This Title VII case raises claims that the very substantial, good faith efforts of the employer and union working together to end racially segregated working practices still fall short of eliminating the present effects of past discrimination for many black workers. The central attack is on a continuation of departmental seniority. The district court after a full hearing prepared a detailed memorandum opinion reasoning that the steps taken constituted sufficient compliance with Title VII. Principally because that result was based on a prima facie case and other burden of proof concepts which did not fit this case, we vacate the judgment appealed from and reverse for further proceedings, including reconsideration of appropriate backpay and other relief.

Pullman-Standard, a division of Pullman, Inc., is the world's largest manufacturer of railway freight cars and parts. Operations at its Bessemer, Alabama plant are geared to special orders, rather than the production of an inventory, and accordingly are conducted much like a custom steel fabricating shop, though on a larger scale.2 This method of production, coupled with sporadic market demands from the railroad industry, results in frequent and dramatic fluctuations in the level of employment.3 Since the early 1940's, most4 of the production and maintenance workers at Pullman-Standard's Bessemer plant have been represented by the United Steelworkers, which also holds representation status at Pullman-Standard's other three plants; and key provisions dealing with seniority rights are largely covered by local rules at the individual plants rather than by the company-wide triennial collective bargaining contracts.

The production and maintenance jobs at Pullman-Standard, over a hundred in number,5, are divided among 25 different United Steelworkers departments of varying sizes.6 In addition to these departments there are two Machinist Union departments and an additional United Steelworkers department having only one employee. All assignments to departments are made by Pullman-Standard. Each job has a specified job class (JC) level, which determines its relative ranking in base pay in comparison to other jobs.7 All but two departments, Janitors and Template, have more than one job classification; and most, but not all, classifications will be worked by more than a single employee during peak employment periods. One such job (Welder) sometimes is worked by over 500 employees.

Under the local rules at Pullman-Standard, seniority is measured by continuous service in a particular department and is exercised in competition with all other employees in that department, there being no lines of promotion or progression in any department. Seniority rosters are maintained by department; and departmental age is basically the sole criterion used to determine who is rolled back or laid-off in the event of reductions, and who is recalled or promoted (assuming ability to do the work) in the event of force increases or other vacancies, in the department. It is somewhat inappropriate to talk about "permanent jobs" at Pullman-Standard, except perhaps with respect to the most senior employees in the department; for the constant fluctuations in job requirements and employment levels cause frequent movement of employees from one job to another. While the seniority rosters do indicate a job classification for each employee, these designations do not reflect his8 permanent job, but rather constitute a recognition by the company that the employee has satisfactorily performed the job and is thereby protected under the collective bargaining agreement against rejection in favor of a junior employee on the factor of relative ability. The rosters are updated annually as of June 1st; and frequently only the highest job class which an employee has satisfactorily performed is shown for him.

From the study of payroll information for the 18-month period ending June 1974, it appears that 49.5% of the work force is black, a ratio which is comparable to that which existed in June 1965.9 Understandably, plaintiffs, do not assert that Pullman's initial hiring policies, as such, are now or ever have been racially discriminatory.10 Plaintiffs do, however, contend that assignment of new hires to the several departments was discriminatory, and continued so for several years following the passage of Title VII. Further, the plaintiffs contend that prior to June 1965 a number of the better jobs, including supervisory positions, were "white only" and a number of the lower-paying jobs were "black only." Pullman's old records, quite incomplete, do reflect a mixing of the races in some of these jobs in the 1920's and 30's. Nonetheless, it is clear that by the late 40's many of the jobs had become racially segregated, and remained so into the mid-60's, not by formal agreement to that effect, but under a custom and practice which the company condoned, if not approved.

In early 1965, spurred by an arbitration decision which opened up the then all-white Rivet Driver Job to blacks,11 the company began implementing a program to eliminate barriers to advancement by blacks and, in general, to conform to the impending requirements of Title VII where possible infractions were detected. Black buckers and welder helpers were given trials to ascertain their abilities as rivet drivers and welders, respectively. Beginning in June 1965 black employees were appointed as hourly foremen. A reporting system of hires and promotions was undertaken, as were efforts to recruit blacks for the more highly skilled positions. In early 1966, an agreement was made with the union to utilize seniority in the filling of temporary vacancies.12 Negotiations were commenced in 1968 with the Department of Labor, Office of Federal Contract Compliance (OFCC), which led in January 1969 to a conditional memorandum of understanding (OFCC agreement) designed to enhance opportunities for blacks.13 Although this memorandum never became fully effective due to lack of union acceptance, Pullman-Standard through its Contract Compliance Officers one black and one white began encouraging black employees in certain "low ceiling" departments to transfer to other departments and monitoring the filling of temporary vacancies to assure a fair allotment to black employees. Black employees and their families were encouraged to participate in vocational education at the company's expense, a program that has been particularly significant in the training of black welders.14

In May 1972, Pullman-Standard entered into an agreement with the Department of Labor to serve as a corrective action program and to bring its employment practices into compliance with Executive Order 11246 as amended.15 Of the many provisions in the 25-page OFCC agreement, the most significant to the issues in this case are those relating to interdepartmental transfers. Black employees with employment dates prior to April 30, 1965, are given preference for vacancies arising in the five traditionally all-white departments (Template, Powerhouse, Airbrake Pipe Shop, Inspection, and Plant Protection); and those hired before April 30, 1965, who had been assigned to four "low ceiling" departments (Janitor, Die & Tool, Truck and Steel Miscellaneous) are given preference for vacancies arising in any of the departments.16 These transfer rights are without limit as to time; and, when exercised, the employee vies for promotions in the new department using his total plant age. For layoff and recall purposes, the employee is given at the time of the transfer the option either to take his plant age into the new department or to keep accruing seniority in his old department while building new age for such purposes in the new department. Retreat rights to the employee's prior job are provided should he fail to qualify for at least a JC 4 job or is disqualified for a promotion in the new department; and in such event the employee may have the right to go to another department rather than return to his original one.

The trial court had before it for determination four claims of class discrimination17 and claims of individual discrimination by two employees. The class issues were:

(1) Does the system of departmental seniority, even with the changes made under the OFCC agreement, perpetuate the effects of past discrimination in the assignment of black employees among the various departments?18

(2) Has there been discrimination in the assignment of work to persons having the same job title or in the assignment of jobs having the same JC level to the persons entitled to jobs of such JC level?

(3) Has there been discrimination in the promotion of persons to supervisory positions?

(4) Has there been discrimination in the failure to post publicly a list of changes in assignments?

One of the named plaintiffs, Louis Swint, and an intervenor-plaintiff, Clyde Humphrey, each claim that their discharges by the company in May 1971 and November 1972, respectively, were racially motivated or were in retaliation for their having earlier filed charges with the EEOC.19

The district court upheld the use of departmental seniority at Pullman-Standard on the premise that this practice may only be found to have perpetuated the effects of past discrimination when there has been a showing of prior discrimination in the assignment of black employees among departments by exclusively or disproportionately assigning white hires to higher-paying departments and black hires to lower-paying ones. This general principle, though correct in the abstract, cannot be translated into an absolute requirement that a plaintiff prove economic harm before he has established a prima facie case of racial discrimination in hiring assignments. Because the effect of the district court's approach improperly placed that burden on plaintiffs in this action, it found no unremedied discriminatory hiring assignments. Its conclusion based on the assumption that the plaintiffs had not shown that they were assigned to less desirable departments must be reversed and the issue remanded to the district court for reconsideration and resolution in a manner not inconsistent with this opinion.20

The lower court concluded that: "With the exception of nine departments, the evidence does not indicate any past or present policy of racially discriminatory assignments." This conclusion primarily rested upon the district court's finding that although in 1965 only five departments had racial compositions approximately equal to the ratio for the total of all departments, "there was no pattern of favoritism to whites in the departmental assignments." At the heart of this finding is a chart constructed by the lower court which ranked each department according to job class range.21 The chart reveals that a greater percentage of blacks than of whites were employed in 1965 in departments with higher job class ranges. Because the chart is crucial to the district court's approval of departmental seniority, we must carefully and completely analyze the attacks made on it by plaintiff.A. Was the Chart Wrong? The Clearly Erroneous Attack

Plaintiffs contend that the theory and use of the chart give an incomplete and misleading impression and that fact findings based on it are clearly erroneous. The chart is designed to show departmental desirability. It accumulates the percentages of blacks and whites in each department in three ways. Column I shows all Steelworker departments except Boilerhouse, Column II additionally excludes the nine departments affected by the OFCC agreement, Column III further takes away the maintenance and welding departments. The accuracy of such an accumulating percentage chart directly depends upon the accuracy of the departmental rankings. Plaintiffs argue that the ranking of departments in the chart is arbitrary and inaccurate for the following five reasons:

1) Departments were ranked by job class range.22 Plaintiffs argue that the key factor, economic desirability, is much more a function of the number of employees who work in each job class than of a department's job class range. For example, the chart would rank a department with job class ranges 1-10 above a department with ranges 1-9; yet, a department with 100 workers in JC-1 and five workers in JC-10 is clearly not as desirable as a department with five workers in JC-1 and 100 in JC-9. Plaintiffs' propose that any charting of desirability should rank departments by median job classes.23 Not surprisingly, plaintiffs' proposed charting shows that blacks were disproportionately assigned to less desirable departments. Pullman-Standard replies that medians are also inaccurate indicia unless substantiated by various statistical techniques (such as standard deviations) which have not been supplied in this instance.

2) The plaintiffs contend the court arbitrarily excluded higher, sparsely populated job classes in some departments. The Welding Department has a JC-16 job class which has less than 1% of the department's employees; therefore, the court excluded JC-16 from the Welding Department's job class range. If the JC-16 job class were included in the Welding Department job class range, the department would rise from rank 16 to rank 4 on the chart. Since Welding is the largest department and one in which whites are disproportionately represented, the accumulated percentages would be changed significantly by this move. Plaintiffs argue that the Wood-Mill, Wood-Erection, Paint, and Steel Construction departments all have less than 1% of their employees working in the top job class; however, the 1% job classes are included in those departments' ranges. The last three departments are particularly important because they are disproportionately black departments which by across-the-board equal treatment would be shown to be less desirable than presently reflected by the chart. Pullman-Standard, however, logically argues that the special statistical treatment given the Welding Department is entirely justified because the JC-10 to JC-16 jump (from the highest job class listed to the unlisted 1% job class) is much greater than the jump in the four departments where 1% job classes were included. Also, Pullman-Standard contends Welding is by far the largest department and deserved the special consideration given by the court's Column III to ensure that it did not unfairly skew the chart.

3) Plaintiffs object to the exclusion from the chart of the Die & Tool IAM, Maintenance IAM, and Boilerhouse departments. Two of these departments are covered by the Machinists instead of the Steelworkers Union. Therefore, they do not have job class ranges and could not easily be assigned ranks on the chart. However, no attempt at even estimating relative rank was made by the court. Boilerhouse is a Steelworker's department whose only worker has always been a JC-10 white employee. Plaintiffs argue that these departments are mostly segregated and are higher paying; therefore, their exclusion directly affects the chart and the conclusions drawn from it.

4) Plaintiffs point to departments which have the same job class ranges. These departments are arbitrarily assigned a rank on the chart even though they are nominally equal. The most obvious example is right in the middle of the rankings where five departments with ranges of JC-1 to JC-11 are listed in order. These are large departments which total 1/3 of all workers. Two of these departments are heavily black. One contains more blacks than any other department. The ranking of these departments among the five could affect conclusions drawn from the chart.

5) Plaintiffs assail the chart's accumulating percentage columns. Plaintiffs are especially concerned with Column III which provides the strongest support for the court's findings. The exclusion of Welding24 and Maintenance was made because of special technical skills needed in some of the job classes in these two departments. But, nowhere in the opinion is there the kind of business necessity finding which would be essential to allow a special-needed-skill exclusion to keep these departments from being considered.25 Also, there is no finding of the nature of these needed skills nor whether they are necessary only in the higher job classes. Finally, if the skills are only needed in the higher job classes, there is no finding as to whether they could be gained while working in the lower ones, either with or without additional special training.26 The court also found that plaintiffs had failed to prove that blacks were qualified to enter either department. This, however, reflects an incorrect placement of the burden of proof. The question of absence of qualifications is relevant, but only after the determination that a pattern of discrimination has been shown; and then, the burden is on the defendant to show the absence of qualifications.27

Prior to the desegregation of the Welder job class in this department, whites were able to take a test (usually after having served as a welder's helper and learning the skill) without either prior formal training or experience to determine whether the employee was qualified to be a welder. Since blacks have been allowed to become welders, the company has required that only those persons with either prior experience or formal training could take the qualifying test. Plaintiffs argued that this should be considered as a facially neutral requirement which in reality operates to perpetuate past discrimination. Since there is some question as to whether this final issue was squarely presented below and since its resolution would not control our decision, we pretermit its resolution.

The concept of plaintiffs' attack on the district court's conclusion that blacks were not disproportionately assigned to less desirable departments based on the chart's deficiencies is framed in terms of a clearly erroneous factual determination. It fails because the district court did not treat the chart as plaintiffs contend. The court made two factual findings relevant to this inquiry. First, the court found that the chart shows that blacks were not disproportionately assigned to less desirable departments. This factual finding is correct to a mathematical certainty. Second, the lower court found that the chart was a "rough index of earnings potentials in the absence of other more reliable data, such as average earnings or average job class worked." In other words, the district court recognized that the chart had problems. Its "rough index" conclusion also is not clearly erroneous. The reason plaintiffs' clearly erroneous attack miscarries is that they misread the court's ultimate legal conclusion as resting upon a factual finding that the chart was both accurate and not arbitrary.

Despite our holding that the district court did not make an erroneous factfinding in its direct references to the chart, it is clear that plaintiffs valid attacks on the structure of the chart as discussed in detail above do impugn the credibility of the court's ultimate conclusions based on departmental desirability.

B. Less Desirable Assignments Prima Facie Case

Plaintiffs in successful Title VII cases have uniformly been able to establish that their assignments produced diminished wages.28 When today's proof was ruled insufficient to demonstrate loss, a novel question was squarely presented. To resolve it requires resort to the statutory language and case analysis of legislative purpose. When this path of reasoning is followed it leads to the conclusion that a Title VII plaintiff does not have to show economic loss to prove discrimination.

The district court concluded that plaintiffs' failure to show that class-wide economic harm resulted from the departmental assignments made meant that no prima facie case had been established. This legal deduction is inconsonant with the Act and its interpretative cases. The Title VII gravamen of the complaint is that departmental seniority perpetuated past discrimination. The key for this case is whether there was past discrimination, i. e., that Pullman-Standard discriminated on the basis of race in making assignments to departments. Going further and requiring plaintiffs to prove that past assignment practices produced lower pay checks is contrary to law and precedent. The district court relied on Franks v. Bowman Transportation Co., 495 F.2d 398 (5th Cir. 1974) and Johnson v. Goodyear Tire & Rubber Co., 491 F.2d 1364 (5th Cir. 1974) for its conclusion that economic harm from departmental assignments must be part of plaintiffs' prima facie proof. This reliance is misplaced. These cases and our contemporary decision in Pettway v. American Cast Iron Pipe Co., 494 F.2d 211 (5th Cir. 1974) are cases in which the plaintiffs had clearly demonstrated that the defendants had assigned them to lower paying positions. When we recognized that the proof of such lower pay confirmed the existence of proscribed discrimination we did not mandate this element as a part of the prima facie case proof every Title VII plaintiff must make.

Title VII contains neither requirement nor implication that economic harm must be shown before a class can be found to have made out a prima facie case of racially discriminatory job assignment. Indeed, the statutory prohibitions of the enactment are explicitly broader than economic harm.

It shall be an unlawful employment practice for an employer

(1) to fail or refuse to hire or to discharge any individual, or otherwise to discriminate against any individual with respect to his compensation, terms, conditions, or privileges of employment, because of such individual's race, color, religion, sex, or national origin; or

(2) to limit, segregate, or classify his employees or applicants for employment in any way which would deprive or tend to deprive any individual of employment opportunities or otherwise adversely affect his status as an employee because of such individual's race, color, religion, sex, or national origin.

42 U.S.C. § 2000e-2(a)(1) and (2).

In the seminal departmental seniority case, Quarles v. Philip Morris, Inc., 279 F. Supp. 505 (E.D.Va.1968), the court made the general observation that employees in the predominately black department were usually paid less than employees in the two predominately white departments. However, a comparison of the wage ranges in all three departments showed one white department offered more and one less pay than the predominately black department.29

Notwithstanding this wage range proof and the fact that Quarles was specifically requesting a transfer to a department with a lower range than the department in which he worked, the court found that past discriminatory departmental assignment practices were perpetuated by departmental seniority. While Quarles does not explicitly hold that a prima facie case was made without showing economic harm, it necessarily ruled that a prima facie case was proven even though wage or job class range showed the predominately black department was not the least desirable.

In United States v. N. L. Industries, Inc., 479 F.2d 354 (8th Cir. 1973) the appeals court reversed the trial court's refusal to find that a Title VII violation had been established. The district court had placed on plaintiffs the burden of proving that the Labor Department was the least desirable department, then held that the burden had not been met by proof which showed racially disproportionate assignments of blacks to the Labor Department despite the fact that many jobs in that department paid less than jobs in most of the plant's predominately white departments. The Eighth Circuit reversed this burden placement. In a footnote it declared that where departmental desirability might be relevant defendants must carry the burden of showing blacks had been assigned to a more or an equally desirable department. Speaking more broadly, the court relied on our decision in United States v. Hayes International Corp., 456 F.2d 112 (5th Cir. 1972) (Hayes II ) to hold that it was more important for the court to assure discriminatorily assigned employees had the opportunity to bid for jobs denied them because of race than to determine subjective factors of departmental desirability. Although the appeals court's principal determination was that the subjective evidence relied on did not establish the department to be the plant's most desirable place to work, its opinion leaves no doubt that departmental desirability is not an essential part of a plaintiffs' prima facie proof in a case such as this.

In Hayes II this court reviewed a lower court's determination that a departmental transfer system cured all possible perpetuation of past discrimination. The plaintiff United States unquestionably had established a prima facie case of racially discriminatory departmental assignment and as part of that proof had shown general economic harm to the affected class. However, one substantial issue in Hayes II concerned the defendant's contention that justification for not including several departments in the group of departments affected class members could transfer into could be found in the fact that the wages in those departments were "generally lower" than the departments in which affected class members were presently employed. The court ordered that qualified blacks be given an opportunity to transfer into these economically less desirable departments because "the Civil Rights Act provides for equal opportunity to select and compete for a job notwithstanding its lower pay or other disadvantages." Hayes II, 456 F.2d at 118.

In Reed v. Arlington Hotel Co., Inc., 476 F.2d 721, 723 (8th Cir. 1973), the Eighth Circuit concluded simply that "statistics which show segregated departments and job classifications establish a violation of Title VII." Absent a single statement regarding either objective or subjective economic desirability the court found a Title VII violation. The court expressed concern because the blacks had suffered "the indignities of segregation" and ordered complete relief including transfer rights and departmental merger. Id at 726. This case clearly indicates that economic harm is not a necessary element of a prima facie case.

The same district judge in United States v. United States Steel, 371 F. Supp. 1045 (N.D.Ala.1973), after finding that a prima facie case of racial discrimination had been made, presumed damages to a particular subclass for purposes of injunctive relief.30 The judge then denied back pay in U.S. Steel because plaintiffs had not proven harm caused by discrimination.

The focus of our appellate reversal of this latter burden assignment in U.S. Steel was not on whether sufficient evidence had been presented to establish a prima facie case of discrimination nor on the presumed damages for injunctive relief, but rather focused on the erroneous assignment to plaintiffs of the burden of presenting evidence sufficient to justify a backpay award to the class.

The backpay issue must of course face the causal relation between discrimination and proof of economic harm. We make reference to this prior action not only because of its presumption of damage from racially skewed assignments, but also because it held that economic harm was not even a required part of a backpay prima facie case unless the defendant had convincingly brought the issue into question.

A holding that plaintiffs here may be entitled to relief absent showing harm to their pocketbook is analogous to the situation which exists in Title VII failure-to-hire cases. Courts uniformly award class-wide relief without any showing that other jobs available in the community were less desirable than the jobs the defendant refused to make available.31 Discriminatory-departmental-assignment Title VII cases are neither more nor less than failure-to-hire cases. The failure to hire here is a failure to hire into a particular department.32 Just because the other job the plaintiff received when he was discriminatorily prevented from being hired into one department was in the same plant is not a sufficient reason for increasing the plaintiffs' prima facie burden of proof.

Title VII of the Civil Rights Act prohibits all forms of racial discrimination in all aspects of employment. The degree of discrimination practiced by an employer is unimportant under Title VII. Discriminations come in all sizes and all such discriminations are prohibited by the Act.

Rowe v. General Motors, 457 F.2d 348, 354 (5th Cir. 1972).

The plaintiffs in the instant case were found to have failed to establish a prima facie case, for both injunctive relief and backpay purposes, because they did not show economic harm i. e., that they were disproportionately assigned to less desirable departments. Quarles, N.L. Industries, Hayes II, Reed, U.S. Steel, and Rowe all lead us to the conclusion, buttressed by the broad statutory language, that, at least for the purposes of injunctive relief, plaintiffs need not show that they were assigned discriminatorily to less desirable departments in order to prove a prima facie case of racial discrimination.

For purposes of backpay relief, U.S. Steel holds that economic harm is not required to be shown as an element of a prima facie case unless the defendant has shown "convincingly" by "statistically fair exhibits" that the class earned "at least as much as a plant-seniority comparable group of whites." The district court's desirability finding was based on a "rough index" chart which it used in "the absence of other more reliable data." As demonstrated above the chart contained patent inaccuracies. It does not amount to a convincing showing by a statistically fair exhibit of earnings equality.33 In addition, the chart did not even attempt to correlate seniority-comparable workers. In the instant case, therefore, the defendant's evidence did not "draw into substantial question the group's entitlement to (backpay)."34

Without attempting to lay down a rule for all cases, it was error to require plaintiffs to prove economic harm as an element of their prima facie case of racially discriminatory departmental assignment.35

C. Scope of the Prima Facie Inquiry on Remand

Since the district court was in error when it ruled that plaintiffs had failed in their proof the case must go back for further consideration. Errors apparent in prior procedures indicate that we should now define the prima facie inquiry to eliminate the likelihood they would recur on the reconsideration we now mandate. It would be inappropriate for this court to decide here whether a prima facie showing was made. Not only do we not decide it, but also we do not intimate how it should have been resolved. This evidence-weighing process should be conducted by the district court in the first instance, especially since the question is so obviously a close one. We emphasize that despite our directions as to procedures and parameters on remand, we intimate no decision as to what the district court's ultimate conclusion should be.

Plaintiffs argue that previously the court limited its inquiry only to Steelworkers Union departments, exclusive of such departments as were covered by OFCC, plus Welding and Maintenance. Also, they contend that neither pre-'65 policies of discrimination nor statistics showing overall lower job class assignments to blacks were sufficiently considered. Finally, they claim undue weight was given to a 1972 yearly wage comparison. The trial court's opinion does not make it apparent that the weighing of this evidence ever actually took place (although several elements were discussed). The lack of proof of harm necessarily sidetracked the prima facie case conclusion.

The most important aspect of determining whether a prima facie case has been proven is to identify the proper scope of the evidentiary examination. The authorities speaking to this scope all conclude that the question should be whether the class has established a history of broad patterns of plant-wide racially discriminatory departmental assignments.36

(T)he class must demonstrate a prima facie case of employment discrimination. Sometimes statistical evidence alone will suffice; on other occasions live testimony or additional exhibits may be necessary. At all events, however, the stress . . . is upon demonstration of the defendant's broad employment policies and practices, the defendant's rebuttal and business necessity defenses, and the inferences which remain at the close of the evidence.

United States v. United States Steel, 520 F.2d 1043, 1053 (5th Cir. 1976); citing, United States v. T.I.M.E.-D.C., 517 F.2d 299, 315-16 (5th Cir. 1975); Rodriguez v. East Texas Motor Freight, 505 F.2d 40, 53-55 (5th Cir. 1974); United States v. Hayes International Corp. (Hayes II), 456 F.2d 112, 120 (5th Cir. 1972).

During the class-wide inquiry, commonly referred to as phase one,37 the court must also define the limits of the class actually entitled to relief and determine the relief necessary. Composing the makeup of the affected class is usually one of the more difficult Title VII tasks.38 The temptation to expedite the handling of a complicated Title VII case such as the one before us here urges a court to combine the determination of the prima facie case phase and the delineation of the affected class. These are separate problems and must be handled separately.39 The combination of a prima facie case determination and an affected class limitation inevitably results in a narrowing of the broad prima facie case inquiry or broadening of the affected class delineation. Neither result is permissible. Title VII adjudicatory procedures are complex. In this maze of shifting of burdens and first-narrow-then-broad areas of trial court concern, it is particularly important that a court proceed only one step at a time.

The first inquiry should be whether a prima facie case has been shown. In the instant case that requires us initially to focus on whether a history existed at Pullman-Standard of broad patterns of plant-wide racially discriminatory departmental assignments.

(1) Disparate Department Assignments

Although there are 28 departments at Pullman-Standard, the court's opinion and the chart from which it draws its major conclusions did not consider the two Machinists Union departments, the single-member Boilerhouse department, the departments covered by the OFCC agreement, and the Welding and Maintenance departments. In addition, the court found five of the remaining departments to be fairly statistically balanced. Thus, under its decisional framework, what was left was to decide whether a showing that nine departments are racially imbalanced makes out a prima facie case of racial departmental assignment. On remand, this entire analysis must be discarded. The appropriate evidentiary examination should focus on the plant as a whole to determine whether there have been racially discriminatory assignments.40

Many courts have found departmental assignments to be discriminatory where statistics showed absolute or almost absolute one-race departments.41 The inference of discriminatory assignment in such complete exclusion or only token inclusion situations is much more direct than we have here.42 But it is not necessary to show complete or substantially complete segregation by departments.43 When the departments are racially mixed but simply do not represent the plant-wide balance, the inference of discriminatory assignment is much more difficult to draw. If the focus were only on the statistical racial disparity in nine departments, one could not conclude that a prima facie case existed. However, these statistics need not be viewed in this sort of vacuum. There exists a great deal of other evidence in the record which must also be entered onto the balance. It is this additional evidence which must be considered in the first instance by the trial court so that the balance can be struck according to the inferences that court chooses to draw.

(2) OFCC Departments

In its initial prima facie case inquiry the court examined only those departments not covered by the OFCC agreement. The logic of its reasoning was sound, as far as it went discriminatory assignments which are shown by these departments' statistics showing heavy segregation have been cured by OFCC.44 But, the scope of the court's examination should have continued to be whether there had been past discriminatory racial assignments instead of stopping at answering whether the most glaring past discriminatory racial assignments were no longer frozen in by department seniority. The fact that discriminatory assignments were made to these nine departments is clearly probative on the issue of whether plant-wide racially discriminatory departmental assignments existed. In this case where percentage discrepancies could indicate subtle discrimination or no discrimination in the remaining departments, the elimination of the unmistakable discrimination in these nine departments skews the analysis. An inference can be drawn from this proof that discrimination in departmental assignments stopped at these most segregated rosters. Another conclusion is that every department was involved. Whether either of these or some inbetween conclusion should be drawn will most probably depend on the other evidence presented and is for the trial court. But, the evidence must be assayed for this purpose to keep the process right.

(3) Machinists Union Departments

Similarly, the racial makeup of the Machinists Union department should also be considered in the prima facie balance. Pullman-Standard made assignments to these departments as it did the others. Although the court states in a footnote that the Machinists Union departments are included in the OFCC agreement, it appears that the departments were not included as all-white departments into which any blacks hired before the cutoff date could transfer. Instead, the departments appear to be opened only to the extent that blacks in all-black departments could transfer into them. The actual effect of the OFCC plan on Machinists Union departments must be clarified on remand. Plaintiffs contend the Machinists Union departments were all-white and should have been included in the OFCC agreement as transfer-in departments for any black hired before the cutoff date, not just blacks who were assigned to all-black departments. Plaintiffs claim that these departments were still all-white in 1974. The district court made no finding regarding the racial composition of these departments in 1965 or 1974 and drew no inference from their composition. On remand it must do so. If blacks were excluded from these departments in the past, the departments must be included in the relief ordered by the court as departments to which any black hired before the cutoff date may transfer.

Moreover, the racial makeup of these departments in 1965, especially if they were all-white and high-paying departments, could be probative of plant-wide discriminatory assignment. The broad plant-wide scope of examination required during the prima facie case determination requires that this evidence be evaluated.

(4) Welding and Maintenance Departments

The court indicated that special skills are required for Welding and Maintenance departments and, therefore, they should not be considered as to whether there was plant-wide racially discriminatory departmental assignment. This conclusion would only be justified if the defendant could show that blacks were excluded from these departments because they lacked minimum qualifications and if such qualifications were justified by a business necessity.45 Neither finding appears in the district court's opinion. Unless these findings can be made, the racial makeup of both departments must also be considered in determining whether a prima facie case has been shown.

(5) Past Policies of Discrimination

Plaintiffs contend that prior to 1965 blacks were discriminatorily assigned to departments and were discriminatorily assigned to lower paying job classes within each department. They point out that in 1964 98.4% of the blacks and only 15.8% of the whites were qualified to work in job classes 8 or below, while 79.7% of all whites were in job classes 10 or above. Since arbitration decisions in 1965 eliminated upward mobility barriers for blacks within each department, it would be logical to assume that in the 8 years between 1965 and 1973 if the disparity in job classes had truly resulted from the limitation of upward mobility within each department and not from discriminatory assignment to lower paying departments that there would be substantial improvement in these figures. But, plaintiffs' exhibits show that in 1973 74.1% of blacks still remained assigned to job classes lower than 8 while 80.7% of all whites were assigned to job classes over 10. If this evidence is not rebutted by substantial evidence that blacks were not qualified to progress within their departments, it indicates that blacks were assigned to less desirable departments and should be weighed in determining whether a prima facie case has been made.46 The district court not only rejected consideration of the evidence, but also inexplicably discussed completely different figures from those shown in the record. If it is pertinent, this statistical discrepancy should be clarified on remand.

The district court recognized that until mid-1965 job classes were segregated by Pullman-Standard. Plaintiff states that 134 out of 148 job classes were all-white or all-black. The district court found that this job class segregation phenomenon has substantially dissipated and, therefore, did not give this past practice any weight. However, the prior history of discriminatory job class assignments is clearly relevant to the issue of whether the present discrepancies in departmental assignments were part and parcel of a broad scheme to treat black and white workers differently. Historical policies of racial discrimination have often been used by other courts as indicia of plant-wide discriminatory conduct.47 The weight to be given and the inferences to be drawn are for the trier of fact.

(6) Average Yearly Wage

The district court buttressed its conclusion that there was no perpetuation of discriminatory racial assignment at Pullman-Standard by referring to 1973 wage proof which showed that the gross yearly income of blacks from Pullman-Standard employment and unemployment benefits was 96.8% of that of whites from the same sources.48 These statistics must be carefully examined. United States v. United States Steel, 520 F.2d 1043, 1054 (5th Cir. 1976), teaches that such wage parity is relevant only if groups with similar seniority are being statistically compared.49 The figure used by the district court does not attempt only to compare groups of blacks and whites with relatively equal seniority. This is especially pertinent here because this record indicates that in general blacks have more seniority than whites at Pullman-Standard; if discrimination had not been perpetuated, blacks reasonably might be making more than whites. In addition, this yearly salary parity does not include two other extremely relevant factors: amount of overtime worked and number of weeks or hours worked. Because the figures did not consider seniority, overtime, and time worked, they were not a reliable indicia of even rough wage parity in 1973.50 Their use by the district court was not proper.

(7) Summary

The district court must examine the evidence presented concerning the plant as a whole to determine whether a prima facie case of discriminatory departmental assignment has been made. If a prima facie case of past discrimination has been shown,51 the court must then determine if it has been perpetuated. If such a determination can be made, then as a second and separate step the court must limit the broad certified class to appropriate affected classes or sub-classes and order necessary relief.

The district court did not find any discrimination which could be perpetuated by the departmental seniority system in use at Pullman-Standard. Although it did find that the nine departments included in the OFCC plan were previously completely segregated, it determined that the OFCC provision allowing carryover departmental seniority for transferring discriminatees broke down any barriers for these blacks in reaching their rightful place in the plant. The approval of the OFCC plan (with slight expansions in effective dates) was a proper judicial recognition that if the departmental seniority system in effect at Pullman-Standard was left to function without modification, it would perpetuate past discrimination. The case law precedent is legion if not unanimous in holding that departmental seniority plans similar to that in use by defendant do perpetuate past discrimination.52 If the trial court should find that the plaintiff has shown plant-wide discriminatory racial assignment has been perpetuated by the departmental seniority system, this barrier to transfer, which bars blacks from their rightful places in the plant, must fall.53

Before the Court is able to order appropriate relief, however, it must define the plaintiff class affected by this discriminatory conduct. In a factual milieu such as that provided at Pullman-Standard, the two limiting factors on the definition of the affected class will be the number of departments covered and the hiring date upon which discriminatory departmental assignment ended. These two considerations department and end date will define the parameters of the affected class.

(1) Affected Class Departmental Limitation

The purpose of this second Title VII decisional step is to narrow the certified class only to those members most probably entitled to relief.54 In performing this task the court should first delineate the elements of the plaintiffs' proof which were sufficient to make out a prima facie case of racially discriminatory departmental assignment. If these elements are clearly and convincingly inapplicable to a particular department, the department must be eliminated from membership in the affected class. In a completely segregated departmental situation, this limitation by departments may be easy.55 In mixed department cases, however, the court may choose to order relief which applies plant-wide to all black employees hired before a certain date.56

Of course a defendant may also attack the inclusion of a particular department for reasons entirely separate from whether the elements of the plaintiffs' prima facie case are applicable. For example, proof that there have been specific requests by all black applicants not to be assigned to a particular department could well exclude a department from membership in the affected class. The burden of proving departmental exclusion in this latter type of situation would be on the defendant.

(2) Affected Class Time Limitation

If the court determines that plant-wide discriminatory assignment exists, it must limit the class to all blacks hired before the date when this historical discriminatory assignment process ceased.

Plaintiffs attack the dates used by the lower court when it expanded the time limitation on the departments under the OFCC agreement. The court used as the significant date the date when the first black or white was appointed to the all-white or -black department respectively. Token integration, however, does not signal the end of discrimination. These dates must be reconsidered. It is improper to end affected class membership at a date upon which the first black or white was assigned to the department unless this date signifies the end of the discriminatory practices.57 Otherwise, the court must use a date which clearly signifies the end of a racially discriminatory departmental assignment.

Should the affected class be broadened on remand, the same analysis must be used whether the court chooses to limit the hiring date time factor of the affected class on a department-by-department or on a plant-wide basis.

If there ever was a time of facile Title VII litigation, it surely ended with the demise of intentional violations of equal employment opportunity. Today's parade of Title VII cases present more and more subtle manifestations of discrimination. Proof of invidious practices becomes more difficult as the ability to separate the real violation from the unfounded suspicion grows harder. This is especially so since many employers and unions, including Pullman-Standard and Steelworkers, have made substantial good faith efforts toward eliminating racial distinctions for the work force. Frequently, bargaining which results in minority employees being hired and allowed to progress in their jobs free of discrimination does not accord the Act's intended relief to older black employees. Although the temptation to enforce the accommodation reached by the parties most directly interested is great, it remains part of the court's task to see that these older workers, who bore the brunt of past disparities, do not continue to work at jobs their white peers would not have and for wages less than those paid to recent hires who were not put down because they were black.

Red circling is the mechanism by which the wage rate of an affected class member who transfers goes with the man even though the new department to which he moves would pay the transferee a wage rate less than the old one. The purpose of the remedy is to protect the employee from financial loss until he has an opportunity to learn the necessary skills to progress in the new department to a wage equal to his old department scale.

The district court held that red circling in the instant case was unnecessary for six reasons: (1) "(V)irtually all of the jobs in the transfer out departments have lower job class levels than the lowest jobs in the transfer in departments." (2) There are no lines of promotion or residence requirements to impede promotion in new departments. (3) Frequent manpower fluctuations may create new opportunities. (4) Some of these manpower fluctuations will obviate old wage rates. (5) No training is provided for new jobs. (6) The key reason for affected class members turning down transfer opportunities is apprehension over retreat rights and not the possibility of diminished wages.

Should the court find plant-wide discriminatory departmental assignments were perpetuated by the departmental seniority system, several of the six reasons discussed above would be inapplicable. However, since we hold that some kind of modified red circling system probably will be necessary even if the affected class is not expanded, we do no more than mention this contingency which an expanded affected class would present.

The red circling remedy was originally developed to eliminate impediments to discriminatees rising to their rightful place in a plant. Therefore, whenever the possibility exists of receiving lower wages in the new department, red circling must be ordered. The remedy must continue until the transferee has had an opportunity to progress to the new department job which he would have held but for the past discrimination. For this reason, it is now clear that red circling is a necessary element of a Title VII remedy in most,58 though not all,59 cases.

The only situation where we have recognized that this part of the remedy would not be mandated is where the plaintiff is unable to show that the absence of red circling presents a practical impediment to any affected class member. Stevenson v. International Paper Co., 516 F.2d 103, 113 (5th Cir. 1975); accord, Watkins v. Scott Paper Co., 530 F.2d 1159, 1174 (5th Cir. 1976). The district court's major premises for refusing to order red circling were its conclusions that (1) "virtually all" affected class members will be able to transfer immediately to higher paying jobs and (2) the key reason for nontransfers has been apprehension over retreat rights and not lower wages. Neither reason appears to meet the Stevenson and Watkins standard. The test is not whether "virtually all" discriminatees will not suffer reduced wages. If any will suffer, red circling is appropriate. On remand, unless the district court determines that the plaintiffs have failed to show that any discriminatees would suffer diminished wages on transfer, red circling must be ordered.60 This would also be true should the affected class be expanded on remand. Apprehension over retreat rights could only result from ignorance of the OFCC agreement's provisions. Full and complete retreat rights are guaranteed. Surely neither Pullman-Standard nor the union should or will allow ignorance of rights already conferred to operate as a bar to departmental transfer. When the fear over retreat rights has dissipated, the possibility of reduced wages surely will be seen as the major impediment to transfer. Neither of these considerations can pass the no -affected-class-member test.

The four other reasons given by the district court are insufficient to justify its refusal to order red circling.61 However, the lack of on-the-job training and lines of progression along with frequent fluctuations in the work force indicate that the standard red circling remedy would be inappropriate and that some modification of that remedy is needed.

The purpose of red circling to eradicate the obvious impediment of lowered wages to transfer to a new department is meant to protect the affected class transferees only until they have been in the new department a sufficient amount of time to have had a realistic opportunity to become recognized and listed as qualified to work in a job class which pays at least as much as the old department wage rate.

Job manpower fluctuations, which are particularly acute in some departments offering greater advancement opportunity, create problems for devising a remedy which will not go too far and convert the attempted remedy into a windfall. The lack of on-the-job training and lines of progression also present problems for the trial court. They are not ones appropriate for resolution or detailed suggestion here. Suffice it to say that if the remedy is required, it must be devised justly but, if required, it must be granted despite difficulty in its formulation.

The union contracts with Pullman-Standard do not provide for a formal bidding procedure or a posting of vacancies either plant-wide or department-wide. In fact, not even an after-the-fact notification of changes in assignments was required.62 Pullman-Standard asserts that the absence of these formal notification and bidding procedures is justified because the daily changes in employment levels make a bidding procedure impossible and the posting of assignments unnecessarily burdensome. The plant operates on a "word of mouth"63 and "personal observation" approach to the notification of job vacancies. In addition the union takes a very active role in monitoring employees' seniority rights.

Plaintiffs requested below that a plant-wide system of posting job vacancies be ordered for the Pullman-Standard plant. The court framed the issue as: "Has there been discrimination in the failure to post publicly a list of changes in assignment?" Its answer was that "the lack of formal procedures for notification neither discriminated nor perpetuated the effects of any past discrimination against blacks." So addressed, the claim is limited to whether the notification system prevented blacks from taking full advantage of their seniority rights. The court's answer is not clearly erroneous. However, it only answers the questions of nondiscrimination and perpetuation of discrimination presented by the posting issue per se. The propriety of ordering posting to facilitate transfer rights has not been considered.

Not only must posting be examined to see if the lack of a formal system itself discriminates or perpetuates prior discrimination against blacks, but the ordering of a formal posting remedy must be considered as a part of the remedy which will facilitate the advancement of the affected class members to their rightful places in the plant. The illegal discrimination found at Pullman-Standard was racially discriminatory departmental assignment. We hold that the neutral departmental seniority system has perpetuated that discrimination into the present. Carry-over seniority, red circling, and backpay are parts of the total remedy. The need of a formal system of notification of plant-wide vacancies to make the remedy efficacious must also be reconsidered. If any affected class member would be prevented from exercising his transfer rights because the present nonsystem failed to notify him of a vacancy then some more formal system of notification must be ordered unless business necessity prevents it.64

The district court did not award any backpay, and except for one small subclass of discriminatees no adjudication regarding backpay has been made in the instant case.65 Plaintiffs' attack on the failure to award any backpay in this action is premature. The pretrial order of June 5, 1974 severed backpay claims for a subsequent trial should they be necessary.

The OFCC agreement identified five all-white departments and allowed blacks hired prior to April 30, 1965, to transfer into these departments with carry-over departmental seniority as vacancies occurred. The trial court ordered that the cut-off date for transfer into three of these departments, plant protection, inspection, and air brake departments, be extended until June 1, 1967, June 1, 1970, and June 1, 1971, respectively, to allow blacks who were discriminatorily prevented from being assigned to those departments adequate transfer rights. The court recognized that any blacks hired between April 30, 1965 and these new dates were entitled to backpay. However, it also recognized that no backpay would be necessary if there were no vacancies in these departments prior to transfer.

The court did not hold that there were no vacancies; instead, it held there were not enough vacancies to allow this class of employees at the bottom of the eligibility group to be assigned to these departments. Here the district court obviously had reference to the priority enjoyed by senior affected class members which would preclude workers in the expanded affected class from eligibility for transfer to one of these departments. This finding by the court is not clearly erroneous. However, the legal conclusion that the expanded class is not entitled to backpay must be reversed since it was reached using an erroneous legal principal.

A finding of no vacancies in these departments would support a conclusion that no entitlement to backpay existed, as would a finding that all vacancies were filled by more senior discriminatees. The court did not make either clear-cut, absolute factual finding. Rather, it inferred that since there were so few vacancies and since the expanded affected class' priority was so low, they would have been precluded by more senior discriminatees. This inference is precluded by Baxter v. Savannah Sugar Refining Corp., 495 F.2d 437 (5th Cir. 1974), cert. denied, 419 U.S. 1033, 95 S. Ct. 515, 42 L. Ed. 2d 308 (1975).

Baxter requires a two phase backpay inquiry. During the first or general phase the class must show its entitlement to backpay. This is presumptively shown once employment discrimination has been found.66 The second or specific phase allows each discriminatee to present the backpay claim to which he is presumptively entitled unless the defendant bears the burden of proving nonentitlement. A general inference of class-wide unavailability of vacancies is insufficient.67 To avoid liability the defendant must show during phase one that no vacancies whatsoever were available or during phase two it must show that the particular discriminatees seeking backpay could not have been assigned to that department nor transferred to it because of the unavailability of vacancies they were eligible to fill.

Of course, we pretermit any indication of appropriate resolution of the severed claims for backpay. We would only note that if on remand the court should expand the affected class, substantial backpay issues would be raised.68

The appointment of supervisory personnel at Pullman-Standard is done totally subjectively. There are no established criteria for selection of new foremen. The plant manager and superintendent choose department heads (C foremen) who in turn select track supervisors (B foremen), production foremen (A or salaried foremen), and hourly (temporary) foremen. Plaintiffs contend that prior to 1965 there were no black foremen. In 1966 the first black salaried foreman was promoted to one of the 143 existing salaried foreman positions. Four years later there were only nine black salaried foremen while there were 151 white foremen. At the time of trial there were 13 departments in which blacks had never been offered either salaried or temporary foreman positions. Since 1966 and until the time of this trial there were at least 59 salaried foreman vacancies. Only 12 of these were filled by blacks.

The district court essentially agreed that these statistics were correct but chose to focus on the selection of hourly foremen since they in turn formed the source group for the selection of salaried foremen. The court found that since 1965, 56 blacks approximately 1/3 of the total number have been selected as hourly foremen. Recognizing that this was a lower percentage than the approximately 50% of black personnel in the plant, the court found four factors which ameliorated the otherwise compelling inference of discriminatory promotion.

1. The court found as a fact that a greater number of blacks than whites were functionally illiterate and, therefore, were less likely candidates for promotion. Educational requirements which operate disproportionately to exclude blacks may not be used unless they pass Title VII validation and business necessity muster.69 The court itself recognized in a footnote that it was not holding that some particular level of education was allowable under Title VII. Despite this recognized absence from the record of validation or business necessity proof, the court found this absence of functional literacy would render black employees "less likely" candidates for promotion. This non sequitur is impermissible.70 Since no minimum educational requirement was proven legal under the strict guidelines of Title VII, the "requirement" may not be used as a generalized inference to explain why blacks were not potential supervisors.

2. The court found that 30 blacks and 17 whites have turned down hourly foremen opportunities since the mid-1960's. This, of course, is a legitimate item of rebuttal to a prima facie case. The most statistically pure representation of discrimination would be the percentage of black vs. white offers of promotion. A comparison of actual appointments with the number of turned down offers would enable the court, if the number of actual offers was not before it, to approach this statistically sterile measure.

3. The court concluded that due to pre-'65 segregation of jobs, it has taken blacks some time to learn the range of job skills necessary to perform supervisory duties. This justification for not promoting blacks to supervisory positions has been uniformly rejected by this court. Before such justification could be considered the defendants would have to prove that unpromoted blacks did in fact lack the necessary skills; that the needed skills are justified by a business necessity; and, if plaintiffs could show that the failure to gain these skills was the result of a neutral policy which perpetuated past discrimination, defendants would have to show that the policy was justified by business necessity.71 Since none of these facts were found by the lower court, it was improper to consider this item as indicative of nondiscrimination.

4. The district court noted the recent dramatic improvement in black promotions to salaried supervisor. Between 1971 and 1974, eight of the twenty newly appointed foremen were black. This figure is clearly indicative of the lack of discrimination in promotion. It is appropriate to consider it as counterbalancing evidence of the statistics indicative of discrimination which were presented by the plaintiffs.

Since two of the four reasons given for finding that the plaintiffs' statistics did not establish a prima facie case of racially discriminatory promotion to supervisory positions were erroneously considered, the court's conclusion that no discrimination was proven is reversed. We do not conclude, however, that the evidence clearly makes out a prima facie case. Instead, we remand to the district court to balance recent promotion statistics and black turn-downs against the overall statistics presented by the plaintiffs.72 The inference to be drawn from this balancing of the evidence is for the lower court. Should the court desire supplemental evidence in its reconsidering it may on remand open the record to evidence concerning promotions made since trial.

With but one exception the management of Pullman-Standard has the sole authority under the collective bargaining agreement to assign work among employees working within a given job class in a given department. The only exception is that members of the welding department are able to bid to their supervisors for "sub-assembly" work according to departmental seniority. Plaintiffs argue that the supervisors in exercising this unbridled discretion give the more desirable assignments to white employees and the less desirable ones to blacks. The lower court not only found that the plaintiffs had failed to prove that blacks were assigned to less desirable tasks within each job class, it also found that the plaintiffs were unable to prove which tasks were more or less desirable. The totality of the proof clearly showed that various employees had varying interpretations of which tasks within each job class were more or less desirable. Given nothing but this kind of subjective opinion evidence, especially since the credibility of each witness was at the core of the issue, we are unable to say that the lower court's factual conclusion was clearly erroneous.

The lower court concluded that Louis Swint and Clyde Humphrey were not discharged for racial reasons as alleged. Insubordination, the principal reason given for the discharge in each instance, was more than amply justified. The court's conclusion that there was not even the slightest evidence anywhere in the record upon which it could indulge the inference that these individuals were fired because they were black is not clearly erroneous.

Pullman-Standard's attempt to limit the issues on appeal solely to these presented by Louis Swint in his individual capacity is without merit. Jones v. Chaney & James Construction Co., 399 F.2d 84, 86 (5th Cir. 1968); Markham v. Holt, 369 F.2d 940, 943 (5th Cir. 1966); see Foman v. Davis,371 U.S. 178, 181, 83 S. Ct. 227, 230, 9 L. Ed. 2d 222, 225 (1962).

The judgment appealed from is vacated and the cause is remanded to the district court for such proceedings as that court may determine are appropriate or necessary to compliance with this opinion.

AFFIRMED IN PART, PART VACATED AND REMANDED.