Doris M. Davis, Appellant, v. Richmond, Fredericksburg and Potomac Railroad Company, Appellee.doris M. Davis, Appellee, v. Richmond, Fredericksburg and Potomac Railroad Company, Appellant.sandra Jean Hylton, Appellee, v. Richmond, Fredericksburg and Potomac Railroad Company, Avirginia Corporation, Appellant, 803 F.2d 1322 (4th Cir. 1986)

The Richmond, Fredericksburg & Potomac Railroad Company ("RF & P" or "railroad") appeals from an order of the district court entering judgment in favor of two RF & P employees, Doris M. Davis and Sandra Jean Hylton, in an action alleging sex discrimination in violation of Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e, et seq. Davis v. Richmond, Fredericksburg & Potomac R.R. Co., 593 F. Supp. 271 (E.D. Va. 1984). Davis and Hylton jointly cross-appeal the district court's denial of their request for expert witness fees. Davis individually cross-appeals the district court's decision denying monetary damages and limiting her to injunctive relief. Finding only one slight error in the district court's disposition of this case, we affirm in part and reverse in part.

RF & P operates a mainline railroad approximately 108 miles long between Northern Virginia and Richmond. The railroad maintains two large classification yards, one at either end of the line, the Acca Yard in Richmond and the Potomac Yard in Alexandria. At the time this discrimination case was filed in 1983, RF & P had 843 employees.

Doris M. Davis was fifty-seven years old at the time of trial below and had been employed by RF & P since December, 1943. During the majority of her employment she had served as a clerk in the mechanical department at the Potomac Yard. Sandra Jean Hylton was thirty-seven years old and had been employed by RF & P for thirteen years. At the time of trial, she was a crew dispatcher at the Acca Yard in Richmond.

Both Davis and Hylton had sought to be admitted to the apprentice locomotive engineer program operated by RF & P. Davis first applied for admission to the program in 1976 with the Richmond personnel office, while Hylton had earlier applied in 1975. Despite their long-expressed interest, it appears that neither Davis nor Hylton was given serious consideration for employment as engineers. At trial, spokesmen for RF & P maintained that apprentice engineers were chosen solely from those employees with prior train service.¹  Although that policy was not formally disseminated to employees such as Davis and Hylton, RF & P relied upon it to employ two males as new engineers in 1983.

Davis and Hylton subsequently filed complaints with the Equal Employment Opportunity Commission ("EEOC") on April 1, 1983 alleging sex discrimination pursuant to Title VII. After investigation, but without making any determination, EEOC issued a Notice of Right to Sue to both women. Following institution of their actions in district court, the two cases were consolidated for purposes of a bench trial on the question of liability. The court set the cases for additional proceedings, if needed, with regard to damages and remedies.

At trial, Davis and Hylton asserted two alternate theories of Title VII liability. They contended that RF & P had exhibited discriminatory intent sufficient to sustain a showing of unlawful disparate treatment. They also argued that RF & P's reliance on the "train service" pool as the sole source of apprentice engineers constituted a policy with a discriminatory disparate impact which could not be justified by business necessity.² 

Both plaintiffs testified to the informal personnel policies of RF & P. Both testified that they were never informed of any specific skills or experience necessary to qualify for the position of apprentice locomotive engineer until the commencement of their lawsuit. The railroad conceded that their employees were not notified by posting of available positions. Nor was it railroad practice to reduce to writing a list of the qualifications necessary to apply for a position such as locomotive engineer. Additionally, the railroad stipulated that there was no gender related physical capability necessary to function as an engineer and that the plaintiffs were capable of being trained for that position.

Plaintiffs presented the testimony of W.P. Cunnane, a road conductor and local chairman of the United Transportation Union. Cunnane testified regarding a 1979 conversation with J.D. Doswell, Superintendent for Transportation with RF & P. During that conversation, Doswell referred to Hylton's application for engineer training and stated that no woman would be hired "as long as he was superintendent."

Plaintiffs also offered the expert testimony of David Decker on the subject of engineer training. At the time of trial, Decker was training director for the Long Island Railroad and had trained 215 engineers for that line. He testified that his training program made no distinction between individuals with or without prior train service. All apprentice engineers began at "ground zero." Decker testified that apprentices could learn to operate a locomotive in five days although the Long Island training program lasted fifteen to eighteen months. The additional time was required to learn the physical characteristics and rules of the road. In response to questioning by the court, Decker conceded that a person who had worked directly around trains and knew a great deal about them would be "an easier piece of raw material to work with as a student."

In its defense, RF & P disputed that plaintiffs had shown even a prima facie case of intentional discrimination. The railroad also sought to show that as a smaller and less sophisticated operation than the Long Island Railroad, its preference for train service personnel was a rational attempt to promote safety and economic efficiency. In support of this contention, the railroad introduced a projection, prepared by their Road Foreman of Engineers in 1973, which estimated that the training time required for individuals with prior train service would be six to eight months. Needed training time for those without train service was estimated at approximately two years.

The Superintendent of the Potomac Yard, John F. McGinley, testified that the present training time for engineers drawn from the train service pool was two months. The railroad was unable to offer any empirical evidence regarding actual training time for non-train service personnel because it had earlier conceded that no such individuals had ever been trained by RF & P.

The district court concluded that Davis and Hylton had prevailed on both a theory of disparate treatment and disparate impact. The court supported its decision with specific findings of fact, the most critical of which related to RF & P preference for prior train service by apprentice engineers. The court stated that:

The duties of a locomotive engineer require specific training. However, while prior experience with the operation of locomotive engines and the railroad system itself may be helpful, such prior experience is not necessary in order to produce a competent locomotive engineer. The relationship between prior experience with the railroad and competency as a locomotive engineer is nil. While it may be easier to teach an individual with knowledge of the railroad and the lay of the tracks, the training program for such individuals is not substantially different or more expensive from that for individuals without any prior knowledge.

593 F. Supp. at 274-75.

Following its determination of liability, the district court held a hearing on the question of appropriate relief. With regard to plaintiff Hylton, the court awarded specific monetary relief in the form of back pay, retroactive seniority, and admission to RF & P's apprentice engineer program. The court also ordered that the railroad take curative action to eliminate the possibility of future discrimination.

With regard to plaintiff Davis, however, the court limited the remedy to injunctive relief. The court concluded that back pay was not appropriate because she would have been unable to satisfy RF & P's sexually neutral requirement that all prospective engineers have on file a back x-ray displaying no abnormalities.

Following the court's decision on relief, Davis moved to amend her complaint to allege age discrimination in RF & P's x-ray requirement. Both plaintiffs also applied for fees and expenses pursuant to 42 U.S.C. § 1988. Following oral argument on plaintiffs' motions, the court denied the request to amend the complaint and awarded fees and costs to plaintiffs' attorneys in the amount of $51,354.04. In making its fee award, the court denied, sub silentio, plaintiffs' request for $7,750.28 as compensation for their expert witnesses.

This appeal and cross-appeals followed.

On appeal, RF & P contends that the district court committed numerous errors in its disposition of this case. Specifically, the railroad argues that (1) Davis' claim should have been dismissed as untimely, (2) Davis and Hylton were allowed incorrectly to proceed on a disparate impact claim that was never properly pleaded, (3) the court's finding that the railroad had engaged in discriminatory conduct was clearly erroneous, and (4) the injunctive relief granted was impermissibly broad. In their joint cross-appeal, Davis and Hylton contend that the court abused its discretion by denying an award for the cost of expert witnesses. Davis individually argues that the court erred by failing to recognize that RF & P's back x-ray requirement had a discriminatory disparate impact upon older females and could not, therefore, serve to deny her monetary relief. With the exception of RF & P's challenge to the scope of injunctive relief, we see no merit in any of the contentions presented by the parties.

RF & P's first two contentions need not delay us long. The railroad's time-bar argument with regard to the Davis claim is based upon the fact that Davis did not file a complaint with EEOC within 180 days of the time that a male employee, M.J. Dinunzi, was selected as an apprentice engineer on May 20, 1980. This argument was specifically rejected by the district court which found that, although Dinunzi received some training in 1980, he was not selected as an engineer until 1983. Dinunzi's 1983 promotion was an act that the railroad was under no legal obligation to perform. The district court concluded, therefore, that the 1983 promotion was an independent affirmative act of discrimination that brought Davis' claim within the bounds of the continuing violation theory recognized in Delaware State College v. Ricks, 449 U.S. 250, 101 S. Ct. 498, 66 L. Ed. 2d 431 (1980). We agree. Davis filed her claim within 180 days of Dinunzi's selection as an engineer in 1983, thereby clearly satisfying the time requirements of 42 U.S.C. § 2000e-5(e).

We likewise find no merit in the railroad's argument that it was unfairly surprised by the district court's decision to allow plaintiffs to proceed on both a disparate impact as well as a disparate treatment theory of discrimination. We are aware of no authority mandating that a Title VII plaintiff mechanically elect a disparate impact theory in the pleadings.³  The key consideration in an EEOC complaint is fair notice. Wright v. Olin Corp., 697 F.2d 1172 (4th Cir. 1982). We agree with the district court that EEOC complaints in this case fully satisfied that consideration by stating the factual allegations necessary to support both theories. We, therefore, see no basis to RF & P's claim of unfair or prejudicial surprise.

We conclude as a threshold matter, that, without question, plaintiffs established a prima facie case of sex discrimination on behalf of RF & P. The railroad's argument that the court's finding of discrimination was clearly erroneous must turn, therefore, upon whether the court's conclusion with regard to the value of prior train service for apprentice engineers can be sustained. Relying heavily upon the testimony of plaintiffs' expert, Decker, the court found that train service was of minimal value to an apprentice engineer and that RF & P's requirement of such service was neither a business necessity nor a nonpretextual explanation sufficient to rebut a prima facie showing of deliberate discrimination. The railroad argues in this appeal that Decker's testimony concerning the training practices of the much more complex Long Island Railroad was essentially irrelevant to the operation of the smaller RF & P. The railroad argues, therefore, that Decker's testimony could not refute RF & P's claim that prior train service was a necessary prerequisite for an apprentice engineer on that line.

We are not persuaded by the railroad's argument. When making findings of fact, a trial court is permitted to draw inferences from other facts and testimony. An appellate court may not overturn findings based on those inferences if the trial court's conclusions are "plausible in light of the record viewed in its entirety." Anderson v. City of Bessemer City, N.C., 470 U.S. 564, 105 S. Ct. 1504, 1512, 84 L. Ed. 2d 518 (1985).

In this instance, the only direct testimony on the subject of instructing non-train service personnel as locomotive engineers was provided by Decker, who indicated that he did not find such prior service to be of significant importance. From the fact that the Long Island Railroad does not seek engineer trainees solely from individuals with train service and does not differentiate in the type of training accorded based on prior experience, it may be reasonably inferred that RF & P's restrictive use of that requirement to bar women was discriminatory. The plausibility of the inference is strengthened by the fact that RF & P has never attempted to instruct anyone without a train service background and, therefore, has no concrete basis for assuming that an individual without that experience would require a substantially greater period of instruction.⁴ 

Thus, the district court's conclusion that train service was not of significant importance to an apprentice engineer is not clearly erroneous. Given that conclusion, the specific application of that finding to the claims advanced by Davis and Hylton must also be affirmed. With regard to the disparate treatment theory, we can see no clear error in the court's holding that RF & P's reliance on an experience requirement of doubtful value was a pretext for deliberate discrimination. Under a disparate impact analysis, the district court's conclusion is likewise unassailable. We have noted that the defendant's burden in a disparate impact case is a heavy one. When a practice has been shown to have a discriminatory impact, the defendants must prove a "compelling" business necessity in order to justify that practice. Betsey v. Turtle Creek Associates, 736 F.2d 983, 988 (4th Cir. 1984). We agree with the district court that the railroad fell far short of demonstrating a compelling necessity for its discriminatory hiring practice.

We also see no error in the award of retroactive seniority to plaintiff Hylton. In light of the Supreme Court's unambiguous approval of such a remedy in Zipes v. Trans World Airlines, Inc., 455 U.S. 385, 102 S. Ct. 1127, 71 L. Ed. 2d 234 (1982), and Franks v. Bowman, 424 U.S. 747, 96 S. Ct. 1251, 47 L. Ed. 2d 444 (1976), RF & P's contention that the award was impermissible borders on being frivolous.

We agree with RF & P, however, that the injunctive relief granted by the district court swept too broadly in this instance. In crafting its injunction, the court directed in paragraph (b) that RF & P refrain from "committing further violations of Title VII." In the subsequent paragraphs (c) and (d), the court set forth certain specific actions that the railroad was required to undertake as a means of remedying its prior discriminatory conduct.

While we see no error in the extent of the relief granted in paragraphs (c) and (d), paragraph (b) is fatally reminiscent of the broad injunction to "obey the statute," the United States Supreme Court found unacceptable in National Labor Relations Board v. Express Publishing Co., 312 U.S. 426, 61 S. Ct. 693, 85 L. Ed. 930 (1941). Such an injunction impermissibly subjects a defendant to contempt proceedings for conduct "unlike and unrelated to the violation with which ... [it] was originally charged." Id. at 436, 61 S. Ct. at 699.

Accordingly, we vacate paragraph (b) of the injunction granted by the district court. In all other respects, however, we affirm the extent of injunctive relief as stated in paragraphs (c) and (d).

Turning now to the joint and individual cross-appeals, we likewise see no basis for disturbing the conclusions of the district court. Davis' and Hylton's joint contention that expert witness fees should have been awarded is refuted by our opinion in Wheeler v. Durham City Bd. of Ed., 585 F.2d 618 (4th Cir. 1978). Although that case dealt with fee awards under 20 U.S.C. § 1617, rather than awards under 42 U.S.C. § 1988, we expressly stated that "what we say about Sec. 1617 applies equally to Sec. 1988, for there is no indication that Congress intended the language of the two statutes to be interpreted differently." 585 F.2d at 618.

In Wheeler, we concluded that although an award of fees should include related litigation expenses, the costs attributable to outside non-legal experts are not traditionally included within such expenses. We, therefore, declined to adopt a rule that witness fees are automatically included in the statutory authorization and remanded to the district court for consideration of a specific claim advanced by the plaintiff in that case.

Applying Wheeler to the present case, it is clear that the district court properly held that Davis and Hylton had no statutory right to compensation for their witness. To the extent that a non-statutory claim for witness fees could be based upon the general equitable powers of the court, the request would be addressed to the sound discretion of the trial court. In that regard, we can see no abuse of discretion that would merit reversal.

Lastly, we consider Davis' cross-appeal from the district court's denial of back pay in her case. The district court concluded that because after 1976 she could not have satisfied the railroad's policy of requiring a back x-ray free from abnormalities, she would have been unable at any time to qualify as an apprentice engineer. Relying upon White v. Carolina Paperboard Corp., 564 F.2d 1073, 1087 (4th Cir. 1977), the court then held that notwithstanding RF & P's discriminatory conduct, back pay is barred when a defendant can show that a particular plaintiff was ineligible for the job sought. We agree with the district court's analysis.

Davis argues that the back x-ray requirement was both irrational and discriminatory in its operation. She notes that long-term RF & P employees in the train service field could use the required x-ray taken at the time of their employment even if they had incurred back injury in the interim. Since non-train service personnel were not required to have x-rays when beginning employment, these employees could not rely upon older x-rays on file to satisfy the requirement. Davis argues that because the x-ray requirement could serve no logical purpose and had a disparate impact upon older female employees, the district court erred in allowing the railroad to escape its responsibility for providing full monetary relief. We find this argument unpersuasive.

It is undisputed that RF & P has maintained a policy of requiring back x-rays as a condition of employment for all individuals in its mechanical and transportation departments since at least 1966. There is no indication in the record that the requirement has ever been applied in a discriminatory fashion. Both men and women seeking employment in those departments would have been required to provide the x-ray. Although we may question the business wisdom of allowing the initial x-ray to satisfy a subsequent requirement for promotion to apprentice engineer many years later, we cannot say that the district court's finding of sexual neutrality is clearly erroneous.

We also find no merit in Davis' contention that the requirement has a disparate impact upon older women. More realistically, the requirement adversely affects only older women who have not been employed in the mechanical or transportation departments and who have sustained back injuries after their employment commenced. That is far too narrow a class to support a contention of disparate impact.

We, therefore, conclude that the district court properly denied Davis monetary relief in the form of back pay. We also see no abuse of discretion in the court's refusal to allow a much belated amendment to the complaint that attempted to state an entirely new cause of action for age discrimination.

For the foregoing reasons, the judgment of the district court is affirmed in part, and reversed and vacated in part.

AFFIRMED in part; REVERSED and VACATED in part.

HAYNSWORTH, Senior Circuit Judge, dissenting:

I respectfully dissent.

The district court rejected the Railroad's business justification for including only experienced trainmen in the pool of those eligible for the locomotive engineer apprentice program upon a finding that a training program for inexperienced persons would not be substantially different or more expensive than one designed for persons with extensive experience as trainmen. I would hold that finding clearly erroneous and the business justification for the practice clearly apparent.

These two training programs were designed to produce locomotive engineers qualified for yard operations in each of the two terminal yards. There was no prospect that they would operate commuter trains, and the programs were not intended to qualify the trainees for the operation of over-the-line trains moving between the two terminals. They were being trained to operate switch engines operating entirely within the confines of the yard in which they were receiving their training.

Analysis should begin with some consideration of what a yard engineer does, and the areas of knowledge and experience he^*  must have in order to function efficiently.

Yard locomotives are principally concerned with making up and breaking up trains. While the train crew receives orders for its successive moves, the engineer must have a thorough understanding of the technique. He must know all of the hand signals to which he must respond, and he should have a thorough understanding of the safety rules designed to protect people in the yard, particularly those who must pass between the cars. He must know the yard in intimate detail, including all its tracks, its switches, its humps and checking devices. Finally, he must know how to operate the locomotive smoothly and efficiently and a good deal about its care and maintenance.

In the course of acquiring years of experience as a brakeman at the Potomac Yard, for instance, one naturally acquires the intimate knowledge of the yard that a locomotive engineer must have. The experienced brakeman knows all the techniques for making up and breaking up trains. He knows the safety rules. He knows how to read and follow the orders that govern the crew's movements.

He would need training in the actual operation and care of a locomotive, but he already knows much of what the yard locomotive engineer must know, and a training program need not be designed to teach him what he already knows or be based upon an assumption that he knows nothing.

But perhaps I should begin with the beginning.

Initially, on the RF & P, yard locomotive engineers were drawn from the pool of locomotive firemen. The fireman's station was in the cab, and, in the course of his work, he learned all about the yard operating techniques, signs and rules. Probably he would have spent some time operating the locomotive. It was the best possible on-the-job training, and firemen could move from one side of the cab to the other with little or no additional training.

I do not believe that anyone would suggest that a stranger could be trained to be a yard locomotive engineer at the Potomac Yard in the same time and as inexpensively as the yard locomotive fireman with years of experience in that yard.

In 1965, by agreement between the Railroad and the union, the position of fireman in yard locomotives on the RF & P was abolished. As the pool of experienced locomotive firemen dried up, the Railroad, of necessity, had to turn to other sources.

Between 1973 and 1983 in the Acca Yard, the Railroad employed several experienced locomotive engineers who had been furloughed by other railroads. That was a practice encouraged by federal officials, though not mandated. It utilized skills already possessed by the furloughed locomotive engineers. They knew how to operate a locomotive, and they knew the techniques for making up and breaking up trains, as well as the signals and the rules they must follow.

No such furloughed engineers were ever employed in the Potomac Yard, however. The Yard superintendent gave a readily understandable explanation. It would take an experienced locomotive engineer from another railroad at least two months to become sufficiently familiar with the complex Potomac Yard in which the RF & P provides switching services for five other railroads as well as itself. He said that he could train his own experienced yard trainmen in less time than that, and he did.

There is more to railroading than just operating a locomotive, and the decision not to employ experienced locomotive engineers from other railroads in the Potomac Yard suggests the importance of the knowledge that a yard brakeman in that yard would already possess.

This history also demonstrates a consistent practice on the RF & P of utilizing previously acquired skills of potential trainees and of adapting training programs to teach trainees only those things they do not already know. An experienced fireman was promoted to engineer with little or no additional training. In the Acca Yard, experienced locomotive engineers from other railroads were put through a program of familiarization with that yard, but were not trained to develop those skills and knowledge they already possessed. First in the Potomac Yard, and later in the Acca Yard, experienced trainmen were put through a program to develop those skills they did not already possess, but an experienced brakeman, for instance, was not put through a program of familiarization with the yard with which he was already intimately acquainted or trained in the techniques of yard operation that he already knew.

The demonstrable consequence was that on the RF & P, locomotive engineer apprentices became productive, operating engineers after a very short training program. At the time of trial, thirty-one apprentices had been through the two programs. Twenty-one of those completed the program in less than two months. Another seven completed the program in less than three months. Of the remaining three, we know that one of them took longer because the need for additional engineers dried up before completion of his program, and he was allowed to return to his former position. When there again was a need for a locomotive engineer, he was given only a thirty day refresher course. The record does not disclose why the remaining two took longer than three months, but does disclose that one of them began training at the same time as the one whose training was interrupted. There may have been an interruption of his training program also.

Superintendent McGinley testified that he thought he could train the plaintiff, Davis, to be a locomotive engineer in six to twelve months. Earlier, there had been an estimate that it would take up to twenty-four months to train a completely inexperienced person. The RF & P had never attempted to train an inexperienced person to be a yard locomotive engineer, but it seems beyond refutation that it would take substantially longer to train a completely inexperienced person than an experienced person who need not be taught what he already knows.

The finding that an inexperienced person could be trained to be a yard locomotive engineer substantially as easily and inexpensively as an experienced person is entirely dependent upon the testimony of Decker who was in charge of training on the Long Island Railroad, but his testimony seems, at best, irrelevant.

With assistants, Decker had trained 215 locomotive engineers. The Long Island is predominantly a commuter railroad, and the large number of trainees suggests that he was training engineers for the operation of commuter trains. It is conceivable that prior train experience may not be as useful in the training of commuter engineers as it is in the training of yard locomotive engineers. For good reason or bad, however, the Long Island Railroad chose to treat all of its apprentices as completely inexperienced. They were put through in groups, and each training period ran between fifteen and eighteen months. Because the training program on the Long Island was probably not concerned with the same kind of locomotive operation as the programs with which we are concerned, it probably has little relevance here, but there is nothing in it to contradict the established fact that, utilizing skills already possessed by experienced trainmen, they could be qualified as yard locomotive engineers after a very short training program, far shorter than one designed to teach a complete novice everything a competent yard locomotive engineer needs to know.

The record discloses that at the RF & P it cost approximately $2,000 a month for each apprentice in the training program. The apprentice, of course, is not productive. Putting a completely inexperienced person through a much longer training program would be much more costly, and it would require forecasting the need of additional yard locomotive engineers for a much longer period of time.

At the Potomac Yard they went to a locomotive apprentice program early, as soon as the pool of experienced firemen dried up, because it was thought that experienced trainmen, already familiar with that complex yard, could be trained to be qualified yard locomotive engineers in less time than it would take to train an experienced locomotive engineer from another railroad. That judgment was clearly based upon an appraisal of the worth of previous relevant experience. There would have been no conceivable room for that if the alternative to the employment of experienced locomotive engineers from other railroads was the training of a completely inexperienced person who knew neither the yard nor what experienced locomotive engineers from other railroads already knew.

I conclude, therefore, that the critical finding of the district court was clearly erroneous. On the RF & P, experienced trainmen were actually trained to be qualified yard locomotive engineers much more quickly and cheaply than inexperienced people could have been trained for those positions, and the projection of need for additional engineers could be made over a much shorter period of time.

Without that erroneous finding, the district court could not have rejected the defense of business necessity or justification. That is, of course, a defense to the disparate impact claim. See Griggs v. Duke Power Co., 401 U.S. 424, 431, 91 S. Ct. 849, 853, 28 L. Ed. 2d 158 (1971); Betsey v. Turtle Creek Associates, 736 F.2d 983, 988 (4th Cir. 1984). Even if there was a permissible inference of intentional discrimination against women, the Railroad also carried its burden of proving that the same decision would have been made in the absence of a discriminatory motive. See Mt. Healthy City School District v. Doyle, 429 U.S. 274, 287, 97 S. Ct. 568, 576, 50 L. Ed. 2d 471 (1977); Patterson v. Greenwood School District 50, 696 F.2d 293, 295 (4th Cir. 1982).

The definition of the pool from which engineer apprentices were drawn was not shown to have a direct relation to gender. Office and clerical employees are excluded, but the exclusion applied to men as well as to women. There were no women in the pool, but jobs in train service and in the mechanical department were open to women. Women did not seek those jobs, and neither the plaintiffs nor any other woman had acquired the experience which would facilitate training as a railroad engineer. As long as the line of progression was open to women at the bottom and restriction of the pool to persons with train experience had a substantial impact upon efficiency and cost containment, the business necessity justification cannot be rejected as pretextual or insubstantial.

Since the crucial finding of the district court is clearly erroneous, I would reverse.