David M. Stagner, Sherman, TX, for Appellant. James L. Moss, Assistant County Attorney, Fannin County, Bonham, TX, Lisa C. McMinn, State's Attorney, Austin, TX, for the State.

OPINION

KEASLER, J., delivered the opinion of the Court, in which KELLER, P.J., PRICE, WOMACK, JOHNSON, HERVEY, COCHRAN, and ALCALA, JJ., joined.

Vaughn Bell contends the court of appeals erred in finding the trial judge's error in ordering him shackled during trial was harmless. We hold that the judge erred in shackling Bell, but conclude the error was not constitutional error because there is no evidence the jury saw his restraints. Despite the court of appeals' erroneous application of a constitutional-error harm analysis, we affirm its judgment.

Over a lunch break during the guilt phase of Bell's trial for the offense of possession of a controlled substance, the judge ordered Bell shackled. Accordingly, Bell was shackled with cuffs and “a chain that is linked between his two ankles.” Bell objected to the shackling, arguing that using the device in front of the jury, if seen, would deprive him a presumption of innocence, fair trial, “and his rights under United States and Texas Constitution.” The State responded by asking the judge to have the court's bailiff sit in various seats in the jury box to see if jurors would be able to see Bell's shackles. Complying, the bailiff told the judge that he could not see the chain or ankle cuffs. Bell countered that the jury would be able to hear the chains rattle if Bell moved during the trial. The judge overruled Bell's objection and stated,

The bailiff's obligation is to the jury, not to an inmate. There is a person here to handle that. Everybody who is in custody has the same necessity of restraint. The difficulty is, the sheriff's office has one discreet [restraint] and, therefore, we only have the one that's been available. We have allowed for y'all to position briefcases, and somebody went down and got an extra one so you have two. The State also has one under there. So, it's the Court's opinion that there is no impact on the presumption, and the defendant will just be mindful about movement of his legs, and we will be sure to give breaks for everybody to use the restroom and he can move his legs at that time.
We presume Bell was shackled for the duration of trial because the record does not indicate otherwise. The jury found Bell guilty, and he was sentenced to twenty years' confinement.

Consistent with his trial objection, Bell asserted on appeal that his shackling violated his right to due process and presumption of innocence. The court of appeals held that the trial judge erred in ordering Bell shackled without finding a particularized reason for such action apart from a general concern for courtroom security and the prevention of escape. Bell contended, as he does here, that the harm of this error should be analyzed under the United States Supreme Court's opinion in Deck v. Missouri, in which the Court held that, when a judge improperly orders a defendant to wear visible restraints, “[t]he State must prove ‘beyond a reasonable doubt that the [shackling] error complained of did not contribute to the verdict obtained.’ ” The court rejected Bell's argument by distinguishing Deck on the basis that “there is no evidence, in this case, that the restraints were perceived by the jury.” Nonetheless, the court of appeals reviewed the error under a constitutional-error harm analysis. After applying the Harris v. State factors (as modified by Snowden v. State )—the nature of the error, whether it was emphasized by the State, the probable implications of the error, and the weight the jury would likely have assigned to it in its deliberations—the court found the error harmless. We granted Bell's petition for discretionary review alleging the court of appeals erred in failing to consider alleged systematic shackling errors by the trial judge in evaluating harm and in placing the burden on Bell to show his shackles were perceived by the jury or otherwise infringed upon his constitutional right to be tried without restraint. However, it is the characterization of the trial judge's error—whether it is constitutional or non-constitutional error—that is central to the court of appeals' erroneous analysis.

Over one hundred twenty-five years ago, the Texas Court of Appeals—this Court's predecessor—embraced the English common-law principle that a defendant has the right to appear at trial unbound by visible shackles; however, “in extreme and exceptional cases, where the safe custody of the prisoner and the peace of the tribunal imperatively demand, the manacles may be retained.” Courts of varying jurisdictions have reached the same conclusion “that the right may be overcome in a particular instance by essential state interests such as physical security, escape prevention, or courtroom decorum.” Similarly, courts have uniformly imposed an express prohibition on routine shackling, as a defendant should only be shackled “as a last resort.”

More recently, we reaffirmed these principles and concluded that while courts permit shackling, it is only justified when, in the trial judge's discretion, it is necessary for a particular defendant in a particular proceeding. To this end, the record must manifest the trial judge's reasons for restraining a defendant. When the record fails to detail the grounds for restraint, a trial judge errs in ordering a defendant shackled. And even when exceptional circumstances or a manifest need for such restraint exists, the trial judge should make all efforts to prevent the jury from seeing the defendant in shackles.

In Deck v. Missouri, the Supreme Court found that “judicial hostility” to shackling is premised on three legal principles: (1) “the criminal process presumes that the defendant is innocent until proved guilty”; (2) “the Constitution, in order to help the accused secure a meaningful defense, provides him with a right to counsel,” and “[s]hackles can interfere with the accused's ‘ability to communicate’ with his lawyer”; and (3) “judges must seek to maintain a judicial process that is a dignified process.” Even in light of the widely held views of shackling and the principles these views espouse, the Deck Court found that due process is only implicated when the jury can see the restraints: “Thus, the Fifth and Fourteenth Amendments prohibit the use of physical restraints visible to the jury absent a trial court determination, in the exercise of its discretion, that they are justified by a state interest specific to a particular trial.” “Visible shackling undermines the presumption of innocence and the related fairness of the factfinding process. It suggests to the jury that the justice system itself sees ‘a need to separate a defendant from the community at large.’ ” The Deck Court was clear that it is not the mere shackling alone, but rather the jury's perception of the shackles, that undermines a defendant's presumption of innocence. And only when the trial judge unjustifiably orders a defendant to wear shackles that will be seen by the jury must the State prove beyond a reasonable doubt that the shackling error did not contribute to the verdict.

We too have focused on whether the jury perceived the defendant's shackles in addressing shackling errors, and Deck confirms the validity of such approach. In Long v. State, we found that the trial judge erred in shackling Long, but because the record did not show that the jury actually saw the shackles, we deemed the error harmless. A year later, in Cooks v. State, we avoided finding that the trial judge erred in shackling Cooks because there was no evidence that the jury saw Cooks's shackles, and therefore concluded that any error was harmless. Read plainly, these cases indicate that the question of whether the jury saw the shackles is a consideration relevant only to deciding harm. But these cases were decided under Texas Rule of Appellate Procedure 81(b)(2), Rule 44.2's predecessor, which set out only one harm standard irrespective of the error's nature. Nonetheless, these cases emphasize that the primary concern in shackling error is the jury's perception of the restraints and the ill effects it may have upon the defendant's presumption of innocence. More importantly, Long and Cooks demonstrate an appropriate unwillingness to make the factual assumption that the jury perceived a defendant's shackles in the absence of any record support.

Bell urges us to adopt the Fifth Circuit's approach in United States v. Banegas, in which it held that the government has the burden of proving beyond a reasonable doubt that the jury could not see the shackles when the record is unclear whether the jury in fact saw the shackles, and because the government could not satisfy its burden, the court assumed the restraints were visible. Because of our own precedent to the contrary, we decline to follow the Banegas approach.

We are mindful that even a well-developed record on appeal may fail to establish with certainty whether the jury perceived the defendant's restraints, much less describe the extent and nature of that perception. In light of this limitation, we must strike a balance between the protection of a defendant's valuable constitutional rights and the inappropriate speculation and assumption of facts on appeal when it is unclear whether a shackling error undermined a defendant's presumption of innocence. Therefore, shackling error may rise to the level of constitutional error when the record reflects a reasonable probability that the jury was aware of the defendant's shackles. We do not intend to suggest that reasonable probability in this context means more probable than not; it simply requires a substantial basis supporting a conclusion that the jury perceived the defendant's restraints.

Turning to the present case, the trial judge did not make any particularized finding articulating the reasons for shackling Bell, nor are any justifiable reasons for shackling clear from the record. The judge's statement that “[e]verybody who is in custody has the same necessity of restraint” evinces, at best, a generalized concern for courtroom security and, at worst, a propensity to shackle defendants in custody during trial as a matter of course. Neither suffices: the former is an insufficient reason; the latter a distasteful practice “[reminiscent] of an era when the accused was brought from prison to the courtroom in chains, unkempt and wearing (at best) prison attire, following which he was exposed to a jury in the worst possible light.” Under these circumstances, it was clearly error to order Bell shackled during his trial.

Whether this error is of constitutional dimension in that it deprived Bell of his presumption of innocence turns on an additional inquiry: whether the record shows a reasonable probability that the jury was aware of the defendant's shackles. After reviewing the record, we cannot conclude that there was a reasonable probability the jury perceived Bell's restraints. In fact, the bailiff's observations support the conclusion that the jury could not see Bell's restraints. It is also apparent that the judge took precautionary measures to shield Bell's shackles from view by placing briefcases in front of counsel's table. Of course this does not speak to whether the jury could hear the rattling of the restraint's chain. But in light of this record, to conclude that it is reasonably probable that the jury heard the rattling of the chain would be purely speculative. If the rattling of the chain was in fact audible, it is reasonable to expect that Bell would have brought these facts to the trial judge's attention, thereby making a record of the rattling, and perhaps, provide an opportunity to reurge his objection.

At trial, Bell never specifically asserted that the shackles would or did interfere with his ability to communicate with counsel or undermined the dignity of the judicial process. However, even if we interpreted his objection at trial and arguments below to make such claims and ignore the substantial number of cases finding that Deck does not control invisible restraints, it does not appear that these principles identified in Deck were compromised here. There is no evidence that the restraints interfered with Bell's ability to communicate with counsel or participate in his trial or was a factor in his decision not to testify. And even if we assume that an undignified judicial process is an independent component of due process equal to the presumption of innocence and the right to counsel, there is simply no evidence that Bell's shackles undermined the dignity of the judicial process. We therefore find that the judge's error in shackling Bell was not of constitutional dimension.

In finding the error harmless, the court of appeal applied the constitutional-error harm analysis, defined by Rule 44.2(a), which required the court to reverse Bell's judgment unless it determined beyond a reasonable doubt that the error did not contribute to the conviction or punishment. Although we would typically remand to the court of appeals to perform the proper harm analysis—under Rule 44.2(b), here—it makes little sense to do so when the court of appeals previously found the error harmless even under the more rigorous constitutional-error harm analysis. By finding the error harmless under the greater standard, the court of appeals logically held that Bell's shackling did not affect his substantial rights. MEYERS, J., filed a dissenting opinion.

MEYERS, J., filed a dissenting opinion.

I agree with the court of appeals that it was constitutional error for the trial judge to order Appellant to be shackled during his trial. I disagree that this error was harmless. Unlawful and uncalled for shackling has a substantial effect on the jury's view of the defendant. The fact that a defendant is shackled without cause gives the jury the perception that he is a much more dangerous criminal and may prevent him from receiving a fair trial. Because we cannot determine beyond a reasonable doubt that the error did not contribute to Appellant's conviction or punishment, I would hold that Appellant was harmed by the trial judge's error and would reverse the court of appeals. Therefore, I respectfully dissent.