dissenting.
I respectfully dissent, for the agency’s failure to follow its own rules for the conduct of random drug tests, including meeting the requirements of the Bargaining Agreement, can not be harmless error when the consequence was loss of Richard Frank’s job.
First, the agency’s departures from the protocols required for the collection and custody of the urine specimen were per se harmful, for the presumption of adulteration — the sole basis on which Mr. Frank was fired — is founded on the premise that the specimen was properly collected and handled. It was admitted that the rules were not followed; indeed, the arbitrator found that some of the statements made under oath by both of the specimen collectors were false.
Second, Frank’s right to union representation was, at best, not conscientiously implemented. Although the arbitrator found that Frank was not prejudiced thereby, he was surely prejudiced by the absence of notice, as required by the Bargaining Agreement, that he was suspected of adulterating the specimen. Frank was given no opportunity to provide another specimen, although the agency ordered another specimen. He was not told that he might be fired if he acted on the manager’s permission to go home at the end of his shift; and he was not told that he was at any risk whatsoever due to the apparent low temperature reading. Indeed, the government agrees that had Mr. Frank been permitted to provide a second specimen, separation from service would not have been required if the second specimen was drug-free. In this context, the somewhat casual treatment of the requirements of the representation agreement, together with the several admitted errors in the collection and processing of the specimen, leave no reasonable support for the drastic disciplinary action that was taken.
TEST PROCEDURES
The urine specimen collection and shipment was consigned to a contractor, Olsten Health Care. Dominick Scalercio and Sandra O’Connor, his supervisor, were both new employees of Olsten, who after two hours of training were sent on December 18, 1992, to the Dutchess County Airport. At about noon Mr. Scalercio and Ms. O’Connor reported to Anthony Capaldi, the acting manager of the airport. Several employees, selected at random, were to be tested that day. The employees were not given advance notice of their selection.
Richard Frank was working a 6:48 a.m. to 2:48 p.m. shift as an air traffic controller. At about 12:30 p.m., as Frank was leaving the men’s room, Mr. Capaldi told him of his selection for the random drug test and asked him to provide a urine specimen. Having just left the men’s room, Frank was unable *1561to do so.1 Capaldi told Frank that he would be called later in the day. Frank returned to work in the airport tower.
At about 2:35 p.m. Capaldi instructed Frank to return to the collection site. Frank was met by Scalercio, who escorted him to the men’s room. Scalercio and O’Connor both testified under oath that they placed a bluing agent in the toilet and secured the faucets. (The arbitrator found that this testimony was not credible, thus overcoming the obstacle of lack of access to water in the restroom. As I shall discuss, Frank was fired based on testimony of Scalercio and O’Connor that they followed all requisite procedures in collecting and processing the specimen, although both witnesses contradicted themselves or each other on highly significant points.)
Scalercio remained outside of the men’s room while Frank provided the specimen, in accordance with agency procedure regarding privacy. Frank testified that he had not had the opportunity to drink fluids during the intervening two hours he worked in the tower, and that he was able to void only “dribs and drabs” of urine at a time. He was in the men’s room for about ten minutes. He emerged at about 2:47 p.m. and gave the specimen to Scalercio, who measured its temperature. This measurement is accomplished by pouring the specimen into a cup with a temperature measuring patch affixed to its side, and must be obtained within four minutes of the time of urination. The agency rule requires that the temperature be within 90.5° and 99.8° F; there is a presumption of adulteration for specimens outside of this range. There appears to have been some difficulty in reading the temperature, which was recorded as 88° F, Frank testifying that it was first read as 92° by him and Scalercio.2 Scalercio notified O’Connor that the temperature was low. She testified that she “[came] across the hallway, open[ed] a new urine collection kit, pour[ed] the urine from the one cup into a second cup”. The second cup gave no reading, perhaps because the lower limit for the patch was 88° F.
Following these events, O’Connor checked the Olsten procedure manual, and then she and Capaldi went to Capaldi’s office where O’Connor called the FAA’s Drug Program Coordinator (DPC) for instructions. The DPC requested that a second specimen be obtained from Frank, in accordance with FAA procedure. The DPC asked Capaldi whether he would authorize the necessary overtime for Frank to remain on site in order to provide a second specimen; Capaldi said that he would.
Frank and Scalercio had remained in the conference room with the specimen. Frank then went to Capaldi’s office. Scalercio followed Frank. When Scalercio caught up with him, Frank was inside Capaldi’s office. Scalercio remained outside of the office. The specimen remained unattended in the conference room. When O’Connor noticed that Scalercio was not with the specimen, she told Scalercio that the specimen was not supposed to leave his sight and instructed him to return to the conference room and secure the specimen. It is significant that Scalercio testified as to the difficulty he encountered in attempting to stay close to Frank because there were “a lot of people in the area,” for Scalercio and O’Connor also testified that neither Scalercio nor Frank was away from the specimen for more than a minute.
Scalercio and Frank then returned to the conference room and Frank initialed the seal for the bottle, which he testified was attached to the chain of custody form.3 In another *1562procedural violation by those in charge of the tests, Frank was not asked to sign the chain of custody form; and indeed the rules for chain of custody were violated several times by O’Connor and Scalercio. Soon after it appeared that the sample did not meet the temperature requirement Capaldi asked Frank if he wanted to stay on paid overtime to provide another specimen. Frank declined, explaining that his wife worked and that he had to go home to look after his two children, ages five and ten. Capaldi told Frank that he could leave; and he did. Ca-paldi testified that he expected that Scalercio and O’Connor would return another day for the requested retest.
At this time, O’Connor was on the telephone with the DPC. She told the DPC that Frank was leaving the building and that he had not completed or signed the chain of custody form. O’Connor was instructed to follow Frank and obtain his signature. However, Frank had already left the premises.
Agency procedure also requires that the collector immediately note anything unusual about the .donor’s behavior or the specimen on the form; a copy of the form is then to be given to the donor. Scalercio made no such notation; for example, he did not record on the form that Frank appeared nervous with his hand shaking when he handed Scalercio the specimen, although Scalercio later so testified. Also, neither Scalercio nor O’Connor recorded at the time that the color was not that of urine or that there was no odor, although O’Connor later testified that the sample was colorless and had no odor. No record appeared on the form that there was no odor of urine. At some time later, O’Con-nor wrote on the form that Frank’s specimen was slightly cloudy and colorless. This notation was absent from Mr. Frank’s copy; there is no explanation of when this information — the only recordation that the sample was colorless — was added to the “official” copy, although it was certainly after a copy was given to Frank. Further, the rules require that color be observed immediately and anything unusual recorded: See section II. B.l:
1. Immediately after collection, the collector shall ... conduct an inspection to determine the color and signs of contaminants. Any unusual findings resulting from the inspection shall be noted on the chain of custody form.
There was no notation concerning color on the copy of this form that was given to Frank. The laboratory report, issued on about January 14, 1993, was that the specimen was not urine.4 The sample was not otherwise identified. The additional sample that the DPC had requested was never obtained, or its necessity under FAA rules made known to Frank.
Mr. Frank was telephoned by Dr. Anderson, the FAA’s Medical Review Officer, on about January 15, 1993 (the briefs and record are not consistent on the date). During their conversation it became apparent to *1563Frank that he was the subject of potential disciplinary action, and he asked that the conversation be terminated until he had the opportunity to speak to either Capaldi or a union representative. Before that telephone call Frank was neither notified that disciplinary action might be taken against him, nor informed by the agency of his right to union representation, as required by Article 6 of the Bargaining Agreement, see infra.
On January 25, 1993 Frank was issued a Notice of Proposed Removal on the ground that he had submitted an adulterated specimen. He was duly removed.
PROCEDURAL VIOLATIONS
A
The several violations of the DOT rules of collection, custody, and sealing, of themselves require reversal of the disciplinary action against Mr. Frank. I can not agree that these errors were “harmless”. U.S. DOT Order 3910.1B and the implementing Drug Testing Guide establish the procedures that are required to be followed by agency personnel in conducting random drug tests. They were violated from the first to the last steps taken by these novice, untrained, and not credible (according to the arbitrator and their own contradictory statements) collectors.
1. Collection
In collecting the specimen the Guide requires, inter alia, that a bluing agent be placed in the toilet bowl and there must not be any other source of water. Scalercio and O’Connor testified that they placed bluing in the toilet, secured the faucets, and that O’Connor verified this was so prior to each specimen collected. The arbitrator overcame this critical aspect simply by finding the testimony of O’Connor and Scalercio not credible on this point, even as the arbitrator credited all other aspects of their testimony that were adverse to Mr. Frank. I deem not only the procedures, but the arbitrator’s treatment of them, seriously flawed.
2. Chain of Custody
The chain of custody requirements of the Guide are extensive and detailed, and include in section III.8.a:
(2) While performing the collection part of the chain of custody procedures, it is essential that the urine specimen and chain of custody documents be under control of the collector. The collector shall not leave his/her work station, even momentarily, without securing the specimen and chain of custody form. The specimen should be packaged for mailing before the collector leaves the site.
(3) The chain of custody form shall be utilized for maintaining control and accountability from the point of collection to final disposition of specimens. With each transfer of possession, the chain of custody form shall be dated, signed by the individual releasing the specimen, signed by the individual accepting the specimen, and the purpose for transferring possession noted....
(4) .... The collector and the individual providing the specimen shall always have the specimen within sight prior to its being seeded and labeled. The collector shall sign and date across the tape label sealing the container and ensure that the chain of custody documentation is completed and attached to each sealed container.
JA-305 (emphases added). None of the emphasized provisions was complied with.
3.Temperature Measurement
Guide section II.B.l includes the following requirements for the measurement of the specimen’s temperature:
1. Immediately after collection, the collector shall measure the temperature of the specimen (avoiding cross contamination of specimens) and conduct an inspection to determine the color and signs of contaminants. Any unusual findings resulting from the inspection shall be noted on the chain of custody form. The time from urination to delivery of the sample for temperature measurement is critical and in no case shall exceed U minutes. The individual giving the specimen will be asked to observe the reading of the temperature and the recording of that reading on the control *1564form. If the temperature of the specimen is outside the range of 32.5-37.7 degrees C/90.5-99.8 degrees F, this gives reason to believe the specimen has been tampered with. After consultation with and approval by the DPC, another specimen shall be collected under direct observation and both specimens forwarded to the laboratory. Any specimen suspected to be adulterated should always be forwarded for testing. (See paragraph B.2., “Direct Observation Collection Procedures.”)
JA-337 (emphases beyond first added). Neither the requirement for delivery for temperature measurement within four minutes, nor the collection of another specimen, were met.
4. Packaging and Sealing
Another relevant provision is:
M. Both the individual being tested and the collector should keep the specimen in view at all times until it has been packaged and sealed for shipment. If the specimen is transferred to a second container, the collector shall request the individual to observe the transfer of the specimen and the placement of the tamper-proof seal over the bottle cap and down the sides of the bottle.
JA-338 (emphases added). Again, the emphasized provisions were not complied with.
B
The panel majority agrees that there were errors in the chain of custody, but calls them harmless, stating that the bottle containing Frank’s specimen was sealed before it was left unattended, and that “there is nothing in the record indicating that the seal was tampered with.” As I have discussed, Scalercio testified that the bottle was not sealed until later, and the arbitrator did not find otherwise. O’Connor testified that she instructed Scalercio to return to the specimen and to “seal the bottle.” She later testified, correcting herself, that the bottle was already sealed. However, her first statement is consistent with Scalercio’s testimony, and Frank’s, that Frank was not present when his specimen was transferred from the second cup to the shipping bottle and then sealed.
There was no finding by the arbitrator on the critical question of whether the specimen was sealed before it was left unattended; and the weight of evidence is that it was not. Whether or not Frank really did adulterate his specimen I do not know; but the purpose of these careful procedures is to avoid just the kind of uncertainty and taint that here have arisen. I do not think that the cavalier breach of these elaborate procedures should be tolerated.
Nor can I agree with the majority that the only people present when the specimen was left unattended were “the three collectors, respondent, and possibly one other would-be donor.” Scalercio testified that at the time that he left the specimen unattended O’Con-nor and Capaldi were in Capaldi’s office. Scalercio also testified that when Frank went to Capaldi’s office, Scalercio’s attempts to remain close to Frank were thwarted because there were “a lot of people in the area.” The testimony of Scalercio and Frank supports no conclusion other than that the specimen was left in the second cup after the unsuccessful attempt to measure the temperature a second time, and that this cup was left unattended in an area where there were “a lot of people”.
Although the arbitrator found that these violations of the required procedures were not prejudicial to Frank because of the absence of evidence that any other person tampered with the specimen, the many breaches of the requisite protocols and procedures require that no prima facie case of adulteration can stand. The fact that the second test was never performed, although requested by the agency, further impeaches the total process whereby a low temperature reading, surrounded by incompetencies, resulted in summary separation from service. I conclude that the presumption of adulteration can not be upheld on this record, and that Frank was unjustly and irregularly denied the opportunity to rebut that presumption by providing a second specimen.
THE BARGAINING AGREEMENT
Union representation for federal employees is a statutory right, 5 U.S.C. § 7114(a)(2)(B); National Labor Relations *1565Board v. Weingarten, 420 U.S. 251, 263, 95 S.Ct. 959, 966-67, 43 L.Ed.2d 171 (1975), which may be modified through collective bargaining. 5 U.S.C. § 7106(b)(2); see Cornelius v. Nutt, 472 U.S. 648, 659, 105 S.Ct. 2882, 2889, 86 L.Ed.2d 515 (1985). The Bargaining Agreement included the following provisions:
ARTICLE 6. REPRESENTATION RIGHTS
Section 1. When it is known in advance that the subject of a meeting is to discuss or investigate a disciplinary, or potential disciplinary situation, the employee shall be so notified in advance. The employee shall also be notified of his/her rights to be accompanied by a Union representative if he/she so desires, and shall be given a reasonable opportunity to obtain such representation before the beginning of the meeting, if requested. If during the course of a meeting it becomes apparent for the first time that discipline or potential discipline could arise, the Employer shall stop the meeting and inform the employee of his/her right to representation if he/she so desires, and provide a reasonable opportunity to obtain representation [from the Union] before proceeding with the meeting, if requested....
Section 2. No disciplinary action may result from any such meeting unless the provisions of this Article have been met.
Frank argues that since he was not given the requisite notice of a potential disciplinary action, as required by Section 1, then in accordance with Section 2 no disciplinary action could properly ensue. This is a serious charge. It can not be disposed of on the basis that it was not raised before the arbitrator, for indeed, as the arbitrator reported at n. 5 of his opinion, it was raised at least in connection with the telephone call from the MRO. Further, Frank explicitly raised the issue of the absence of notice at the December 18 meeting that he was suspected of tampering, and the absence of notice of potential disciplinary action, in his February 24, 1993 written response to the Notice of Proposed Removal, Frank stating:
Although I was requested to stay overtime so that a direct observation sample could be taken, you never indicated to me that I was being suspected of tampering with my sample. I was unable to stay because of a prior commitment. However, if I had known or been advised that I was being suspected of tampering, I would certainly have made other arrangements.
JA-264 (emphasis added).
Our review is of the record before the arbitrator:
In any case filed in the United States Court of Appeals for the Federal Circuit, the court shall review the record and hold unlawful and set aside any agency action, findings, or conclusions....
5 U.S.C. § 7703(c) (1988) (emphasis added). Oshiver v. Office of Personnel Management, 896 F.2d 540 (Fed.Cir.1990), cited by the majority, holds simply that the appellant can not submit supplemental papers on appeal. Frank’s February 24, 1993 letter is of record and was referred to by the arbitrator. Further, the agency is charged with knowledge of the Bargaining Agreement; any lapse in compliance is not a technicality that can be ignored.
Although Capaldi was present when Frank’s specimen was declared out of the specified temperature range, raising the presumption of adulteration and the possibility of removal, none of the safeguards of Article 6 was followed in Capaldi’s subsequent meetings with Frank. Instead, Capaldi authorized Frank to go home, believing that another test would be made, knowing that the DPC had requested that another test be made. But for the ensuing events, this meeting between Capaldi and Frank may not have violated the Bargaining Agreement. Although in retrospect it is clear that Frank should have been told of possible disciplinary action, it appears that Capaldi in good faith believed that another test would be taken.
The harm to Frank is now apparent, for had he been advised of possible disciplinary action he might have sought immediate union representation, and in turn he might have been advised of the importance of staying on site for a second specimen. Thus it was error for the agency not to have viewed these events as a “potential disciplinary situation”, for the error clearly harmed Frank. See Cornelius v. Nutt, 472 U.S. at 656 n. 7, 105 *1566S.Ct. at 2888 n. 7 (“the requirement that harmful error have some likelihood of affecting the outcome of the agency’s decision”). As the government’s counsel stated during argument of this appeal:
COUNSEL: If [Frank] had given a second sample, both samples would have been tested. And the fact that the first sample still turned out to be water would have given grounds to instigate a removal proceeding. Whether or not the removal went forward after that, maybe it would have and maybe it wouldn’t have.
Indeed, maybe Mr. Frank would have been removed anyway. But maybe not, were the second sample drug-free and the first tainted, as it is, by the gaps in the chain of custody and the other procedural lapses.
Frank says that had he had the advice of his union representative, he would have made arrangements for his children and provided a second specimen that afternoon. The provisions of Article 6 reflect the purpose of timely representation for, as the Court stated in NLRB v. Weingarten, 420 U.S. at 263-64, 95 S.Ct. at 967, “it becomes increasingly difficult for the employee to vindicate himself [after the employee has been discharged or disciplined], and the value of representation is correspondingly diminished.”
CONCLUSION
The agency’s failure to follow its procedures in collecting and processing the sample, the failure to obtain the authorized second sample, and the failure to give notice to Frank that he was at risk, were all error, and not harmless. It is not disputed that the agency’s decision might have been affected by a promptly given, drug-free test. As the Court stated in Cornelius v. Nutt, 472 U.S. at 655, 105 S.Ct. at 2886-87,
in an appeal of an agency disciplinary decision to the Board, the agency’s failure to follow bargained-for-proeedures may result in its action being overturned, but only if the failure might have affected the result of the agency’s decision to take the disciplinary action against the individual employee.
When all of these errors are viewed in concert, the agency action is not defensible. Thus, respectfully, I dissent from the affir-mance of the arbitrator’s decision.
. This undisputed explanation of why Frank did not provide the first specimen requires fan-weight, for the government's brief leaves the impression that adverse inferences are warranted. Indeed the record shows that three of the five persons selected for testing that day did not provide a specimen at the first request.
. Robert Fisher — an air traffic controller who was in the collection area for a second attempt to provide a specimen — also testified that Scalercio first read the sample temperature to be 92° F, and that Frank verified this measurement.
.The panel majority states that the sample was sealed "immediately” after the temperature was taken, a fact on which Scalercio and O'Connor did not agree, and that Frank testified did not happen. Following is an example of Scalercio’s testimony under oath, elicited on cross-examination:
Q Didn’t you testify that Mr. Frank had already left the building when you prepared the samples for shipment to the lab?
A We — no. I told you that he gave me the specimen. The temperature was out of range. We tried another cup. There was no temperature on it at all. We transferred to a *1562shipping bottle. I sealed the bottle, signed it, put it in the box, sealed the box.
Q You're testifying now that Mr. Frank was there when this happened?
A I don't believe he was exactly when—
Q Answer yes or no. Was Mr. Frank present when you did this?
A No.
On the same point, O’Connor testified:
Q Was Mr. Frank with you?
A Yes, yes. He must be and he was. And we poured the specimen from the cup into the sealed bottle. We placed a cap over the sealed bottle. Then you put the custody, bottle custody seal on top of it.
O’Connor later made and corrected a contradictory statement, which the majority describes as nothing more than a misstatement.
Frank’s testimony matched Scalercio's:
A Well, when I initialed the tag, it was on the form that I believe they peeled it off. Now, I don't remember where the tag came from, to be honest with you.
Q But the tag was not on the bottle?
A No, not that I recall.
I have no way of knowing the truth. However, I do think that O’Connor's testimony, contradicted by Scalercio and Frank, is a slim basis indeed for depriving Mr. Frank of his job.
. This determination was made by testing the sample's pH, specific gravity, and creatinine concentration. There is an unexplained discrepancy between those results on the lab data sheets and those reported to the agency. The pH of this specimen was recorded as 7.3 on the lab data sheets, but the pH reported to the agency was 6.8. The record offers no explanation of this mystery; but it is one more reason for deeming unreasonable the agency’s action in firing Mr. Frank without the authorized second test.