Albert Francis Perrault was convicted on two counts of armed bank robbery ( 18 U.S.C. § 2113 (a) and (d)), the first count charging a robbery of the MacArthur and 96th Avenue branch of the Bank of America in Oakland “on or about” October 8, 1961, and the second a robbery of the same branch on January 31, 1962. Gordon Arthur Lyda was charged and convicted only under the second count. With one exception, the errors assigned on these appeals are peculiar to each appellant, and we will discuss each appeal separately.

I. The Perrault Appeal

Perrault’s first challenge is to the sufficiency of the evidence. The weakness of this challenge is apparent in the manner in which it is phrased — “The evidence tying Mr. Perrault to the crime was unreliable and misleading”- — -and we find no merit in it.

In both robberies two men entered the bank, one standing guard with a shotgun and the other picking up the money from the tellers’ cages. On each occasion partial disguises were worn; hats, false noses, and glasses were used for the first robbery, and silk stocking masks for the second. In the second robbery, a stolen Chevrolet later recovered by the police was used as the getaway ear.

Perrault is “tied” to these circumstances by three main bodies of evidence. One McCarter testified that he held the shotgun during both robberies, and that Perrault was his partner in crime. McCarter described the commission of the robberies in great detail, as well as the events leading up to and following them. He was impeached by proof of prior convictions and prior inconsistent statements, and at the time of the trial he was awaiting sentencing on his plea of guilty. Six witnesses identified Perrault in the courtroom as participating in one or both of the robberies. Five of these witnesses were bank employees. The sixth was a shopkeeper who saw three men in the getaway car shortly before the robbery of January 31. Most of the identifications are tentative and some suspicion can be cast upon them, but two at least are more positive. Finally, Perrault’s fingerprints were found on the getaway car. This evidence is more than sufficient, “viewed, as it must be, most favorably to the government.” (Twitchell v. United States, 9 Cir., 1963, 313 F.2d 425, 429 )

Perrault next contends that the prosecutor misled the jury as to the extent to which Perrault was identified with the robberies. His contention is based on the testimony of four witnesses, each of whom was asked to, and did,, identify Perrault in the courtroom. Two-of the witnesses testified to the repayment of loans by Perrault, the third to his paying a large car repair bill, and the fourth to his purchase of a car for cash,, all of the transactions being on or shortly after the dates of the robberies. Evidence of the sudden acquisition of money is usually relevant to prove a crime for-which pecuniary gain is the motive (see United States v. Jackskion, 2 Cir., 1939, 102 F.2d 683 ), but in some cases its probative value may be outweighed by the-possibility of confusing and prejudicing-the jury. Perrault argues that the identity of the bank robbers was the crucial: issue in the case, and that four identifications of him by those with whom he transacted business would influence the jury in evaluating the testimony of the bank employees identifying him as a thief.

The testimony of the four witnesses was brief. The firms they served were-identified, and none was a bank. The-business context of the testimony was obvious throughout. To find-the identifications misleading would attribute to the-jury a profound lack of discrimination.. There was no error.

Perrault asserts he was denied his constitutional right to trial by jury, alleging that one of the jurors was deaf. As proof he relies upon an affidavit of a fellow-juror, stating that the assailed juror did' not initially engage in conversation with other panel members, failed to respond, to one question from the judge and one-from another juror, and “leaned forward, in her seat while the trial was in process, and appeared to be straining to hear the • words of counsel.” Perrault would excuse his failure to raise the matter until' after verdict because it was the trial-judge who conducted the voir dire, as the-rules of the Northern District of Cali--fornia provide.

We need not decide whether the Sixth Amendment would be violated by trial before eleven good men and true and one who is hard of hearing. Section 1861(3) of title 28 disqualifies one who “is incapable, by reason of mental or physical infirmities to render efficient jury service,” and failure to abide by the statute would be reversible error. We shall assume that the error was timely raised by a motion for a new trial. (But see United States v. Baker, S.D.N.Y.1868, 24 Fed.Cas. 952 (No. 14,499); compare the state cases collected in Annot., 15 A.L.R.2d 534 ) Nonetheless, Perrault cannot prevail.

“The trial court is invested with a wide discretion in determining the competency of the jurors, and the court’s judgment in this respect will not be interfered with except for an abuse of the discretion.” (Lias v. United States, 4 Cir., 1931, 51 F.2d 215, 217 , aff’d per curiam, 1931, 284 U.S. 584 , 52 S.Ct. 128 , 76 L.Ed. 505 ; see Beck v. United States, 9 Cir., 1962, 298 F.2d 622, 629 (prejudice) ; King v. Leach, 5 Cir., 1942, 131 F.2d 8 (advanced age and infirmities)) Half the incidents relied upon to show deafness took place in the courtroom, in full view of the trial judge, one even occurring during a colloquy between the judge and the assailed juror. The affidavit was before the judge on the motion for a new trial, which he denied. There has been no sufficient showing of incompetency as to establish an abuse of discretion by a judge both apprised of the problem and in observation of the juror.

According to Perrault, the prosecutor twice improperly suggested to the jury that Perrault had a police record. On neither occasion did Perrault’s counsel indicate to the court with precision that he objected on this ground. However, since counsel for Perrault stated to the court and the prosecutor at the commencement of the trial that he was most anxious to avoid any mention before the jury, “accidently or otherwise,” of his client’s prior criminal record, we will give the alleged errors closer scrutiny than we might otherwise.

At one point the prosecutor elicited from a witness, an F.B.I. agent, the information that Perrault told him he had been absent from his home the week following the second robbery “to avoid police harassment,” and from this it is argued that a jury would infer that only a person with a record would so behave. This would indeed be a strained inference to attach to one short line of testimony; if the jury inferred anything from this avoidance of the police, it would be a consciousness on Perrault’s part of guilt of this crime. The probative value of such evidence would be slight (see Vick v. United States, 5 Cir., 1954, 216 F.2d 228 ), but any error in its admission was waived by failure to object on the ground of irrelevancy.

The prosecutor offered in evidence the fingerprint card used by a criminologist to match up the prints on the getaway car. The fingerprint card was identified by the witness as Perrault’s because “his name is recorded [on it] and also his Oakland Police Department identification number.” Perrault’s counsel objected on the ground that the card contained extraneous matter — presumably prior arrests or convictions— and stipulated that the prints on the card were those of Perrault. The card therefore was not received in evidence. Since a fingerprint card made after Perrault’s arrest had previously been offered and received, it was clear to the jury that the second card predated the second robbery. To infer a police record from the existence of a fingerprint card is conceivable, but not persuasive; fingerprint records exist for former members of the armed forces, most public employees, and, for that matter, members of the Bar of the State of California. The reference to the Oakland Police Department identification number goes a little further, and perhaps had better been stricken, but we cannot see that it so alerted the jury to the blemishes in Perrault’s past as to deprive him of his substantial rights. (F.R.Crim.P. 52(a); see United States v. Henderson, 7 Cir., 1950, 185 F.2d 189 )

Perrault claims that the trial judge, through his questioning of several witnesses, gave the jury the impression that the court sided with the prosecution. We do not so read the record; further, an adequate cautionary instruction was given.

II. The Cross-Examination of Defense Witnesses

Both Perrault and Lyda complain of the prosecutor’s conduct in cross-examining two defense witnesses, Darrow and Dubey. Darrow was put on the stand to impeach the self-confessed accomplice, McCarter. Darrow testified to a conversation in a “holding cell” used by the government, occupied at that time by the appellants, McCarter, Darrow, Dubey, and several others. Perrault allegedly asked McCarter why he didn’t reveal the true identity of his fellow robbers, and McCarter is said to have replied that “he couldn’t because he was in fear for his family, that they were on the street and so were the others.” Dubey’s version of the incident is less complete, only that he overheard McCarter say “I can’t do that; I might hurt my wife and family,” or words to that effect. It might be added that McCarter denied making these statements. His version is this: “We were discussing the Fifth Amendment that I could invoke and I said I couldn’t because I had already pled guilty and this wouldn’t make me no grounds, so I said, ‘The only way this could come about is if someone was on the street that could harm my wife and children and I could claim my wife and child’s lives were in danger.’ ”

Neither appellant specifies just what in the cross-examination of Darrow could be error, and indeed we find none. Dubey was asked if he were not using the name of Martin in jail, whether he was the same Dubey who had obtained a Mexican tourist permit in the name of Donald Ira Price, whether he had registered in a motel as Price with one Darlene McGraw and her daughter, and how many names he had used since November 3, 1961. Dubey answered only the first question, claiming his privilege against self-incrimination as to the second and fourth; it would appear that the court sustained an objection to the third. Perrault’s counsel objected to the last three ques tions as immaterial and going beyond the scope of the direct examination.

We might begin by pointing out that a trial court has broad discretion in setting the limits of cross-examination (See Marteney v. United States, 10 Cir., 1954, 218 F.2d 258, 265 ), and that the scope of the direct examination may be exceeded in an effort to test the truthfulness of the witness. (Young Ah Chor v. Dulles, 9 Cir., 1959, 270 F.2d 338 ; United States v. Lawinski, 7 Cir., 1952, 195 F.2d 1 ). The issue is whether the use of false names bears directly enough upon the witness’ veracity as to outweigh the general prohibition against cross-examining about particular acts of misconduct other than convictions of a crime. We think it does. If a man lie about his own name, might he not tell other lies ? (See Touhy v. United States, 8 Cir., 1937, 88 F.2d 930, 935 (semble); cf. Simon v. United States, 4 Cir., 1941, 123 F.2d 80 . Compare Newman v. United States, 9 Cir., 1928, 28 F.2d 681 ).

But insofar as the question dealing with registration at a motel suggested that Dubey spent the night with a woman other than his wife, it was improper. A witness may not be degraded by accusing him of improprieties not directly relevant to his veracity. (Abdul v. United States, 9 Cir., 1958, 254 F.2d 292 ; Ingram v. United States, 9 Cir., 1939, 106 F.2d 683 ) However, we do not think this misconduct by the prosecutor is reversible error. The question is open to an innocent interpretation. The court did not allow it to be answered. Dubey’s reputation had previously been tarnished by his admission that he was awaiting trial for an unspecified crime; and his value to the defense as a witness was limited both because he could give only a fragmentary and inconclusive account of the conversation and because his behavior on the stand was somewhat unconvincing.

III. The Lyda Appeal

Lyda contends that the evidence is insufficient to sustain his conviction. The telling evidence against him was given by McCarter, who testified that Lyda drove the getaway car to the bank, remained in it during the robbery, and drove Perrault and McCarter away after the robbery was completed. For this he was allegedly paid $500 by McCarter.

McCarter’s testimony indicates that Lyda’s inclusion in the scheme was casual, to put it mildly. Lyda was in no way involved in the first robbery or in the preparations for the second. McCarter and Perrault began considering a second robbery before Christmas in 1961, and around January 24, 1962, settled on the same bank they had previously robbed. On January 29, they stole the getaway car from a parking lot. On January 30, they set off to rob the bank but “we got the jitters so we decided we would wait until the next day.” On January 31, shortly after 11 a. m. — the robbery occurred at noon — McCarter and Perrault were driving along when they saw Lyda crossing the street. They hailed him and Lyda got into the car. While they were driving to McCarter’s house, McCarter told Lyda that the two were going to rob a bank, and asked Lyda to come along. “He said, ‘Yes, sure,’ but he didn’t act like he was serious * * After Mc-Carter and Perrault picked up the stolen car at McCarter’s house, Lyda “finally began to realize it was serious.” Mc-Carter offered him $500 to do the driving, and Lyda accepted, because “he needed the money. He was on bail now * * *. He borrowed the money to pay the bail.” At this time McCarter had known Lyda for three or four months, “fairly well.” Perrault and Lyda “never did know each other very well. Just for a month or so that they had seen each other more or less.” According to McCarter, Perrault had no part in making the offer to Lyda, and Lyda was paid out of McCarter’s share of the loot.

Evidence to corroborate this story is slight. A bystander testified that a Chevrolet, with its motor running and a man in the driver’s seat, was in front of the bank during the robbery, and that the two men who robbed the bank rode off in it. The witness could not identify the driver. Lyda and McCarter were together in a bar when they were arrested on the night of the robbery. At that time Lyda was carrying $178.50 in cash; he claimed he had been lent this sum, but would not tell the arresting officer from whom he had borrowed it.

McCarter’s testimony is suspect for a number of reasons. He is a convicted felon. He was awaiting sentencing for his part in the crime, and his remarks make clear that he hoped for a lighter sentence because he testified against the appellants. Statements he made before trial inconsistent with his testimony there were offered, and there is an apparent inconsistency within his testimony at the trial. His version of Lyda’s entrance into the conspiracy is somewhat bizarre.

On the other hand, McCarter’s story of Lyda’s part in the crime is not unbelievable. It obtains some slight corroboration from the evidence recited above. McCarter’s description of the commission of the two crimes, apart from Lyda’s recruitment, is detailed and has the ring of truth. It was not seriously shaken on cross-examination, and is amply corroborated. Further, the jury was instructed that accomplice testimony was to be received with caution and weighed with great care.

We find ourselves in something of a dilemma. The credibility of witnesses is a matter for the jury, not an appellate court. (See United States v. Verra, 2 Cir., 1962, 301 F.2d 381 ) If McCarter were an ordinary witness, though a felon and impeached by inconsistencies, we could not reverse. But McCarter is an accomplice, and as such he has a “built-in” untrustworthiness. It is true that in federal courts a conviction may be based on the uncorroborated testimony of an accomplice. (See Marcella v. United States, 9 Cir., 1960, 285 F.2d 322 , and cases there cited) We-think this to be the rule even though the. accomplice is in a position to gain favors-from the government by his testimony (United States v. Carengella, 7 Cir., 1952, 198 F.2d 3 (alternative holding)), and even though there are inconsistencies in. his story (see McQuaid v. United States, 1952, 91 U.S.App.D.C. 229 , 198 F.2d 987 ), so long as it is not “incredible or unsubstantial on its face.” (Haakinson v. United States, 8 Cir., 1956, 238 F.2d 775, 779 ) Obviously there comes a point when the witness’ qualifications are so shoddy that a verdict of acquittal should have been directed. (See United States v. Haderlein, N.D.Ill., 1953, 118 F.Supp. 346 ).

Bearing in mind the limited role of an appellate court in a criminal appeal, we do not think that point was reached here. But we think the sufficiency of the evidence so close a question as to bring into sharp relief the other errors which Lyda assigns.

Lyda objects to references to his prior misconduct unconnected with the offense charged. After McCarter described the meeting with Lyda on the street, the prosecutor put this question:

Hater in his testimony McCarter gave as Lyda’s reason for taking part in the robbery that he needed the money to pay bail for the earlier offense. Ample objections were made to both these remarks, by motions to strike and for a new trial, but these were denied. The government also adverted to Lyda’s arrest for burglary in its closing argument.

Lyda did not take the stand, and his character therefore was not in issue. The government puts forward two arguments to justify its attack on his character: that this was not, after all, a reference to a prior conviction, but merely an arrest; and that even if such evidence were usually inadmissible, it was here admissible to show a close and continuing relationship between Lyda and McCarter and thus explain Lyda’s abrupt inclusion in the bank robbery.

The first argument is hardly persuasive. It may be that the difference in prejudicial effect on the jury between an arrest and a conviction would have weight in our decision whether this was reversible error, but error it clearly is, unless it is within one of the excepted areas where prior misconduct can be shown. See Erwing v. United States, 9 Cir., 1961, 296 F.2d 320 “[T]he rule about the proof of other crimes is but an application of the wider prohibition against the initial introduction by the prosecution of evidence of bad character.” (McCormick, Evidence § 157, at 327 (1954)).

We also reject the government’s second argument. It is quite true that evidence of other criminal acts by the accused may be admitted to piece out the proof of the present crime, by showing such matters as a continuing scheme, motive, malice, and the like. The evidence here does not fall within one of the common exceptions. McCarter’s testimony refutes any notion that the earlier arrest was part of a continuing plan of bank robberies, and there is no evidence that the same modus operandi was employed in both robberies.

The government, however, would have us carve out a special exception, that a common bond in crime may be shown where an issue in the case is the likelihood of the accused suddenly agreeing to take part in a robbery with which he had no previous connection. We need not decide whether such evidence would ever be admissible. The government cites us to a discussion by Judge Learned Hand, with whom we agree: “[I]f evidence is relevant to prove one crime, it does not become inadmissible because it also proves another. * * * [T]he competence of the evidence in the end depends upon whether it is likely, all things considered, to advance the search for the truth; and that does not inevitably follow from the fact that it is rationally rele vant. As has been said over and over again, the question is always whether what it will contribute rationally to a solution is more than matched by its possibilities of confusion and surprise, by the length of time and the expense it will involve, and by the chance that it will divert the jury from the facts which should control their verdict.” (United States v. Krulewitch, 2 Cir., 1944, 145 F.2d 76, 80 , rev’d on other grounds, 1949, 336 U.S. 440 , 69 S.Ct. 716 , 93 L.Ed. 790 ). See also Erwing v. United States, supra, 296 F.2d at 324 : “A study of many of the decisions which apply exceptions to the general rale reveals that whether the general rule or an exception thereto should be applied depends upon the facts and circumstances of each case.”

In this weighing process, we think that the government occupies the light end of the scales. The government’s need of the testimony was slight. McCarter had already testified that he knew Lyda fairly well, that he had been out with him seven or eight times, and that Perrault knew Lyda slightly; and independent evidence of McCarter’s friendship with Ly-da surely would not have been hard to come by. The probative force of the evidence is not much greater. It may be claimed that a man may be more likely to join a friend in a robbery if he has been previously arrested with him, but the proposition would be difficult to prove. The prejudicial effect on the jury could be great; it would be a simple step from “Lyda was arrested for one burglary” to “Lyda committed this robbery.” This would be doubly unfair because Lyda was only said to have been arrested for the first burglary, not convicted, and for all the record shows he may have been acquitted, or never even tried.

Admission of the testimony was error. Since the other evidence against Lyda was scanty, it was clearly reversible error. (See Sang Soon Sur v. United States, 9 Cir., 1948, 167 F.2d 431 ).

The judgments of conviction of appellant Perrault under counts one and two are affirmed; the judgment of conviction of appellant Lyda is reversed.

. This date was a typographical error in the indictment, as the proof showed the robbery was committed on October 18, 1961. On this appeal Perrault makes no contention that the typographical error prejudiced him. The jury was instructed: “The indictment charges that the acts were committed on or about certain dates. It is not necessary tliat the proof establish with certainty the exact date of the alleged offense. It is sufficient if the evidence shows beyond a reasonable doubt that the offenses, or either of them, were committed on dates reasonably near the date alleged.” . “During the course of the trial I have occasionally asked questions of a witness in order to bring out facts then not fully covered in the testimony. Do not assume that I hold any opinion on the matters to which my questions related. Remember at all times that you, as jurors, are at liberty to disregard all comments of the Court in arriving at your own findings as to the facts.” . Lyda couples this assignment of error with a charge that the prosecutor made improper references to the two witnesses in his closing argument. Perrault uses no language showing that he assigns the second matter as error, unless it be in this sentence from his brief: “* * * for the reasons set forth in appellant’s opening brief for defendant Lyda, * * * the inflammatory and irrelevant cross-examination by the prosecutor must have debased the true value of Darrow’s and Dubey’s testimony.” This is hardly “a specification of errors relied upon which * * * shall set out separately and particularly each error intended to be urged.” (Rule 18, Court of Appeals for the Ninth Circuit) We therefore do not, on Per-rault’s appeal, consider the references to these witnesses in the government’s dosing argument. (See Pinkston v. United States, 9 Cir., 1960, 278 F.2d 833 ). . “Q. I will ask if you are the same Donald Dubey who on November 8 and 9,1961, registered at the Sahara Motel in Los Angeles with one Darlene McGraw and her daughter under name of Don Price? [Counsel for Perrault] * * * Your Honor, I am not only going to object to the question, but I ask that the jury be admonished to disregard the question of counsel and move for a mistrial on the ground of the flagrant misconduct of counsel for the Government. [Counsel for the government] * * * Well, I think we are entitled to know who this man is, Your Honor, for the purpose of his identity. The Court: What does the registration in a motel in Los Angeles have to do with it? You can ask him how many differ-erent names he has used during the course of his career.” . Dubey testified that he overheard the quoted remark and little else. On cross-examination he was asked: “When was tins conversation next brought to your attention? “A. This Monday. “Q. By whom? “A. Myself. ***** “Q. How was it that you happened to recall this conversation? “A. Mr. Perrault and I now are in the same cell with about ten other men, and we came back from court one Monday, he said that Mr. McCarter was saying that he and the other gentlemen with him were his accomplices, and I just naturally blurted out and said, ‘How could he say that when he said downstairs that they might hax-m his wife and children’ ? “Q. But you didn’t hear any of the other conversation, is that right? “A. I wasn’t paying any attention. I try to keep by myself. “Q. And as I understand your testimony, you don’t even know what the subject matter was they were talking about; is that right? “A. I can’t honestly say I know the subject matter-.” . When he was first on the stand, McCar-ter was asked on cross-examination: “Do you ever recollect any statements that, after your arrest and in the hearing- of several other people that were in the jail being held with you, you told these two defendants that you couldn’t name the people who actually had worked with you on these robberies because of your wife and children?” McCarter replied: “Yes.” At a later point he recited the version of tbe bolding cell conversation set forth in the opinion. Counsel did not follow up Ms original line of questioning, so McCarter had no-opportunity at that time to clarify his answer; and it is worth noting that he was recalled to testify on the same point because the court did not think a sufiicient foundation had been laid for McCarter’s impeachment by Darrow and Dubey.