ON REHEARING
GALLAGHER, Associate Judge, Retired:In its per curiam decision in this case (Epperson v. United States, 471 A.2d 1016 (D.C.1984)), this court held, essentially, that a trial judge should not repeatedly give an “anti-deadlock” instruction to a “hung jury.” Sometime later, the government *1172filed a motion to recall the mandate and to consider its petition for rehearing. Upon the government’s representation that, in its view, the court’s opinion would have harmful effects “on the administration of justice in this jurisdiction,” we recalled the mandate and afforded the parties further opportunity to be heard on the issue.1
Upon reargument, it appears that the government has in some respects misinterpreted our decision in this case. Perhaps we did not elaborate sufficiently because we were simply synthesizing what we felt to be an existing custom among most trial judges in this jurisdiction. First of all, in our earlier opinion we addressed ourselves only to a “hung jury,” not just simply a jury.2 By “hung jury,” we mean a jury which the trial judge has concluded is deadlocked, giving due consideration to such things as the nature and complexity of the trial issues, the duration of the trial and the length of the jury deliberations, as well as the representations of the jury to the court about the state of its deliberations. See Thompson v. United States, 354 A.2d 848, 851 n. 8 (D.C.1976). If the court decides it has before it a genuinely “hung jury,” that is the type of jury to which our decision is addressed. The court should not repeatedly give an “anti-deadlock” instruction to a deadlocked jury, bearing in mind that a deadlocked jury is one which the trial judge has concluded is genuinely deadlocked, as previously related. It is for the trial judge to resolve this factor.3
The government states in its Motion to Recall Mandate that:
When we considered that the Court had found the mild ABA instruction [see infra ] ... to be the equivalent of a Winters charge, and had held that two such charges could not be given, we realize that the safest course for a trial judge to take under the holding was initially to refrain from giving any guidance to a jury which announces itself deadlocked and simply allow the deliberations to continue, no matter how premature or ambiguous the announcement of deadlock may be. Only after a substantial amount of time had passed and the jury had still not reached a verdict should any instruction be given. Then, the strongest charge possible — the Winters charge — should be given, since the judge will be able to address the jurors only once. [Emphasis added.]
The court’s decision was concerned only with the matter of anti-deadlock instructions to a “hung jury.” That being the case, the government’s reference to instances “no matter how premature or *1173ambiguous the announcement of deadlock may be” immediately takes the question out of the category of a genuinely “hung jury” and consequently outside the ambit of the court’s opinion. The same would hold true of the government’s apprehension that the court may not give the jury any “guidance.” The court’s opinion is restricted to “anti-deadlock” instructions to a jury which the court has concluded is a genuinely “hung jury.” Naturally, as we point out in this opinion, it would not foreclose a judicial response to a request for guidance by a jury, nor, for example, would it foreclose a repetition of a “hung jury” instruction to a jury which indicates it did not understand the first “hung jury” instruction. While these things may seem obvious, it is just as well to point them out as a matter of caution.
Which “anti-deadlock” instruction it gives is in the discretion of the trial judge. One can mention three instructions of this nature in use in this jurisdiction: (a) the American Bar Association suggested anti-deadlock instruction, American Bar Association Standards for Criminal Justice, No. 15-44 (2d ed. 1980); (b) the instruction suggested by this court in Winters, supra note 1; and (c) the instruction suggested in the concurring opinion in Winters, supra. While they vary to some degree, all are directed toward “hung juries.” These instructions all came into existence in relation to the nationwide movement4 of courts to abandon the so-called “dynamite charge” known as the Allen charge,5 and were designed as substitutes for the Allen charge. See, e.g., Winters, supra.
As we have indicated, in requesting rehearing it appears the government may have misinterpreted our opinion as not being restricted to “anti-deadlock” instructions to “hung juries.” For instance, the government in support of rehearing cites us to decisions such as United States v. Washington, 144 U.S.App.D.C. 338, 340-41, 447 F.2d 308, 310-11 (1970), and Fulwood v. United States, 125 U.S.App.D.C. 183, 369 F.2d 960 (1966), cert. denied, 387 U.S. 934, 87 S.Ct. 2058, 18 L.Ed.2d 996 (1967). Those cases are cited to us for the proposition that a trial court may instruct a “hung jury” more than once. In those cases, however, the courts referred to a hung jury instruction given before the jury commences its deliberation. This is hardly then an instruction to a “hung jury.” Those decisions held the court may properly give the instruction again after the jury actually becomes “deadlocked.”
It is evident that those decisions have no application here. We are addressing ourselves entirely toward a practice of repeatedly giving “anti-deadlock" instructions to a “hung jury.” A jury can hardly be hung before it begins deliberations. Consequently, that line of cases does not assist us here.
When the nationwide trend toward abolition of the “dynamite charge” (the Allen charge) began, it was apparent that in the public interest there should be a substitute instruction directed at breaking the deadlock when a jury becomes “hung.” The first nationwide anti-deadlock instruction to supplant the Allen charge was the American Bar Association (ABA) suggested instruction for this purpose, to which we have referred. It seems quite plain that while it does not contain the “dynamite” of the Allen charge, it is nevertheless regarded nationally as an “anti-deadlock” instruction and was so intended by the ABA. See Winters, supra, 317 A.2d at 534-39 (concurring opinion). We say this because there seems to be a misunderstanding to the effect that since the American Bar Association instruction is more moderate than the Allen charge, for example, this means it is not an anti-deadlock instruction or, if it is, the court should be able to repeat an anti-deadlock instruction to a hung jury and, in so doing, graduate to a stronger “anti-deadlock” instruction the succeeding *1174time (or times). This would be another avenue back to the disapproved situation in this jurisdiction resulting from the previously rejected Allen charge. By disapproved situation, we mean instances where the integrity of jury verdicts were in question due to the unwise coercive elements of the Allen charge.
In 1971, the federal circuit court for this jurisdiction, among other courts throughout the country (state and federal), made clear that the ABA charge is a hung jury instruction. Speaking for the court, now Chief Judge Robinson said:
[W]e conclude that ... future rendition of Allen-type charges[6] must conform to the standard which has been approved by the American Bar Association and adopted by the Third and Seventh Circuits.
United States v. Thomas, 146 U.S.App.D.C. 101, 103, 449 F.2d 1177 (1971) (en banc) (footnote omitted).
It is elementary that a defendant is entitled to a jury in disagreement. It is in the public interest, however, that a jury reach a verdict if it can reasonably do so. Moreover, it is fundamental that a trial judge may not coerce a verdict. It is also apparent that a trial judge should give a temperate prod to a “hung jury” so as to bring out a verdict. It is our view, however, that to repeatedly give “anti-deadlock” instructions to a “hung” jury may well coerce a verdict from the jury. For this reason, we conclude that a jury believed by the trial judge to be “hung,” as we related earlier in this opinion, should ordinarily be given an “anti-deadlock” instruction only once.7 As the court stated in United States v. Seawell, 550 F.2d 1159, 1163 (9th Cir.1977):
Repetition of the charge, together with rejection of the jury’s second report of deadlock, is almost certain to convey the thought that by failing to come to an agreement — by once again reporting themselves at impasse — the jurors have acted contrary to the earlier instruction as that instruction was properly understood. (“Apparently you didn’t listen to what I said before, so I’ll repeat it.”) Given a second time, not at the request of the jury but at the instance of the judge, the charge no longer serves as an instruction; no matter how it may be softened it becomes a lecture sounding in reproof. [Emphasis added.]
See also United States v. Fossler, 597 F.2d 478 (5th Cir.1979) (reversed on ground that trial judge erred in twice giving anti-deadlock instruction to hung jury). The Ninth Circuit in Seawell went on to say:
On those rare occasions when the use of two Allen charges has been upheld by other courts the second charge was at the request of the jury, United States v. Kahaner, 317 F.2d 459, 484 (2d Cir.), cert. denied, 375 U.S. 836 [84 S.Ct. 73, 11 L.Ed.2d 65], reh. denied, 375 U.S. 926 [84 S.Ct. 263, 11 L.Ed.2d 169] (1963); White v. United States, 279 F.2d 740, 750 (4th Cir.), cert. denied, 364 U.S. 850 [81 S.Ct. 96, 5 L.Ed.2d 74] (1960); or the first charge was a part of the general jury instructions delivered before the jury had begun its deliberations and thus was not in response to a report of deadlock. United States v. Washington, 144 U.S.App.D.C. 338, [340], 447 F.2d 308, 310 (1970). Moreover, the vitality of these decisions might seriously be questioned. The first two decisions were rendered well before federal courts of appeals began to prohibit the use of the Allen charge and the third decision no longer represents the law of its circuit. ...
*1175Seawell, supra, 550 F.2d at 1163 n. 10 (emphasis added).
It must be remembered that in this jurisdiction the trial judges have been enjoined not to give the Allen charge to a “hung jury” even once, let alone repeatedly. In our en banc decision in Winters, supra, exercising its supervisory power, this court adopted an instruction in place of the Allen charge, and disapproved the future use of the latter charge. The United States Circuit Court for this jurisdiction had previously abandoned the Allen charge, United States v. Thomas, supra, 146 U.S.App.D.C. at 103, 449 F.2d at 1179.
This court in Winters, supra, specifically stated it was there setting “the high water mark for an anti-deadlock charge,” 317 A.2d at 534. What this means is, the court laid out its approval charge but left discretion with the trial judges to fashion their own anti-deadlock instructions as long as they do not exceed in pressure for a verdict the Winters charge, while excluding the Allen charge and similar “dynamite” type anti-deadlock instructions. In addition to the Winters charge, this left for approved usage the ABA anti-deadlock instruction and the instruction in the Winters concurring opinion, among others. While the dissent in this case refers to the ABA instruction as being dubiously “an anti-deadlock charge,” the dissenting opinion stands alone in this view. The ABA charge was fashioned to be, and is nationally regarded by the courts and scholars, as being an anti-deadlock instruction.8 It matters not whether it is given before deliberation or after a trial judge concludes the jury is genuinely “hung,” the instruction is addressed to a “hung jury” situation. When it is given before a jury begins deliberation it is thought by some to be potentially educational if the jury later approaches a deadlock. What the actual benefit may be to a jury when instructed in this matter prior to deliberation is a matter for speculation. It certainly can do no harm and may conceivably head off a “hung jury” situation later on.9
An instructional error which materially affects a jury verdict, is not an error to pass off lightly. If the integrity of a jury verdict is called into question to the extent that error is found, it is rarely amenable to a finding of harmlessness. We conclude a trial judge should not repeatedly give an anti-deadlock instruction to a “hung jury” after it has already been given an “anti-deadlock” instruction when the court decided it was actually a “hung” jury.10 Only if there are extenuating circumstances, e.g., if there is confusion and there is a request by the “hung jury” for a repetition of the anti-deadlock instruction, as previously related, or there is some ex*1176ceptional circumstance which makes evident it is not likely to be coercive to rein-struct the “hung jury” with another “anti-deadlock” instruction (because of the exceptional circumstance), should an “anti-deadlock” instruction be repeatedly given. Otherwise, repeatedly giving “hung jury” instructions to a “hung jury” risks affecting adversely the integrity of a verdict. This accords with the viewpoint of the courts in this jurisdiction in disapproving the Allen charge even once.
Basically, the majority view in this case is that a trial judge should most certainly try to prod a verdict from a “hung jury,” but should do so with a certain moderation to avoid going over the boundary into coercion. Barring exceptional circumstances, such as we have related, repeatedly confronting a genuinely “hung jury” with instructions which most objective observers regard as “anti-deadlock” instructions may endanger the integrity of a jury verdict.
With the qualifications along the lines we have noted, we hold that an “anti-deadlock” instruction should not repeatedly be given to a “hung jury,” as defined. From all that appears, the jury in this case was apparently found by the trial court to be hung when it was given the first “anti-deadlock” instruction (the ABA version). It was error, over objection, to give the second “anti-deadlock” instruction (the Winters version). The lawlfapd practice) in this jurisdiction does'n'ot admit of this. As this court said in its recent eh banc opinion when speaking to an.instructional issue, an error which calls “into question the integrity of the verdict” warrants reversal. Allen v. United States, 495 A.2d 1145 (D.C.1985) (en banc).
Reversed.
.Actually, at the trial level the prosecutor had an approach in this case with which we find no fault. When after one hour of deliberation, the jury notified the court it could not reach a unanimous verdict, the court asked the position of the parties. The prosecutor stated: "The government would, at this time, unless we are hopelessly deadlocked, continue the deliberations." (Emphasis added.)
After a colloquy, the court rejected the government's approach and decided to give the American Bar Association (ABA) anti-deadlock instruction (see infra). When the jury later notified the court that it was still hung, the government then suggested it be given the Winters anti-deadlock charge (see Winters v. United States, 317 A.2d 530, 534 (D.C.1974) (en banc)). This was done, over objection, and this time a verdict of guilty was returned.
. It seems almost too obvious to mention that trial judges routinely send juries back for further deliberation who prematurely send a message to the court that they are unable to reach a verdict. Juries are known to deliberate for an hour or so and send such a message. This ordinarily is not a genuinely "hung jury" at this point.
. From a distance, one would have had to have misgivings about whether the jury in this case was deadlocked since it appears to have been rather precipitous in returning to the trial judge so soon. Be that as it may, the trial judge appears to have considered it to be a “hung jury,” and as indicated, this is ordinarily a trial court determination. So, for the purposes of this case there was then a "hung jury” when the first anti-deadlock (the ABA) instruction was given.
. See Winters, supra, 317 A.2d at 534-39 (Gallagher, J., concurring).
. Allen v. United States, 164 U.S. 492, 17 S.Ct. 154, 41 L.Ed. 528 (1896).
. We assume it is not open to question that the Allen-type charge is a "hung jury” instruction.
. As we stated earlier, we are not inferring that if a jury indicates it did not understand the first "anti-deadlock" instruction it may not again be so instructed in an effort to bring understanding of the charge; nor are we inferring that a jury request for guidance should be denied. These are matters of common sense. Perhaps there will be arising other such compelling factual situations in the future.
. See, e.g., cases cited in Winters, supra, 317 A.2d at 536-38 (concurring opinion).
. The dissent points us to United States v. Taliaferro, 558 F.2d 724 (4th Cir.1977), cert. denied, 434 U.S. 1016, 98 S.Ct. 734, 54 L.Ed.2d 761 (1978); United States v. Fossler, supra, 597 F.2d at 478, and United States v. Robinson, 560 F.2d 507 (2d Cir.1977), cert. denied, 435 U.S. 905, 98 S.Ct. 1451, 55 L.Ed.2d 496 (1978). We find those authorities unpersuasive here.
In Taliaferro the record is not exemplary. After being more than generous in the number of "anti-deadlock” instructions, the trial judge then sent the jury to dinner at a private club, instructed them that they could continue deliberations during dinner and authorized alcoholic beverages with dinner. The jury announced a verdict fifteen minutes after returning to court. We find the decision affirming in Taliaferro rather unimpressive.
In Fossler, we are puzzled by the dissents’ reliance because the court held it was error to repeat the Allen charge to the jury.
As to Robinson, the jury announced by note an 11-1 split; the note to the trial judge was not revealed to counsel; and the court repeated a modified Allen charge, though it was apparent it was necessarily directed at only one juror. We do not find that decision affirming very convincing, especially when one stops to think about the different attitude in this jurisdiction. Here, the Allen charge even once is disapproved.
.As we have related earlier, a jury is not necessarily a "hung jury” simply because it says it has been unable to reach a unanimous verdict. Some juries make this representation prematurely. It is for the trial judge to decide if, and when, there is a genuine "hung jury.”