Federal jurisdiction in this case rests on diversity of citizenship. The action sought damages for the death of Mrs. Hudson, who was the plaintiff’s mother and of whose estate she was appointed administratrix. Mrs. Hudson was killed by being run over by the defendant’s motor bus due, it is alleged, to the bus driver’s negligence. The trial resulted in a verdict for the defendant. From the ensuing judgment the plaintiff has appealed, claiming errors in the charge given to the jury.

The accident occurred at about 11:40 P. M. on April 8, 1939. The bus had been proceeding easterly along Newkirk' Avenue in Brooklyn; it stopped at the southwest corner of Newkirk and Flatbush Avenues to let off passengers, among whom1, was Mrs. Hudson. When the bus stopped the traffic light at the corner showed red against traffic on Newkirk Avenue. After it changed to green the bus started forward and made a wide swing to the right into Flatbush Avenue. In the meantime Mrs. Hudson had walked over to the westerly curb of Flatbush Avenue and when thei traffic light was green she started to cross the street. She came into contact with the bus during its turn and after the accident her body was found about 15 feet south of the curb line of Newkirk Avenue, between the southbound rails of the trolley track on Flatbush Avenue, and 18 or 20 feet from the rear of the bus. The principal question of fact was whether she walked into the side of the moving bus or whether she was sufficiently in front of it so that the driver should have seen her. The only eyewitness to the accident testified that she was hurrying across the street, looking toward the south, and that he “saw the woman walk into the side of the bus.” One of the passengers, however, gave testimony from which the jury could conceivably have found that just before the accident she was directly in front of the bus and should have been seen by the driver, who could have stopped the bus had he then seen her. Admittedly he did not see her. His testimony that she was never in front of the bus was corroborated by a passenger who occupied a seat directly behind the driver. Obviously, the issue of how the accident happened and whether it was caused by the carelessness of the driver or the decedent’s own negligence were mat*512ters for the jury to determine under proper instructions. It found for the' defendant; on the evidence it is difficult to see how it could properly have reached any other verdict.

One of the appellant’s complaints relates to the trial court’s interpretation of Article 4, section 29, subdivision B of the Traffic Regulations of the Police Department. This requires that the driver of a vehicle “equipped with a mechanical or electrical signalling device” shall, “before turning to the right or left or slowing down,” give timely warning by said signalling device. It was the contention of the appellant that this regulation required the bus driver to sound his horn before turning into Flat-bush Avenue, but the court refused so to instruct the jury and explained to them that in his opinion the device to which the regulation referred was a mechanical arm or an electrically operated rear light signal with which trucks or other vehicles are often equipped. We agree with the trial court’s interpretation; every automobile is equipped with a horn and the idea that the horn must be sounded whenever the vehicle slows down or turns a corner appears fantastic.

The only point really meriting discussion is the charge respecting burden of proof. Early in his instructions the judge correctly stated that the plaintiff had the burden of proving that her mother died because of the negligence of the defendant. He then proceeded to refer to the evidence and concluded this portion of the instructions with the statement that, in order to entitle the plaintiff to recover, the bus must not only be at fault but Mrs. Hudson must have been in the exercise of due care on her own account. This also was clearly correct, but the charge continued: “If the bus was at fault and she also was at fault, if she failed to exercise due-care, then the plaintiff has failed to sustain her burden of proof. On the other hand, if the bus was at fault and Mrs. Hudson was not at fault under all the evidence in the case, then the plaintiff has sustained her burden of proof * * *

It is urged that this erroneously instructed the jury that the plaintiff had the burden of proving the decedent’s freedom from contributory negligence, which is contrary to Rule 8(c) of the Federal Rules of Civil Procedure, 28 U.S.C.A. following section 723c, as well as contrary to the law of New York. Decedent Estate Law, Consol.Laws, c. 13, § 131; Code of Civil Procedure, ch. 9, § 841-b; Schrader v. New York, C. & St. L. R. Co., 254 N.Y. 148, 150, 172 N.E. 272. The quoted language does not in so many words say that the plaintiff has the burden of proving that Mrs. Hudson exercised due care, but we shall assume that the jury might have so understood -it and that the judgment would have to be reversed had the matter ended here. But it did not. The last instruction given the jury was the following:

“Mr. Rothbard: Would your Honor permit me to make one oral request and that is that the defendant is under the duty of convincing the jury in this case — not plaintiff but the defendant — of convincing the jury that the deceased was guilty of contributory negligence.

“The court: I so charge,"

This emphatic declaration.that the defendant, not the plaintiff, must convince the jury of the deceased’s contributory negligence, was the court’s final word before the jury retired for deliberation. It definitely put upon the defendant the “duty” of “convincing” the jury as to negligence on the part of Mrs. Hudson. A majority of the court entertain no doubt that it was sufficient to remove any misconception which may have been engendered in the jury’s mind by the preceding instructions concerning the burden of proof if she was at fault in failing to exercise due care.

In the federal courts technical errors which do not affect the substantial rights of the parties cannot be made the basis of reversal. 28 U.S.C.A. § 391. The test of the correctness of instructions is whether upon the whole charge the jury will gather the proper rules to be applied ’in arriving at decision. Taxi Service Co. v. Phillips, 1 Cir., 187 F. 734, 736-738; People v. Goldstein, 285 N.Y. 376, 383, 34 N.E.2d 362. Applying that test we are satisfied that the jury could have entertained no uncertainty that the burden of convincing them of Mrs. Hudson’s negligence was on the defendant. This conclusion is in no respect inconsistent with this court’s decision in Cummings v. Pennsylvania R. Co., 2 Cir., 45 F.2d 152, 153. There we held that an incorrect instruction constituted prejudicial error notwithstanding a subsequent correct charge because the later statement “would, in all probability, have been treated only as a restatement of what had gone before. Quite likely the jury was unaware of any change.” The same criticism of the correction cannot be made in the present case. If the jury understood the first charge to place the burden of proof on the plaintiff, *513they could riot have understood the later charge as a mere restatement of that. It expressly told them that “not plaintiff but the defendant” had the duty of convincing them of Mrs. Hudson’s contributory negligence. Whether a subsequent correct charge is sufficient to cure a prior inadvertent error is necessarily a matter to be determined in the sound judgment of the appellate court. On such a question judges will not always agree. In the present case a majority of the court is convinced that the inadvertent error was cured and no prejudice to the plaintiff survived. Judgment affirmed.

(dissenting).

As I am far less experienced in deciding jury cases on appeal than my very wise colleagues, for whose views I entertain the sincerest respect, I have hesitated to disagree with them in this case. But this seems to me an instance of Homer nodding. I think the majority opinion unwisely overrules former decisions of this court, disregards rulings of other Circuit Courts of Appeals, and has the effect of improperly substituting — contrary to the Seventh Amendment as construed recently by the Supreme Court — a judicial for a jury verdict in a case in which, as my colleagues say, the evidence is conflicting.1 My reasons for that conclusion are as follows:

The trial judge in his charge gave two conflicting versions of the burden-of-proof rule. The first was wrong, sufficiently so that, if not retracted, the majority opinion concedes it would compel reversal; the second was correct. If it were clear that the jury understood that the judge meant to substitute the second for the first, then the error would be harmless. But I perceive no basis for the belief that the jury understood that he had changed his mind, since he did not so state or imply. See what he did: Immediately after he uttered his earlier mistaken version, he ignored an exception thereto, made by appellant’s counsel. Then he went on to give instructions on other matters. After a considerable interval— five folios later in the transcript — he was requested to and did state the correct rule. But he did not then say or imply: “This is in place of what I previously said on this subject, which was mistaken and which you must ignore.” How, then, can we say that the jury knew that the second was a retraction?

For reasons noted below, we are obliged to assume that they listened intently to all his remarks. If they accepted him — which again we are obliged to assume — as the infallible source of all legal learning relevant to the case, they must have thought that his legal dissertation was intended to be logical and self-consistent. It would, indeed, have been presumptuous for them to have decided, without advice from the judge to do so, that part of his legal discourse was wrong. And, if they did not, then, if they were conscientious, and worked as juries are supposed to do, they must have been baffled. To illustrate: Suppose that my two brother judges and I read a paper by an eminent physicist on radioactivity. If we found what appeared to us to be inconsistent statements, we would, in all likelihood, reject the suggestion that the savant had at any point been in error; we would reason that the appearance of inconsistency was due to our limited knowledge of a subject as to which we were but amateurs, and we would try, in some way, to reconcile his seemingly discrepant remarks. In similar fashion, the jury here— if, I say, they acted as juries are supposed to act, heeding the legal instructions from an expert and according to him appropriate respect for his superior legal wisdom — must have tried to harmonize the discrepant parts of his charge. On that assumption, they were asked to achieve the impossible.

As an alternative, we might conjecture that the jury thought (a) that the judge was merely, in new verbiage, repeating what he had said before or (b) that they were at liberty to choose either rule. On any of these hypotheses, the defendant was deprived of a fair trial by jury.

Of course, we should not be fly-specking in our examination of what goes on in a trial. Perfection is no more to be sought in courtroom conduct than in any other human activity. I agree that we should never engage in “an over jealous scrutiny of every word that may fall from the judge’s mouth,”2 else there would be few unreversed judgments. Judges “are all human and it is unreasonable to demand that any human process be absolutely flawless. The work-a-*514day world is no place for the perfectionist.”3 The harmless error doctrine is so eminently sensible that it should freely he used. But its use in connection with jury trials, where, as here, the jury brings in merely a general verdict, often presents a troublesome problem.

Some State courts have been exceedingly liberal in their application of that doctrine even in jury cases. But State courts are not bound by the Seventh Amendment. Doubtless because the federal courts are subject to that Amendment, the Supreme Court has been more strict — despite the enactment in 1919 of the “harmless error statute” (28 U. S.C.A. § 391) — in federal jury cases where erroneous instructions have been given to the jury.4 As recently as Bruno v. United States, 1939, 308 U.S. 287, 294, 60 S.Ct. 198, 200, 84 L.Ed. 257, the court said that that provision was intended merely “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.” That case involved an erroneous instruction. There were many earlier similar decisions. In Fillippon v. Albion Vein Slate Co., 250 U.S. 76, 82, 39 S.Ct. 435, 437, 63 L.Ed. 853, in a case which arose before that statute went into effect, it was said: “And of Course in jury trials erroneous rulings are.presumptively injurious, especially those embodied in instructions to the jury; and they furnish ground for reversal unless it affirmatively appears that they were harmless.” That such rulings were not altered by the passage of the Act is shown by the decision in McCandless v. United States, 298 U.S. 342, 347, 56 S.Ct. 764, 766, 80 L.Ed. 1205, where the court, after quoting part of the statute, said: “This, as the language plainly shows, does not change the well-settled rule that an erroneous ruling which relates to the substantial rights of a party is ground for reversal unless it affirmatively appears from the whole record that it was not prejudicial.” It is noteworthy that the Supreme Court, in this connection, cited the Fillippon case and italicized the word “affirmatively.” 5

.Wigmore deplores such rulings.6 But our master’s voice is that of the Supreme Court, not Wigmore’s. For a time, the Eighth Circuit Court of Appeals, in a series of decisions, held that the statute had changed the earlier rule so that, it said, the burden is on the complaining party to show that an error was harmful.7 Those decisions won the applause, of Wigmore.8 But in 1941, in Fort Dodge Hotel Co. v. Bartelt, 8 Cir., 119 F.2d 253, 259 (a case involving an erroneous in*515struction), that same court reversed those rulings in obedience to the intervening decisions of the Supreme Court. Our court, too, I think, must fall in line. Heretofore, it has done so. Several times it was recognized that, where there is a general verdict, it is frequently impossible to ascertain, with any sufficient degree of assurance, what errors did or did not harmfully affect the jury’s verdict. Cf. Christian v. Boston & M. R. R., 2 Cir., 109 F.2d 103, 105; Schilling v. Delaware & H. R. Corp., 2 Cir., 114 F.2d 69, 72.

That problem is particularly difficult of solution where the judge’s charge is confusing and contradictory. Deserant v. Cerillos Coal Co., 178 U.S. 409, 420, 20 S.Ct. 967, 44 L.Ed. 1127; Fillippon v. Albion Vein Slate Co., 250 U.S. 76, 83, 84, 39 S.Ct. 435, 63 L.Ed. 853; Cummings v. Pennsylvania R. Co., 2 Cir., 45 F.2d 152, 153; Schroble v. Lehigh Valley R. Co., 2 Cir., 62 F.2d 993, 996; J. H. Sullivan Co. v. Wingerath, 2 Cir., 203 F. 460; Mideastern Contracting Corp. v. O’Toole, 2 Cir., 55 F.2d 909, 911; Memphis Furniture Mfg. Co. v. Wemyss Furniture Co., 6 Cir., 2 F.2d 428, 432.

It is, indeed, difficult for me to distinguish the case at bar from Cummings v. Pennsylvania R. Co., 45 F.2d 152, 153, decided by this court in 1930. There a judgment on a verdict was reversed where, after an incorrect charge as to presumption of negligence, the trial judge later gave a correct charge. This court said: “If the latter cured the error in the former, all is well, but it is impossible to reach that conclusion. Having been told plainly that the plaintiff was cast unless it explained how the injury occurred and established that it was free from negligence, nothing short of an express repudiation of that charge coupled with a correct statement of the law can be thought to have erased the erroneous impression from the minds of the jurors. The subsequent charge given, not as an express correction and with no attempt to point out to the jury the difference between it and what had previously been said, would, in all probability, have been treated only as a restatement of what had gone before. Quite likely the jury was unaware of any change. At best, it did know of it and was left to take its choice between two inconsistent statements of the law, one of which was wrong and one right.” In 1933, this court, in Schroble v. Lehigh Valley R. Co., 2 Cir., 62 F.2d 993, 996, said: “Consequently the instruction that the defendant was bound to ‘explain away the presumption of negligence’ was not correct. Nor was it fairly cured by the instruction that ‘the burden is on the plaintiff to prove negligence,’ for the instruction about the burden of proof was confused by the prior statement that it was the duty of the defendant to ‘explain away the presumption of negligence.’ ” In Chicago, B. & Q. R. Co. v. Kelley, 8 Cir., 74 F.2d 80, 84, 85, it was said: “Where the court has given an erroneous instruction, it may be withdrawn or explained by the court so as to remove error which otherwise might be present. But the withdrawal must leave no doubt in the minds of the jury as to what the court ultimately declares the law to be.” Cf. St. Louis, I. M. & S. Ry. Co. v. Needham, 8 Cir., 52 F. 371, 377, 378; Louisville & Nashville R. Co. v. Johnson, 7 Cir., 81 F. 679, 681.9

I have found no decisions of this court or of the Supreme Court expressly or impliedly over-ruling the Cummings case, the reasoning of which, I think, is both persuasive and controlling. However, as my colleagues seem to me to reject it here, it is desirable to add this reasoning:

The judge’s charge is a markedly important component of the postulate underlying the general-verdict-jury system, which postulate may be briefly described thus: The jury listens with strict attention to what the judge tells them concerning any particular applicable legal rule; they understand thoroughly the meaning of that rule; they, accept it as a major premise, and, with nice logic, “apply” it to the minor premise (i.e., the “facts” as the jury “finds” them after careful deliberation) ; they thus arrive at their verdict. A serious error in the judge’s instructions, likely to confuse the jury, can be regarded as harmless only if one rejects that basic postulate.

There are many who are sceptical about its correspondence with the actualities of jury behavior. Whether or not that scepticism is justified, we, while acting in our judicial capacity, are forbidden, as a matter of “law,” to inquire: It is well settled that, generally, it is improper, after *516the trial' is concluded,, fór a court to learn, even from the jurymen themselves, how any jury went about its business.10 As a consequence, the manner in which any particular jury discharged its functions is, for us an unknowable, when that jury was permitted to return a general verdict unaccompanied by special findings of fact. As Sun-derland (one of the ablest students of trial procedure)' says: “In a vast majority of cases, the [general] verdict is a complete mystery throwing a mantle of impenetrable darkness over the operations of the jury. * * * No one but the jurors can tell what was put into it, and the jurors will not be heard to say. * * * The court protects the jury from all investigation and inquiry. * * * The peculiarity of the general verdict is the merger into a single indivisible residuum of all matters, however numerous, whether of law or fact. It is a compound made by the jury which is incapable of being broken up into its constituent parts. No judicial reagents exist for either a qualitative or quantitative analysis. The law supplies the means for determining neither what facts were found, nor what principles of law were applied, nor how the application was made. There are, therefore, three unknown elements which enter into the general verdict: (a) the facts; (b) the law; (c) the application of the law to the facts. And it is clear that the verdict is liable to three sources of error, corresponding to these three elements. It is also clear that, if error does occur in any of these matters, it cannot be discovered, for the constituents of the compound cannot be ascertained. * * * ”11

Such is the situation here. Had there been a special or fact-verdict, or if the general verdict had been coupled with, answers to' fact-interrogatories, we would almost sur.ely know whether the mistaken portion of the judge’s charge contributed to the jury’s conclusions. As it is, we are unable so to determine. Not that special or fact-verdicts (or jury answers to fact-interrogatories) can be cure-alls. But they can do-much to bring the behavior of juries as fact-finders into line with that fortunately increasing tendency to improve fact-finding manifested in the requirement that the orders of administrative agencies, and of judges sitting without juries, must be accompanied by adequate findings of fact. Cf. N.L.R.B. v. Virginia Electric Power Co., 314 U.S. 469, 479-480, 62 S.Ct. 344, 86 L.Ed. -; Interstate Circuit, Inc., v. United States, 304 U.S. 55, 58 S.Ct. 768, 82 L.Ed. 1146; Matton Oil Transfer Corp. v. The Dynamic, 2 Cir., 123 F.2d 999, 1001; United States v. Forness, 2 Cir., 125 F.2d 922, 942; Petterson Lighterage & Towing Corp. v. New York Central R. Co., 2 Cir., March 10, 1942, 126 F.2d 992.

The adequate determination of the “facts” in law suits is the weak spot in the judicial process.12 Many thousands of dollars and years of effort have been spent in devising “Restatements” of the legal rules; but, even if they were faultless; they would be of little worth in the administration of justice, if courthouse fact-finding were not substantially bettered.13 As an important step in that direction, Rule 49 of the new Federal Rules of Civil Procedure was designed to encourage the use of the special verdict or, in the alternative, the general verdict accompanied by the jury’s answers to interrogatories as to issues of fact.14 It is strange to insist on niceties in *517fact-finding by administrative officials and trial judges, all men trained in the art of sifting evidence, and, at the same time, to fail to do what is possible to require jurors, untrained in that art, to disclose more accurately their reactions to the evidence. But, sitting in an appellate court, we have no power to direct the trial judges to call for fact-verdicts. When, however, a trial judge fails to do so, we ought to be singularly careful to see to it that the resultant unexplained lump-verdict of the jury does not cover up substantial errors at the trial.

Although some judges (and I am one) may be highly doubtful as to the value of the jury in most non-criminal cases, especially when the verdicts are general, and although it is entirely appropriate that we should inform the public and Congress of our doubts, we must, until Congress more drastically alters the jury system, accept the general verdict as one of the facts of life. We have no right, because of our doubts, to disregard, either consciously or unconsciously, what I have called its basic postulate. That postulate must govern us, no matter how lacking in reality we may think it. While performing our judicial duties, we must often enforce rules we deem unwise. And so, when on the bench, our private views concerning the wisdom of the general verdict and its necessary implications are as irrelevant as our attitudes towards bimetalism or the transmigration of souls.15 Cf. Holmes, J., in Aikens v. Wisconsin, 195 U.S. 194, 206, 25 S.Ct. 3, 6, 49 L.Ed. 154: “It was urged farther that to make a right depend upon motives is to make it depend upon the whim of a jury, and to deny the right. But it must be assumed that the constitutional tribunal does its duty. * * * ”16

If, as I believe we are required to do, we adhere to the basic postulate of the general verdict system, then, where there is a conflict in the evidence, this, I think, follows: (1) Substantial errors in the judge’s instructions as to the legal rules must be assumed to have resulted in an erroneous verdict unless those errors were so adequately corrected by the judge that we can say that, without doubt, the jury knew that they were to be ignored. (2) If the instructions are confusing and contradictory on, a material issue, the jury must be assumed to have been confused and unable to reach a proper verdict. The application of some such rules of thumb to the instant case induces my dissent.

For I see no basis for the belief that, in some unexplained way, the jury penetrated the judge’s mind and thus learned that, in the last sentence of his charge, he meant to withdraw his former faulty advice. Not that there is no place for repentance; there is room for a more or less unequivocal correction by the judge of a previous mistake, of course; but here the fact of any intention to correct was, to say the least, highly equivocal.

My colleagues, to be sure, stress the fact that the proper statement of the “law;’ came at the very end of the judge’s charge. But what justification is there for believing that the last words in a speech are invariably the ones most heeded by an audience? I take it that that is not an irrebuttable presumption, a “rule of law,”’ i.e., a generalization about facts — founded on observation and experience — no longer open to debate.17 Nor is it, I believe, an established canon of psychology18 (an art which, as yet, is far from becoming a science and is possessed of very few generally accepted canons).

I find it impossible, therefore, to decide that the jury were correctly guided. To say that the jurymen were not misled is to indulge in unverified and unverifiable guessing. I fail to perceive any foundation for my colleague’s statement that they “entertain no doubt” that the second correct instruction “removed any misconception which may have been engendered in *518'the jury’s mind by the preceding” erroneous instruction. • Who are we that we should so confidently probe the mental • interiors of the jurors? A sagacious English judge, almost five centuries ago, observed that the devil himself knoweth not the mind of man. Unable, as we are, to interrogate the jurors,19 any effort on our part to determine, merely from their general verdict, which of the two conflicting' instructions influenced them, is an undertaking which would be regarded as hopeless’ by all psychologists 20 — except perhaps those devoted to that crude type of “behaviorism” which might be called “veterinary psychology” in the light of the following story told by Wigmore: “It is said that Prince von Bismark was once ruffled by the number of questions put to him by his medical attendant, an eminent physician of no less individuality and self-possession than the Chancellor. The' latter intimated that’ the physician could do his duty without putting so many intrusive questions. ‘Very well, Highness,’ said the other, ‘but, if you wish to be cured, without questions asked, you had better send for a veterinary surgeon.’ ” 21

The best that can be said is that the chances that the jury were not misled are as good as the chances that they were. In such circumstances, it does not seem to me that we should conclude that the error was without substantial harm to plaintiff and that she received a fair trial. The majority opinion says “Whether a subsequent correct charge is sufficient to cure a prior inadvertent error22 is necessarily a matter to be determined in the sound judgment of the appellate court.” I agree, where it is clear that the second charge could have left “no doubt in the minds of the jury as to what the court ultimately declares the law to be.” 23 But this is not at all such a case. To say here that the jury’s mind was cleared of all doubt is to indulge in pure conjecture. And such conjecturing is not a matter for the appellate court.

There are possible two extreme and opposing positions concerning such a case as this: (1) Nothing in the judge’s charge makes any difference to the jury; therefore even the grossest errors or the most pronounced inconsistencies in the charge are harmless. (2) Everything in the charge must be assumed to be of importance; therefore, even the slightest error is prejudicial. ■ Common sense suggests a middle category: If the evidence is such that all doubts as to the “facts” of the case are negligible and it is fairly apparent that no other verdict than that which was rendered could have been reached by a jury in its senses, then the verdict should not be disturbed except for the grossest inconsistencies in the charge. But that is not this case, for here the evidence was in conflict.

Indeed, my colleagues say, “There were matters for the jury.” Of course there were, since the witnesses flatly disagreed with one another as to the pivotal facts in the case. Accordingly, had the jury brought in a verdict for the plaintiff, I cannot believe that my colleagues would have reversed the judgment on the ground that such a verdict was against the weight of the evidence. I fail, therefore, to understand what they mean when they say “it is difficult to see how” the jury “could properly have reached any other verdict” than that rendered for the defendant. If there had been no. prejudicial error, of course that verdict would have to stand. But where, because of misleading instructions, there has been substantial error' in a'jury trial, it is not, I think, for upper court judges to say that that error should be overlooked because, reading the record, they think that the testimony of one witness is more credible than that of another, especially when the witnesses testified not by deposition but wholly in open court. The Supreme Court decisions, cited above, show that, in jury cases, federal courts must not thus apply the harmless error doctrine.

*519Consequently, if we affirm here, we will be substituting our finding for a jury’s. This appears from the following analysis: The verdict occurred after the jury had heard a confusing charge. As no jury has considered the evidence in the light of a proper charge, our determination that the verdict rendered should stand will mean, in practical effect, that we are deciding, in a case where the evidence is in conflict, what verdict should be rendered by a properly instructed jury. And that means, in effect, that the court is bringing in a verdict. But, where, as here, there is conflicting evidence, the Seventh Amendment forbids judicial verdicts in jury cases. The Supreme Court last year reminded us of “the exclusive power of juries to weigh evidence and determine contested issues of fact — a jury being the constitutional tribunal provided for trying facts in courts of law.” Berry v. United States, 312 U.S. 450, 452, 453, 61 S.Ct. 637, 638, 85 L.Ed. 945, reversing 2 Cir., 111 F.2d 615; Conway v. O’Brien, 312 U.S. 492, 61 S.Ct. 634, 85 L.Ed. 969, reversing 2 Cir., 111 F.2d 611; Gunning v. Cooley, 281 U.S. 90, 95, 50 S.Ct. 231, 74 L.Ed. 720; Slocum v. New York Life Ins. Co., 228 U.S. 364, 33 S.Ct. 523, 57 L.Ed. 879, Ann.Cas.1914D, 1029; Richmond & Danville Railroad v. Powers, 149 U.S. 43, 13 S.Ct. 748, 37 L.Ed. 642; Sioux City & P. Railroad Company v. Stout, 17 Wall. 657, 663, 664, 21 L.Ed. 745. To use an expressive street-word, judges must not be “verdictators.”

The previous decisions of this court as to the prejudicial effect of confusing instructions are, I think, in accord with my conclusion. There are, in other federal courts and in state courts, dozens of similar decisions, but it would be tedious to cite them. The majority opinion cites two cases said to be contra: (a) The opinion of Judge Aldrich, in 1911, in Taxi Service Co. v. Phillips, 1 Cir., 187 F. 734, which is distinguishable on its facts; if not, it gives no persuasive reasons for its contrary conclusion, (b) People v. Goldstein, 285 N.Y. 376, 383, 34 N.E.2d 362; it lays down no legal rule but turns on its peculiar facts and does not purport to over-rule earlier New York cases (such as Chapman v. Erie Railway Co., 55 N.Y. 579, 587; Ulrich v. Ulrich, 136 N.Y. 120, 122, 123, 32 N.E. 606, 18 L.R.A. 37; and Johnson v. Blaney, 198 N.Y. 312, 317, 91 N.E. 721) which state the rule I think applicable here. Moreover, as I read Berry v, United States, supra, and related cases, we are not required, on the issue in the instant case, to follow New York decisions pursuant to Erie R. Co. v. Tompkins, 304 U.S. 64, 58 S.Ct. 817, 82 L.Ed. 1188, 114 A.L.R. 1487, since we, unlike the state courts, are controlled by the Seventh Amendment.