.

Appellants contend that the layoff by ESB of July 24, 1970 was discriminatory in that more blacks than whites were laid off. They argue that blacks were concentrated in the lower job classifications due to discriminatory job assignments and promotions.

However, we do not find that the layoff was discriminatory. It was conducted on the basis of plantwide seniority in compliance with Robinson v. Lorillard Corp., 444 F.2d 791 (4th Cir. 1971), cert. den., 404 U.S. 1006, 92 S. Ct. 573, 30 L. Ed. 2d 655 (1971). There is not here the heavy low end job placement of blacks that was found in Robinson. The court in Robinson focused on the use of departmental seniority to perpetuate past discriminatory practices in placement. There was no such effect here since ESB used a system of plantwide seniority in its layoff. As an example of the general black and white plant distribution, in September 1972, there were blacks working in 10 of the 11 ESB labor grades and in 24 of the 26 departments. In the two departments without black employees, one had only one person, the other three.

The statistical ratio of the July 24, 1970 layoff shows a higher percentage of blacks laid off than whites, but this must be considered in conjunction with the fact that some 54% of the ESB work force was non-white prior to the layoff, and remained 53% non-white after the layoff. While some 63% of those laid off were black, in light of the small statistical sampling involved (83 persons), this figure does not prove a discriminatory layoff, especially considering the manner in which the layoff was effected. Only a small change (8) in the numbers of blacks and whites involved would yield a ratio the same as that of blacks to whites in the plant. We also take note of the undisputed economic necessity that compelled the cutback in employees, the reduction in the number of tube loaders needed after the change in the production process, and the uncontradicted evidence as to the selection of those to be laid off on the basis of the affected positions rather than on any personal criteria.6

Appellants contend that the defendant ESB is guilty of racially discriminatory hiring practices. We do not find this to be true.

Statistical data as to hiring by ESB is clearly in favor of ESB. It had a non-white employee percentage of 54% prior to the layoff, well above the percentage of the black population of the Sumter community.

Appellants seek in this case to focus rather on the isolated period of March, April and May 1971 in which 28 whites and only one black were hired. They limit the statistics they wish accepted as proof of the whole case as to hiring after the layoff in terms of a one-year period, August 1970 to August 1971. However, hiring was done by ESB in only the three-month period just referred to during the one-year period which plaintiffs insist must control this point. In light of the evidence presented as to the hiring ratio before and after this period, consideration of the period in isolation is unsound. In December 1971, the first month after May 1971 in which ESB hired, there were 11 blacks hired and only two whites. In January 1972, there were 14 blacks hired and eight whites. The ratio for the following eight months is well in line with the racial ratio of the Sumter community, 75 white (59%) to 53 black (41). Prior to the time of the layoff, 1214 whites had been employed (73%), and 444 blacks (27%), and overall plant hiring through September 1972 was 1327 white (71%) and 523 black (29), again in line with the population of the Sumter community (65.3% white, 34.4% black, .3% other). And we again note that the plant employment was more than 50% black.

The district court examined the three-month period of March to May 1971, and made a finding of no discrimination. We noted earlier its finding that three blacks refused jobs during this period, that three blacks were reinstated from temporary layoffs and maternity leave, and that 10 of the 28 whites hired were only temporary employees. We do not feel, in light of the overall hiring statistics, coupled with the factual findings of the district court, the hiring figures for the period of March to May 1971 are sufficient to support a finding of discrimination. Indeed, far from supporting an inference of discrimination, the overall statistics tend to support a finding of lack of discrimination.

Appellants also challenge the failure of ESB to rehire those blacks who were laid off in July 1970. They assert as discriminatory the application by ESB of the 30-day application rule and its policy not to rehire at a lower job classification. Due to the absence here of any showing that such rules were applied to black employees and not to whites, or were discriminatory as to blacks, or evidence that they were rules adopted with the intent to discriminate, or which had the effect of discriminating, we do not find these rules to be discriminatory in their nature or effect. We agree with the district court that no discrimination was present in the hiring and rehiring practices of ESB.

Appellants assert racial discrimination by the defendant in the job assignments given blacks and in the number of blacks promoted. In general, they charge that blacks are more heavily distributed in the lower job classifications and not adequately represented in the high classifications. The district court found no showing of racial discrimination in the job assignments.

Overall, whites do outnumber blacks in the higher job classifications and blacks outnumber whites in the lower classifications, but any imbalance is not striking. In October 1970, there were 175 blacks and 108 whites in the five lower of the 11 ESB job grades, and in the upper six grades there were 22 blacks and 60 whites, while, in September 1972, the five lower grades contained 119 blacks (62%) and 71 whites (38%), and in the upper six grades there were 24 blacks (41%) and 34 whites (59%). But it should be noted that in the lowest pay grade, in September 1970, there were 25 black and 24 white employees, and in September 1972 there were 23 blacks and 23 whites.

Of the 115 job classifications at the plant, blacks worked in 73 and whites in 83. The district judge, after considering all the evidence, including the statistics offered by both sides, held that the plaintiffs had failed to show the differences in job placement were the product of racial discrimination, nor that qualified blacks were denied employment or promotion to the higher labor grades.

We agree that the plaintiffs have failed to show racial discrimination by ESB in regard to assignments to jobs and promotions. There is not here a statistical imbalance such as this court found in Brown, nor a factual background of segregation as encountered in that case. In Brown, there was a long history of segregation of workers by race with a denial of job opportunities and promotion possibilities, which setting is not present here. Nor are the statistics offered here of the nature of those in Brown, where there were only 29 blacks employed and 187 whites, with an overwhelming concentration of the few blacks in the lowest labor grades.

The spread here of blacks through the labor grades is not such as would alone create a prima facie case of discrimination. This is especially true in light of the highly skilled work done by the higher pay grade workers such as tool and die.

There is no showing of discrimination in promotions. For the period of May 1969 to September 1972, there were 347 promotions among hourly paid employees. Of these, 199 or 57.3% went to blacks. These figures include job reevaluations made by ESB of its employees, which resulted in the assignment of a higher labor grade to a particular job classification. Of 160 such job reevaluations, 92 or 57.5% involved jobs held by black workers.

The district court found that ESB, in making promotions, considered the requirements of a job when seeking an employee to fill a vacancy, using seniority as the determining factor in those cases where applicants had substantially equal qualifications. ESB, in its employee handbook, had listed the factors to be considered in making promotions, e.g., performance on present job, willingness to accept responsibility, and dependability.

Appellants also challenge ESB's lack of system of posting of job vacancies. However, there is no showing that the absence of job posting resulted in any racial discrimination in the plant. While we have held that the lack of objective guidelines for promotions and the failure to post notices of job vacancies may be badges of discrimination, Brown, 457 F.2d at 1383, their absence does not necessarily make a prima facie case of discrimination, although in a proper case they may be considered evidence thereof. This is especially true as here when there were guidelines for promotion; there was no evidence of a discriminatory promotion; the black employees received a part of the promotions slightly more than their percentage of the work force; and the hiring practices are free from discrimination.

Appellants point to tube loading7 in particular in their argument as an example of a department which is racially exclusive due to discrimination. Their position is contradicted by the record. The record shows, and the district court so found, that more whites were hired into tube loading than blacks. Between the opening of the plant and November 1971, 97 blacks and 110 whites were hired into the tube loading department. Ninety-five whites quit and left the job, and 15 were discharged. Of the blacks, 53 quit and 14 were discharged, with others being transferred from tube loading as the demand decreased in that department. It should be noted that tube loading was the highest paid production job using semi-skilled employees. The district court also found that most of the vacancies in the department were being filled by transfers from other departments by employees seeking the higher wages.

Appellants assert the supervisors, office, clerical, and secretarial workers were chosen in a discriminatory manner. Due to the skilled nature of these jobs, the business necessity test, see Robinson, 444 F.2d at 797, may allow such practice. The district court concluded it did, based on the absence of qualified black applicants who have sought the positions and the scarcity of such qualified personnel in the Sumter community. On the evidence produced, this was a proper holding.

The plaintiffs have failed to prove that qualified blacks applied for these jobs and were refused them in favor of less or no better qualified whites. See Green v. McDonnell Douglas, 411 U.S. 792, 93 S. Ct. 1817, 36 L. Ed. 2d 668 (1973). Green, of course, was not a case involving a class action, and we have taken this into account. Nor is there a showing of how the craftsmen came into the company, whether from outside the plant or by working their way up. The same is true of supervisors, office, clerical, and secretarial workers. Absent a showing of qualified black applicants being available and not being hired or promoted into these skilled jobs, there is no discriminatory practice proven.

Craftsmen in the company occupy 13 jobs. There are 57 total craftsmen employees, of which 15 (26%) are black. A list of transfers from hourly to salaried jobs is included with appellants' brief. Some of such transfers were promotions; some were not. Of those transfers considered promotions which we can track through the exhibits, 8 out of 11 since May 3, 1969, 3 of the 11 not being promotions, six (75%) were white and two (25%) were black. These promotions by way of transfer to salaried jobs have been included in the promotion statistics otherwise mentioned which commence with May 1969 and end July 1972. Although the Company furnished a list of each employee hired since the plant opened, with his employment history, the appellants make no attempt to show the source of employees employed as office workers, in salaried jobs, or as craftsmen.

We have just above mentioned that out of 57 craftsmen, 15 (26%) are black. The uncontradicted evidence is that 3% of the minority in the labor market are qualified for such work. Applying this figure to those employed, it is apparent that 1.7 black persons should have been in craftsmen positions, yet 15 were so employed, more than eight times the number statistical nicety would have required.

There were no black officials and managers. 27 employees were in this category. 2% were available in the labor force, so the number of black managerial employees should have been .54. One black employee in this category would have more than compensated for the statistical imbalance.

There were 20 office and clerical employees of which 2 (10%) were black. 3% were available in the labor force, so a statistical balance would have shown .6 employee. The number actually employed was thus more than three times that required by statistical precision.

The above statistics taking into account the availability of blacks in the work force is from an EEOC report dated August 1, 1971. Since only one employee reported there and considered here was hired between March 31 and July 31, 1971, and that not a black employee, their relevance is apparent. See Patterson, infra.

Appellants contend that ESB's wage scale favored whites over blacks. The district court examined these contentions and rejected an attempt by the appellants to establish such a disparity by average wages of whites and blacks on a plantwide scale without regard to differences in level of skill, education and training. We agree with the district court that there is no discernible pattern of paying blacks less than whites in the same job classification or grade for the same work done. Indeed, leaving out craftsmen, the difference in pay of production employees in the plant is negligible, $2.23 for white and $2.17 for black employees in 1970.8

Nor is there any discrimination shown by the plaintiffs in ESB's providing training opportunities to its employees. The training programs were equally open to blacks and whites, and employees were encouraged to take part. No instance is shown of an employee, black or white, who tried to take part and was turned down on account of race or for any other reason.

We emphasize that the decision of the district court is binding only on the forty-four named plaintiffs. The ruling of the district court would have res judicata effect only as to those individual plaintiffs since the final judgment did not have any class application. Should other employees of the Sumter plant of ESB wish to bring another action charging racially discriminatory practices in employment, they may do so without any binding restriction being imposed on them on the merits of their claim by the decision in this case.

In this connection, we also note that even had this action been a class action at the time of judgment, it might very well have not been binding on the other class members due to the inadequacy of the representation. There is authority that before a judgment in a class action is held binding on the members of a class the representatives of the class should have performed adequately. As Wright & Miller state, basic notions of fairness and justice demand that absent class members be adequately represented if they are to conclusively bound by the result. 7 Wright & Miller, Federal Practice and Procedure, § 1765, pp. 617-18; see Hansberry v. Lee, 311 U.S. 32, 44-46, 61 S. Ct. 115, 85 L. Ed. 22 (1940).

The representation in this action was inadequate as to the class, as was observed by the district court. The evidence offered by the plaintiffs, other than that involved in the layoff, failed to prove but very little. Except for the layoff, the plaintiffs attempted to prove their case almost wholly on statistical data which was inconclusive, for the statistics which were offered tended as a whole to support the defendant's case as well as they did the plaintiffs'.

More importantly, what glares from the record is the failure of the plaintiffs to develop evidence of any consequence as to those parts of the employment situation at ESB (not involved in the layoff) which may arouse suspicion.

For example: Where did the employees come from who are employed in grades 10, 11, and 12, from the plant or from the outside? Where did the employees come from who are employed as supervisors, office, clerical and secretarial workers, from the plant or from the outside? Where did the employees come from who are employed in the tool and die and plant service departments, from the plant or from the outside? The implied allegation is that whites of equal seniority are in higher hourly pay grades, yet, despite much testimony, no evidence as to such claimed discrepancies was developed.

These unanswered questions, among many others which suggest themselves from the record, do not support an overturning of the findings of fact by the district court, nor do they excuse the plaintiffs' failure to prove their case. We conclude, however, that they do add to the district court's questioning of whether there was adequate representation of the broad class, and furnish added support for the holding of the district court that the broad class should have been limited because it was inadequately represented.

In some situations in which a final class determination is made after a significant period of time has elapsed under a conditional class order, the possibility of prejudice may warrant remedial action by the court. See 3B Moore's Federal Practice, 2d Ed. 1974, P 23.50; Frankel, Some Preliminary Observations Concerning Civil Rule 23, 43 F.R.D. 39, 42 (1967). Prejudice to dismissed parties in such cases may be in such forms as the running of a statute of limitation or the passage of time precluding the filing of an administrative complaint. However, there is apparently no such prejudice here, and none has been brought to our attention. Since the alleged discriminatory practices would be continuing in nature, it is doubtful if a statute of limitation would be a bar to contend with in bringing a later action. Nor, for the same reason, should there be any obstacle to the filing of a complaint with the EEOC. See Cox v. United States Gypsum Company, 409 F.2d 289, 290 (7th Cir. 1969); Sciaraffa v. Oxford Paper Company, 310 F. Supp. 891 (D.Maine 1970).

Indeed, it is likely that members of the broad class other than the named plaintiffs in this action will be able to bring a timely action based on alleged discriminatory practices in all events without a bar of a statute of limitations that had not run at the time this action was instituted. The Supreme Court has only recently stated that the rule in such a case is that "the commencement of a class action suspends the applicable statute of limitation as to all asserted members of the class who would have been parties had the suit been permitted to continue as a class action." American Pipe & Construction Company v. Utah, 414 U.S. 538, 554, 94 S. Ct. 756, 766, 38 L. Ed. 2d 713 (1974).

The district court noted in its opinion that the white office staff "could cause defendant problems in some future case or with different evidence before the court." The same might be said for the supervisory positions. The claim of the plaintiffs as to the predominance of white employees as craftsmen is ameliorated, so far as the statistics show, to a large extent by the fact that, except in particular jobs as opposed to the classification, there is little statistical imbalance, 26% of the craftsmen being black. But we note it is true that in some of the craft jobs there are no black employees. Remarkably, in one of the craft jobs, there is no white employee. These matters, however, are not before the court except on this record, and here, with the only real attempt at offering evidence made with respect to the layoff, the record supports the decision of the district court.

We express no opinion as to the merits of any other action which might be brought by parties other than the named plaintiffs in this suit.

Accordingly, the judgment of the district court is

AFFIRMED.

WINTER, Circuit Judge, dissenting.

In my view our prior holdings, especially those in Brown v. Gaston County Dyeing Machine Co., 457 F.2d 1377 (4 Cir. 1972), cert. denied, 409 U.S. 982, 93 S. Ct. 319, 34 L. Ed. 2d 246 (1972), and Barnett v. W. T. Grant Co., 518 F.2d 543 (4 Cir. 1974), which the in banc court does not purport to overrule, require us to conclude that plaintiffs proved a prima facie case of racial discrimination in employment. I cannot agree that when their principles are properly applied to the facts established in this record, ESB is entitled to exoneration. I also find error on the part of the district court in restricting the class. I would reverse and remand. I therefore respectfully dissent.

* In Brown, we clearly recognized and held that statistical evidence showing a marked racial imbalance in various categories of employment, as where blacks are excluded from higher-paying positions and are concentrated in those which are lowest-paying, established a prima facie case of unlawful racial discrimination in employment.1 We held further that an absence of objective criteria for hiring, promotion, transfer and pay increases, strengthens the prima facie case established by statistics demonstrating racial imbalance. Such absence "serve(s) to corroborate, not to rebut, the racial bias pictured by the statistical pattern of the company's work force." 457 F.2d at 1383.2 A prima facie statistical case, whether or not corroborated by an absence of objective standards, shifts the burden to the employer to prove lack of racial discrimination. Affected employees need not prove concrete and particular acts of discrimination toward them in order to make a prima facie showing, as the district court apparently thought.

In Brown, the factor we found compelling in the statistics was stratification of the work force. Slightly less than half of all black employees were confined to two jobs, both of which paid comparatively little and offered very small opportunity for promotion. At the other end of the scale, in jobs where pay ranged over $3.00 per hour, there were 102 white and three black employees. This stratification of the work force made out a prima facie case despite the fact that the overall percentage of blacks employed was not dramatically different from the percentage of black population in the vicinity.

Barnett is to the same effect. There, we permitted a warehouseman, occasional clerk and switcher, who was also a would-be over-the-road driver, to mount an "across the board" attack on all discriminatory actions by defendants on the ground of race, notwithstanding that we concluded that Barnett was not discriminated against when he was denied a transfer to over-the-road driver employment and that not all of defendant's acts of racial discrimination were directed to him. Notwithstanding the district court's contrary conclusion, we held, on the basis of statistical evidence and proof of the lack of objective hiring criteria, that Barnett had proved discrimination in violation of Title VII, and we directed a further evidentiary inquiry to pass upon the need for injunctive relief to obviate any continuing effects of past discrimination.

The factual proof in Barnett was that blacks were employed in only one of defendant's seventeen non-supervisory job classifications, that a warehouseman, and no blacks were employed in a supervisory classification. Defendant's eighteen black warehousemen represented nineteen percent of its total non-supervisory work force of ninety-five, and defendant's operation was in a community which was approximately twenty-five percent black. We thought significant the discrepancy between a twenty-five percent black community and a nineteen percent black non-supervisory work force, especially because all black employees were clustered in a single job category. This discrepancy, coupled with the total absence of blacks from the supervisory force, proof of a lack of objective employment policies, proof of discriminatory seniority provisions in the collective bargaining contract, and proof that at least one qualified black had been twice passed over for supervisory status in favor of whites, one of whom had less seniority, led us to conclude that discrimination had been proved.

While the statistics in Brown and Barnett probably showed a somewhat starker picture of discrimination than the statistics here, an analysis of the ESB work force by race shows a comparable pattern. The statistics here manifest racial stratification although the overall employment statistics for blacks are not strikingly disparate from overall black population statistics. The record shows that at its Sumter plant, the only ESB operation which is the subject of this litigation, the racial balance of the area from which employees are drawn was, according to the 1970 census, 57.7% white, 41.7% black, and .6% other. While blacks represented fifty-four percent of ESB's hourly employees in July, 1970, nearly ninety percent of ESB's hourly-rated black employees were in the lowest paying jobs, and none of ESB's Sumter top-management, middle-management, supervisory or clerical employees was black. There were one general foreman and twelve other foremen, all of whom were white. Of the fifty-eight highest-paid jobs in the top fifteen hourly classifications, fifty-three were held by whites. No blacks held skilled craft jobs. The Tool and Die Department was all white except for two janitors. The average pay rate of white hourly employees was $2.41 per hour; the average pay rate of black hourly employees was $2.11 per hour.3

In addition to the apparent racial discrimination reflected by these statistics, the record shows that ESB had no procedure for posting or notifying employees of vacancies or promotional opportunities. Also, it had no formalized procedure for employees to apply for promotion or transfers.4 All promotions were initiated by a white foreman and, unless a white foreman initiated the action, the employee would have no chance at all of being promoted. Prior to June, 1970, there were no written standards or procedures to guide the foremen deciding whom to select for promotion or transfer.

Although ESB had a stated policy of promoting from within wherever possible, there were, within the period August 9, 1965 to January 1, 1970, sixty-nine persons hired into the Quality Control Department and all were white except four. Many of the salaried positions were Inspector positions, and as of August 1, 1971, ESB had six Inspectors in the Quality Control Department, all of whom were white.

Between 1965 and 1969, twenty-one whites were promoted from hourly to salaried positions. Only after a complaint was filed with EEOC in 1969 did ESB promote the first black to a salaried job. Promotions from the lower labor grades to higher labor grades show the same pattern. From August, 1969, to September, 1972, sixteen whites and only one black were promoted from lower labor grades to labor grades 11 and 12. On July 24, 1970, there were thirty-seven whites and only one black in the three highest labor grades 10, 11 and 12.

Finally, the record establishes that ESB's employees included some exceptionally well-qualified black employees and that many of its black hourly employees had more seniority than white employees who were promoted to higher positions. Typical of the former is the fact that six of the named plaintiffs had completed a considerable amount of college work; two had previously been employed as school teachers, of which one had prior clerical experience in a department store. One plaintiff had completed several courses in typing and bookkeeping at a business college. Where the named plaintiffs had been evaluated by their supervisors, they received mostly high ratings. With respect to seniority, the record reflects that in the Positive Plate Department which had fifty-seven black employees and twelve white employees, ten black employees had more company seniority than the white Leader. In the Negative Plate Department which had eight black and two white employees, four black employees had more company seniority than the white Leader. In the Electro-Mechanical Department which had twenty-six black employees and twenty-one white employees, eight black employees had more seniority than the Group Leader and the Leader. In the Parts Plating Department which had seven white employees and seven black employees, six of the black employees had more seniority than the white Leader. ESB offered no evidence to establish that the white employees were better qualified.

Consistent with Brown and Barnett, I think the conclusion inescapable that plaintiffs proved a prima facie case of racial discrimination in employment at ESB's Sumter plant. Not only does the statistical proof reflect the pattern of clustering and limitation of black employees to the less attractive lower-paying positions found where there has been racial discrimination in fact, the proof also shows the lack of objective standards under which it is so easy and so likely that white supervisory and white management personnel may consciously or unconsciously discriminate against black employees because of their race. Finally, the proof shows that, despite the purported policy of promotion from within, few blacks were promoted and a not insubstantial group of black employees with significant qualifications for better positions or with greater seniority than more highly placed white employees have not realized their apparent potential for promotion. This proof corroborates the inference, to be drawn from the statistics and lack of objective standards, that race has been a factor in promotions. The burden of proving that racial discrimination was not practiced in fact clearly shifted to ESB. The district court was significantly in error in concluding that plaintiffs were required to offer proof of individual instances of discrimination in order to entitle them to relief; the majority perpetuates this error.

Nor could the district court, consistent with Brown, properly have found that ESB's proof was sufficient to overcome plaintiffs' prima facie case. The district court did not require ESB to show, nor did ESB attempt to show, the actual qualifications required of its white employees selected for supervisory, clerical, salaried, and higher-paying, hourly-rated positions, and whether the white incumbents which had been selected to fill them were better qualified than blacks. And this failure was in the telling context of affirmative proof that job vacancies were not posted and promotions were initiated exclusively by ESB's supervisory personnel. At most, ESB sought to prove that, in accordance with its Employee Handbook, its supervisory personnel made a good-faith, subjective effort to apply such factors as willingness to accept responsibility, dependability, and ability to get along with others in determining promotions to the higher-paying hourly, or the salaried, positions. These are standards found to be utterly inadequate in United States v. Jacksonville Terminal Co., 451 F.2d 418 (5 Cir. 1971), as well as by us in Brown; Moody v. Albemarle Paper Co., 474 F.2d 134, 139 (4 Cir. 1973); reversed on other grounds, 422 U.S. 405, 95 S. Ct. 2362, 45 L. Ed. 2d 280 (1975); and Barnett. Were the case to be finally decided on the present record, there is no question in my mind that plaintiffs would merit substantial relief.

From what has been said, it should be manifest that I neither read the record as does the majority, nor do I place the same significance on what it unquestionably contains. I view the case as one in which greater significance should be placed on the facts existing at the time the complaints of racial discrimination were lodged than on what ESB may have done subsequently in a cosmetic effort to improve its defense. Particularly is this so when the type of past racial discrimination which the record reflects can be expected to have continuing deleterious effects. While subsequent efforts to redress discriminatory practices may have great relevance to the fashioning of relief, they have little bearing on whether plaintiffs have proved a prima facie case. Barnett, 518 F.2d at 550. The suggestion that an unrebutted prima facie case does not require relief, it only permits relief, is a concept unsupported by authority and one that I totally reject.

Although the majority is ambivalent as to whether to place its decision on the ground that plaintiffs failed to make out a prima facie case or whether, having done so, plaintiffs' case was rebutted, I disagree that in either event the district court's conclusions can be sustained as not clearly erroneous. First, the district court's conclusions were findings of ultimate fact to which the not clearly erroneous test is inapplicable, Causey v. Ford Motor Co., 516 F.2d 416 (5 Cir. 1975); and second, in requiring proof of specific acts of discrimination to warrant relief, the district court applied the wrong test so that its ultimate findings may be rejected and ultimate findings may be made anew. Finally, I think that the portion of our decision in Patterson v. American Tobacco Company, 535 F.2d 257 (4 Cir. 1976), relied on by the majority, has no application to the instant case at its present stage. That part of Patterson was concerned with the fashioning of appropriate relief to redress discrimination in the selection of supervisors; it had nothing to do with what proof constitutes a prima facie case. Of course, if, in rebuttal of plaintiffs' prima facie case, ESB had offered proof that it had no black supervisors because none were available in the area from which the work force was drawn, the proof would be highly relevant. ESB did not do so and it is a subversion of Patterson to suggest that it placed the burden of proving the availability of qualified black supervisors on plaintiffs.III

The present litigation arose when ESB, in 1970, substantially reorganized its Sumter plant. On July 24, 1970, eighty-three of its 365 hourly employees were discharged. Of those laid off, fifty-three, or approximately sixty-three percent, were black. Immediately prior to the layoff, fifty-four percent of ESB's hourly employees were black. Layoffs were made in accordance with a policy set forth in the Employee Handbook, but a substantial factor in the policy was seniority in the job classification where a reduction in personnel was to be made.5 It is at once readily apparent that if racial discrimination in promotions in fact was practiced, its effect was to confine blacks to the lower-paying classifications most vulnerable to layoffs. Thus, even if benignly made,6 layoffs, too, would constitute a form of illegal racial discrimination in employment.

Plaintiffs were victims of the layoff and sought to maintain their action as a class action for a class defined as all black applicants for employment and all present and former black employees of ESB's Sumter plant. Broadly stated, their theory of their case was that since they had been restricted to positions vulnerable to discharge because of racial discrimination, they, individually and in a representative capacity, could attack the entire pattern of racial discrimination. Thus, fifty-three, or sixty-four percent of the eighty-three discharged employees, came from six departments; seventy-five percent of the employees in these departments were black. Because the case arose in the context of layoffs, the district court erroneously limited the purported class to those laid off on July 24, 1970, and, after eliminating certain named plaintiffs for failure to respond to discovery procedures,7 concluded that the class was not so numerous that actual joinder was impracticable, with the result that the district court denied class status to the suit. Signficantly, the ruling that the action could not be maintained as a class action was not made until after the close of the evidence in a trial which had proceeded on a conditional order that it was a class action. I think that the district court abused its discretion and its decision is inconsistent with our holding in Barnett. As previously stated, in Barnett we permitted an "across the board" attack on discrimination by a representative of the class whose only complaint of specific discrimination was that he was not promoted to a position of over-the-road driver for which he was found not qualified. We reached this conclusion because we thought it "more consonant with the broad remedial purposes of Title VII . . . and that the district court's less charitable view, under which Barnett could as a class representative challenge only those specific actions taken by the defendants toward him, would undercut those purposes." 518 F.2d 547-48.

Of course the significant event preceding the litigation was the July, 1970 layoff, but it was simply the latest manifestation of an overall pattern of discrimination against all black employees which the statistical evidence and evidence of lack of objective standards prima facie established. In complaining about the layoff, the named plaintiffs of necessity were obliged to prove the overall pattern. At least 197 black employees were concerned with the overall pattern. Manifestly, they had a community of interest. For a named plaintiff to prevail, he must of necessity prove the case for each member of the class, and, thus, I have no doubt about his ability to represent the class. This case is an even stronger one than Barnett in which to define a broad class. Manifestly, also, the members of the class were too numerous to require individual joinder and to deny class action status.

Unlike the majority, I cannot approve restriction of the class on the theory that there was not an abuse of discretion. I think discretion, to the extent present, was abused.8

Thus, I would conclude that the district court's ruling restricting the class should be reversed and it should be instructed to certify the class as requested by plaintiffs. On the merits, I would vacate the district court's judgment and remand for further proceedings. Its rulings do not comport with our subsequent decisions in Brown and Barnett and failed to recognize that plaintiffs had proved a prima facie case which shifted the burden of disproving racial discrimination to ESB. The district court should reappraise the evidence and make findings of fact anew in the light of Brown and Barnett. It should also receive any additional evidence from ESB, and rebuttal evidence from plaintiffs, that ESB might adduce to meet the burden of proof which our decisions imposed on it.

Judge CRAVEN and Judge BUTZNER authorize me to say that they concur in this dissenting opinion.