On this appeal from a conviction of bank robbery, it is contended that reversible error appears in the trial judge's supplementary charge to the jury.

During the course of its deliberation the jury sent two written communications to the judge. The first read: "It appears that we cannot arrive at any decision in this case. We have one member who refuses to enter into deliberation on the basis that all of the witnesses are not to be believed." The second note read: "The same juror still states that she has doubts that the defendants were the men in the bank and also doubts other parts of the testimony. The remaining jurors cannot carry on effective deliberation due to this unyielding attitude. The question seems to revolve on her understanding of 'reasonable doubt.' Any helpful suggestion would be appreciated."

With the matter in this posture and after the jury had been out for about four hours, the trial judge had the jury returned to the courtroom where he gave the challenged supplementary instructions.

First, the judge repeated his clearly correct charge on "reasonable doubt." Next, he summarized the testimony of the witnesses who had identified the defendants as the robbers and in so doing indicated, inferentially at least, that to him the identification testimony seemed strong and persuasive. Finally, the court said this to the jury:

"Another thing you should bear in mind is this: If this jury cannot agree, the case is going to have to be tried again. It took almost a week-did take a week-and all those witnesses are going to have to be called back to testify again. They will be subject to examination, cross-examination, and so forth. So, it is not an insignificant event when a jury does not agree in a case like this. It can produce great additional expense to the government and additional-well, I think I have said enough."

The appellant excepted to this supplementary charge.

We have recently disapproved the use of a supplementary charge, in that case the so-called Allen charge, "to blast a hung jury into a verdict." United States v. Fioravanti, 3 Cir. 1969, 412 F.2d 407, 419. We reasoned there that a juror should not be subjected to judicial pressure to vote contrary to his honest best judgment because a large majority of his fellow jurors have reached a conclusion to which he is not persuaded.

In this case we think that the concluding sentences of the supplementary charge, however well intended, were incorrect and prejudicial. To the dissenting juror the charge must have meant that she should consider the expense of a new trial to the government and its imposition upon the time of many people as a significant factor that could and should persuade her to change her vote. But a juror's responsibility is to evaluate the evidence and the credibility of witnesses and, if a reasonable doubt as to proven guilt persists, to vote for acquittal. The possibility of a hung jury and a retrial is not relevant to that determination. The jurors should not be told that this circumstance should influence them. United States v. Smith, 4th Cir. 1962, 303 F.2d 341. Contra, Hale v. United States, 5th Cir. 1970, 435 F.2d 737.

In our judgment it was reversible error to give such an instruction.

We are also concerned, though our decision does not turn on this point, that the judge strongly indicated that in his view the identification testimony was adequate and credible without thereafter saying, in substance, that a juror should not vote contrary to his own conscientious best judgment because of any indication of the judge's views. In the setting of a supplemental charge to a hung jury, it is very important that the judge not seem to be imposing his view of the strength of a party's case upon the holdout or holdouts. He should at least minimize this risk by coupling any statement of his views in a supplementary charge with a clear statement, or restatement, of each juror's responsibility to exercise independent judgment.

We appreciate the judge's apprehension that time and money might wastefully be expended in a retrial as a result of jury disagreement in a case in which the evidence of guilty was very strong. But occasional necessity for such a new trial is an unavoidable consequence of a system of trial by jurors who are the sole judges of the facts and can properly convict the accused only if each of them is persuaded by the evidence that each essential fact has been proved beyond reasonable doubt.

The judgment will be vacated and the cause remanded for a new trial.

I respectfully dissent. It is my view that the alleged error was harmless and that the conviction should be affirmed.

The contents of the notes sent by the jury to the trial judge are set forth in the majority opinion and need not be repeated here. The jury ended the second note with this comment, "any helpful suggestion would be appreciated." The trial judge in open court gave the jury supplemental instructions including a definition of "reasonable doubt," reviewed some of the testimony and finally made the statement alleged to be fatal error. He said this:

"Another thing you should bear in mind is this: If this jury cannot agree, the case is going to have to be tried again. It took almost a week-did take a week-and all those witnesses are going to have to be called back to testify again. They will be subject to examination, cross-examination and so forth. So, it is not an insignificant event when a jury does not agree in a case like this. It can produce great additional expense to the government and additional-well, I think I have said enough."

The jury knew that the case would, of necessity, be tried again if they failed to reach a verdict. It was a serious case not likely to be neglected. They knew a good deal about the expense involved. They had seen the trouble and inconvenience caused by the trial. The judge was not critical nor abusive, nor threatening nor angry. He had tried to respond to the jury's request and he had said what was obvious and only what was obvious.

The evidence of guilt in the trial had been very strong. It was charged that Burley, along with two other men, had robbed a bank in Philadelphia and had escaped on foot with $25,535.00. They were unable to reach their car due to approaching police officers. They split from one another and Burley attempted to seize a car from a woman whose screams attracted a bystander who came to her aid. It was charged that Burley shot the bystander with a .38 caliber revolver, the bullet passing through him and into the body of another man. As the man who was first wounded fell he grabbed the money which Burley was carrying. Burley fled on foot but was apprehended within a few blocks by a police officer. Burley had in his possession a .38 caliber revolver containing four live rounds and one empty cartridge. Burley was taken back to the bank forthwith and was identified.

During the trial Burley was identified by two bank employees and a customer as one of the robbers. He was identified by the passerby as the man who shot him. The revolver found on him by the police officer was received in evidence. There was identifiable money (bait money) in the total sum stolen which had been grabbed by the bystander. It was received in evidence. Finally, a co-defendant who had pled guilty testified against Burley.

It is true that in United States v. Fioravanti, 1969, 412 F.2d 407, this court instructed trial judges not to give instructions that directed a juror to distrust his own judgment if he found a majority of the jurors taking a view different from his. It is true that this court said that such an instruction in the future would be deemed error, normally reversible error, although there were circumstances where the error might be found inconsequential.

However, in Fioravanti the instruction which was criticized was much more explicit than the one before us. There the trial judge said this, "the jury should listen with deference to arguments of fellow jurors and distrust his own judgment if he finds a large majority of the jury taking a different view of the case from that what he does himself."

In my view telling a juror to distrust his judgment is far more serious than telling a juror that failure to come to a verdict will result in a retrial with its attendant expense because I think the juror already knows this and has already made leeway for this result in his thinking. I think most of us worry about the expense of government. We are reminded of it every day and another reminder is not shocking even though it comes from a judge during a trial. The dissenting juror may just as well have been influenced by the redefinition of "reasonable doubt" or the review of the testimony.

I think the majority seeks too great a degree of trial perfection. Even in Fioravanti the conviction for the possession of counterfeit money was affirmed although the government's case rested largely on the testimony of a secret service agent. There are many cases where similar language by a trial court has been held harmless under the circumstances as in United States v. Tolbert, 406 F.2d 81 (7th Cir. 1969).

Frequently in talking to a jury a trial judge treads a thin line between due process and good sense. The infraction here was inconsequential in the light of the whole record.