with whom SCHROEDER, REINHARDT, WARD LAW, and PAEZ, Circuit Judges, join, dissenting:
I agree with the majority that the jury instructions given in this case were erroneous, because they relieved the government of its burden of proving that the object in question was a “dangerous weapon” within the meaning of the federal assault statute. The given instructions purported to define “dangerous weapon,” but, critically, failed to instruct the jury that it must find, as an element of the offense — and, therefore, beyond a reasonable doubt — that Smith used a dangerous weapon. Given that error and the trial record, I cannot agree with the majority that it is clear beyond a reasonable doubt that a rational jury would have returned the same verdict had a proper instruction been given. See Chapman v. California, 386 U.S. 18, 23, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967). I therefore respectfully dissent.
We have held that an object is a “dangerous weapon” within the meaning of 18 U.S.C. § 113(a)(3) if it is “dangerous per se ” or “used in a manner likely to endanger life or inflict great bodily harm.”1 *943United States v. Riggins, 40 F.3d 1055, 1057 (9th Cir.1994) (quoting United States v. Guilbert, 692 F.2d 1340, 1343 (11th Cir.1982)). For objects that cannot “obviously” be described as dangerous, Riggins, 40 F.3d at 1057, then, a conviction under § 113(a)(3) must rest not on the object’s abstract capabilities, but instead on evidence that the accused used the object during the commission of an assault in a seriously endangering manner. We have therefore emphasized that determining whether an object was so used in a particular case is a fact-intensive inquiry reserved for the jury. See id.
The erroneous jury instructions removed this critical' — and dispositive — issue of fact from the jury. Smith’s defense at trial questioned the dangerousness of the Styrofoam “prison-made knife” by attacking its inherent capabilities, by emphasizing the relatively minor extent of the injuries it allegedly caused, and by contesting the prison guard witnesses’ version of how it was used. Unlike the majority, I cannot say that the evidence of the object’s dangerousness, either inherently or as it was used, was so overwhelming that the jury instructions’ failure to direct the jury to decide this critical issue of fact was harmless beyond a reasonable doubt.
If anything, the bulk of the testimony at trial reasonably suggested that the prison-made knife, as used during the altercation, was unlikely to cause serious bodily injury. The jury heard from two correctional officers who observed the fight in the prison yard. Both officers testified that while Jeffries was restrained, Smith struck him with a sharpened object in a downward motion with a high degree of force. One officer offered a more specific account: he testified that Smith struck Jeffries in the small of his back “really forcefully ... it looked like he was putting all of his effort into it.” But the testimony also was that the object broke under that pressure and that, despite Smith’s use of his utmost force, the injuries to Jeffries’s back required only minor first aid. Defense counsel emphasized this point at closing. Thus, it would have been reasonable for the jury to conclude that the object, even when used as violently as possible, was capable of causing only minor injuries.
True, Jeffries’s physical examination after the incident revealed a laceration on his right eyelid. There was no testimony, however, that Smith directed the prison-made knife towards Jeffries’s face, nor any testimony, from Nísperos, the medical ex*944aminer, or any one else, that Jeffries “nearly lost an eye.”2 Maj. Op. at 940-41. Nor, contrary to the majority’s account, see Maj. Op. at 938, 939, did Nísperos testify that the injuries to Jeffries’s eyelid resulted from the prison-made knife or the use of any other sharp object. In fact, the medical term Nísperos used to describe the injury, “laceration,” signifies “a torn or jagged wound,” as opposed to a clean cut, suggesting that the wound was not inflicted by a sharp object. See PDR Medical Dictionary 958 (2d ed.2000); see also http://www.healthatoz.com/he altha-toz/Atoz/common/standard/transform.jsp? requestURI =/healtha-toz/Atoz/ency/wounds.jsp (distinguishing “cuts,” “slicing wounds made with a sharp instrument,” from “lacerations,” which are “produced by a tremendous force against the body ... from an external source like a punch”). There was testimony that in addition to striking Jeffries with the sharpened object, Smith struck him with his clenched fist. The photographic depiction of Jeffries’s injury, from my lay perspective, reveals nothing more than a pool of blood above his right eye, hardly a showing sufficient to support the conclusion that a stabbing is the only plausible explanation for the injury. The jury reasonably could have concluded, instead, that the injury to the eyelid was the result of punches thrown by Smith, not of the prison-made knife.
In short, in the absence of any testimony establishing that Smith directed the prison-made knife toward Jeffries’s eye, I cannot conclude that it is clear beyond a reasonable doubt that, had they been properly instructed, every member of the jury would have found beyond a reasonable doubt that Smith used the object in a manner likely to cause great bodily harm.
The medical examiner’s testimony that the object “could cause very fatal injuries” does not convince me otherwise. Níspe-ros, a medical professional who had twice failed the American medical doctor board exams and had never before testified as a forensic expert, testified that the object could cause fatal injuries if directed at a vital organ. The jury could well have decided not to rely on Nisperos’s opinion, given his less-than-impressive medical background, the vagueness of his testimony about the dangerousness of the object, and the contradiction between his very general assertion and the record fact that the object broke when used on the victim’s back, after causing no more than minor lacerations.
Moreover, Nisperos’s testimony is only marginally pertinent to the facts of this case, because there was paltry evidence *945that Smith attacked Jeffries in any vital organ. Defense counsel stressed this point in closing argument, in part to defend against the attempted murder charge. As the definition the jury should have been given focuses on how the object was used, the jury could have viewed Nisperos’s testimony as off-to-the-side on the facts of this case.
Tellingly, the jury acquitted Smith on the attempted murder charge, suggesting either that it did not believe that the object was used in the manner the prosecutor suggested or that it did not believe that the object was, as Nísperos testified, capable of causing “very fatal injuries” (or both). Either way, the jury’s actual verdict indicates that it did not believe a great deal of the evidence the majority relies on as indisputably supporting the assault with a dangerous weapon verdict.
Finally, although Smith’s counsel, as well as the prosecutors, referred to the object as a “weapon” and a “knife,” the jury was required to find that the object is a “dangerous weapon,” not just any weapon. As the majority observes, the use of such words may connote dangerousness, but this observation only highlights the prejudicial effect of failing to give a correct jury instruction on dangerousness. Under the case law, dangerousness is a critical question for the jury, not for counsel or the court.
I cannot conclude that the failure to give the correct instruction on dangerousness was harmless beyond a reasonable doubt. I respectfully dissent.
. I am not at all sure that this standard is an appropriate interpretation of the statute. Were the issue before us, I would probably conclude that a "dangerous weapon” must be *943an object designed to injure someone through the use of force, not an object — like a shoe or a pot or a chair — that could seriously injure someone but is not meant for or likely to be used for that purpose. Cf. Medley v. Runnels, 506 F.3d 857, 863-64 (9th Cir.2007) (en banc) (recognizing that the California statute in question requires that a "firearm” be "designed to be used as a weapon”). It seems evident to me that a "weapon” describes a specific kind of object, not any object that can injure someone. Congress's purpose in providing an enhancement for use of a dangerous weapon, I would likely conclude, was to deter possession of, access to, and use of objects particularly dangerous in themselves, not the use of everyday objects that are not meant as objects of violence but can be used for that purpose. In other words, Congress wanted to deter people from having and using guns, switchblades, and brass-knuckles, but not from wearing shoes, cooking in pots, or sitting in chairs and then deciding to use them to hurt someone. The use of everyday objects to exert additional force does not seem sufficiently different from throwing a good punch to come within the statutory language or Congress's purpose.
There has been no direct challenge in this case, however, to the Riggins line of cases holding otherwise. Also, an appropriately instructed jury probably could have found that the prison-made knife falls on the "dangerous weapon” side of the line I suggest, as it was at least designed to inflict injury, whether capable of doing so or not. I therefore do not dissent on this ground.
. Unlike the majority, I do not understand Smith’s appellate counsel's statements at oral argument as concessions about the jury’s likely understanding of the record in this case, and would not rely on them even if I did so understand them. Our task in conducting the harmless error analysis is to determine whether every member of the jury hearing the evidence in this case would have concluded, beyond a reasonable doubt, that Smith used the prison-made knife in a manner likely to inflict great bodily harm. See Sullivan v. Louisiana, 508 U.S. 275, 279, 113 S.Ct. 2078, 124 L.Ed.2d 182 (1993) (”[T]he question [Chapman ] instructs the reviewing court to consider is not what effect the constitutional error might generally be expected to have upon a reasonable jury, but rather what effect it had upon the guilty verdict in the case at hand.”). For that reason, Smith’s appellate counsel's statements, made nine years after the fact, are irrelevant to our inquiry, as they reflect only her assessment of the same trial evidence before us now. Neither her characterization of the testimony, nor, for that matter, any post-trial factual concession about what actually happened during the altercation, bear on our task. The only question posed is whether this jury, in its assessment of the trial evidence, would have reached the verdict of guilty beyond a reasonable doubt had a correct instruction been given.