STARR, Circuit Judge:

This case arises from an administrative denial of a within-grade salary increase to a government physician. Dr. Ayub K. Ommaya was at all times relevant to this case a GS-15 neurosurgeon employed by the National Institute of Neurological and Communicative Disorders and Stroke (NINCDS), a component of the National Institutes of Health (NIH). Dr. Ommaya was denied a within-grade increase in December 1979 because of his alleged failure to perform at an acceptable level of competence for the period from December 1976 to December 1, 1979. However, inasmuch as Dr. Ommaya had not received a 60-day advance notice of performance deficiencies required by the governing regulation, 5 C.F.R. § 531.407(b), he was provided a 60-day review period by NIH within which to improve his performance. 5 C.F.R. § 531.407(c)(5). Joint Appendix (J.A.) at 255.

After the reevaluation of his performance at the conclusion of the sixty-day period, Dr. Ommaya was again denied the within-grade salary increase on February 6, 1980. Petitioner sought reconsideration of the two denials within NIH. The Director of NINCDS, however, sustained the denials. J.A. at 255.

Petitioner thereupon appealed the decisions to the Merit Systems Protection Board ("MSPB" or "the Board"). J.A. at 183. In an initial decision issued in May 1981, the MSPB presiding official reversed NIH's December 1979 denial of petitioner's within-grade increase and ordered that the increase be granted retroactively. In a series of findings, the presiding official, first, upheld the NIH's charge that Dr. Ommaya failed to publish an adequate number of scientific articles on his research. J.A. at 186, 197. Second, the presiding official reversed the NIH on the charge that Dr. Ommaya had failed to focus his research in an adequately narrow fashion, concluding that Dr. Ommaya had never been instructed specifically to do so. J.A. at 191. Third, as to the charge that Dr. Ommaya had inadequately supervised clinical and research functions, the presiding official upheld the NIH's assertion that Dr. Ommaya had improperly mixed research protocols, but rejected the charge that he had failed to maintain proper patient records or to supervise research personnel. J.A. at 191-93. The presiding official concluded that because Dr. Ommaya's "responsibility for publishing his clinical findings does not appear to have been a major part of his duties," he was entitled to his within-grade salary increase. J.A. at 197. NIH filed a timely petition for review before the MSPB.

The Board issued its final opinion and order on June 16, 1982, reversing the presiding official's initial decision insofar as it overturned the agency's December 1979 denial of the salary increase. The Board overturned as not supported by the evidence the presiding official's conclusion that Dr. Ommaya was not specifically told to focus his research activities more narrowly. J.A. at 256-58. The Board also rejected the presiding official's finding that "publication of scientific findings was not an important part of [Dr. Ommaya's] duties." J.A. at 259. Finally, the Board reversed the presiding official's conclusion that Dr. Ommaya had not failed to maintain patient records or to supervise research personnel. J.A. at 279-80. The petitioner thereupon appealed the Board's adverse decision to this court.

While Dr. Ommaya's appeal was pending, this court decided White v. Department of the Army, 720 F.2d 209 (D.C. Cir. 1983). In White, the court reviewed an MSPB decision upholding the Department of the Army's denial of a within-grade salary increase to a civilian employee. The court concluded that the applicable statute, 5 U.S.C. § 7701(c)(1) (1982), mandates application of a "preponderance of the evidence" standard, rather than a "substantial evidence" standard, when the MSPB reviews agency denials of within-grade salary increases. Id. at 212-14. The court, accordingly, reversed and remanded the case for reconsideration, since the Board had applied the substantial evidence test in reaching its decision. Id. at 214.

As in White, both the presiding official and the MSPB applied the substantial evidence test in reaching their respective decisions in this case. J.A. at 184, 197-98, 260-61. In no uncertain terms, the presiding official stated that the substantial evidence test, rather than the preponderance of the evidence test, was being applied. "In order for the agency's decision denying an appellant's within-grade increase to be upheld on appeal, it must be supported by substantial evidence." J.A. at 184 (citing Parker v. Defense Logistics Agency, 1 MSPB 489 (1980)). See also J.A. at 197-98. Further, the MSPB, in its decision reversing the presiding official's grant of the petitioner's salary increase, concluded that "the evidence of record . . . meets the agency's burden of supporting the charges by substantial evidence." J.A. at 260-61 (emphasis added). It is thus clear that, under the supervening decision in White, the MSPB employed the incorrect test in deciding the propriety of NIH's personnel action with respect to Dr. Ommaya.

At oral argument, respondent, while acknowledging both White's clear holding and its applicability to the case at hand, suggested that this court apply the doctrine of harmless error and affirm the Board's decision. See Doe v. Hampton, 566 F.2d 265, 277-78 (D.C. Cir. 1977) (doctrine of harmless administrative error applies when "in all likelihood [the error] would not have affected the result"), Delta Air Lines, Inc. v. CAB, 564 F.2d 592, 598 (D.C. Cir. 1977) (courts should not upset a decision because of errors, such as a subsidiary finding of fact, that are not material); cf. Kotteakos v. United States, 328 U.S. 750, 776, 66 S.Ct. 1239, 1253, 90 L.Ed. 1557 (1946) (error is harmless unless it has a "substantial and injurious effect" on the verdict).

Under settled law, application of the harmless error doctrine is appropriate when it is clear that the error complained of would not have affected the result in the case. Thus, in Doe v. Hampton, supra, 566 F.2d at 277-78, the erroneous admission of a letter into evidence was deemed harmless when the letter contained only cumulative evidence. Similarly, in Delta Air Lines, Inc. v. CAB, supra, 564 F.2d at 598, this court held that the CAB's route award to Eastern Airlines should not be overturned simply by virtue of minor factual errors by the Board on such matters as the competing airlines' respective economic predictions and the projected load factor of the successful airline. See also Cohen v. United States, 369 F.2d 976, 988, 177 Ct.Cl. 599 (1966) (introduction of remarks of trial judge concerning indictment was harmless error because there was no evidence that the trier of fact had relied upon the remarks).

While the MSPB's suggestion is not without appeal, we are nonetheless constrained to conclude that White precludes the application of the harmless error doctrine to cases such as the one at hand. In White, we specifically rejected the very similar argument that the MSPB's decision at issue there should be affirmed "because an independent assessment of the evidence would show that the personnel action would in any case be sustainable under the preponderance of the evidence standard." White, 720 F.2d at 210. We stated in unmistakable language that such an inquiry "is properly the province of the agency, not this court." Id. We did not mince words in this respect. "If MSPB relied on incorrect legal grounds, it would be error for this court to enforce without first remanding for agency examination of the evidence and proper fact-finding." Id. (emphasis added).

In White's wake, remand in this case seems inescapable, notwithstanding the obvious care with which the Board reached its decision. Indeed, remand seems particularly called for where, as here, the presiding official had ruled in favor of the petitioning employee in the first instance. Moreover, where, as here, the error goes to the entire proceeding, inasmuch as all of the evidence was considered under the substantial evidence test, we are reluctant to conclude that the harmless error doctrine should be applied.

We are fortified in this disposition by the classic statement of the rationale of the harmless error doctrine, in which the Supreme Court suggested that departures from an express command of Congress should not be deemed harmless. In Kotteakos v. United States, a criminal case, Justice Rutledge, speaking for the Court, suggested two categories of cases to which the doctrine should not apply, namely, "where the departure is from a constitutional norm or a specific command of Congress." 328 U.S. 750, 764-65, 66 S.Ct. 1239, 1247-48, 90 L.Ed. 1557 (1946) (citation omitted).

As demonstrated above, the Board here has clearly departed, in light of the supervening decision in White, from the congressional command that agency denials of a within-grade salary increases be reviewed under a preponderance of the evidence standard. In such circumstances, the dictum in Kotteakos, coupled with the holding in White, lead us to decline the invitation to invoke the harmless error doctrine in the context of this case. We, of course, intimate no view on the question to be resolved by the Board on remand, namely whether the NIH's actions challenged by the petitioner are supported under the preponderance of the evidence standard.

Reversed and remanded.