The defendant appeals his conviction and sentence for robbery, burglary, kidnapping, and murder. He argues, among other issues, that the trial court erred in allowing handwriting experts to bolster their testimony by describing the peer review process. We agree and reverse.
Late one evening in 1999, two law enforcement officers, working as security guards, saw two nervous young men outside a mall. One of the officers approached them. One of the individuals, an employee of a restaurant inside the mall food court, said that he was concerned because one of his friends, a restaurant supervisor, had not yet come out of the restaurant.
Both officers went to the closed restaurant. There they found the supervisor, duct taped to a folding chair and shot in the head. In the area, the officers found a file cabinet with its hasp open and a metal cash box located on top, a locked safe, and a desk. The daily receipt records and a yellow notepad were located on top of the desk. The notepad contained the following writing: “55, 65, 9, 10,” “4 time stop,” “left right left right.”
The officers also found a wad of duct tape in a trash basket. The cash earned from the day was still in the safe, but $625 was missing from the cash box. A black nylon skull cap was recovered from a dumpster outside the restaurant’s back door.
A partial print, sufficient for identification, was obtained from the duct tape found on the victim. No identification was *582made despite numerous comparisons, including the defendant’s. The latent fingerprints obtained from various surfaces at the crime scene matched those of other employees, but not the defendant. The only latent fingerprint on the yellow notepad belonged to the restaurant manager.
In 2000, the State charged the defendant with committing the offenses, but later nolle prossed the case due to insufficient evidence. The State filed new charges against the defendant in 2006; specifically, first degree murder with a firearm, robbery with a firearm, burglary with an assault or battery with a firearm, and armed kidnapping. In 2009, the case went to trial and resulted in a hung jury. The case was retried later that same year.
The State sought to establish motive by presenting the testimony of several restaurant employees. Their testimony revealed that the defendant had worked at the restaurant and had been promoted to crew leader. The victim was a supervisor. In early 1999, the victim reported the defendant for stealing money from the restaurant. The defendant denied the allegation, but the manager demoted him to crew member and reduced his hours. The defendant’s work performance deteriorated, and the manager terminated him two months prior to the murder.
One employee testified that the defendant was upset about his demotion, angry with the victim, and wanted to fight him. Another testified that the defendant referred to the victim as a “squealer” or “rat,” but he had not seemed angry or upset.
A third employee testified that although the defendant was upset and wanted to fight the victim, the two simply went their separate ways. The defendant had once talked about robbing the restaurant. And, although the defendant had a hand gun with him on one occasion, he did not display it when he argued with the manager. Between the time of his termination and the murder, the defendant returned to the food court a few times. During one visit, the defendant formed his hand into the shape of a gun and pointed it at the victim.
Another employee confirmed that the defendant was upset by his demotion, and had commented about robbing the restaurant. This witness did not take the defendant seriously. However, he had overheard the defendant arguing with the manager when he was terminated.
Yet another employee testified that she saw the defendant return to the restaurant three times after his termination. On one visit, the defendant “trash talked” a few feet from the counter. On another visit, the defendant told the victim to “watch his back” and threatened to “kick his ass.” On a third visit, the defendant walked up to the .counter and told the victim that he was going to kill him.
Using the disgruntled employee motive, the State used the following evidence to connect the defendant with the crimes. In 1999, the detective originally assigned to the case obtained writing samples from the defendant and five other employees. He had each individual write what was written on the yellow notepad fifteen times.
In 2000, a second detective took over the investigation and obtained additional handwriting samples from the defendant. In 2003, he repeated what was done in 1999 and obtained four more samples from the defendant. He obtained samples of business writings from the other persons. Ultimately, most of the sample writings were from the defendant.
In 2003, the second detective retained a criminal investigator with the United States Postal Inspection Service to obtain yet more handwriting samples from the defendant. In 2006, the second detective *583and the State retained a forensic document examiner, who was provided with the writing on the notepad and handwriting samples of twelve persons.
In 2004, the second detective retained a second forensic document examiner with the United States Postal Inspection Service to examine the writings. The second examiner was provided with the same materials as the first. In total, both examiners received seventy-five pages of handwriting samples from the defendant.
At trial, the first examiner explained that handwriting identification is based on the premise that everyone has an individual set of writing characteristics. With enough of a particular person’s writings, the individual characteristics can be found. Handwriting analysis was similar to fingerprint analysis. However, he explained that the notepad contained “a rather small quantity of writing” and a small number of characteristics, which limited the analysis that could be done.
When the State asked the first examiner to explain the protocol employed in issuing a final report from his laboratory, defense counsel objected on the ground of “bolstering.” During a sidebar conference, defense counsel argued that “it’s impermissible for him to talk about his peer review in his office that as a protocol, someone else reviews the work that he has done to make sure that it’s succinct.... He can certainly testify to how he came to his conclusion, but to talk about somebody else in his office, who’s not going to be here to testify, reviewing that work as protocol is certainly bolstering his testimony.” Defense counsel cited our opinion in Bunche v. State, 5 So.3d 38 (Fla. 4th DCA 2009). The trial court overruled the objection and allowed the examiner to give the following testimony:
Q. Okay. Does your conclusion have to be peer reviewed by another analyst within the context of your laboratory?
A. It does.
Q. And is that part of your protocols and procedures?
A. Yes.
After explaining more of the protocols, defining some industry-specific terms, and how a document analysis is conducted, the first examiner testified more specifically about his review of the documents presented to him. He determined to a “high degree of probability” that the defendant executed the text on the yellow notepad. He explained that in the world of document examiners, “high degree of probability” meant it was a virtual certainty that the defendant had written the text. The limiting factor was the small number of characteristics that were in the questioned document. When the State asked if his conclusion was contained in the report issued by the lab, he testified:
Q. Now, let me ask you this question: Was that the opinion within the purview of your laboratory protocols?
A. The ultimate conclusion that I just gave, the high degree of probability, was the conclusion that exited the laboratory due to the peer review process that we follow.
Defense Counsel: Objection, bolstering.
Court: Overruled.
The first examiner then went on to clarify that it was a “positive” identification, which was a higher level of certainty than that contained in the final report. It was only through the peer review process that the conclusion went down a slight notch to “high degree of probability.”
The State also presented the testimony of the second examiner, who testified that the volume of submitted writing had no bearing on the results. He testified that the failure to have both requested writings *584and business writings for comparison does not preclude an examiner from rendering a valid opinion. In his expert opinion, the defendant wrote the words and numbers on the yellow notepad to the exclusion of everyone else. He made a full identification, which was the highest conclusion in the field of forensic documents examination. When the State asked him about the protocols in his field, he testified:
Q. When you formulate that opinion within the context of the laboratory, for whom you worked at the time that you did this work, are there quality assurance controls in place within the context of that laboratory?
A. Yes, we have—
Defendant: Objection, Judge, bolstering.
Court: Okay, I’m sorry. The objection is bolstering?
Defendant: Bolstering, I’m sorry.
Court: Overruled.
Q. And could you explain to the members of the jury what those quality assurance controls are?
A. We have two different types of quality assurance processes, one is a technical review and one is a confirmation. Technical review, every tenth case normally that is examined is reexamined by another examiner to ensure that the protocols have been followed, the standards have been applied, and that the notes and findings or that the notes, diagrams or whatever actually support the conclusion that is being offered. Additionally, in our laboratory, whenever there was an identification made, a separate examiner is required to independently examine the evidence associated with that case and reach his or her own conclusion. The identification would only go out of the laboratory based on our own protocols if there’s confirmation.
Q. And did you have that in this case?
A. Yes, sir, I did.
The morning after the document examiners testified, a juror reported that they had looked up the word “bolstering.” The juror was individually questioned and said they were discussing the meaning of bolstering and looked it up on a cell phone. All but three jurors were present. The juror said the defense attorneys had used the word two to three times. The question came up when they were going to write a note asking the court for the meaning, but a juror looked it up instead. Each juror was questioned and none said it would affect their jury service. The defendant moved for a mistrial; the court denied the motion.
The remaining evidence consisted of testimony that a ski mask found outside the restaurant had a mixture of DNA on the inside of it. According to the State’s DNA experts, the defendant could not be excluded as a minor contributor of the DNA. The major contributor of the DNA was another individual. One expert testified that it was 180 million times more likely that the DNA stain came from the defendant and Otto Wright — an individual who gave a confession and was convicted of the crimes — than Otto Wright and some unknown person.
It was the defense, however, that presented the taped confession of Otto Wright. Mr. Wright admitted to being present on the night of the crime, but indicated that he had been the lookout. He testified that two other individuals other than the defendant had committed the crime. He did not know the defendant. On cross-examination, the State pointed out that, at his 2003 trial, Mr. Wright had *585testified that his confession was coerced and was false.
During closing argument, defense counsel questioned the handwriting identification. He argued that the identifications were tainted by the tactics of the second detective, who submitted excessive amounts of samples of the defendant’s handwriting, but only limited samples from other individuals — many of which were business writings. The defendant highlighted the limited nature of the writing at issue and the limited nature of the identifying characteristics.
The defendant was found guilty as charged in the indictment and sentenced to four concurrent life terms. The defendant, who was seventeen years old at time of the offense, filed a motion to correct sentencing error, which the trial court denied. The defendant filed a timely notice of appeal.
We review trial court decisions on the admissibility of evidence for an abuse of discretion. Dessaure v. State, 891 So.2d 455, 466 (Fla.2004).
The defendant argues the trial court abused its discretion by allowing the State “to bolster the testimony of its two document examiners with testimony that other non-testifying examiners verified the identification.” The State responds that the defendant waived the issue by failing to file a motion in limine, and by failing to raise the “bolstering” objection during the first trial.1 The State further responds that the error was invited. The State then suggests there was no improper bolstering because the experts only discussed peer review as a means of explaining the process through which handwriting is analyzed at the lab. Lastly, the State argues that any error was harmless.
We agree with the defense that the testimony constituted improper bolstering and was not harmless. We therefore reverse.
In Schwarz v. State, 695 So.2d 452 (Fla. 4th DCA 1997), we held that an expert may not testify that he formed his opinion by consulting with others in the same field. Id. at 455; see also Telfort v. State, 978 So.2d 225, 227 (Fla. 4th DCA 2008) (holding that the trial court committed reversible error by allowing a fingerprint expert to testify that his identification of the defendant’s fingerprint was verified by two other examiners). Subsequently, we extended the reach of that concept to preclude an expert from bolstering “his own opinion on direct examination with that of another expert.” Bunche, 5 So.3d at 40. We explained that this prohibitive rule applies equally whether the expert relies on a second expert’s opinion or testifies about other experts’ opinions in explaining the process employed. Id. at 40-41.
Here, the defendant specifically lodged a bolstering objection to both examiners testifying about the process and called the court’s attention to our decision in Bunche. Nevertheless, the court overruled the objections and permitted the testimony. In doing so, the trial court erred. We must next decide if the error was harmless.
In State v. DiGuilio, 491 So.2d 1129 (Fla.1986), our supreme court explained that:
The harmless error test ... places the burden on the state ... to prove beyond a reasonable doubt that the error complained of did not contribute to the ver-*586diet or ... that there is no reasonable possibility that the error contributed to the conviction.
Id. at 1135.
Here, as in Bunche, the handwriting examiners testified that they “use[d] ... a second examiner in the verification process.” Bunche, 5 So.3d at 40. The first examiner testified that, “in our laboratory, whenever there was an identification made, a separate examiner is required to independently examine the evidence associated with that case and reach his or her own conclusion. The identification would only go out of the laboratory based on our protocols if there’s confirmation.” (Emphasis added). The second examiner testified that, “the high degree of probability ... was the conclusion that exited the laboratory due to the peer review process that we follow.” (Emphasis added). Both of these comments improperly bolstered the testifying experts’ findings because they relied on the “use of a second examiner in the verification process.” Id.
The State argues that no improper bolstering occurred because the experts “were simply providing a general explanation of the [peer review] process.” While that may be true, it does not eliminate the harm of admitting the opinions of non-testifying experts to bolster the testimony of those testifying. Instead, it deprives the opposing party of the opportunity to cross-examine the non-testifying experts. This was the reasoning in Bunche, notwithstanding that we found the error harmless there.
The State attempts to distinguish Telfort and Bunche because it presented two independent experts instead of one. The State argues the expert examiners’ testimony was properly admitted because their conclusions “were consistent with each other.” Two wrongs do not make a right. Two instances of inadmissible bolstering testimony do not ameliorate the harm and prejudice simply because the experts come to the same conclusion. Rather, two instances double the harm.
The State also suggests that Telfort, Bunche, and Potts v. State, 57 So.3d 292 (Fla. 4th DCA 2011), conflict with cases from the Third and Fifth Districts. See J.V. v. Dep’t of Children & Family Servs., 967 So.2d 354 (Fla. 3d DCA 2007); Masters v. State, 958 So.2d 973 (Fla. 5th DCA 2007). We disagree. J.V. and Masters involved expert witnesses rendering an opinion after reviewing “facts and data.” Such testimony is entirely permissible under the Florida Evidence Code. There simply is no conflict.
Because the defendant contested the experts’ conclusions that he wrote the words and numbers on the yellow notepad, and those opinions are the only real direct evidence placing the defendant at the scene of the crime, we cannot find the error harmless.
Next, the defendant argues that his life sentence for first-degree murder is unconstitutional because he was a minor at the time the crime was committed. The State responds that the trial court properly imposed four concurrent life sentences. Because we reverse the conviction and sentence, this issue is moot. However, the United States Supreme Court has recently held that a mandatory life sentence without the possibility of parole “for those under the age of 18 at the time of their crimes violates the Eighth Amendment’s prohibition on ‘cruel and unusual punishments.’ ” Miller v. Alabama, — U.S. —, —, 132 S.Ct. 2455, 2460, 183 L.Ed.2d 407 (2012).
Florida imposes a mandatory sentence of life in prison without the possibility of parole if convicted of first degree murder *587when the death penalty is not imposed. § 775.082(1), Fla. Stat. (2005). Upon retrial of the case, should the defendant be found guilty, the trial court should be mindful of Miller.
Reversed and Remanded for a New Trial.
DAMOORGIAN, J„ and TUTER, JACK, B., JR., Associate Judge, concur.. Because Bunche was decided on February 18, 2009, it was not in existence at the time of the first trial.