concurring.
I concur in the principal opinion. I write separately to emphasize my belief that the time has come for this Court to clarify what standard it intends to apply in determining whether error in a criminal trial is harmless.
In Chapman v. California, 386 U.S. 18, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967), the Supreme Court concluded that there may be instances in which a federal constitutional error might be considered harmless. It held, however, that “before a federal constitutional error can be held harmless, the court must be able to declare a belief that it was harmless beyond a reasonable doubt.” Id. at 24, 87 S.Ct. at 828. Under that test, the “beneficiary of a constitutional error,” the government, must “prove beyond a reasonable doubt that the error complained of did not contribute to the verdict obtained.” Id. The Court simply applied the level of certainty necessary for conviction of a criminal defendant, see In re Winship, 397 U.S. 358, 361-64, 90 S.Ct. 1068, 1070-72, 25 L.Ed.2d 368 (1970), to the determination whether a federal constitutional error is harmless. The reasonable doubt standard, the Court said, is a “workable” one because it is “familiar ... to all courts.” Chapman, 386 U.S. at 24, 87 S.Ct. at 828.
Because of the high nature of federal constitutional rights, it would seem that a federal constitutional error would more grievously infringe the rights of the accused than would an error of state criminal law or procedure. It follows that the burden of demonstrating that a federal constitutional error is harmless should be the most difficult harmless .error burden the government should be made to bear. Yet this Court has embraced an even more rigorous harmless error standard applicable to those cases, such as this one, in which evidence was admitted when it should have been excluded. That test, we have said, is that “error in the admission of evidence should not be declared harmless unless it is so without question.” State v. Degraffenreid, 477 S.W.2d 57, 64 (Mo. banc 1972). This has been characterized as “for all practical purposes the same as” the reasonable doubt standard, State v. Brown, 549 S.W.2d 336, 345 (Mo. banc 1977), and there have also been hints that it is lower than the reasonable doubt standard, see State v. Cook, 628 S.W.2d 657, 662 (Mo. banc 1982) (Seiler, J., concurring); id. at 665 (Donnelly, C.J., dissenting). Neither, of course, is accurate. It is clear that the Degraffenreid *622standard is higher, for the requirement that any error be harmless without question leaves no room for reasonable doubt.
It is perhaps because the Degraffenreid standard places an onerous burden on the state that this Court has eschewed application of it in favor of the reasonable doubt standard. For example, we have applied the reasonable doubt standard, in accordance with Chapman, in cases allegedly involving federal constitutional error. E.g., State v. Bonuchi, 636 S.W.2d 338, 341 (Mo. banc 1982), cert, denied, - U.S. -, 103 S.Ct. 1206, 75 L.Ed.2d 446 (1983); State v. Johnson, 618 S.W.2d 191, 195 (Mo.1981); State v. Baskerville, 616 S.W.2d 839, 844 (Mo.1981); State v. Kent, 515 S.W.2d 457, 460 (Mo. banc 1974); State v. Adams, 497 S.W.2d 147, 152 (Mo.1973); State v. Crow, 486 S.W.2d 248, 252 (Mo.1972); State v. Granberry, 484 S.W.2d 295, 300 (Mo. banc 1972). We also have applied the reasonable doubt standard in cases involving erroneous exclusion, rather than erroneous admission, of evidence. Burton v. State, 641 S.W.2d 95, 100 (Mo. banc 1982). See State v. Clark, 592 S.W.2d 709, 718-19 (Mo. banc 1979), cert, denied, 449 U.S. 847, 101 S.Ct. 132, 66 L.Ed.2d 57 (1980).
More telling is our vacillation in application of the harmless error rule in cases in which Degraffenreid was clearly applicable because the issue involved erroneous admission of evidence as a matter of state law. In some cases we have applied the Degraf-fenreid standard. E.g., State v. Haggard, 619 S.W.2d 44, 48 (Mo. banc 1981), vacated and remanded on other grounds,-U.S. -, 103 S.Ct. 1171, 75 L.Ed.2d 423 (1983). In others we have applied the reasonable doubt standard instead. For example, the defendant in State v. Wright, 582 S.W.2d 275 (Mo. banc 1979), was convicted of first degree robbery. The dispositive issue was whether the trial court erred by admitting into evidence a wallet seized from the defendant that did not belong to the defendant, his accomplices, or the victim. The defendant argued successfully that the wallet constituted evidence of a crime with which he was not charged and that he was prejudiced by its admission. In reaching its conclusion that the error was not harmless, the Court recited the Degraffenreid standard, id. at 277, but it noted that “the burden is upon the respondent-state to show this evidence was harmless error beyond a reasonable doubt,” id. at 278. In another case, State v. Stephens, 507 S.W.2d 18, 19 (Mo.1974), the Court said that any error in admission of the challenged in-court identification was harmless beyond a reasonable doubt.
The foregoing illustrates the confusion and inconsistency of application of tests for harmless error that Degraffenreid has spawned. I am persuaded that the Degraf-fenreid standard, even limited in application to cases in which evidence was erroneously admitted, is flawed. First, as indicated above, the Degraffenreid standard places a greater value on a proper application of state criminal law and procedure than is placed on rights guaranteed by the federal constitution. Second, that standard anomalously imposes upon the state a greater burden in demonstrating error to be harmless than the state must bear in proving guilt in the first instance. Third, it is unclear just when the Degraffenreid standard should be applied. A tension exists between the Degraffenreid standard and the reasonable doubt standard when the erroneous admission of evidence impinges upon a federal constitutional right. De-graffenreid did not involve a federal constitutional error, but nothing in the Court’s language in that case limits its application to nonconstitutional eases. Degraffenreid should be overruled to the extent that it applies a harmless error rule more stringent than the reasonable doubt standard, which I believe should be applied in all cases.
I concur in the principal opinion because I believe that the error in this case was harmless. The evidence against appellant was overwhelming, and “error which in a close case might call for a reversal may be disregarded as harmless when the evidence of guilt is strong.” Degraffenreid, 477 S.W.2d at 65. The evidence of which appellant complains was cumulative. It was virtually identical to the testimony of the two witnesses, one of whom was a law enforcement officer, to whom appellant admitted his guilt in a conversation prior to his testimo*623ny in the trial of co-defendant Stufflebean. “Even observing the caution noted in De-graffenreid, with reference to cumulative evidence, the admission of the [testimony in the previous case] was both cumulative and harmless.” Haggard, 619 S.W.2d at 48. On this state of the record, the error was harmless even under the erroneous Degraffen-reid “harmless without question” standard. A fortiori, it was harmless beyond a reasonable doubt.
We do a disservice to the lower trial and appellate courts of this state when we fail to articulate clearly which harmless error rule we intend to apply. The principal opinion, like Wright, recites the Degraffen-reid standard and then applies the reasonable doubt test. Our lower courts confront issues of harmless error much more frequently than do we, and they expect, and deserve, more guidance in this important area than we have given. The signals we send should be consistent rather than conflicting. That could be accomplished by overruling Degraffenreid and applying the reasonable doubt standard to all questions of harmless error.