(dissenting):
Although I agree with much in the Court’s opinion, I firmly believe we should remand the record in this case for supplementation on certain factual issues, discussed in more detail below, before making any final decision on whether Lloyd Lee’s conviction should be affirmed.
The Court as a general matter would leave issues of credibility of “shabby witnesses” such as we confront in this case to the realm of argument by trial counsel. The Court restricts the responsibilities of the trial judge because over-use of cautionary instructions on credibility might “oversteer members of a jury . . . notwithstanding a general disclaimer that theirs is the fact function.” While I am very sensitive to judicial actions which might prejudice the jury,1 I find the Court’s concern that cautionary instructions might have *126that effect highly problematic. The crucial fact is that the whole range of “shabby witnesses” present ascertainable and serious issues of credibility.2 Judicial responsibility for the effect of untrustworthy evidence is, I submit, traditional.3 One need refer only to the rule against the admissibility of hearsay, the limitations on the admissibility of evidence of prior crimes, the cautionary instruction on identification testimony,4 the instruction that evidence be admitted only for a limited purpose5 and to some extent the use-immunity instruction.6 A failure to give these instructions, and I believe a failure to give the addict-informant instruction, “understeers” the jury, since the trial court itself (and this Court on appeal) must take responsibility for the threat to justice posed by unreliable evidence. To the extent the Court’s opinion is based on a delegation of the responsibility for unreliable evidence to counsel, I must dissent.7 I am fortified in my conviction that the sensitive issue of witness credibility cannot be left totally to argument of counsel by Judge Leventhal’s recent assertion that improper comment on credibility by the presecutor is “not significant in terms of influencing a conviction.” 8 If prosecutorial comments on credibility will not significantly affect the jury, it is hard to see how defense comments on credibility can so affect the jury. It follows that the dangers of “shabby witnesses” will go unmet unless the trial judge assumes responsibility for deflating any false credibility such witnesses might have.
Another theme running through the Court’s opinion is that the trial judge should have some discretion in the decision to issue a cautionary instruction on credibility. This discretion is coordinate to the degree of unreliability of a particular witness. Thus, a mere accomplice may testify without an instruction if there is substantial corroboration of his testimony. At that point, I take it, there is a delicate issue as to whether the witness is in fact unreliable and the trial judge who has the opportunity to view the witness is best qualified to judge the need for a cautionary instruction. On the other hand, the testimony of an addict-informant is so suspect that an appeals court should reverse whenever the addict-informant’s testimony is not corroborated in a respect material to the source of his unreliability. This distinction is based on the fact that an accomplice is considered unreliable because of a suspicion that he may have a motive to lie and an addict-informant is considered unreliable because of a documented existence of an actual motive to lie.
*127Turning first to the accomplice issue in this case, I note that I have recently expressed my reservations about affirming convictions when no accomplice instruction is given and the accomplice’s testimony is not corroborated in every material respect.9 I realize that this is a minority position, to say the least, and that the Court correctly states that the weight of precedent supports its result. I need not now decide whether the Court’s arguments for relegating discretion to the trial judge 10 on the grant of the accomplice instruction where substantial corroboration exists persuade me; I believe that the record in this case should be remanded for supplementation on the issue of the propriety of an addict-informant instruction.
The addict-informant instruction, the Court asserts, need only be given where the government presents a paid addict-informant who has criminal charges pending against him. I do not believe the class of witnesses subject to the addict-informant cautionary instruction is so limited. The Court correctly identifies the specific motive to lie which necessitates the addict-informant instruction. However, it presents no convincing reasons why that particular motive is limited to the class of witnesses it describes. While I am not prepared to define with precision the outer boundaries of the class of witnesses which are subject to this specific motive to lie, I can with confidence assert that an addict witness who bargains his testimony against another defendant specifically in return for immunity or a plea to reduced charges is subject to this motive to lie.11 As the Court notes, the need for the addict-informant instruction is obviated only when the witness’s testimony is corroborated in every material respect. Since Faith Wise’s testimony is not so corroborated,12 I would reverse and remand for a new trial if it were clear that Wise were an addict who bargained her testimony against Lee for a plea to reduced charges.13
Unlike the Court, I find that the factual issue of whether Wise did make such a bargain cannot be resolved on the basis of this record. For that reason I would remand the record for supplementation. The evidence in the record which leads me to entertain doubt about this issue is as follows. First, we have Wise’s plea bargain: she pleaded to one count of intent to distribute marijuana and at least three other counts, including accessory after the fact in the Gordon King robbery, intent to distribute heroin and intent to distribute cocaine were apparently dropped. Furthermore, *128on the stand she admitted participation in two other drug crimes for which she has not, according to the record before us, been prosecuted. This plea bargain should be viewed against the backdrop of a grand jury investigation into the drug traffic in the District of Columbia conducted by the prosecution. Indeed, the prosecutor offered Lee immunity in return for his testimony before the grand jury, an offer Lee refused. The sequence of Wise’s arrest on November 1, the search of Lee’s home three weeks later and his arrest a week after that support the inference that Wise was the initial source of evidence against Lee, evidence given while she was apparently in plea negotiations. Finally, against this background, the prosecutor’s request for an addict-informant instruction is significant. This combination of circumstances is such that a remand for supplementation of the record would not be tantamount to requiring the instruction whenever a government witness has recently pleaded guilty to reduced charges.
It is true that Lee’s attorney did not fully explore the issue of Wise’s status as an addict-informant, apparently limiting himself to an unsuccessful cross-examination of Wise herself. I am not convinced that this lost opportunity to lay the factual predicate for the instruction requires affirmance. I hold this view for two reasons. First, under the specific facts of this case, it is apparent that the District Court limited Lee’s cross-examination of the police officers who searched Lee’s home and arrested him.14 This limitation was important since cross-examination of those officers as to the source of their information against Lee — the precise subject Lee’s counsel sought to explore — would have developed whether Wise was the sole source of the original evidence against Lee. That evidence, if favorable to Lee, could have been used to more effectively cross-examine Wise. It might also have led to further investigation by Lee’s counsel. Furthermore, the District Court’s summary rejection of the defense request for an instruction may also have deterred any further attempts to develop the factual predicate for the addict-informant instruction.15
Second, and most important, I am unwilling to place the entire responsibility for developing this factual predicate on defense counsel. My views on the responsibilities of the trial judge for unreliable evidence necessitate a holding that on the facts of this case the District Court should have received an assurance on the record from the appropriate prosecutorial official that no specific bargain had been made for Wise’s testimo*129ny against Lee. This is not an onerous responsibility and I would not impose it unless the defense has come forward with some evidence that suggests a specific bargain. Furthermore, it is certainly not unknown for the prosecution to silently permit its witness to perjure himself with regard to potential plea bargains.16 If the trial court fails to assume this responsibility, this Court need not reverse and grant a new trial. We can, as I suggest here, remand the record for supplementation on this issue. I, therefore, would dispose of Lee’s case for the time being in that manner.
The Court properly closes its opinion with general observations on the fairness of Lee’s trial. But what is its conclusion? Not really that Lee had a fair trial but that his counsel had a fair opportunity to argue on the addict-informant issue. Neither the Court nor I can conclude as to the fairness of Lee’s trial until we know whether Wise was in fact an addict-informant.17 As to that issue, we can only speculate 18 and ponder on the performance of the District Court and Lee’s attorney. I would forego such an exercise and opt instead for a hearing on whether Wise was in fact an addict-informant. The dangers of this speculation are compounded by the fallibility of even the best of the trial bench and bar. Another trial judge or counsel might well have recognized the importance of developing the facts as to Wise’s status as an addict-informant. So the Court adds up the elements of chance and speculation, producing a combination which always totals up to affirmance. When we contrast the uncertainty of this exercise with the small effort needed to eliminate it — a remand for supplementation of the record — I fear we do a great injustice, more to our own ideals, perhaps, than Lloyd Lee, by affirming. without further inquiry. Furthermore, in weighing the total costs and benefits of a remand for supplementation of the record, we should properly consider the future effects of our failure to require information which would lay to rest the danger of serious error in Lee’s trial. Such a failure could well encourage a fatal laxity in the administration of criminal justice. This laxity will produce more, and more difficult, errors than the one the Court avoids in this case. The prophylactic effect of a strategic remand thus will not only serve the interests of fairness to defendant Lee but will conserve judicial resources in the long run.19 Finally, I think it important in conclusion to recall that the issue which divides the Court in this case does not concern some issue collateral to the purposes of the criminal justice system; rather it concerns an issue which raises a serious question about Lee’s guilt on the charges of intent to distribute cocaine and heroin. Therefore, I must respectfully dissent.
ORDER
On consideration of appellant’s petition for rehearing, it is
Ordered by the Court that appellant’s aforesaid petition is denied.
Statement of BAZELON, Chief Judge, as to why he voted to deny rehearing:
I see no need for reconsideration of the holding in this case. However, I do think that the Court’s opinion should be modified by striking the paragraph numbered 4 on page 125. This paragraph appears to have been added as an afterthought. However, it purports to eon*130tain a supplementary holding that there is sufficient corroboration of Wise’s testimony such that no instruction on credibility need be given even if she were an addict-informant. I see no reason for reaching out to decide this issue, since it is completely unnecessary to the result, the major part of the Court’s opinion being devoted to the proposition that there is insufficient evidence that Wise is in fact an addict-informant. Furthermore, paragraph 4 obfuscates the holding on this major point concerning the state of the record on Wise’s status as an addict-informant. Is all that discussion mere dicta? But more important than this, this afterthought in paragraph 4 purports to establish a rule of law on the degree of corroboration necessary to obviate the need for the addict-informant instruction, thus resolving the conflicts between the various opinions in United States v. Kinnard, 150 U.S.App.D.C. 386, 465 F.2d 566 (1972) on that issue. I think such a resolution should await a case in which the issue is squarely confronted and necessary to disposition of the cause.
Passing the issue of the propriety of the holding, I furthermore think that paragraph 4 represents an incorrect view of the facts in this case. The Court notes on page 122 that “the existence of material corroboration would not obviate the cautionary instruction where there was a lack of corroboration on a central issue as to which informants as a class have a motive to lie, namely the issues of disposition and entrapment.” It is not necessary in this case to explore the full ramifications of this general proposition. The facts of Lee’s case require us to consider only the issue of corroboration of Lee’s disposition to distribute certain quantities of drugs found in his apartment.1 Both the Court and myself agree that there must be corroboration on the issue of disposition. Whether we might disagree on the necessity of corroboration in other respects is an issue not raised by this case and which may be confronted in some later case. Such a potential dispute is not resolved by the holding of this case.
Turning to the evidence on disposition, I think it clear enough that the prosecution must prove Lee specifically intended to market the drugs found in his apartment. This burden can not be sustained merely by proving that a marketable quantity of drugs was found in Lee’s apartment with the paraphernalia incident thereto. There must be some proven connection between Lee and the drugs. The prosecution sought to establish that connection by proving that Lee occupied his apartment with full knowledge of the drugs. However, I think the prosecution failed in this effort since the overwhelming mass of evidence tends to show that Lee moved out of his apartment when “Roach” Henry and Wise moved in, because of his fear of Henry. To be sure, Lee did return to his apartment for clothes and most likely did know that Wise and Henry had large quantities of drugs in the apartment. However, that is not proof of a specific intent on the part of Lee to distribute the drugs. The only evidence tying Lee to those drugs was Wise’s testimony that she, Henry and Lee had made a deal in which Lee was to market the drugs in return for a share of the profits.
The Court in footnote 46 states that there is sufficient corroboration of Lee’s disposition because of three facts: (1) the presence in Lee’s apartment of a marketable quantity of drugs; (2) the presence of the attendant paraphernalia; and (3) the presence of a large sum of cash. I agree that this is sufficient corroboration that someone intended to market the drugs. But I fail to perceive how it is corroboration of testimony that Lee was that someone. The *131only corroboration on that score is the fact that the items were in Lee’s apartment, but as stated above the bulk of the evidence establishes that Lee moved out of his apartment after Henry and Wise moved in. There is thus no corroboration of Wise’s testimony that Lee agreed with her and .Henry to distribute the drugs in return for a share of the profits. It follows that if Wise were an addict-informant, the Kinnard instruction must be given.
ORDER
Appellant’s suggestion for rehearing en bane having been transmitted to the full Court and no Judge having requested a vote thereon, it is
Ordered by the Court en banc that appellant’s aforesaid suggestion is denied.
Statement of BAZELON, Chief Judge, as to why he would vote to deny rehearing en banc.
While I adhere to the views expressed in my dissent, I do not believe this case presents a significant question of law such that rehearing en banc would be appropriate. I do not think the result of the majority opinion necessarily binds a future panel in either its sensitivity to the need for supplementation of the record in close cases or its concern with judicial responsibility for the effect of untrustworthy evidence. Rather on the particular facts of Lee’s case the majority was of the view that there was insufficient doubt about Wise’s testimony and its effect on Lee’s conviction to justify further inquiry. Such decisions are not, except in unusual circumstances, appropriate for reconsideration en banc. See International Union, Electrical Workers v. NLRB, 163 U.S.App.D.C. 347, 502 F.2d 349, 361, cert. denied, 417 U.S. 921, 94 S.Ct. 2629, 41 L.Ed.2d 226 (1974) (Statement of Bazelon, C. J., in which Wright and Robinson, JJ., concur).
. See Young v. United States, 129 U.S.App.D.C. 312, 346 F.2d 793 (1965); Blunt v. United States, 100 U.S.App.D.C. 266, 244 F.2d 355 (1957).
. See United States v. Leonard & Sarvis, 161 U.S.App.D.C. 36, 494 F.2d 955, 974-975 (1974) (Bazelon, C. J. concurring in part, dissenting in part); United States v. Kinnard, 150 U.S.App.D.C. 386, 390-391, 465 F.2d 566, 570-571 (1972).
. See generally I Wigmore on Evidence § 21, at 374 (3d ed. 1940); A. Goldstein, Reflections on Two Models: Inquisitorial Themes in American Criminal Procedure, 26 Stan.L.Rev. 1009, 1022 (1974). Compare Carrado v. United States, 93 U.S.App.D.C. 183, 193, 210 F.2d 712, 722 (1953) cert. denied sub nom., Atkins v. United States, 347 U.S. 1018, 74 S.Ct. 874, 98 L.Ed. 1140 (1954); United States v. Ragsdale, 438 F.2d 21, 27 (5th Cir.), cert. denied, 403 U.S. 919, 91 S.Ct. 2231, 29 L.Ed.2d 696 (1971).
. See Macklin v. United States, 133 U.S.App.D.C. 139, 409 F.2d 174 (1969); United States v. Shelvy, 148 U.S.App.D.C. 1, 458 F.2d 823 (1972).
. See United States v. McClain, 142 U.S.App.D.C. 213, 440 F.2d 241 (1971); United States v. Gilliam, 157 U.S.App.D.C. 375, 484 F.2d 1093 (1973).
. See United States v. Leonard & Sarvis, 161 U.S.App.D.C. 36, 494 F.2d 955, 961 (1974).
. The significance of placing responsibility for the reliability of evidence on the trial judge is largely reflected in the scope of the plain error rule. See United States v. Leonard & Sarvis, 161 U.S.App.D.C. 36, 494 F.2d 955, 976-977 (1974) (Bazelon, C. J. concurring in part, dissenting in part). I thus disapprove of the holdings in Virgin Islands v. Hendricks, 476 F.2d 776 (3d Cir. 1973) and United States v. Collins, 472 F.2d 1017 (5th Cir. 1972). Compare Tatum v. United States, 88 U.S.App.D.C. 386, 190 F.2d 612 (1951).
. United States v. Young, 150 U.S.App.D.C. 98, 102, 463 F.2d 934, 938 (1972).
. United States v. Leonard & Sarvis, 161 U.S.App.D.C. 36, 494 F.2d 955, 977 (1974) (Bazelon, C. J., concurring in part, dissenting in part).
. I note that there is no evidence in the record before us that the trial judge’s refusal to give the accomplice instruction was an exercise of informed discretion. Bather, it does not appear from the record what, if any, consideration the District Court gave to the need for the instruction.
In this context, I note my continuing puzzlement at a rule of law that states it is the “better practise” to give an instruction and then refuses to reverse a conviction on that basis. Is this a rule of harmless error? I think not, unless there is complete corroboration of the suspect witness. The Court’s point seems to be that it isn’t always the “better practise” to the give the instruction; while I might disagree with that, it is an intelligible position.
. See United States v. Leonard & Sarvis, 161 U.S.App.D.C. 36, 494 F.2d 955, 961 (1974) which in effect extended the principle of United States v. Kinnard, 150 U.S.App.D.C. 386, 390-391, 465 F.2d 566, 570-571 (1972) to witnesses testifying under use-immunity. If Wise did make a bargain in this case before us now, she would be testifying under a limited use-immunity and the Leonard & Sarvis case should apply. Furthermore, the various factors which give rise to a specific motive to lie, as presented in Kinnard, are virtually all applicable to an addict who bargains a favorable plea in return for her testimony. See 150 U.S.App.D.C. at 391, 465 F.2d at 571.
. In particular, there is no corroboration of Faith Wise’s testimony that she and Henry made a bargain with Lee in which Lee would market certain drugs in return for a share of the profit.
. I find there is sufficient evidence in the record to establish that Wise was an addict for purposes of applying the Kinnard factors.
. Trial Tr. at 74-76. The Court correctly notes that nothing in the record establishes that Lee’s counsel was attempting to uncover the source of the government’s information against his client by his examination of Agent Seibert. However, neither does the record establish the opposite conclusion. Furthermore, it is not necessary for us to establish counsel’s motivation in order to give some weight in decision to the probability that untrammeled examination might have produced information bearing on the source of the government’s information against Lee. The implication in the Court’s opinion to the effect that the District Court did not really limit cross examination must contend with the fact that the District Court very definitely sustained a prosecution objection to a line of questioning by Lee’s counsel.
. See Trial Tr. at 554, 557; Butler v. United States, 88 U.S.App.D.C. 140, 188 F.2d 24 (1951). The Court’s suggestion that this abrupt denial could not have deterred efforts by defense counsel to further develop information on Wise’s status because it came at the end of the trial is not persuasive. A sensitive approach by the District Court at the time of the instruction request might have produced at least two beneficial results. First, Lee’s counsel might have responded to a suggestion by the Court that insufficient evidence existed to determine whether Wise was an addict-informant by moving to re-open the trial for the limited purpose of adducing proof on Wise’s status. Second, Lee’s counsel might have requested the District Court to order the prosecution to verify Wise’s testimony that she had received no deals in exchange for her testimony. Such an order would seemingly be justified by the prosecutorial duty enunciated in Napue v. Illinois, 360 U.S. 264, 79 S.Ct. 1173, 3 L.Ed.2d 1217 (1959).
. See Napue v. Illinois, 360 U.S. 264, 79 S.Ct. 1173, 3 L.Ed.2d 1217 (1959).
. Of course, this assumes that if Wise were an addict and had made a specific bargain for her testimony, Kinnard would be applicable and thus reversal a necessity; the Court, as I indicate in the text, seems to be arguing that Kinnard is not applicable in such circumstances. I may, therefore, be overstating the Court’s position.
. Cf. United States v. Henson, 159 U.S.App.D.C. 32, 50, 486 F.2d 1292, 1310 (1973) (Bazelon, C. J. concurring in part, dissenting in part).
. See United States v. Leonard & Sarvis, 161 U.S.App.D.C. 36, 494 F.2d 955, 977 (1974) (Bazelon, C. J. concurring in part, dissenting in part).
. I think there is sufficient evidence quite apart from Wise’s testimony to sustain the convictions for’ possession of heroin, cocaine, marijuana and a sawed-off shotgun.