United States v. Rubenstein, 151 F.2d 915 (2d Cir. 1945)Annotate this Case
August 25, 1945
Circuit Court of Appeals, Second Circuit.
*916 Joseph Leary Delaney, of New York City (John L. Goldstone, of New York City, of counsel), for appellant.
Martin J. McLaughlin and John F. X. McGohey, U. S. Atty., both of New York City, for appellee.
Before L. HAND, AUGUSTUS N. HAND, and FRANK, Circuit Judges.
Writ of Certiorari Denied November 13, 1945. See 66 S. Ct. 168.
L. HAND, Circuit Judge.
Rubenstein appeals from a judgment of conviction under an indictment charging him with conspiracy to bring into the country an alien by false representations, by concealment of material facts and by false documents. He raises four objections on this appeal: (1) That the prosecution was permitted to prove that he was party to a fraudulent divorce of the alien after the crime had been completed by her entry; (2) that the court refused to declare that the alien and one, Sandler, had been validly married; (3) that the prosecution was allowed to ask the defendant upon the stand whether he had been disbarred or suspended from the bar; (4) that the indictment did not charge, and the evidence did not prove, any crime. The evidence was such that a jury might have found the following facts. A Czechoslovakian, named Alice Spitz, entered the United States on a temporary visa on November 17, 1938, which allowed her a sojourn of only six months: until May 17, 1939. She got an extension for another six months in the spring of 1939, but wished to remain longer; and in July of that year consulted *917 one, Muller, of whose wife she was a family connection, as to how she might do so. She suggested to him that she might get married to an American, just to stay in the country, and after entry she would get a divorce. Thereafter Muller induced one Sandler to take up this proposal; and introduced him to Spitz, who asked him to marry her "because I would like to have my parents from Czechoslovakia here and the only way I could do it, I thought, was to marry, and then I asked him if he would do it, and he said yes." She told him she did not wish to live with him, but that she would give him $200. Sandler's account of the conversation was substantially the same: If she married him, she could remain in this country, "and after six months time there will be a divorce, and she said there will be something for me, and that she will give a sum of $200 to me." The couple were married in Newark, New Jersey, on July 29, 1939, by a judge; they at once separated, always lived apart, and the marriage was never consummated. The evidence does not show by what means Spitz succeeded in staying in the country after the end of 1939, but she was still here in the spring of 1940 and she consulted Rubenstein about getting a visa for her as the wife of Sandler. She gave him her own home address in Manhattan, and Sandler's in Brooklyn; and thereafter she and Sandler had an interview, or interviews, with him, at which they told him that they had agreed that the marriage was to last only six months, after which they were to be divorced.
Sandler objected to signing a paper which Rubenstein presented to him at one of these interviews, saying that that was not part of the bargain, but upon Rubenstein's assurance that it was only a formality and that "the divorce would take place right after that," he consented. The paper which he signed was a petition for the grant of a visa, upon a form on which the specific details were unfilled. Sandler never swore to it at all; the details were typed in after he signed it; and a jurat was added. The statements filled in were false in a number of particulars: e.g. his address, his net worth and his income. The petition, when completed, declared that Spitz was his wife and asked that she might have an immigration visa. Rubenstein sent this petition to the Commissioner of Immigration, to whom it was addressed, supported by an affidavit, purporting to have been sworn to by Muller that he had known Sandler for at least fifteen years. That was false and Muller had signed the affidavit in blank. He also submitted an affidavit purporting to have been sworn to by one, Isaac, stating that he had known Sandler for twenty years. Sandler knew no such person. All three documents bore the date, August 20, 1940; and on December 7, 1940, Rubenstein wrote a letter to the American Consul at Montreal in which he enclosed a fictitious letter stating the present employment and earnings of Sandler. Upon these papers Spitz received a visa in Montreal on December 20, 1940, by means of which she entered the United States.
On January 27, 1941, Rubenstein procured one Haimowitz, to sign a false affidavit of service upon Sandler of a summons and complaint in an action of divorce brought by Spitz against him on the ground of adultery. Haimowitz never served Sandler and Sandler never answered. Rubenstein also on April 5, 1941, procured Haimowitz to sign a false affidavit, that Sandler was not in the army at the time. Finally, when the cause came on for trial as an undefended divorce, Haimowitz, at Rubenstein's instigation, swore falsely before the official referee that he had discovered Sandler and a woman together at night in a room in circumstances that were sufficient to support a decree. On this a divorce was granted.
Rubenstein's complaint of the admission of all evidence concerning the divorce is that it was an independent and disconnected crime, the conspiracy having ended when Spitz entered under the immigration visa. He is right in saying that the crime ended with the entry; but it by no means follows that evidence of the divorce was not relevant to the crime. Before she obtained the fraudulent visa Spitz had told Rubenstein that there was to be a divorce; Sandler had said the same thing; and Rubenstein had assured him that there would be a divorce almost at once. If the spouses' intent at the time of the ceremony was probative of fraud upon the immigration officials (as we shall show that it was) it was relevant to prove that they were later divorced, because it went to confirm their testimony as to what they had originally intended and agreed upon. We are therefore all agreed that the fact of the divorce was relevant; indeed this becomes at once apparent if we consider *918 how damaging to the prosecution's case it would have been, if no divorce had followed. However, Rubenstein's objection goes further: as we understand it, it is that, even though the divorce per se was admissible, it was not necessary in order to establish it to prove in addition that Haimowitz's two affidavits and his testimony before the referee at the hearing were perjured, or that Rubenstein suborned that perjury. Were that true, the evidence would have been so damaging, that we are not sure that the verdict should stand. We need not say, because we do not think that it was erroneous to admit the evidence. Sandler had already testified that he never was served with the summons and complaint; that he was in the army on April 5, 1941; and that he never knew of the divorce until September, 1941. He had also testified that Rubenstein had suggested to him that he should "get a girl and go to a hotel room"; that he had refused; and that he had never been found with any woman at the place mentioned in the divorce testimony. With this testimony in the case it was impossible to prove the fact of the divorce without incidentally proving that Haimowitz's affidavits and his testimony were perjured; and that led inevitably to Rubenstein himself as the author. The prosecution at least had to prove the complaint, its service, the order of reference, the findings and the interlocutory and final, decrees: nothing short of these would have shown that the parties had really put through a collusive divorce of the kind that they had agreed upon. But those documents on examination (especially the recitals of the interlocutory decree) disclose Haimowitz's perjuries, when coupled with the earlier testimony. Thus, Haimowitz introduced testimony as to no new crime; true, he did confirm the inferences to be drawn from the testimony of Sandler, but we are not aware that it has ever been held that it is error for the prosecution to buttress its facts as much as possible. It had to prove a collusive divorce and the precise form which the collusion took it could not control; Rubenstein had chosen what that should be, and any proof relevant to it was relevant to the charge.
Besides even if the affidavits and the testimony had not been competent, Rubenstein was not injured. No fraud was practised on Sandler; he was indifferent to the form of the divorce provided it went through. The fraud was on the court, and there would have been a fraud, though Sandler had actually committed adultery, for collusion or condonation invalidates a divorce whatever the defendant's conduct. Moreover, Rubenstein never intended that Sandler should do more than go through the form. We cannot understand how he was more deeply stained because he suborned Haimowitz to swear to the two affidavits and the hearing, than if he had put through the divorce as he proposed. The outstanding fact is that the divorce was too deeply interwoven with the conspiracy not to involve in its proof the means by which it was accomplished; and it is well settled law that evidence which is relevant to the proof of one crime is not incompetent because it discloses the commission of another. Finally, after a sedulous examination of the record, we have been unable to find any objection at the trial which distinguished between the papers showing the divorce, and Haimowitz's affidavits and testimony as to his perjury. The single objection was that all were "collateral": that was not enough to call the judge's attention to the exceedingly refined distinction which is brought forward for the first time on this appeal.
Rubenstein's second objection rests upon the claim that the intent to divorce was irrelevant and that the marriage was valid. The statute condemns not only a false representation, but a "willful concealment of a material fact." § 180a, 8 U.S.C.A. Rubenstein knew that the parties proposed a divorce within six months, and that was a fact most material to the granting of the visa. The statute is not concerned with marriage, merely as marriage; one, perhaps the chief, reason why it allows the wife of a citizen to enter is because the husband will be responsible for her support. If the spouses at the time of the wife's entry intend that that responsibility shall end as soon as possible, they have evaded the statute by suppressing a material fact; and the suppression is a fraud, even though the marriage is valid. But, that aside, Spitz and Sandler were never married at all. Mutual consent is necessary to every contract; and no matter what forms or ceremonies the parties may go through indicating the contrary, they do not contract if they do not in fact assent, which may always be proved. New York Trust Co. v. Island Oil & Transport *919 Corporation, 2 Cir., 34 F.2d 655; In re Hicks & Son, 2 Cir., 82 F.2d 277; Theiss v. Weiss, 166 Pa. 9, 31 A. 63, 45 Am.St. Rep. 638; Bruce v. Bishop, 43 Vt. 161; Graves v. Northern N. Y. Publishing Co., 260 App.Div. 900, 22 N.Y.S.2d 537; Williston on Contracts, § 21, note 11. Marriage is no exception to this rule: a marriage in jest is not a marriage at all. This is the law of New Jersey as well as elsewhere. McClurg v. Terry, 21 N.J.Eq. 225; Girvan v. Griffin, 91 N.J.Eq. 141, 108 A. 182 (semble). It is quite true that a marriage without subsequent consummation will be valid; but if the spouses agree to a marriage only for the sake of representing it as such to the outside world and with the understanding that they will put an end to it as soon as it has served its purpose to deceive, they have never really agreed to be married at all. They must assent to enter into the relation as it is ordinarily understood, and it is not ordinarily understood as merely a pretence, or cover, to deceive others.
It was plainly proper to impeach Rubenstein's credibility on cross-examination by asking him whether he had ever been disbarred or suspended. United States v. Buckner, 2 Cir., 108 F.2d 921, 927; People v. Dorthy, 156 N.Y. 237, 50 N.E. 800.
We do not understand the meaning of the last point raised. The crime was proved beyond the faintest peradventure of doubt; it was a deliberate fraud upon the immigration authorities, without excuse or palliation. We do not stop to consider the indictment; it advised the accused adequately of the crime with which he was charged.
FRANK, Circuit Judge (dissenting).
I agree, of course, that it was proper to receive evidence tending to show the sham nature of the marriage, that defendant originally arranged that there should be a divorce, and that he aided in obtaining a divorce; for that evidence went to prove defendant's complicity in a plan to violate the statute by means of a fake marriage and by concealment of a material fact.
But I cannot agree that it was proper to receive the additional testimony which (if taken as true) showed (1) that defendant in procuring the divorce used a false affidavit and suborned perjury, (2) that defendant had suggested to Sandler that he should "get a girl and go to a hotel room," and that he refused, (3) that Sandler was not served, and knew nothing of the divorce until after the entry of the divorce decree. None of that evidence was needed to show the fictitious character of the marriage or the illegal plan to conceal the intention that it was soon to be followed by a divorce. For these purposes, it sufficed to prove merely that defendant, from the first, joined in the plan for the sham marriage and divorce and that, with his assistance, Miss Spitz and Sandler were divorced not long after the marriage ceremony.
Judge Hand suggests that the prosecutor "had to prove a collusive divorce," that therefore the "precise form which the collusion took" was unimportant, and that consequently the additional testimony as to the false affidavit and defendant's subornation of perjury, etc. was, in effect, cumulative and did defendant no appreciable harm. I cannot agree. No doubt the prosecutor had to prove defendant's part in a pre-arranged divorce relating to a marriage the unreality of which defendant did not disclose to the divorce court. But it is one thing to show such misconduct constituting a fraud on the divorce court (justifying disbarment) and perhaps some sort of crime under New York law and quite another to add proof of the commission of the distinct and far more serious crime of suborning perjury. That added proof was by no means merely cumulative. So that while, as Judge Hand states, "it is well settled law that evidence which is relevant to the proof of one crime is not incompetent because it discloses the commission of another," that doctrine cannot here warrant reception of the additional evidence, especially that tending to show the crime of subornation of perjury in the divorce proceedings. See United States v. Krulewitch, 2 Cir., 145 F.2d 76, 80, 156 A.L.R. 337.
To be sure, that additional evidence might properly have been admitted for its limited bearing on defendant's credibility, since he took the witness stand; but it was not offered for that purpose and absent a limiting instruction should not be so accepted on appeal. Shepard v. United States, 290 U.S. 96, 54 S. Ct. 22, 78 L. Ed. 196. As this evidence was of a kind highly likely to arouse the jury's animosity against defendant, his failure to object specifically *920 to its admission (as distinguished from his objection to the reception of any evidence concerning the divorce) or to request a limiting instruction, did not cure the error. I feel sure that if either of my colleagues had been sitting as the trial judge, he would, without prompting from defendant's counsel, either have excluded the additional evidence or given such an instruction. We have often held (and our own Rule 10 so provides) that we should take notice of and reverse on account of a markedly serious error to which a defendant did not object at the trial.
Here we have markedly serious error. It cannot be characterized as "harmless," if the test of harmlessness be that employed by the Supreme Court and most other courts. The only conceivable remaining basis, therefore, for not reversing this judgment must be the tacit application of the unique interpretation of the phrase "harmless error" which has unfortunately become current in this Circuit. A superficial reading of Judge HAND'S opinion might make it appear that he does not employ the "harmless error" doctrine, since he says that, were it true that the additional evidence was inadmissible, "the evidence would have been so damaging that we are not sure that the verdict should stand," and adds, "We need not say, because we do not think that it was erroneous to admit the evidence" (going on in that same paragraph to assign reasons for that conclusion which, as above noted, I think untenable). But in the very next paragraph he abandons the suggestion that reversal might follow if this evidence was incompetent: He now says that, even if the additional testimony was not competent, "Rubenstein was not injured" by its admission; and in the last paragraph of the opinion, he says, "The crime was proved beyond the faintest peradventure of doubt * * *." Those two remarks contain, in abbreviated form, the Second Circuit doctrine of "harmless error," i. e., the test to be used in determining whether inadmissible evidence is harmless is not the probable effect of its admission on the jury but whether the Court considers the defendant guilty.
I have the very highest respect for Judge Hand. To sit with him is an inestimable privilege, a constant source of education. Consequently, I usually suspect my own tentative opinions, when they vary from his. But on this one subject I find myself recurrently and unregenerately at odds with him. In short, I cannot accept this view which my five colleagues have adopted and which he has enthusiastically endorsed: If we, sitting on a reviewing court, believe, from merely reading the record that a defendant is guilty, then we must hold that an error, in admitting evidence, even if it may seriously have prejudiced the jury against the defendant, is to be regarded as "harmless." Several times heretofore I have stated my reasons for opposition to that doctrine. What I shall say here is partly by way of summary and partly by way of amplification of those previous comments.
It is seldom possible with even moderate competence to conjecture solely from perusal of a written or printed record whether or not a defendant is guilty. As the judges of an appeal court have not heard or seen the witnesses, they have no reasonably adequate way of judging whether any of the witnesses lied or because of unconscious bias or faulty memory testified with inadvertent inaccuracy.*921 On that account alone, the rule is well settled that the credibility of the witnesses is exclusively a question for the jury. When, then, an upper court, with an air of emphatic assurance, says in an opinion, as Judge Hand, speaking for the majority, says in this case, "The crime was proved beyond the faintest peradventure of a doubt," what it actually means, for all its magisterial tones, comes to no more than this: "As we were not present at the trial, we did not listen to, and observe, the witnesses and cannot determine their credibility. However, we shall make the assumption which may not be at all correct in fact that the government's witnesses testified honestly and with entire accuracy, and that the testimony of defendant's witnesses, in so far as it contradicted that of the government's witnesses, was wholly untrustworthy. Only by making that assumption (and, accordingly, paying no heed to the evidence offered by the defendant) do we come to the conclusion that defendant's guilt is unquestionably clear." Were that unspoken assumption explicitly stated in such an opinion, the seeming air of indubitable certainty about its pronouncement of defendant's guilt would disappear, and the weakness of this court's peculiar harmless error doctrine would be exposed to scrutiny.
For see: My colleagues' rule is to overlook errors in admitting evidence, when (without regard to the jury's verdict) the record, in their opinion, shows that defendant is guilty. But why does the record so show? Only because of an assumption which may easily be fictitious that the testimony of defendant's witnesses is unreliable. Had my colleagues been present at the trial, they might have concluded that, as to pivotal items, the testimony of the government's witnesses was not worthy of belief. My colleagues employ the contrary assumption precisely because they were not present at the trial which highlights the unwisdom of their sustaining the conviction on the ground that, independent of the jury's verdict, they find the defendant guilty.
The unjustified use, in such a case, of the assumption about the government's evidence must be distinguished from its appropriate use in another context: In a case where no prejudicial evidence was erroneously received, and a defendant asks an upper court to reverse his conviction on the sole ground that the record evidence is such that no reasonable jury could have found him guilty, the judges are compelled to resort to a sort of fiction: Since they were absent from the trial and since the credibility of the witnesses is a question for the jury, the judges, perforce, must assume it to be true (whether or not it is true in fact) that all the testimony pointing to guilt was given by honest, reliable, credible witnesses and that all other testimony was not. That assumption (a fiction, an "as if") in such a case is unavoidable; therefore necessary; therefore proper. Be it noted, however, that, in such circumstances, the judges at least have before them, although in printed form, the identical evidence neither more nor less which had been presented to the jury. But that assumption (fiction) is not necessary and is therefore, I think, improper, when (as here) the record contains prejudicial *922 evidence improperly admitted. For the issue on appeal is then not whether, making that assumption and on the very same record which was before the jury, the jury reasonably found against defendant. If, in a case like this, the judges conclude that defendant is guilty, they reach that conclusion by ignoring important matter which had been presented to the jury and which may well have induced its verdict. In those circumstances, the judges decide against defendant, not by affirming a jury's verdict on the record which the jury had before it, but on a strikingly different record. For that reason, I said that the judges are deciding such a case "independent of the jury's verdict."
The important fact is that in this Circuit "harmless" error does not mean merely that the improperly admitted evidence was such that in all probability it made no difference to the jury; were that the rule, I would find it unobjectionable. The rule here is that, even if that evidence was such that it may well have affected the jury's verdict, yet it is no ground for reversal if my colleagues believe the defendant guilty. The rule in this Circuit thus boils down to this: If the jury, on the basis of certain evidence, has brought in a verdict of guilt, then, on quite different evidence (i.e., different because the judges delete prejudicial evidence) the judges may render their own independent verdict of guilt, despite the fact that they neither saw nor heard the witnesses on whose testimony they rest their verdict.
In so doing, the judges convert themselves into a jury. By thus substituting themselves for the legally authorized jury, I think they exercise a power beyond their legitimate their constitutional scope. Without warrant in statute or Constitution, the judges find the facts. I cannot believe that such a procedure satisfies the constitutional requirement of a jury trial. The defendant has been convicted by the judges, not by a jury. He has been unconstitutionally deprived, I think, of the privilege of a trial by jury fully as much as if, in the first instance, he had been compelled to go to trial before a juryless court.
To be sure, some lawyers maintain that only through such a judge-made device can the jury-system be made "workable." But such a device makes the jury system workable by not working it. (Strangely, those lawyers who endorse this judicial circumvention of the Constitution on the ground of practicality are, in general, those who freely criticize the administrative agencies as "usurpers"; those lawyers would cry out indignantly were any such agency to rely on that ground to justify itself in evading a statutory requirement.) If any judges happen to regard as "impractical" the constitutional obligation to give a defendant a jury trial, they should not, by indirection, amend the Constitution. They should frankly state their position and invite our citizens to bring about a constitutional amendment in the manner prescribed by the Constitution.
*923 A jury trial unquestionably has defects. At best, such a trial, especially as now conducted i. e., as if it were a game or sporting event is an imperfect, all-too-human, instrument for ascertaining the true facts of a case. As Borchard reported several years ago and as current discussion in the press of the recent Campbell case dramatically reminds us, occasionally it is discovered that an innocent man, after a jury trial, has been convicted and sent to jail or put to death by the government. No one can doubt that there have been undiscovered instances (no one knows how many) of convictions of the innocent. Unfortunately, some tragedies of that kind are bound to occur. I, for one, do not care to accept responsibility for any such miscarriage of justice which, with reasonable precautions, could have been avoided.
The conventions of judicial opinion-writing the uncolloquial vocabulary, the use of phrases carrying with them an air of finality, the parade of precedents, the display of seemingly rigorous logic bedecked with "therefores" and "must-be-trues" give an impression of certainty (which often hypnotizes the opinion-writer) concealing the uncertainties inherent in the judging process. On close examination, our legal concepts often resemble the necks of the flamingos in Alice in Wonderland which failed to remain sufficiently rigid to be used effectively as mallets by the croquet-players. In a case like this, all our complicated judicial apparatus yields but a human judgment, not at all sure to be correct, affecting the life of another human being. If we are at all imaginative, we will comprehend what that judgment will mean to him, and what a horror it will be if we wrongly decide against him. To be sure, one can say that it does not pay to take too seriously the possibility that one man, more or less, may be unjustly imprisoned, considering the fact that, in the recent war, millions have died and that the Atomic Age just begun may end any minute in the destruction of all this plant's inhabitants. Yet (perhaps because I am growing old or because, despite my years, I have not yet fully matured) it seems to me that, if America's part in the war was meaningful and if mankind's development has any significance against the background of eternity, then the dignity of each individual man is not an empty phrase. If it is not, then we judges, part of a human arrangement called government, should proceed with great caution when we determine whether a man is to be forcibly deprived of his liberty. Recognizing the fallability of juries, we should, I think, be vigilant to prevent the jailing of an innocent person through an erroneous appeal to jury prejudices. In such a case as this, where seriously damaging testimony has been erroneously admitted, we should not assume that the government's witnesses necessarily reported the facts accurately, and should order a new trial.
In fairness I must observe that, to the working principle of harmless error which I think my colleagues have employed, they make these exceptions: (1) Where important material evidence has been excluded; (2) where the trial judge has, in a peculiarly outrageous way, displayed "a prosecutor's zeal"; (3) where defendant has been deprived of a basic constitutional or statutory privilege (e. g., the right to be advised by counsel). (4) There is a fourth exception which has at least once *924 been utilized, i. e., where, because of the excessive severity of a sentence, the court reverses for an error which it would otherwise have held harmless; but usually that exception, I think, is more apparent than real. It is true that my colleagues will often generously note an error even if it has not been asserted in the lower court or in this court; but (subject to the exceptions to which I have referred) they will not hold an error whether claimed or unclaimed by a defendant to be ground for reversal unless they have serious doubts about defendant's guilt.
It is there that I disagree. To repeat, I think the correct rule is this: We should reverse where error has been committed, regardless of our belief as to guilt or innocence, unless we conclude that in all probability the error had no effect on the jury; or, to phrase it differently, that the record is such that, if there had been no error, no reasonably sane jury could have acquitted, or that there is no reasonable ground for thinking that the jury was misled by the error. Such a rule does not demand perfection. Nor will it work automatically; some limited guesswork is unavoidably involved, but there are many instances where one can say with a fairly high degree of assurance that the error did not affect the jury and was therefore harmless.
It will not do to say that, were my thesis accepted, delay and expense would ensue. They are of far less importance than a fair trial. And they will be incurred infrequently because of errors if, by reversals in cases like this, we educate prosecutors and trial judges to prevent unfairness to a person on trial. Instead of so educating them, this court seems to me to invite such unfairness. For it announces that, if the district attorney succeeds in persuading a jury to say "guilty", this court will usually (subject to the exceptions I noted above) disregard improprieties in a case in which the properly admitted evidence is compatible with guilt. That thereby such improprieties have been encouraged is suggested by the frequency with which this court has found it necessary to condone them by invoking, tacitly or openly, its unique harmless-error rule.
Although usually I feel constrained to concur in a ruling with which I disagree if it has been endorsed in previous decisions by a majority of this court, I think that, in respect of this "Second Circuit doctrine" of harmless error, because of what may be its consequences, I may appropriately continue to dissent. I feel less hesitant to do so, as I think that this doctrine is patently out of line with several recent Supreme Court decisions and has been rejected by every other Circuit in which the matter has been considered.
In United States v. Liss, 2 Cir., 137 F.2d 995, 999, this court said that there is a "modern disposition to assume that an error *925 has been harmless * * *." But previously, in two cases, the Supreme Court had, I think, held to the contrary. In McCandless v. United States, 298 U.S. 342, 347, 56 S. Ct. 764, 766, 80 L. Ed. 1205, the Court said that the "harmless error" statute (28 U.S.C.A. § 391) "does not change the well-settled rule that an erroneous ruling which relates to the substantial rights of a party is a ground for reversal unless it affirmatively appears from the whole record that it was not prejudicial" (citing United States v. River Rouge Co., 269 U.S. 411, 421, 46 S. Ct. 144, 70 L. Ed. 339; Fillipon v. Albion Vein Slate Co., 250 U.S. 76, 82, 39 S. Ct. 435, 63 L. Ed. 853; Williams v. Great Southern Lumber Co., 277 U.S. 19, 26, 48 S. Ct. 417, 72 L.Ed. 761). In Bruno v. United States, 308 U.S. 287, 293, 294, 60 S. Ct. 198, 200, 84 L. Ed. 257, the Supreme Court said that this statute was intended "to prevent matters concerned with the mere etiquette of trials and with the formalities and minutiae of procedure from touching the merits of a verdict." It is suggested that in Palmer v. Hoffman, 318 U.S. 109, 116, 63 S. Ct. 477, 482, 87 L. Ed. 645, 144 A.L.R. 719, the Court took a different position. But there the Court merely held that appellant could not complain on appeal of an error in refusing to allow him to inspect a document in the possession of his adversary when, because appellant had not identified and incorporated the document in the record, the upper court court not possibly ascertain whether its contents were material and therefore could not determine whether the error was harmful; in that limited context the Court said that a party seeking reversal for error has "the burden of showing that prejudice resulted." That, in that limited ruling, the Supreme Court did not intend to repudiate its ruling in Bruno v. United States, is made clear, I think, by the fact that subsequently, in Weiler v. United States, 323 U.S. 606, 611, 65 S. Ct. 548, 551, 156 A.L.R. 496, the Court cited the Bruno case with approval and said, "We are not authorized to look at the printed record, resolve conflicting evidence, and reach the conclusion that the error was harmless because we think the defendant was guilty. That would be to substitute our judgment for that of the jury and, under our system of justice, juries alone have been entrusted with that responsibility."NOTES
 See, e.g., United States v. Trypuc, 2 Cir., 136 F.2d 900, 902; United States v. Haug, 2 Cir., 150 F.2d 911, 914; Wiborg v. United States, 163 U.S. 632, 658, 16 S. Ct. 1127, 1197, 41 L. Ed. 289; United States v. Atkinson, 297 U.S. 157, 160, 56 S. Ct. 391, 80 L. Ed. 555; Brasfield v. United States, 272 U.S. 448, 450, 47 S. Ct. 135, 71 L. Ed. 345; New York C. R. Co. v. Johnson, 279 U.S. 310, 318, 49 S. Ct. 300, 73 L. Ed. 706; Crawford v. United States, 212 U.S. 183, 194, 29 S. Ct. 260, 53 L. Ed. 465, 15 Ann.Cas. 392; Van Gorder v. United States, 8 Cir., 21 F.2d 939, 942; Meadows v. United States, 65 App.D.C. 275, 82 F.2d 881, 884.
 This rule seems to be inapplicable where the error consists of the exclusion of important evidence. Cf. United States v. Andolschek, 2 Cir., 142 F.2d 503; United States v. Hoffman, 2 Cir., 137 F.2d 416.
 See, e.g., United States v. Liss, 2 Cir., 137 F.2d 995, 1001, 1005; Keller v. Brooklyn Bus Corp., 2 Cir., 128 F.2d 510, 513, 514, 515.
 The ascertainment of the true facts of a case, even in the best of circumstances, is but a conjecture or guess: Since that process involves events which occurred in the past, the fact-finder must rely on the narratives of other persons. That means that the "finding of facts" rests on the fact-finders' guess as to whether and to what extent those narratives are accurate. A jury (or judge), in attempting to learn the facts, seeks to reconstruct the past, and thus acts as an historian. A great historian has called history-writing a "conjectural science."
 Observation of the witnesses' demeanor is by no means an infallible method of determining the accuracy of their testimony. But, no perfect method having been devised, such data are of considerable value. The printed record necessarily omits such data. The testimony of a glib liar may show up in print far more persuasively than that of an honest, cautious witness.
Perhaps, if on appeals we used records consisting of talking motion pictures of trials, this particular difficulty could be largely overcome.
 There are other (constitutional) reasons.
 One gets a hint of it in a separate part of Judge Hand's opinion where he says: "The evidence was such that a jury might have found the following facts." (Italics added.)
 "Flowers of expression like the following have grown up in the garden of jurisprudence and are the specific property of the lawyer: `We must assume as proven'; * * * `we cannot justly doubt'; * * * etc. Such phrases serve to render the difference between the real degree of probability and the cogency of the inference drawn, as inconspicuous as possible. Usually they amount to this, that in place of direct assertion we put an obligation to believe, and this modifying factor is afterwards disregarded." Wurzel, Methods of Juridical Thinking, in Science of Legal Method (1917) 286, 419-420.
 It is a fiction because it involves taking something as true regardless of whether or not, in fact, it is true.
 The assumption of the exclusive credibility of evidence telling against the defendant has here been needlessly and unjustifiably extended.
In all fields of thought, fictions must be cautiously employed. They are "useful lies," statements which, although contrary to truth or incapable of verification, nevertheless are immensely convenient and often necessary. But while a fiction may have such value in one context, it may be harmful in another. "Extrapolation", always tricky business, is peculiarly so when applied to fictions.
Cf. Holmes, Collected Legal Papers (1920) 49ff; Hammond-Knowlton v. United States, 2 Cir., 121 F.2d 192, 199, 200; United Shipyards, Inc. v. Hoey, 2 Cir., 131 F.2d 525.
 True, even with the rule thus limited, the appeal judges are conjecturing as to whether the evidence affected the jury. But the area of guessing is severely restricted.
 It is important to differentiate between specific constitutional provisions and those which deliberately employ vague phrases such as "due process." The latter, unlike the former (such as the jury-trial provision), justify indeed compel liberal and developing judicial interpretation. See, as to this differentiation, United States v. Carolene Products Co., 304 U.S. 144, 152, 58 S. Ct. 778, 82 L. Ed. 1234, note 4; United States v. St. Pierre, 2 Cir., 132 F.2d 837, 840, 847.
The jury-trial provision does not preclude all elasticity in construction; see Ex parte Peterson, 253 U.S. 300, 309, 310, 40 S. Ct. 543, 64 L. Ed. 919. But it surely does not authorize a construction which, in effect, eliminates the function of a jury's verdict of guilt except as a preliminary to a verdict by appellate judges based upon a record significantly different from that considered by the jury.
 See, e.g., Morgan, Book Review, 49 Harv.L.Rev. 1387, 1389, to the effect that a trial usually is not "a proceeding for the discovery of truth by rational processes" but "a game in which the contestants are not the litigants but the lawyers."
See, also, Osborn, The Mind of the Juror (Student's ed. 1937) 12, 83, 86; Wigmore, Evidence (3d ed.) § 1845; Seagle, The Quest For Law (1941) 84ff (as to the "judicial duel"). Cf. In re Barnett, 2 Cir., 124 F.2d 1005; Postgate, Verdict of Twelve (1940); Frank, If Men Were Angels (1942) 80-97.
 Borchard, Convicting The Innocent (1932).
 I am here borrowing and adapting an image of Edmund Wilson used in his criticism of the verse of certain modern poets.
 See footnote 2.
 United States v. Marzano, 2 Cir., 149 F.2d 923; cf. United States v. Hoffman, 2 Cir., 137 F.2d 416, 422.
That, where my colleagues believe defendant guilty, they will not reverse unless the judge's conduct is extravagantly unfair, see United States v. Liss, 2 Cir., 137 F.2d 995; United States v. Mitchell, 2 Cir., 137 F.2d 1006 and 2 Cir., 138 F.2d 831.
 See, however, cases where this court believed defendant guilty. United States v. Mitchell, 2 Cir., 137 F.2d 1006 and 2 Cir., 138 F.2d 831; United States v. Gutterman, 2 Cir., 147 F.2d 540.
 Amendola v. United States, 2 Cir., 17 F.2d 529, 530.
 Thus in United States v. Hoffman, 2 Cir., 137 F.2d 416, the court, while noting the severity of the sentence (page 422), said (pages 420, 421) that the evidence was not sufficient to "conclude defendant's guilt." In United States v. Trypuc, 2 Cir., 136 F.2d 900, 902, the severity of the sentence prompted the court to reverse for an "unclaimed" error under our Rule 10, but nothing indicates that this factor induced the court to regard as prejudicial an error which it would otherwise have held harmless. In Nash v. United States, 2 Cir., 54 F.2d 1006, 1008, the Court criticized the sentence but affirmed.
 See footnote 1.
 See United States v. Liss, 2 Cir., 137 F.2d at page 1004 and note 8a; Keller v. Brooklyn Bus Corporation, 2 Cir., 128 F.2d at pages 513, 514.
 See footnote 11.
 A similar problem arises when the judge's charge to the jury lacks clarity: If the chances are as good that the jurors were bewildered as that they were not, then I think there is prejudicial error, regardless of whether the upper court believes the defendant guilty. See cases cited in United States v. Liss, 2 Cir., 137 F. 2d at page 1003 note 7; cf. Keller v. Brooklyn Bus Corporation, 2 Cir., 128 F. 2d at page 518.
 Bruno v. United States, 308 U.S. 287, 294, 60 S. Ct. 198, 84 L. Ed. 257; McCandless v. United States, 298 U.S. 342, 347, 348, 56 S. Ct. 764, 80 L. Ed. 1205; Weiler v. United States, 323 U.S. 606, 611, 65 S. Ct. 548.
 Little v. United States, 10 Cir., 73 F.2d 861, 866, 96 A.L.R. 889; Lynch v. Oregon Lumber Co., 9 Cir., 108 F.2d 283, 285, 286; Fort Dodge Hotel Co. v. Bartelt, 8 Cir., 119 F.2d 253, 259; Worcester v. Pure Torpedo Co., 7 Cir., 127 F.2d 945, 947, 948; Farris v. Interstate Circuit, 5 Cir., 116 F.2d 409, 412; Evansville Container Corporation v. McDonald, 6 Cir., 132 F.2d 80, 85; Ah Fook Chang v. United States, 9 Cir., 91 F.2d 805, 810.
 I think Palmer v. Hoffman merely followed the usual rule that an appellate court cannot consider matters not in the record before it. See, e. g., Henneford v. Northern Pac. Ry. Co., 303 U.S. 17, 19, 58 S. Ct. 415, 82 L. Ed. 619; Drake v. General Finance Corporation, 5 Cir., 119 F.2d 588, 589; 4 C.J.S., Appeal and Error, § 1206.