join, dissenting.
The heart of Judge Kalodner’s opinion for the Court is the language:
“Wife murder is an atrocious crime, revolting and abhorrent t'o the conscience of our society. No cautionary instruction could purge the jury’s mind and memory of the devastating impact of the question ‘You say your wife was killed. You killed her, didn’t you?’ The question irretrievably seared itself into the conscious and subconscious minds of the jury.”
As is often the case, the reviewer of the drama of United States v. Larry J. Gray has injected far more drama into the review than may be found in the script. The judges who join in the majority were not present to observe the irretrievably searing effect of the prosecutor’s question. The trial judge was present, observed the demeanor and inflection of the prosecutor and the witness, and the reactions of the jurors. Yet from carefully selected and isolated portions of the printed record the majority concludes that the prosecutor’s question “irretrievably seared itself into the conscious and subconscious minds of the jury.” On the record as a whole, I see no reason whatsoever to conclude that in denying a motion for a mistrial the trial judge, on the scene, abused his discretion. Appellate courts should not substitute dramatic description or poetic license for reasoned analysis. Reasoned analysis, I suggest, requires an affirmance in this case.
At the trial Gray was positively identified by the teller who had been the victim of the holdup. Other witnesses testified that Gray resembled the holdup man. A handwriting expert employed by the Federal Bureau of Investigation testified that Gray had written the holdup note which was used in the commission of the robbery. In the Government’s case an agent of the Federal Bureau of Investigation, in order to establish the authenticity and chain of possession of handwriting samples to which the expert made comparisons, testified without objection that he obtained them from Gray at Holmesburg Prison. There was no offer by the defendant to stipulate to the authenticity and chain of possession of these samples. There was no testimony at this point about the reason for Gray’s incarceration in Holmesburg. There was nothing more than the establishment, without objection, of the time and place of obtaining the samples. The majority opinion describes this as “plain error” within the meaning of Rule 52(b), Fed.R.Crim.P. If, when on retrial of this case, the defendant declines to stipulate and the Government seeks to establish the authenticity and chain of possession of the samples the trial court, forewarned by the majority opinion, will undoubtedly rule that it is sufficient for the F.B.I. agent to answer the questions “from whom and when did you obtain the samples”, rather than the questions “from whom, when and where.” But to say that in the first trial the district court committed plain error, when the de*264fendant made no offer to stipulate respecting the samples, although he knew very well they were obtained from him while he was imprisoned, is to require a degree of foresight by trial judges far greater than can be anticipated from mere mortals. The question sought relevant evidence, and the ability to guard against possible collateral prejudice from the admission of that evidence lay primarily within the control of the defendant.
Gray’s defense was alibi. He took the stand to establish that alibi and to contest the Government’s identification evidence. Up to the time he took the stand there was no testimony as to the reason for his incarceration in Holmesburg Prison. In fact, however, he was being held on a murder charge growing out of his wife’s death. In order to cast doubt upon the identification testimony of a government witness, Gray denied that on June 17, 1968, the date of the robbery, he wore glasses. He then testified:
“Q. Do you have glasses now?
A. Yes, sir, I have glasses now.
Q. Now, when Mrs. Lindsay was in the courtroom you produced a pair of glasses. Do you have them with you?
A. Yes.
Q. Would you take them out and show them to us ? Hold them up.
A. (Indicates)
Q. Where did they come from? Where did you get those ?
A. I had trouble reading in Holmes-burg Prison while I was locked up waiting trial on my wife’s death.
* * * * * *
Q. When was that?
A. Around January of ’69.
* * * * * *
Q. You did not have these glasses in 1968?
A. No, I did not have these glasses in 1968.”
Gray also challenged the testimony of the government identification witness concerning his weight. Gray claimed that he always weighed 174 to 175 pounds rather than the 157 pounds which the witness described. On cross examination the Government inquired whether in September 1969 Gray weighed 157 pounds. At that time Gray was in Holmesburg Prison. He answered:
“A. Well, let me see, there was one period there after my wife was killed, I lost quite a lot of weight. Well, not before she was killed, but when I was locked up from March 31, June, July, August, three months I was locked up. I lost quite a bit of weight at that time, at that period.
* * * * * *
Q. So what you said earlier, that you were always 174 or 175 pounds, wasn’t quite correct?
A. It was under ordinary circumstances. Those were very extraordinary circumstances. I was in a lot of mental anguish at the time.
Q. So in 1968 you could have weighed 157 pounds, too, couldn’t you, in June, June 17?
A. No, no.
Q. In other words, you lost 17 pounds between June, ’68 and September, ’69?
A. Well, that could be checked from the official files of the Budd plant, because I get weighed there.
Q. You say your wife was killed, you killed her, didn’t you ?”
As Judge Kalodner’s opinion points out, the last question was objected to and the jury was removed from the courtroom. The Assistant United States Attorney represented that Gray had in fact been found guilty of the voluntary manslaughter of his wife and had been sentenced. He proposed to establish this conviction and a conviction for robbery. A defense motion for a mistrial was denied, and the trial court gave the cautionary instruction which, with editorial emphasis, Judge Kalodner’s opinion quotes. After Gray was found guilty he made a motion *265for a new trial on the ground, among others, that the question “You say your wife was killed, you killed her, didn’t you?” was in the circumstances so inherently prejudicial that a mistrial was required. The trial court denied this motion, finding specifically that considering the magnitude of the evidence against the defendant, if the use of the question by the Assistant United States Attorney was error at all, it was harmless beyond a reasonable doubt. The majority completely disregards that finding and substitutes its own finding that the question “irretrievably seared itself into the conscious and subconscious minds of the jury.”
The “searing” question came at a point in the trial when, in an effort to cast doubt on the Government’s identification testimony, Gray had twice in his own testimony made reference to his incarceration awaiting trial because of his wife’s death. The trial judge was in a position to observe the effect on the jury of Gray’s self-servicing references to his incarceration, and the effect, if any, on the jury of the prosecutor’s cumulative question. The trial judge observed the effect of his cautionary instruction. Substitution for his finding that the question was harmless beyond a reasonable doubt of an appellate court finding that the question “irretrievably seared itself into the conscious and subconscious minds of the jury” is both poor appellate practice and poor amateur psychiatry. At best both the appellate court and the trial court are speculating as to what actually “seared” the minds of the jury. The trial court at least had the benefit of the opportunity for direct observation.
The form of the question asked by the Assistant United States Attorney was objectionable. Yet, for impeachment purposes the Government in the district courts in this circuit ordinarily may establish that the witness has been convicted of a felony or misdemeanor amounting to crimen falsi. United States v. Mitchell, 427 F.2d 644, 647 (3d Cir. 1970); United States v. Evans, 398 F.2d 159, 164 (3d Cir. 1968); United States v. Montgomery, 126 F.2d 151, 155 (3d Cir.), cert. denied, 316 U.S. 681, 62 S.Ct. 1268, 86 L.Ed. 1754 (1942). Gray was not asked about a crime for which he had not been convicted. Cf. United States v. Nettl, 121 F.2d 927 (3d Cir. 1941). Thus, for impeachment purposes, at the time of the trial of this ease, the Government could, by asking the proper question, have established Gray’s conviction for manslaughter. Indeed the majority acknowledges as much.
Here, I submit, is where the analytical inadequacy of the majority opinions is most apparent. As the transcript reveals,1 the objection by defense counsel to any reference to the manslaughter conviction was based on his assumption that the federal courts should follow the rule of Commonwealth v. McIntyre, 417 Pa. 415, 208 A.2d 257 (1965). The majority declines to subscribe to the McIntyre rule for the good reason that it is unsound. It remands the ease for a retrial in which the prosecutor will be permitted to establish what he wás not permitted to establish in the original trial — that the defendant had been convicted of voluntary manslaughter. Thus the difference between this case and the case that will be presented on retrial is solely in the form of the prosecutor’s question. This is claimed to be such fundamental error as to require a mistrial, even though the prosecutor’s reference to the defendant’s wife was merely cumulative of the defendant’s own prior volunteered reference to the fact he had killed her. That the Court which found harmless error in United States v. Mitchell, 427 F.2d 644 (3d Cir. 1970), and which affirmed in United States v. DeCarlo, 458 F.2d 358 (3d Cir. 1972), should find such plain error here as to have required a mistrial is puzzling indeed. A district court judge *266ruling on objections to evidence or to the form of questions or on mistrial motions, contrasting the three decisions can only conclude that what is so prejudicial as to require a mistrial will depend upon the visceral reaction of a majority of this Court; a visceral reaction having no foundation in empirical evidence about how jurors react to one type of question rather than another.
Judge Kalodner argues:
“To hold that the cautionary instruction was an effective detergent which completely laundered out of the jury’s mind the impact of the prosecutor’s question, is unrealistic and a little less than fantasmagoría. A trial judge can ‘strike’ evidence from notes of testimony; it is something else again to ‘strike’ its searing impress from a juror’s mind.”
This is a misstatement of the issue. The rules permitting correcting, limiting, and cautionary instructions are not based on the assumption that jurors will forget, but on the assumption that by and large they will follow the court’s instructions, and on the necessities of an adversary fact finding process conducted by human beings. The harmless error rule is a recognition that mere mortals cannot be expected in every instance to conduct a “perfect” trial. Although some of the sweeping language in the majority opinions might suggest it, I do not read them as holding that cautionary instructions can never cure a trial error. Thus in ruling on mistrial motions district court judges will continue to be called upon to exercise subjective judgments based upon their overall impression of the impact of the given imperfection. When this court undertakes, without the benefit of direct observation, from the cold record, to draw so fine a line as has been drawn by the majority, between what was harmful and what on retrial will not be harmful, it merely substitutes its subjective judgment for that of the judge responsible for conduct of the trial. It does so in this case despite the finding by the district court that the impact of the question was harmless beyond a reasonable doubt.2 Rejection of that finding does not seem to me an appropriate exercise of appellate jurisdiction.
If the majority opinions proposed the anticipatory adoption of Rule 609(a) of the Revised Draft of Proposed Rules of Evidence for United States Courts and Magistrates, 51 F.R.D. 315, 391, (March 1971), they would at least have the virtue of reasoned analysis in place of subjective reaction. I would not vote to adopt that rule now, while the Draft is still under consideration, especially at the cost of a reversal and a new trial on a record containing fairly overwhelming proof of guilt and, at most, harmless error. Cf. United States v. Fiorvanti, 412 F.2d 407, 415, (3d Cir.), cert. denied, Panaccione v. United States, 396 U.S. 837, 90 S.Ct. 97, 24 L.Ed.2d 88 (1969).
. The objection was stated:
“I want to remind the Court that I would agree to some extent to prior, but the killing took place in ’69 and the conviction in ’70. This is subsequent. It has no bearing at all on what happened in ’68, Your Honor.” (Transcript at 17).
. In view of that finding the majority makes far too much of the district court’s initial reaction to the question. “Frankly, this is beyond belief. ...” The trial judge knew, better than we can ever know, when his own expression out of the presence of the jury was hyperbolic.