(dissenting).
Unable to conclude that under the circumstances of this case, there was a “manifest necessity” for a mistrial, I dissent.
The Double Jeopardy Clause provides that no person shall “be subject for the same offence to be twice put in jeopardy of life or limb.” As interpreted by the Supreme Court, the guiding principle of this deceptively simple command “is that the State with all its resources and power should not be allowed to make repeated attempts to convict an individual for an alleged offense, thus subjecting him to embarrassment, expenses and ordeal and compelling him to live in a continuing state of anxiety and insecurity, as well as enhancing the possibility that even though innocent he *260may be found guilty.” Green v. United States, 355 U.S. 184, 187-88, 78 S.Ct. 221, 223, 2 L.Ed.2d 199 (1957). When a jury or court has acquitted a defendant of the offense charged, this principle governs without question. Arizona v. Washington, 434 U.S. 497, 503, 98 S.Ct. 824, 829, 54 L.Ed.2d 717 (1978). But when a defendant seeks to foreclose reprosecution following a first trial that ended, despite his objection, prior to final verdict, the state is not absolutely barred from another attempt at securing a conviction. Factors present in the first trial may warrant subordinating a defendant’s “valued right to have the trial concluded by a particular tribunal ... to the public interest in affording the prosecutor one full'and fair opportunity to present his evidence to an impartial jury.” Id. at 505, 98 S.Ct. at 830; Wade v. Hunter, 336 U.S. 684, 688-89, 69 S.Ct. 834, 837, 93 L.Ed.2d 974 (1949). Since 1824 the Supreme Court has categorized the various limited situations that permit such subordination under the opaque rubric, “manifest necessity.” United States v. Perez, 22 U.S. (9 Wheat.) 579, 6 L.Ed. 165 (1824); see Illinois v. Somerville, 410 U.S. 458, 461-66, 93 S.Ct. 1066, 1069-71, 35 L.Ed.2d 425 (1973). Though recognizing that a defendant presumably suffers regardless of the reasons for aborting the first trial, see United States v. Jorn, 400 U.S. 470, 483, 91 S.Ct. 547, 556, 27 L.Ed.2d 543 (1971) (plurality opinion), the Court has held that the prosecution may avail itself of a second opportunity if it can “shoulder the burden” of showing that a manifest or “high degree” of necessity justified the earlier mistrial. Arizona v. Washington, 434 U.S. at 505-506, 98 S.Ct. at 830-831.
In reviewing the trial court’s decision to declare a mistrial, over a defendant’s objection because the impartiality of one or more jurors may have been impaired, we are bound to accord substantial deference to that decision. See Arizona, 434 U.S. at 507-15 & n.33, 98 S.Ct. at 831-35 & n.33. To ascertain whether the conditions of a particular trial have created a manifest necessity for a mistrial depends not on the facile application of an established formula but on the careful weighing of the various conditions specific to that proceeding, a task for which the trial court is typically best situated. Id. at 513-14, 98 S.Ct. at 834. Thus, our role is confined to ensuring that the trial court’s decision reflects a “sound discretion,” id. at 514, 98 S.Ct. at 835, quoting United States v. Perez, 22 U.S. (9 Wheat.) at 579, 6 L.Ed. 165, sensitive to the defendant’s interest in “being able, once and for all, to conclude his confrontation with society through the verdict of a tribunal he might believe to be favorably disposed to his fate.” United States v. Jorn, 400 U.S. at 486, 91 S.Ct. at 558. Even though, viewing the same facts de novo, we might have chosen a course different from that elected by the trial court, so long as the court has adequately considered the available alternatives to a mistrial we must accept its conclusion that the mistrial was manifestly necessary. See Harris v. Young, 607 F.2d 1081, 1085 & n.4 (4th Cir. 1979), cert. denied, 444 U.S. 1025, 100 S.Ct. 688, 62 L.Ed.2d 659 (1980); United States v. Pierce, 593 F.2d 415, 417-19 (1st Cir. 1979); United States v. Sanders, 591 F.2d 1293, 1298-99 (9th Cir. 1979); United States v. McKoy, 591 F.2d 218, 222 (3d Cir. 1979); United States v. Starling, 571 F.2d 934, 941 & n.10 (5th Cir. 1978).
I do not join the majority because in my opinion the trial court failed to sufficiently assess the alternatives available under the facts of this case.
The trial court’s decision to declare a mistrial hinged on its belief that the jury would probably have learned of the advertisement and that this knowledge would likely have undermined the impartiality of at least some of the jurors. Whether or not knowledge of the advertisement would have biased the jury’s judgment, I am troubled by the court’s underlying supposition that at some time during the trial the jurors would have gained that knowledge. The advertisement was published in a periodical distributed for the television audience of a limited area in which only three of the sixteen jurors resided. Given these facts, at most the trial court might reasonably have assumed that those three jurors had *261learned of the potentially prejudicial advertisement. Even if the court had so assumed, however, it could have removed the three jurors and proceeded with a jury of thirteen.1 To have also presumed that the advertisement eventually would have come to the attention of the other thirteen jurors was, on the basis of this record, extremely fragile support for the decision to abort the trial.
Ordinarily, when a trial court is made aware of potentially prejudicial publicity, it should first poll the jury to ascertain whether the jurors have also learned of that publicity. See United States v. Perrotta, 553 F.2d 247, 249 (1st Cir. 1977); Commonwealth v. Jackson, 376 Mass. 790, 383 N.E.2d 835, 841-42 (1978). If they have not, then the appropriate course is to caution the jury to shun any publicity about the trial and to remind it of its obligation to judge the defendant solely on the evidence introduced at trial. The trial court in this case, however, assumed, without explanation, that a voir dire and cautio iary instructions would have exacerbated the perceived problem, increasing the probability that the jurors would have learned of the advertisement at some point during the trial. I think the trial court seriously underestimated its ability to frame, with the help of the parties, satisfactorily neutral questions for a voir dire and appropriate cautionary instructions. Trial courts are regularly required to compose such questions and instructions in such a fashion as to not adversely affect the jury’s judgment. I see no reason why the court could not have done so in this case.2
The majority suggests that “if the metropolitan-wide press, which had been much interested before, should get wind of [the ad], the jig would be up.” But nothing in the record or in the majority’s opinion reveals any circumstances to warrant that assumption. The jig would no more have been up in this case than it would in the numerous other criminal trials that are subject to media attention and publicity. To presume that the jury has been exposed to prejudicial publicity requires conditions such as those that surrounded the trial at issue in Sheppard v. Maxwell, 384 U.S. 333, 86 S.Ct. 1507, 16 L.Ed.2d 600 (1966). But the circumstances of this case were a far cry, even potentially, from the incredible media blitz that enveloped the Sheppard trial. In the absence of such extraordinary conditions, I think we must assume that the jury would have complied with instructions from the trial court to shun all publicity. Only if the trial court had justifiably lost “confidence in the jury’s integrity and responsibility,” United States v. Pierce, 593 F.2d at 419, might it be acceptable to dismiss cautionary instructions as a fruitless alternative. The trial court did not, however, express a loss of confidence and even if it had, there is nothing in the record to support such a finding. See id. Whether or not the jury could have remained impartial knowing of the contents of the advertisement should concern us only if there was good reason to presume that they would have acquired that knowledge. And in my view that reason did not exist.
In certain circumstances a trial court properly may assume that information known to the jury will so bias its judgment that a mistrial must be ordered. See, e. g., Arizona v. Washington, 434 U.S. 497, 98 S.Ct. 824, 54 L.Ed.2d 717; Grooms v. Wainwright, 610 F.2d 344, 346 (5th Cir.), cert. *262denied, 445 U.S. 953, 100 S.Ct. 1605, 63 L.Ed.2d 789 (1980). But the propriety of such an order rests on the court’s knowledge that the jury has in fact been exposed to the potentially prejudicial information. Without such knowledge, a trial court reasonably cannot, on the basis of the mere possibility of jury exposure to potentially prejudicial material, override a defendant’s right to the completion of a trial to which jeopardy has attached, particularly where, as here, the defendant has strongly opposed a mistrial and actively sought a voir dire.
Only one factor might distinguish this case from the numerous trials that are subject to publicity: the potentially prejudicial advertisement was the work of a blatantly partisan group whose publicity efforts, the trial court believed, had ended when the court obtained Reinstein’s commitment. Having proceeded to trial with an unsequestered jury in the expectation that no further publicity would flow from the defendant’s camp, the trial court expressed a sense of betrayal when the advertisement appeared. Although the court expressly did not hold Reinstein personally responsible, it did state that the commitment had been breached. But what little evidence there is suggests that Reinstein did what he had promised: He refrained from public comment and attempted to prevent others from disseminating publicity on his behalf.3
The majority asserts that Reinstein’s failure to renew his request to the publisher of TV FACTS to stop publication of the advertisement evidences a less than adequate effort to fulfill his commitment. But it can just as fairly be inferred that Reinstein reasonably believed that the publisher’s assent to his initial request obviated any further action. And it can also reasonably be inferred that Reinstein felt confident about the publisher’s compliance with his request precisely because Reinstein’s mother was part owner of TV FACTS. For all we know from the record, his mother, in deference to her son’s commitment, opposed the Committee’s campaign.
Other than the rebuttable inferences adopted by the majority, there is no evidence to justify saddling Reinstein with the assumed consequences of the Committee’s action. The Commonwealth has failed to produce any evidence in support of the “findings” it now claims justified the declaration of a mistrial. Most importantly, it never attempted to rebut the publisher’s affidavit, which, because unrebutted, I think this Court is bound to accept.4 With*263out evidence sufficient to conclude that Reinstein breached his commitment, the proReinstein character of the ad and its source should play no role in determining whether the mistrial was manifestly necessary.
The trial proceeded with an unsequestered jury on the basis of Reinstein’s commitment, not on an agreement that even if Reinstein did what he could to silence his supporters, publication of another advertisement would free the trial court to declare a mistrial, regardless of Reinstein’s objection. So long as Reinstein was blameless, which I think must be accepted on this record, the trial court remained obligated to adequately explore all the alternatives to a mistrial. Perhaps the court erred in agreeing to go forward with the jury unsequestered. But if this was an error, which I doubt, it was the court’s, not Reinstein’s error. And it is the Commonwealth, not Reinstein, which under the Double Jeopardy Clause must accept the consequences of the court’s error.
Given Reinstein’s earlier participation in the Committee’s publicity campaign, the majority’s suspicions about the adequacy of Reinstein’s efforts to stop the publicity are understandable. Unsubstantiated suspicions are not, however, the equivalent of a manifest necessity. The Commonwealth had the opportunity to substantiate those suspicions and it has come forth with nothing. From Reinstein, however, we have not only the publisher’s affidavit but also Rein-stein’s representation that he fulfilled his commitment. And this should not be lightly disregarded. Reinstein has been under indictment since 1976, has had to prepare twice for trial and has already started two trials. This is a formidable burden which I think lends credence to his assertion that he sincerely attempted to fulfill his commitment so that at last he could obtain a final verdict. This is not an irrefutable conclusion, but it is as plausible as the negative conclusion accepted by the majority.
From the sparse record available in this case the majority consistently relies upon those inferences least favorable to Rein-stein, disregarding the equally plausible inferences favorable to him. In doing so, I think the court fails to place the burden where it belongs. It is the Commonwealth that must “shoulder the burden” of justifying a mistrial declared over the defendant’s objection. Arizona v. Washington, 434 U.S. at 505-06, 98 S.Ct. at 830-31. In requiring the prosecution to prove that a high degree or manifest necessity warranted the mistrial, the Supreme Court has struck what it considers the appropriate balance between the public interest in criminal enforcement and the defendant’s “valued right” to see the trial through to a final verdict. With one exception,5 however, the majority here chooses to rely on those inferences and assumptions most favorable to the Commonwealth. The apparent reason for this choice is the court’s perceived obligation to “be particularly slow” in disagreeing with the state court’s findings of fact. But a federal court is obligated to defer only to those findings of a state court which are “fairly supported” by the record. 28 U.S.C. *264§ 2254(d)(8); see Sumner v. Mata, 449 U.S. 539, 101 S.Ct. 764, 66 L.Ed.2d 722 (1981). To defer to findings for which there is not fair support, as I conclude in the case here, effectively shifts the burden from the prosecution to the defendant. The ultimate, and unjustifiable, result of this shift is to depreciate the defendant’s constitutionally protected right and to retreat beyond the proper boundaries of our habeas corpus jurisdiction. Because I cannot concur in such a result, mindful that ground once lost is regained only with great difficulty, I respectfully dissent.
. Unlike the majority, I do not consider this possibility so subtle that the court should not be expected to have entertained it, regardless of the parties’ silence.
. In United States v. Perrotta, 553 F.2d 247 (1st Cir. 1977), and Commonwealth v. Jackson, 376 Mass. 790, 383 N.E.2d 835 (1978), this court and the Supreme Judicial Court, in the exercise of their respective supervisory powers, established guidelines for trial courts confronted with the problem of potentially prejudicial pubIicity. As the Commonwealth correctly observes, those guidelines were cast in light of the constitutional guarantee of an impartial jury, without reference to the Double Jeopardy Clause. Nevertheless, in the absence of special conditions militating against compliance with those guidelines, of which there were none in this case, I consider the guidelines to define “sound discretion,” Arizona v. Washington, 434 U.S. 497, 514, 98 S.Ct. 824, 835, 54 L.Ed.2d 717 (1978), in this context.
. The trial court’s comments, at trial and in its subsequent memorandum, suggest to me that the court believed that with the commitment it had secured the silence of the defendant’s camp. Thus, the breach of that silence was, from the court’s perspective, a breach of the commitment. But Reinstein’s commitment was to try in whatever way possible to restrain the others. He did not promise to be absolutely effective: there was no guarantee. Moreover, it is far from clear that the trial court found that Reinstein had breached the commitment. Although the court darkly suggested in the memorandum that Reinstein was “vitally affected” by the publication, it implicitly left the question open, stating, “If the defendant had in fact complied with his undertaking, either his family, relatives and friends had chosen to ignore his request or they were essentially uncontrollable.” This is not a finding that Reinstein had breached the commitment. Unable to decisively conclude that Reinstein had broken his promise, the trial court essentially penalized Reinstein for his past association with the Committee, an avenue which I think was foreclosed by the court’s decision to proceed with an unsequestered jury. See p. 259 infra.
. Because I differ with the majority’s vaguely deprecatory evaluation of the publisher’s affidavit, I think it useful to set out the pertinent parts of that document:
3. On several occasions in the months of October and November, 1979, an advertisement concerning the prosecution of William Reinstein was published in TV Facts on behalf of the Committee for Human Rights.
4. Several days before November 21, 1979, William Reinstein asked me not to publish any further advertisements regarding his trial.
5. On November 21, 1979, the TV Facts for the week of December 2, 1979, was submitted to the printer in Framingham, Massachusetts. At that time, an advertisement concerning the prosecution of William Rein-stein was inadvertently placed in the material to be printed contrary to the request of William Reinstein.
6. The mistake made in publishing the advertisement was due to a combination of factors at the publisher’s office. Normally, the *263publisher has a two-day deadline, Wednesday and Thursday of each week, to assemble the material to be published and to send it to the printer. During the week that the mistake was made, the Thanksgiving holiday reduced the deadline time period to one day, the Wednesday before Thanksgiving. In addition, the publisher had recently changed from using an out-of-state printer to using a local printer in Framingham and certain confusion developed in the office as a result of this change. This confusion and the one-day deadline lead to our mistake in placing the advertisement in the material sent to the printer.
7. On November 28, TV Facts containing the advertisement was distributed to the communities of Revere, Chelsea, Winthrop, East Boston, Lynn, Nahant and Saugus.
8. On November 29, I informed William Reinstein of the advertisement published by mistake.
9. Upon Mr. Reinstein’s request, I went to the various distribution outlets and retrieved approximately 600 copies of TV Facts for the week of December 2, 1979.
Signed this 25th day of January, 1980, under the pains and penalties of perjury.
/s/Donald J. Rolinson
. The majority properly assumes that a voir dire “would have disclosed that no juror had learned of the TV publication.”