concurring in part and dissenting in part:
I agree with the majority that Harry-man’s Miranda rights were violated by the *884prosecution’s repeated use of his statement at trial, and I completely concur in Part I of Judge Johnson’s thorough opinion. I part company with the majority, however, in its conclusion that the error was harmless. My reading of the Supreme Court cases upon which the majority’s analysis relies convinces me that the wrong standard of review has been applied to Harryman’s claim of prejudice. I therefore dissent from Part II of the majority opinion.
The standard of review the majority uses requires the reviewing court to “decide whether, absent the so-determined unconstitutional effect, the evidence remains not only sufficient to support the verdict but so overwhelming as to establish the guilt of the accused beyond a reasonable doubt.” Supra, p. 876. This is an easy enough standard to employ: The appellate judge, sitting, as it were, as a thirteenth juror, reviews the sufficiency of the nonobjectionable evidence introduced at trial and either affirms or reverses as he or she would vote either to convict or to acquit. The majority refers to six cases in support of its position on harmless error review: Brown v. United States, 411 U.S. 223, 93 S.Ct. 1565, 36 L.Ed.2d 208 (1973); Milton v. Wainwright, 407 U.S. 371, 92 S.Ct. 2174, 33 L.Ed.2d 1 (1972); Schneble v. Florida, 405 U.S. 427, 92 S.Ct. 1056, 31 L.Ed.2d 340 (1972); Harrington v. California, 395 U.S. 250, 89 S.Ct. 1726, 23 L.Ed.2d 284 (1969); Chapman v. California, 386 U.S. 18, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967); and Fahy v. Connecticut, 375 U.S. 85, 84 S.Ct. 229, 11 L.Ed.2d 171 (1963). A close reading of these cases reveals instead that they do not support the majority’s interpretation. Indeed, the earliest of these, Fahy, expressly rejects the approach taken by the majority in favor of a yet more exacting standard, and none of the cases following can be said to have qualified the policy set forth in Fahy.
Fahy v. Connecticut, supra, applied a common sense rule of harmless error, one which looks to whether the objectionable evidence might have contributed to the conviction. There the Supreme Court concluded
that the erroneous admission of this unconstitutionally obtained evidence at this petitioner’s trial was prejudicial; therefore, the error was not harmless! and the conviction must be reversed. We are not concerned here with whether there was sufficient evidence on which the petitioner could have been convicted without the evidence complained of. The question is whether there is a reasonable possibility that the evidence complained of might have contributed to the conviction.
375 U.S. at 86-7, 84 S.Ct. at 230 (emphasis added). The majority concedes the prejudicial impact of the erroneous and repeated use of Harryman’s statement by the prosecution. But in relying on the strength of the rest of the State’s case against Harry-man, the majority ignores the Supreme Court’s express directive that we must consider whether the erroneously admitted evidence might have contributed to the conviction. Judge Hill’s concurrence points accurately to the impact of a confession on a jury.
None of the cases following Fahy has relaxed the standard of review of similar errors of constitutional criminal procedure. Four years after Fahy the Court had to decide in Chapman v. California, supra, whether a California murder conviction which had been obtained in violation of the Court’s recent ruling in Griffin v. California, 380 U.S. 609, 85 S.Ct. 1229, 14 L.Ed.2d 106 (1965), had also amounted to harmless constitutional error. In reaffirming the standard enunciated in Fahy, the Court concluded that “[a]n error in admitting plainly relevant evidence which possibly influenced the jury adversely to a litigant cannot, under Fahy, be conceived of as harmless.” 386 U.S. at 23—4, 87 S.Ct. at 828 (emphasis added). Thus, even though the rest of the State’s case is overwhelming, constitutional error in the admission of evidence cannot be harmless if it nevertheless has a prejudicial impact on the deliberations of the jury. The Chapman Court expanded on the requirements for a showing of harmless error, adding the elements of who had the burden of proof on the issue of harmless error (the “beneficiary of a constitutional error,” id.) *885and how great a burden that was on review (proof of harmlessness “beyond a reasonable doubt” id).
In Harrington v. California, supra, the issue was whether the California trial court had committed harmless error in admitting into evidence against Harrington the confessions of two nontestifying codefendants who were, consequently, not subject to cross-examination, contrary to the Court’s recent decisions in Bruton v. United States, 391 U.S. 123, 88 S.Ct. 1620, 20 L.Ed.2d 476 (1968), and Pointer v. Texas, 380 U.S. 400, 85 S.Ct. 1065, 13 L.Ed.2d 923 (1965). In finding no prejudicial error in Harrington’s conviction, Justice Douglas did not qualify the Chapman standard of review. Indeed, he expressly reaffirmed it:
Our judgment must be based on our own reading of the record and on what seems to us to have been the probable impact of the two confessions on the minds of an average jury. We admonished in Chapman . . . against giving too much emphasis to “overwhelming evidence” of guilt, stating that constitutional errors affecting the substantial rights of the aggrieved party could not be considered to be harmless. .
We do not depart from Chapman ; nor do we [delete] it by inference. We reaffirm it.
395 U.S. at 254, 89 S.Ct. at 1728 (emphasis added).1
Each of those cases coming after Harrington was decided on precisely the same basis as was Harrington’s. See Schneble v. Florida, 405 U.S. at 429, 92 S.Ct. at 1058,2 Milton v. Wainwright, 407 U.S. at 373, 92 S.Ct. at 2175,3 and Brown v. United States, 411 U.S. at 226, 93 S.Ct. at 1568.4
The proliferation of dicta in these cases has led the majority into applying the wrong standard of harmless error review. Unnecessary language seems to invite review of the sufficiency of the nonobjectionable evidence alone, even though our attention over and over again is properly called to the prejudicial impact of the objectionable evidence. Thus, we read in Harrington that the case against Harrington was “so overwhelming that unless we [can] say that no violation of Bruton can constitute harmless error, we must" leave this state conviction undisturbed,” 395 U.S. at 254, 89 S.Ct. at 1729; in Schneble that “[i]n this case, we conclude that the ‘minds of an average jury’ would not have found the State’s case significantly less persuasive had the [nontestifying codefendant’s] admissions been excluded,” 405 U.S. at 432, 92 S.Ct. at 1060; in Milton that “[our] review . . . leaves us with no reasonable doubt that the jury at petitioner’s 1958 trial would have reached the same verdict without hearing [the interrogating officer’s] testimony,” 407 U.S. at 377, 92 S.Ct. at 2178; and in Brown that “[t]he testimony erroneously admitted was merely cumulative of other overwhelm*886ing and largely uncontroverted evidence properly before the jury,” 411 U.S. at 231, 93 S.Ct. at 1570. Yet each of these cases cites Chapman approvingly, applying the Chapman standard of review.
This interpretation of the standard of review for constitutional error is buttressed by the fact that review of nonconstitutional error in the federal courts has focused for the past thirty years on the impact of the alleged error in light of the surrounding circumstances, notwithstanding the sufficiency of the independent evidence.5 In construing the statutory predecessor to the harmless error provision in the Federal Rules of Criminal Procedure,6 the Supreme Court in 1946 regarded the question of harmless error as not whether “they [the jury] were right in their judgment, regardless of the error or its effect upon the verdict. It is rather what effect the error had or reasonably may be taken to have had upon the jury’s decision.” Kotteakos v. United States, 328 U.S. 750, 764, 66 S.Ct. 1239, 1247, 90 L.Ed. 1557.
If, when all is said and done, the conviction is sure that the error did not influence the jury, or had but very slight effect, the verdict and the judgment should stand, except perhaps where the departure is from a constitutional norm or a specific command of Congress. . But if one cannot say, with fair assurance, after pondering all that happened without stripping the erroneous action from the whole, that the judgment was not substantially swayed by the error, it is impossible to conclude that substantial rights were not affected. The inquiry cannot be merely whether there was enough to support the result, apart from the phase affected by the error. It is rather, even so, whether the error itself had substantial influence. If so, or if one is left in grave doubt, the conviction cannot stand.
Id., 328 U.S. at 764-65, 66 S.Ct. at 1248 (citation and footnote omitted, emphasis added). See, generally, Wright, Federal Practice and Procedure, Criminal §§ 852-55. The Fahy and Chapman Courts were not writing on a clean slate when their turn came to assess the harm in constitutional errors, for the standard of review they applied was no more rigorous, nor less exacting, than that already applicable to errors of a nonconstitutional variety.7
In any case the amount of the independent, nonobjectionable evidence is certainly relevant in assessing whether an asserted error is harmless. The same error would, of course, figure more largely in a close case than in an easy one. But although highly relevant, overwhelming independent evidence of the guilt of the accused is not enough, under Fahy and Chapman, for a finding of harmless constitutional error. However overwhelming the evidence against the accused, error cannot be harmless if prejudicial impact is conceded. Any more lax standard of harmless error review, such as that used by the majority, does not follow the rule in the cases discussed above. Moreover, such a standard of harmless error candidly concedes that the rule of law, here Harryman’s right to have the jury weigh the evidence against him free of the damaging impact of the admission proscribed by Miranda, varies as between the clearly guilty and the not-so-clearly guilty.
The majority in Part I of the opinion rightfully states that the rigidity of the Miranda rule is its core strength. Policemen, prosecutors, and trial courts should *887have rules that are not bent by appellate courts granting so many exemptions that the rule becomes meaningless. In Part II of the opinion I think the majority bends the Chapman harmless error rule. We become the thirteenth juror. We should confine ourselves to answering the question: Was there a reasonable possibility that the inadmissible evidence contributed to the conviction? The majority concludes in this case that there was by saying: “There is no question that the statement admitted against Harryman was incriminating. That it was so [was] evidenced by the decision of one of the prosecutors to refer to it no less than three times in his closing argument to the jury.” Supra, pp. 876-877. If the Chapman rule remained rigid, prosecutors would win their cases with their additional “overwhelming evidence of guilt” without infecting the case with the unneeded constitutionally questionable evidence.
The panel majority concluded that the State failed to carry its burden of demonstrating beyond a reasonable doubt that the error complained of did not contribute to the verdict. Since this standard was correct both in principle and in authority, I would affirm the panel’s decision reversing the district court’s denial of habeas relief. Accordingly, I respectfully dissent.
. Confining his analysis to “these special facts," 395 U.S. at 253, 89 S.Ct. at 1728, Mr. Justice Douglas concluded that Chapman was satisfied: All four defendants had confessed to more or less of the State’s case. A nonobjectionable admission by Harrington placed him at the scene of the crime. The confessions of two codefendants, Bosby and Cooper, however, merely placed someone of the same race as Harrington at the scene of the crime. (Harrington was white, all three codefendants were black.) It was to those portions of those codefendants’ confessions, that “a white guy” was also at the scene of the crime, that Harrington based his Bruton objection, on the ground that those codefendants did not take the stand. But, as the court concluded, any error in admitting the uncontradictable confession-testimony of Bosby and Cooper was completely submerged in Harrington’s own admissible confession, admitting in whole the contents of those codefendants’ testimony against him.
. Where the Court expressly assumed that the content of the codefendant’s admission was merged into certain admissions made by Schneble and that “these admissions were properly before the trial court.”
. Where the testimony objected to was merged into “no less than three full confessions” whose validity and, apparently, admissibility were not challenged on appeal.
. Where the Bruton error was harmless in light of the fact that “[tjhose considerable parts of each petitioner’s confession which did not implicate the other were admitted without objection.”
. Harryman’s appeal, of course, is from a denial of habeas relief from a state court conviction. There is no reason why harmless error in the context of federal criminal appeals is any less applicable here, where the error committed by the Texas trial court was a failure to apply federal constitutionally mandated criminal procedure.
. “Any error, defect, irregularity or variance which does not affect substantial rights shall be disregarded.” F.R.Crim.P. 52(a).
. In Kotteakos, e. g., the Court reversed convictions for conspiring to obtain fraudulent loans under the National Housing Act. The Court found reversible error under the standard it enunciated even though the lower court, Learned Hand, J., had “painstakingly examined the evidence relating directly to each of the petitioners!,] found it convincing to the point of making guilt manifest [, . . ] and concluded that reversal would be a miscarriage of justice.” 328 U.S. at 775, 66 S.Ct. at 1253.