PER CURIAM:

Plaintiff-Appellant Doug Burchfield sued CSX Transportation, Inc. ("CSX") seeking damages for injuries that he sustained in an accident involving a railcar CSX delivered to his employer. After a jury trial, judgment was entered in CSX's favor. Burchfield now challenges a number of the district court's evidentiary rulings and refusal to give jury instructions that he requested. After thorough review, we reverse and remand the case for a new trial.

I.

A.

On June 3, 2005, CSX delivered a railcar identified as "AEX 7136" to General Mills' cereal processing plant in Covington, Georgia. The railcar was loaded with grain. Two days later, Plaintiff Doug Burchfield and a coworker, Rodney Turk, moved the AEX 7136 from one track to another at the plant. Turk operated a Trackmobile railcar mover.1 Burchfield worked on the ground.

After moving the AEX 7136, Burchfield and Turk began working downhill from the railcar. A short time later the AEX 7136 rolled down the track and crashed into two other railcars. All three railcars ran over Burchfield. Burchfield suffered a number of injuries in the accident, including loss of the use of his legs.

In June 2007, Burchfield filed suit against The Andersons, Inc. ("Andersons"), the owner of the AEX 7136, and CSX. Burchfield sought to recover general and special damages under theories of negligence and negligence per se. He also asserted a claim for punitive damages. Burchfield alleged that his injuries were caused by the defendants' failure to deliver the AEX 7136 to General Mills' Covington plant with an efficient hand brake, in violation of the Federal Safety Appliance Act and Federal Railroad Administration regulations. Before trial, Burchfield entered into a settlement agreement with Andersons, which was therefore dismissed from the action with prejudice.

Burchfield's case went to trial with a jury in October 2009. At trial, CSX introduced into evidence, over Burchfield's objection, a video made for General Mills by Gary Wolf, a railroad expert, for purposes of an Occupational Safety and Health Administration investigation of Burchfield's accident. The video had no sound. It purported to show the AEX 7136 railcar involved in Burchfield's accident at two different locations at General Mills' Covington plant. The first location was the site where the AEX 7136 railcar was alleged to be located before Burchfield's accident. The second location was a site with a greater slope. At both locations, the video depicted the railcar with an activated hand brake. The railcar was then separated from the Trackmobile railcar mover and stayed in place for more than twenty minutes. The video also purported to show seven other instances in which the AEX 7136's hand brake was applied and released.

In addition to the video, CSX also introduced the video deposition testimony of Bryan Nutt, an expert hired by Andersons, who performed a "brake shoe force test" on the AEX 7136 railcar's hand brake system. In response to an objection by Burchfield, the district court only allowed CSX to introduce portions of Nutt's deposition testimony that involved his factual observations during testing. The district court barred CSX from introducing Nutt's expert testimony.

As part of its defense, CSX also introduced the video deposition testimony of Charles Jobes. Jobes was the designated Fed.R.Civ.P. 30(b)(6) corporate representative for Barloworld Handling, a company hired by General Mills to train its employees how to properly operate and maintain a Trackmobile rail car mover.

After the close of the evidence, the jury returned a verdict for CSX. Burchfield now challenges a number of the district court's evidentiary rulings and its refusal to give jury instructions that he requested. Specifically, Burchfield argues on appeal that the district court erred by admitting: the video created by Gary Wolf; Bryan Nutt's video deposition testimony; and Charles Jobes' video deposition testimony. Burchfield also challenges the district court's refusal to give his requested jury instruction on negligence per se and instructions involving the exercise of reasonable care by CSX.

B.

"We review a district court's evidentiary rulings for abuse of discretion." Proctor v. Fluor Enters., 494 F.3d 1337, 1349 n. 7 (11th Cir.2007). "[W]hen employing an abuse of discretion standard, we will leave undisturbed a district court's ruling unless we find that the district court has made a clear error of judgment, or has applied the wrong legal standard." Corwin v. Walt Disney Co., 475 F.3d 1239, 1249 (11th Cir.2007) (alteration in original) (quotation marks omitted). Moreover, "we will not overturn an evidentiary ruling unless the moving party establishes a substantial prejudicial effect." Perera v. U.S. Fid. & Guar. Co., 544 F.3d 1271, 1274-75 n. 1 (11th Cir.2008). The moving party makes that showing by demonstrating that the error "probably had a substantial influence on the jury's verdict." Proctor, 494 F.3d at 1352 (quotation marks omitted).

A district court's refusal to give a requested jury instruction is also reviewed for abuse of discretion. Beckford v. Dep't of Corrs., 605 F.3d 951, 957 (11th Cir.2010). An abuse of discretion is committed only when "(1) the requested instruction correctly stated the law, (2) the instruction dealt with an issue properly before the jury, and (3) the failure to give the instruction resulted in prejudicial harm to the requesting party." Goldsmith v. Bagby Elevator Co., 513 F.3d 1261, 1287 (11th Cir.2008) (quotation marks omitted).

II.

We will first address Burchfield's argument that the district court erred in admitting a video made by Gary Wolf, a consulting expert, for General Mills.2 He argues that the video depicted a recreation of his accident. In order for the video to be admissible, Burchfield asserts that CSX was required to prove that the testing on the video was performed under substantially similar conditions as those surrounding his accident. To make that showing, Burchfield maintains that expert testimony from Gary Wolf was necessary. CSX argues in response that the video was not a recreation such that it would be subject to a heightened foundational standard. Rather, CSX contends that the video was properly admitted because it was authenticated under Fed.R.Evid. 901(a), which requires a lesser showing from a witness laying a foundation for a photograph or motion picture. See United States v. Belfast, 611 F.3d 783, 819 (11th Cir.2010); United States v. Clayton, 643 F.2d 1071, 1074 (5th Cir.1981).3

Building on this argument that the video was not offered as a recreation of the accident, CSX goes on to assert that it was unnecessary to establish substantial similarity of conditions between the testing on the video and Burchfield's accident. In making this argument, CSX relies on case law suggesting that the demonstrations of mere "general scientific principles" do not require a showing of substantial similarity. See Muth v. Ford Motor Co., 461 F.3d 557, 566 (5th Cir.2006) ("When the demonstrative evidence is offered only as an illustration of general scientific principles, not as a reenactment of disputed events, it need not pass the substantial similarity test."); McKnight ex rel. Ludwig v. Johnson Controls, Inc., 36 F.3d 1396, 1401 (8th Cir. 1994) ("[W]here the experimental tests do not purport to recreate the accident, but instead the experiments are used to demonstrate only general scientific principles, the requirement of substantially similar circumstances no longer applies."). To resolve the issue, we examine the way in which CSX used the video at trial.4

Our review of the record demonstrates that CSX used the video to discredit Burchfield's theory of the case: that Burchfield applied the hand brake on the AEX 7136 but it was inefficient5 and failed to hold the railcar in place. Although CSX contends that the video was not offered as a recreation of Burchfield's accident, CSX's statements at trial belie that assertion. CSX repeatedly emphasized the purported similarities between the events depicted on the video and the circumstances surrounding Burchfield's accident. During opening statements, CSX told the jury:

During closing arguments, CSX stated: "[T]his video utilized the same car with the same load and the same mechanical condition at two locations."

While we recognize that opening statements and closing arguments are not in themselves evidence, their purpose "is to assist the jury in analyzing the evidence." United States v. Hasner, 340 F.3d 1261, 1275 (11th Cir.2003). Here, CSX highlighted the video as proof that Burchfield did not apply the hand brake on the AEX 7136. CSX did not characterize the video's contents as establishing merely the way in which a hand brake is properly applied. See Four Corners Helicopters, Inc. v. Turbomeca, S.A., 979 F.2d 1434, 1442 (10th Cir.1992) ("[I]f the evidence is offered to merely show physical principles, the experiment should be conducted without suggesting that it simulates actual events.").

CSX also used the video during its cross examination of David Engle, one of Burchfield's expert witnesses. CSX instructed Engle:

After playing the video, CSX asked Engle whether the railcar on the video appeared to have an efficient hand brake. Engle responded: "Based on what I see there, yes."6

During its cross examination of General Mills safety manager John Henderson, CSX asked about the circumstances surrounding the video's creation. Henderson testified that he witnessed the video being made and stated that the AEX 7136 railcar depicted in the video was in the same condition "mechanically and load-wise" as on the date of Burchfield's accident. He also testified that the AEX 7136 railcar was filmed at two different locations at General Mills Covington plant: (1) where Burchfield allegedly left the railcar before the accident; and (2) a location further down the track with a steeper slope.

In addition to what was quoted from the closing argument above, CSX showed the video to the jury and summarized its contents as follows:

Thus, at various points throughout the trial, CSX expressly argued that the video recreated the incident at issue in this case, and proved that Burchfield did not properly apply the hand brake. Thus, this was not a case of "evidence ... offered to merely show physical principles." Four Corners Helicopters, Inc., 979 F.2d at 1442; see also Muth, 461 F.3d at 566; Ludwig, 36 F.3d at 1401. Rather, it is one in which "[t]he results of the experiment purported to coincide with [CSX's] theory of how the accident occurred." Barnes v. Gen. Motors Corp., 547 F.2d 275, 277-78 (5th Cir.1977); see Muth, 461 F.3d at 566 (explaining that "resemblance [to the disputed accident] ... gives rise to the requirement of substantial similarity") We therefore conclude that the substantially similar conditions standard the former Fifth Circuit announced in Barnes applies to the video's admission. See Barnes, 547 F.2d at 277-78.

Under that precedent, a district court generally has "wide discretion to admit evidence of experiments conducted under substantially similar conditions." Barnes, 547 F.2d at 277; see also Bish v. Emp'rs. Liab. Assurance Corp., 236 F.2d 62, 70 (5th Cir.1956) (explaining that evidence of an out of court experiment is admissible "provided it will enlighten and assist, rather than confuse the jury, by directly illustrating and tending to establish or to disprove a material issue, and provided the experiment was fairly and honestly made under circumstances and conditions substantially similar to those attending the alleged occurrence"). The burden is on the party offering the evidence to lay a proper foundation establishing a similarity of circumstances and conditions. Barnes, 547 F.2d at 277. For the experiment to be admissible, "it is not required that all the conditions shall be precisely reproduced, but they must be so nearly the same in substantial particulars as to afford a fair comparison in respect to the particular issue to which the test is directed." Id.; see also United States v. Gaskell, 985 F.2d 1056, 1060 (11th Cir. 1993) (explaining that the conditions of the experiment and the event at issue must be sufficiently similar to provide a fair comparison). "Further, experimental or demonstrative evidence, like any evidence offered at trial, should be excluded `if its probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury.'" Gaskell, 985 F.2d at 1060 (quoting Fed. R.Evid. 403).

We conclude that CSX failed to satisfy the Barnes substantially similar conditions test. To be sure, CSX attempted to lay a proper foundation for the admission of the video through John Henderson's testimony. That testimony, however, did not establish a similarity of conditions between the testing on the video and Burchfield's accident. Although Henderson provided specifics about the AEX 7136 used in the tests and where the tests were conducted, he provided no specifics about the actual tests themselves. For example, when asked about the way that the hand brake was applied on the video and, specifically, whether it was "torqued to higher than a normal human being can do," Henderson replied: "I don't know." Because he only witnessed the testing, Henderson's testimony was limited to his visual observations. Cf. United States v. Spoerke, 568 F.3d 1236, 1249-50 (11th Cir.2009) (finding that the district court did not abuse its discretion in admitting videos of pipe bombs exploding where the government offered testimony of the explosives expert who constructed the pipe bombs stating that the devices on the video were virtually identical to those used by the defendant).

The scope of Henderson's testimony made it impossible to assess whether the video "was fairly and honestly made under circumstances and conditions substantially similar to those attending" Burchfield's accident. Bish, 236 F.2d at 70 (quotation marks omitted). We need not decide whether, as Burchfield contends, Gary Wolf's testimony was necessary for the video to be properly admitted. Because Henderson's testimony did not lay a proper foundation for the video's admission, we conclude that the district court abused its discretion by admitting it under the specific facts present here.

We must next inquire as to whether the district court's erroneous admission of the video was prejudicial. We well recognize that no trial is ever perfect, and acknowledge that there are likely circumstances in which the erroneous admission of a video could be harmless. But this circumstance is not one of harmless error. Burchfield's theory of liability was premised entirely upon a defective hand brake. The video depicted events which the jury was repeatedly told were the same as the accident resulting in Burchfield's injuries. Thus the video spoke directly to the ultimate disputed issue in the case, even though it had not been subject to the standards required for such a reenactment. CSX used the video to argue to the jury that Burchfield's accident resulted from his own negligence in failing to apply the hand brake. CSX's repeated references to the video underscore its centrality to CSX's defense of the case, and the important role assigned to the video in rebutting Burchfield's theory of the case. By showing the AEX 7136 staying in place when the hand brake was applied, the video purported to rebut Burchfield's theory that the brake was defective. As we have recognized, "demonstrative exhibits tend to leave a particularly potent image in the jurors' minds." Gaskell, 985 F.2d at 1061 n. 2; see also United States v. Wanoskia, 800 F.2d 235, 237-38 (10th Cir. 1986) (explaining that a court "must take special care to ensure that [a] demonstration fairly depicts the events at issue" because demonstrative evidence is highly persuasive). As in Barnes, the video was "calculated to cause the jury to accept [CSX's] theory over [Burchfield's] evidence." 547 F.2d at 278. By unfairly prejudicing the jury on the pivotal issue in the case, it is likely that the video's admission had a substantial prejudicial effect, warranting reversal. See Gaskell, 985 F.2d at 1061-62 (finding reversible error where the erroneously admitted evidence went to the key issue in the case).

Having concluded that the district court committed reversible error by admitting the Wolf video, we need not address Burchfield's other points of error. The judgment of the district court is reversed and the case remanded for a new trial.

REVERSED AND REMANDED.