SUBSTITUTE MAJORITY OPINION
Tracy Christopher, JusticeWe withdraw our opinion dated May 28, 2015, and we issue this substitute opinion in its place. We deny the State’s motion for rehearing.
This is an appeal from a Class A misdemeanor conviction for driving while intoxicated. At the trial court level, the State argued to the jury that a finding of intoxication under the per se theory of intoxication could be based on the alcohol content of appellant’s blood plasma rather than his whole blood. Now the State confesses error on that point.
At the trial court level, the trial judge failed to submit a question to the jury as to whether appellant’s blood alcohol level was *691at least 0.15 to support the Class A misdemeanor conviction. Now the State confesses error on that point.
At the trial court level, the State con-idnced thé trial judge that a Class A misdemeanor conviction could be based on blood plasma results without regard for the alcohol concentration of appellant’s whole blood, and, based only on the unconverted blood plasma results, the trial judge made a finding that appellant had an alcohol concentration of at least 0.15. Now the State confesses error on that point.
Despite misleading both the trial judge and the jury, through both an improper charge and an improper closing argument, the State argues that appellant is not entitled to a new trial. The State believes that the judgment should be reformed to reflect a conviction for a Class B misdemeanor, and the case should be remanded for a new punishment hearing only. We disagree. We reverse and render a judgment' of acquittal on the Class A misdemeanor, and we remand for a new trial on the Class B misdemeanor.
BACKGROUND
Appellant and his fiancée were involved in a single-vehicle accident, the cause of which was disputed at trial. Only one eyewitness testified at trial. The eyewitness testified that he was driving down the interstate late at night when a truck in front of him suddenly veered to the left and struck a concrete barrier. Traffic had been light at the time, it had not been raining, and there was no evidence of an obstruction in the road that would have required an evasive maneuver.
The truck rolled over three times and landed upright on its tires. When the eyewitness pulled up next to the truck, he saw appellant climbing out of the driver’s side window, apparently because the door had been jammed. Appellant crawled on the pavement towards his fiancée, who had been ejected from the truck during the rollover. She was unconscious, severely injured, and lying about thirty-five feet away from where the truck had finally stopped.
Appéllant told first responders that he was driving at the time of the accident. He explained that.he and his fiancée were arguing inside the truck, that she grabbed the steering wheel at one point, and that he overcorrected. This story gradually changed over time. Appellant later claimed that his fiancée was feeling sick, that she slid over to be next to him, and that she accidentally hit the steering wheel. At another point, appellant reported that the accident happened because a tire blew out.
The police did not initially suspect that the accident was alcohol-related. In fact, the officer who questioned appellant at the scene opined in his original police report that appellant was not intoxicated.
There was evidence of intoxication, however. According to an emergency medical technician, appellant admitted that he had consumed at least four beers on the night of the accident. There was also blood evidence taken more than an hour after the accident, and the evidence indicated that appellant had alcohol in his system.
The blood evidence was obtained at the hospital where appellant and his fiancée were treated. The hospital collected a vial of appellant’s whole blood, which was then placed into a centrifuge. The vial was spun, causing the blood cells to separate from the blood plasma. A test of the blood plasma revealed that appellant had a blood-alcohol concentration (“BAC”) of *6920.158.1
Charges were eventually brought against appellant for driving while intoxicated. The charging instrument included an additional allegation that, at or near the time of the ■ commission of the offense, appellant’s blood showed “an alcohol concentration level of at least 0.15.”
The State’s expert, William Arnold of the Houston Police Department, testified that the concentration of alcohol is higher in blood plasma than it is in whole blood. The expert opined that a BAC of 0.158 in blood plasma could be converted to a BAC of 0.132 in whole blood. Assuming that appellant had been eliminating alcohol from his system, instead of absorbing it, the expert believed that appellant had a BAC of 0.133 in whole blood at the time of the accident.
Appellant did not testily at trial, but he asserted several defensive theories, one of which was that his fiancée had been driving when the accident occurred. Appellant argued that his statements to first responders had been false, and that he had only taken the blame for the accident because he felt a duty to protect his fiancée.
There was affirmative evidence to support this theory. The record showed that appellant was excluded as a contributor of a DNA sample collected from the driver’s side air bag. The DNA revealed a partial profile belonging to a female of unknown origin, who could have been the fiancée.
Other evidence included testimony from the fiancée herself, who recovered from her injuries after being in a coma, for nearly a month. The fiancée admitted that she did not remember much from the accident, but she testified that there was “a good chance” that she was the driver.
The fiancée explained that the truck belonged to her, not appellant. She was possessive over the truck, and only allowed appellant to drive it on rare occasions. She also observed photographs of the truck after the accident, and said that the position of the driver’s seat would have been uncomfortable for appellant, who is more than a foot taller than she is and has a much larger frame.
The trial court instructed the jury that it could make a finding of intoxication in either of two ways: (1) if appellant did not have the normal use of his mental or physical faculties by reason of the introduction of alcohol into his body, or (2) if appellant had an alcohol concentration of 0.08 or more. “Alcohol concentration” was defined in the court’s charge as “the number of grams of alcohol per 100 milliliters of blood.” Appellant objected to this definition because it did not specify that the blood must be whole blood. Appellant requested that the definition be revised to read as follows: “the number of grams of alcohol per 100 milliliters of whole blood.” The trial court denied the request.
During closing arguments, the issue over blood evidence arose again when the prosecutor advised the jury that it was not limited by the type of blood that it could consider. The prosecutor said:
As you know, we have heard a lot of evidence today and we heard a lot about blood evidence today. I just want to draw y’all’s attention to the charge, that you will.be given the definition of what an alcohol concentration is and that definition is the number of grams of alcohol per 100 milliliters of blood. You won’t find anywhere in here in the law where *693it needs to be whole blood or plasma blood. This is the law /all follow.
Appellant objected to this argument as a misstatement of law. He explained, “The law requires whole blood.” But the trial court overruled the objection. The prosecutor later expanded on her argument, saying directly and without objection that the jury could find that appellant was intoxicated based solely on the testing of his blood plasma.
The jury convicted appellant of driving while intoxicated, but it did not make the additional finding that he had “an alcohol concentration level of at least 0.15,” as had been alleged in the charging instrument. The trial court did not submit that issue in the jury charge.
Appellant elected to have the trial court assess his punishment. No new evidence was offered during this phase. The trial court began the hearing by treating the additional allegation in the charging instrument as an enhancement paragraph, which the court found to be true. By making an affirmative finding that appellant had “an alcohol concentration level of at least 0.15,” the court applied a range of punishment applicable to a Class A misdemeanor, instead of a Class B misdemeanor, which is what appellant would have faced without the perceived “enhancement.”
SUFFICIENCY OF THE EVIDENCE
I. The Jury’s Finding
Appellant was charged under section 49.04(a) of the Texas Penal Code, which provides: “A person commits an offense if the person is intoxicated while operating a motor vehicle in a public place.” In his first issue, appellant contends that the evidence is legally insufficient to show either (1) that he was the driver of the truck, or (2) that he was intoxicated.
When reviewing the legal sufficiency of the evidence, we examine all of the evidence in the light most favorable to the verdict and determine whether a rational trier of fact could have found the essential elements of the offense beyond a reasonable doubt. See Temple v. State, 390 S.W.3d 341, 360 (Tex.Crim.App.2013). The evidence is insufficient when the record contains no evidence, or merely a “modicum” of evidence, probative of an element of the offense. See Garcia v. State, 367 S.W.3d 683, 687 (Tex.Crim.App.2012).
Although we consider everything presented at trial, we do not reevaluate the weight and credibility of the evidence or substitute our judgment for that of the fact finder. See Williams v. State, 235 S.W.3d 742, 750 (Tex.Crim.App.2007). Because the jury is the sole judge of the credibility of witnesses and of the weight given to their testimony, any conflicts or inconsistencies in the evidence are resolved in favor of the verdict. See Wesbrook v. State, 29 S.W.3d 103, 111 (Tex.Crim.App.2000). Our review includes both properly and improperly admitted evidence. See Clayton v. State, 235 S.W.3d 772, 778 (Tex.Crim.App.2007). We also consider both direct and circumstantial evidence, as well as any reasonable inferences that may be drawn from the evidence. Id. Circumstantial evidence is as probative as direct evidence in establishing the guilt of an actor, and circumstantial evidence alone can be sufficient to establish guilt. See Hooper v. State, 214 S.W.3d 9, 13 (Tex.Crim.App.2007).
A. Driver
Appellant acknowledges that he initially admitted to first responders that he was driving the truck at the time of the accident. He suggests, however, that his admissions must be disregarded because he was actually lying to protect his fiancée.
*694We do not agree that the admissions must be disregarded. The jury was free to believe appellant’s original statements, and under the applicable standard of review, those statements must be credited because they support the jury’s verdict. See Guess v. State, 419 S.W.3d 361, 366 (Tex.App.-Tyler 2010, pet. ref d) (holding that the jury was entitled to believe the defendant’s initial statement to police that he was the one who drove off the road).
Appellant’s admissions are also corroborated by other evidence.- For instance, blood was found on the driver’s side' headrest, but unlike the sample that was collected from the driver’s side air bag, DNA testing revealed that a male of unknown origin had contributed the blood. Because appellant was the only male in the vehicle, the jury could have reasonably believed that appellant was the contributor, and that the blood transferred to the headrest because appellant had been sitting in the driver’s seat.
Additionally, an eyewitness testified that he saw appellant climbing out of the driver’s side window immediately after the accident. For the same reason as before, the jury could have inferred that appellant was exiting from that side because he had previously been driving. Cf Dickson v. State, 642 S.W.2d 185, 189 (Tex.App.Houston [14th Disk] 1982, pet. ref'd) (evidence supported a finding that the defendant had been driving a vehicle when the defendant was seen emerging from the driver’s side immediately after the vehicle was stopped).
We do not doubt that the jury could have made the opposite finding that appellant was merely a passenger at the time of the accident. Appellant certainly produced evidence that created a fact issue on this point. But in such circumstances, it is not the role of this court to determine which evidence the jury should have believed. See Wicker v. State, 667 S.W.2d 137, 143 (Tex.Crim.App.1984). Viewing the evidence in the light most favorable to the jury’s verdict, we conclude that a rational finder of fact could have determined beyond a reasonable doubt that appellant was driving the truck at the time of the accident.
B. Intoxicated
The jury received the statutory definition of “intoxicated,” which has two alternative meanings: “(A) not having the normal use of mental or physical faculties by reason of the introduction of alcohol ... into the body; or (B) having an alcohol concentration of 0.08 or more.” See Tex. Penal Code § 49.01(2). We have described these meanings as providing alternative methods of proving that a person is intoxicated. See Bradford v. State, 230 S.W.3d 719, 721-22 (Tex.App.-Houston [14th Dist.] 2007, no pet.). The two methods are known, respectively, as the impairment theory of intoxication and the per se theory of intoxication. Id. Because the two methods of proof are not mutually exclusive, evidence offered under the per se theory can also support a finding that a person is impaired. See Crenshaw v. State, 378 S.W.3d 460, 467 (Tex.Crim.App.2012). As indicated above, proof under the per se theory requires an alcohol concentration expressed in grams of alcohol per 100 milliliters of blood. See Tex. Penal Code § 49.01(1)(B).
Appellant contends that the evidence is insufficient to show that he was intoxicated, citing multiple reasons. He points out first that an investigating officer originally reported that appellant was not intoxicated. Along this same track, appellant refers to testimony that field sobriety tests were never administered on him, that he was never found to have been slurring his words, and that he was actually found to *695be alert and oriented immediately after the accident.
On a different track, appellant asserts that the blood evidence failed to meet certain standards. Appellant complains that the expert who cálculated the BAC of his whole blood used a conversion ratio based on a scientific average, rather than a consideration of appellant’s individual characteristics. Appellant points out that conversions depend heavily on the health of the person, and if his health were shown to be outside the average, then a certain conversion would have established that he was not intoxicated.
The expert’s testimony was primarily used to prove that appellant was intoxicated under the per se theory. Even if we were to conclude that the expert’s methodology was inexact, the conviction would still be supported if there were legally sufficient evidence offered under the impairment theory. We conclude that the record contains such evidence.
The jury heard that appellant admitted to having consumed at least four beers on the night of the accident. Thus, there is some proof that alcohol was introduced into appellant’s body. The jury could have reasonably determined that the accident occurred because appellant lost the normal use of his faculties by reason of that introduction.
The absence of certain environmental factors supports that implied finding. The record shows that there were few cars on the road, and that it had not been raining. There is also no evidence of an obstacle in the road that would have required a sudden turn.
There was some testimony that the accident could have occurred for reasons other than intoxication: appellant may have ov-ercorrected, his fiancée may have accidentally hit the steering wheel, or a tire may have blown out. But the jury was free to disbelieve this evidence and conclude that appellant was impaired by his consumption of alcohol. We conclude that there is legally sufficient evidence of appellant’s intoxication, and that a rational jury could have found every element of the offense beyond a reasonable doubt. See Lorenz v. State, 176 S.W.3d 492, 495 (Tex.App.-Houston [1st Dist.] 2004, pet. ref'd) (concluding that the evidence was sufficient to prove the defendant’s intoxication when, among other factors, there was evidence that the defendant admitted to having consumed portions of three alcoholic beverages).
II. The Trial Court’s Finding
In its brief, the State makes a “Confession of Unassigned Error,” contending that there is legally insufficient evidence to support the trial court’s finding that appellant had “an alcohol concentration level of at least 0.15.” The State bases this confession on authority that the statutory meaning of “blood” is restricted to whole blood, as appellant argued at trial, and there is no evidence that the BAC of appellant’s whole blood was at least 0.15. Even if there were legally sufficient evidence, the State asserts that the trial court should not have made its finding because appellant’s alcohol concentration level was an element of an offense, and therefore, it should have been submitted to the jury during the guilt-innocence phase of trial.
Appellant did not respond to the State’s confession in his reply brief, but we may still address an unassigned error if it was preserved below. See Sanchez v. State, 209 S.W.3d 117, 120-21 (Tex.Crim.App.2006). Because a defendant need not preserve error as to a claim that the evidence is insufficient to prove an element of the offense for which he was conducted, we choose to address this issue. See Flanary *696v. State, 166 Tex.Crim. 496, 496, 316 S.W.2d 897, 898 (1968) (op. on reh’g).
A. Element or Enhancement?
We must first decide whether a person’s alcohol concentration level provides a basis, for enhancement, as the trial court believed, or whether it functions as the element of a completely separate offense, as the State asserts on appeal.
In Calton v. State, the Court of Criminal Appeals explained the differences between elements and enhancements. See 176 S.W.3d 231 (Tex.Crim.App.2006). The court said that the elements of an offense are defined as “the forbidden conduct, the required culpability, any required result, and the negation of any exception to the offense.” Id. at 233. A reviewing court must look to the plain language of the statute when discerning whether any given fact constitutes an element of the offense. Id.
An enhancement, by contrast, is a fact that increases the punishment range . to a certain range above what is ordinarily prescribed for the crime that was charged. Id. “It does not change the offense, or the degree of. the offense, of conviction.” Id.
With those considerations in mind, we now turn to the relevant statute, the full text of which provides as follows:
(a) A person commits an offense if the person is intoxicated while operating a motor vehicle in a public place.
(b) Except as provided by Subsections (c) and (d) and Section 49.09, an offense under this section is a Class B misdemeanor, with a minimum term of confinement of 72 hours.
(c) If it is shown on the trial of an offense under this section that at the time of the offense the person operating the motor vehicle had an open container of alcohol in the person’s immediate possession, the offense is a Class B misdemeanor, with a minimum term of confinement of six days.
(d)If it is shown on the trial of an offense under this section that an analysis of a specimen of the person’s blood, breath, or urine showed an alcohol concentration level of 0.16 or more at the time the analysis was performed, the offense is a Class A misdemeanor.
Tex. Penal Code § 49.04.
There is only a single reference in this statute to “an alcohol concentration level,” and it is located under Subsection (d). A plain reading of that subsection reveals that its effect is to convert an offense from a Class B misdemeanor to a Class A misdemeanor whenever a person charged with driving while intoxicated is shown to have “an alcohol concentration level of 0.15 or more.” Because this conversion represents a change in the degree of the offense, rather than just an enlargement of the punishment range for a Class B misdemeanor, we agree with the State that a person’s alcohol concentration level is not a basis for enhancement. See Cal-ton, 176 S.W.3d at 233 (an enhancement does not change the degree of the offense of conviction). It is instead an element of a separate offense because it represents a specific type of forbidden conduct — operating a motor vehicle while having an especially high concentration of alcohol in the body. Cf. Mapes v. State, 187 S.W.3d 655, 658 (Tex.App.-Houston [14th Dist.] 2006, pet. ref'd) (stating that in a prosecution under Section 49.09, where the defendant is accused of driving while intoxicated and of having prior convictions for driving while intoxicated, the prior convictions are elements of the offense, and not bases for enhancement, in part because they affect the degree of the offense).
*697B. Meaning of “Blood” and Evidence of Alcohol Concentration
The evidence shows that a vial of appellant’s whole blood was obtained at the hospital more than an hour after the accident occurred. From that sample, the blood plasma was separated from the blood cells, and testing of the blood plasma revealed that appellant had a BAC of 0.158. This was the only blood sample that was ever tested for its alcohol content.
The State’s expert converted the blood plasma results using a scientifically accepted method and concluded that appellant’s whole blood had a BAC of 0.132. There was no testimony that appellant’s whole blood had a BAC that was 0.15 or more at the time the analysis was performed.2
Based on this evidence, a finding that appellant had “an alcohol concentration level of 0.15 or more” could be supported only if a person’s alcohol concentration was measured in units of blood plasma, and not whole blood.3 The statute defining “alcohol concentration” does not provide express guidance on this point. It does not say, for instance, whether proof must be submitted in units of whole blood or blood plasma. Rather, it generally provides that alcohol concentration means the number of grams of alcohol per 100 milliliters of “blood,” without specifying which type. See Tex. Penal Code § 49.01(1)(B),
The word “blood” is not defined in the Penal Code, but the Court of Criminal Appeals has indicated that it can only mean whole blood. In Bigon v. State, the court discussed the reliability of certain methods for converting the BAC of blood serum into the BAC of whole blood. See 252 S.W.3d 360, 368 (Tex.Crim.App.2008). If our criminal statutes did not require proof of a person’s intoxication as expressed in units of whole blood, there would have been no need for the conversion testimony. Therefore, Bigon supports the conclusion that “blood” means “whole blood.”
The common definition of “blood” enforces the conclusion that the word encompasses more than just blood plasma. A leading dictionary describes blood as “the usually red fluid, consisting of plasma, red and white blood cells, etc., that circulates through the heart, arteries, and veins of vertebrates.” See Webster’s New World College Dictionary 150 (3d ed. 1996). Medical texts contain similar definitions. See Gray’s Anatomy [5] (15th ed: 1995) (“Blood-'consists of a faintly yellow fluid, the plasma or liquor sanguinis, in which are suspended numerous minute particles, the blood corpuscles, the majority of which are coloured and give to the blood its red tint.”).
The Legislature has also signaled its intent that blood should not be synonymous with blood plasma. In another statute proscribing the sale of human organs, the Legislature expressly provided that the term “human organ” does not include “hair or blood, blood components (including plasma), blood derivatives, or blood reagents.” See Tex: Penal Code § 48.02(a). Because the Legislature de*698scribed plasma as a “blood component,” and juxtaposed that term with “blood” itself, we must conclude that plasma is a only subset of blood, and that the two terms are not congruent.
If we apply that same understanding to the chapter proscribing the offense of driving while intoxicated, then “blood” as used in the definition of “alcohol concentration” must not mean “blood plasma” or any other “component” of blood. See Tex. Gov’t Code § 311.011(b) (providing that words that have acquired a technical or particular meaning, whether by legislative definition or otherwise, should be construed accordingly). Instead, it must mean blood with all of its components, which is otherwise known as “whole blood.”
Because there is indeed no evidence that the BAC of appellant’s whole blood was ever 0.15 or greater at the time of the blood draw, we agree with the State’s confession of error that the trial court’s finding is unsupported by the record. If that were the only issue in the case, we would reform the trial court’s judgment to reflect a conviction for a Class B misdemeanor, which is supported by sufficient evidence as explained above, and remand for a new punishment hearing. See Calton, 176 S.W.3d at 233; Bowen v. State, 374 S.W.3d 427, 432 (Tex.Crim.App.2012). Appellant has asserted other issues, however, and as we explain below, those issues require a new trial.
CHARGE INSTRUCTION
In his second issue, appellant contends that the trial court committed charge error when it denied a request to clarify the definition of “alcohol concentration” so that it was expressed in terms of “whole blood,” rather than just “blood.” We review this complaint under a two-step process, considering first whether error exists. See Ngo v. State, 175 S.W.3d 738, 743 (Tex.Crim.App.2005). If error does exist, we then analyze that error for harm under the procedural framework of Al-manza v. State, 686 S.W.2d 157 (Tex.Crim. App.1984).
I. Error
The trial court must give the jury “a written charge distinctly setting forth the law applicable to the case.” See Tex. Code Crim. Proc. art. 36.14. The State contends that the trial court fulfilled this duty because its charge gave a definition of “alcohol concentration” that exactly tracked the language of Section 49.01. The charge provided: “ ‘Alcohol concentration’ means the number of grams of alcohol per 100 milliliters of blood.”
The Court of Criminal Appeals has previously stated that a jury charge that tracks the language of a statute is “a proper charge on the statutory issue.” See Riddle v. State, 888 S.W.2d 1, 8 (Tex.Crim.App.1994). This rule has been expressed in other settings as well, but the court has not always adhered to it. For instance, when the issue is whether an indictment should be quashed for failing to provide adequate notice, the court has held that tracking the language of the statute may not always be sufficient. In Haecker v. State, the court explained that a charging instrument does not provide adequate notice if it tracks the language of the statute and the statute itself is not “completely descriptive of the offense.” See 571 S.W.2d 920, 921 (Tex.Crim.App. [Panel Op.] 1978). Similarly, in State v. Mays, the court held that an indictment will require greater specificity when a statute uses “an undefined term of indeterminate or variable meaning.” See 967 S.W.2d 404, 407 (Tex.Crim.App.1998). Relatedly, in the charge context, the court has recognized that the jury should be given a definition of terms that have acquired a tech*699nical or established legal meaning. See Medford v. State, 13 S.W.3d 769, 771-72 (Tex.Crim.App.2000); see also Middleton v. State, 125 S.W.3d 450, 454 (Tex.Crim.App.2003) (plurality op.).
The principles behind Haecker, Mays, and Medford guide us when reviewing the correctness of a jury charge. Just as the defendant must receive adequate notice of the charges against him, the jury must understand which law to apply, and the wording of a statute may not be enough. As the Court of Criminal Appeals recently reiterated, “It is not the function of the charge merely to avoid misleading or confusing the jury: it is the function of the charge to lead and to prevent confusion.” Reeves v. State, 420 S.W.3d 812, 818 (Tex.Crim.App.2013) (quoting Williams v. State, 547 S.W.2d 18, 20 (Tex.Crim.App.1977)) (emphasis added). A charge will not prevent confusion if the statutory text on which it is based has a variable meaning in the eyes of the jury.
Standing alone, the statute here is not ambiguous or confusing. “Blood,” as that term is used in Section 49.01, can have only one meaning, and that meaning is “whole blood” as explained above. But during the trial, the jury was not advised of this specific definition, and even before closing arguments began, the jury received conflicting messages regarding its meaning.
When the State’s expert took the stand, appellant cross-examined him about the proof needed to establish that a person is legally intoxicated. The jury heard the following testimony:
Q. And in terms of the law and forensics, Texas requires that the sample be whole blood, correct?
STATE: Objection, Your Honor, that’s not the law.
COURT: Overruled.
A. Not that I’m aware.
Q. Let me rephrase. Not that the — it doesn’t require that the sample itself be whole blood, but the law mandates that the levels of alcohol be related to whole blood, correct?
STATE: Objection, Your Honor, calls for a legal conclusion.
COURT: Overruled.
A. It dictates units. Off the top of my head, I’m afraid I can’t answer regarding the exact statement in the law.
Q. Okay. You’re familiar with the term “BAC”?
A. Yes.
Q. And what does that stand for?
A. BAC can be used — typically it’s used for blood alcohol concentration.
Q. Okay. And if I said to you in Texas the law is that you can’t have .08, I believe it’s milligrams? ■
A. It would be grams per one hundred milliliters.
Q. Okay. So, of whole blood, correct?
A. Again, I would have to go back and look. It’s .08 grams per 100 milliliters is what I remember, at this point.
Q. So, you’re not even aware whether the law in Texas requires whole blood or plasma blood?
A. Without — we report all values as blood alcohol whole blood equivalent. I know we do run serum and plasma and then convert them to the whole blood equivalent and that is standard practice throughout the United States. I’m not going to go off and venture out to say what Texas law specifically delineates in their statutes.
The jury never received a definitive answer from the expert, and the State obfus*700cated the issue by objecting to any suggestion that whole blood is the sole unit of measurement under the statute. As discussed further below, the State also suggested to the jury in its closing argument that whole blood did not need to be the unit of measurement, and the trial court overruled appellant’s objection to this argument.
This is not a case where the jury was free to assign its own understanding to an undefined term from a statute. Cf. Kirsch v. State, 357 S.W.3d 645, 652 (Tex.Crim. App.2012) (holding that the jury should have been free to decide on its own whether the evidence showed that the defendant was “operating” a motor vehicle). The word “blood” is capable of only one meaning under the Penal Code, and the jury should have been instructed of that meaning because it was the law applicable to the case. Because the jury heard testimony regarding both whole blood and blood plasma, we conclude that it was error to refuse appellant’s requested instruction, which removed the possibility of confusion by specifying the type of blood evidence that was available for consideration.
II. Harm
Under Almanza, the level of harm necessary for reversal depends on whether the defendant timely and specifically objected to the jury charge. See Almanza, 686 S.W.2d at 171. If the defendant did not object, then reversal is required only if the trial court’s error was so egregious and created such harm that the defendant did not have a fair and impartial trial. Id. Because the defendant in this case properly objected, reversal is required if there was just “some harm.” Id.
To determine harm, we weigh the following factors: (1) the jury charge as a whole; (2) the arguments of counsel; (3) the entirety of the evidence; and (4) any other relevant factors present in the record. See Warner v. State, 245 S.W.3d 458, 461 (Tex.Crim.App.2008). Even though the “some harm” standard is a low threshold, it nonetheless requires the reviewing court to find actual harm, rather than just theoretical harm. Id. Neither party bears the burden on appeal to prove harm or harmlessness. Id. at 462.
A. The Jury Charge
The jury received a single instruction, which asked it to determine whether appellant had “unlawfully operate[d] a motor vehicle in a public place while intoxicated.” As stated above, the definition of “intoxicated” tracked the language of Section 49.01, meaning that the jury was allowed to convict appellant on either the impairment theory of intoxication or the per se theory of intoxication. However, the jury was never asked to identify which theory it used in the event of a conviction. Thus, we cannot know whether the jury found that appellant had lost the normal use of his faculties as the result of his consumption of alcohol, whether it found that the alcohol concentration in his blood was above the legal limit under the legally flawed portion of the charge, or whether it made findings under both theories. Accordingly, we turn to the other factors to determine whether the erroneous denial of appellant’s requested instruction might have prejudiced the jury’s consideration of the evidence or substantially affected their deliberations. See Bagheri v. State, 119 S.W.3d 755, 763 (Tex.Crim.App.2003).
B. Closing Arguments
Appellant’s main defensive theory was that he was not the driver. During closing arguments, appellant spent most of his time explaining that his fiancée was responsible for the accident, and that he had initially taken the blame because he wanted to protect her.
*701In the alternative, appellant also argued that he was not intoxicated, a point that would have been relevant only in the event that he was found to be driving. Appellant limited his argument to just the impairment theory of intoxication. He asked the jury to acquit him because testimony from the first responders supported a finding that he was not impaired.
The State argued that appellant was the driver and that he should be found guilty under either theory of intoxication. As for the impairment theory, the State recited testimony that appellant had consumed several beers on the night of the accident. As for the per se theory, the State pointed to the expert’s opinion regarding the alcohol concentration in appellant’s whole blood and blood plasma.
Appellant asserts in his third issue that the State made an improper closing argument. We will consider this issue as part of our harm analysis for charge error. At the beginning of its closing argument, the State told the jury: <cYou won’t find anywhere in here in the law [the charge] where it needs to be whole blood or plasma blood. This is the law y’all follow.” Appellant objected to this argument and asserted that the “law requires whole blood,” but the trial court overruled the objection. By overruling appellant’s objection, the trial court gave the incorrect impression that a conviction could be had by evidence other than whole blood, thereby magnifying the possibility for harm. See Good v. State, 723 S.W.2d 734, 738 (Tex.Crim.App. 1986); Kincaid v. State, 534 S.W.2d 340, 342 (Tex.Crim.App.1976). Later, the State further magnified this erroneous ruling by telling the jury that it could convict appellant because the BAC of his blood plasma exceeded 0.15.
The State’s argument was a misstatement of law.4 Proof under the per se theory must be based on evidence of a person’s BAC in whole blood. The trial court put its imprimatur on the improper argument by overruling an objection that “the law requires whole blood.” Had the trial court granted appellant’s requested instruction, the State would have known not to make its improper argument, and the jury never would have heard those prejudicial remarks.
C. The Evidence
Appellant presented affirmative evidence in support of two defensive theories: (1) that he was not the driver, and (2) that he was not intoxicated. The jury rejected the evidence as it related to the first defensive theory. The jury found that appellant was the driver, and we have no reason to suspect that this finding was a byproduct of the trial court’s charge error. If the error had any impact at all on the jury’s deliberations, it affected the manner in which the jury assessed the evidence of appellant’s intoxication.
Insofar as the impairment theory of intoxication is concerned, the evidence was conflicting. The State produced evidence *702that appellant was impaired: he admitted that he consumed several beers, he had alcohol in his system, and there was no environmental cause for explaining the accident. Appellant emphasized a different view of the evidence: the first responders believed that he was not intoxicated, he was not slurring his speech, and he was described as being alert and oriented. There was also evidence suggesting that the accident could have resulted from a cause other than intoxication, such as an overcorrection. This alternative explanation created a fact question for the jury to resolve.
As for the per se theory, the evidence was complex. A hospital technician testified that appellant’s blood sample was obtained solely for purposes of treatment, not forensics. The technician recognized that there were differences between whole blood and blood plasma, but he could not explain how the concentrations of alcohol varied between the two samples. The only witness who could explain that relationship was the State’s expert, William Arnold.
The expert testified that the BAC of a person’s blood “serum” is 16% higher than the BAC of his whole blood.5 Applying that ratio to the BAC of appellant’s blood plasma, the expert opined that appellant’s whole blood had a BAC of 0.132 at the time of the blood draw.6
During a voir dire examination, the expert conceded that his 16% ratio was not universally accepted throughout the scientific community. He explained that 16% is just an average, and that the range of ratios can be “quite dramatic” depending on the health and gender of the person. If a person has a high blood cell count — for example, if he has leukemia — then the BAC of the person’s blood plasma can be 50% or 60% higher than the BAC of his whole blood.
The expert testified that he had “no idea” what ratio should apply in appellant’s case. The expert never considered appellant’s individual characteristics, and there was no live testimony establishing whether appellant had a blood cell count that was within normal ranges.7 The expert ad*703vised the jury, however, that if a higher conversion ratio did apply to appellant because he had a high blood cell count, then the BAC of his whole blood would have been 0.079, below the per se limit.
In addition to this conversion testimony, the expert provided a retrograde extrapolation analysis. The expert explained that when a person consumes alcohol, his body will absorb the alcohol into his system until it is eliminated by the liver. During the absorption phase, the concentration of alcohol in the body will increase at a rate that depends on a number of factors, such as how many drinks were consumed, how quickly they were consumed, and whether they were consumed on a full or empty stomach. According to the expert, once all of the alcohol has been fully absorbed, the liver will eliminate the alcohol at a constant rate of 0.015 grams per 100 milliliters per hour.
The expert made two assumptions: (1) that appellant was eliminating alcohol from his system at the time of the blood draw, and (2) that the accident occurred one and half hours before the blood draw.8 From these assumptions and his stated formula, the expert estimated that the BAC of appellant’s whole blood at the time of the accident was 0.133.
The expert did not explain in detail how he arrived at that figure, nor did he provide a factual basis for assuming that appellant was in the elimination phase. The expert testified that a person normally enters the elimination phase within two hours of his last drink, but there was no evidence indicating when appellant last consumed an alcoholic beverage. Nor was there any evidence of whether appellant had been drinking on a full or empty stomach. By assuming that the accident happened an hour and a half before the blood draw, the expert failed to realize that, without more information, a conclusion could not even be made that appellant had entered the elimination phase by the time he arrived at the hospital.
D. Other Relevant Factors
We first note that the expert struggled to perform his calculations, which were “on the fly,” as he described them. The trial court had to excuse the expert from the courtroom, and the expert returned to give his calculations after another witness took the stand.
We also note that, if appellant had been in the absorption phase at any point between the time of the accident and the blood draw, then the BAC of his whole blood at the time of the accident would have been equal to or less than the BAC of his whole blood at the time of the blood draw. This is true regardless of the conversion ratio that is used.
E. Analysis
It is clear that the jury determined that appellant was driving, but it is not clear under which theory it found that *704he was intoxicated. There was a conflict in the evidence under the impairment theory. There was also a conflict under the per se theory; the jury heard testimony that the BAC of appellant’s whole blood could have been above or below 0.08, depending on which sorts of assumptions were made.
The expert’s testimony demonstrated that many variables must be accounted for when trying to establish that a person is intoxicated per se. But under the trial court’s erroneous view of the law, none of those variables needed to be considered in this case. The trial court’s charge error, when combined with the State’s improper argument, allowed the jury to convict appellant solely on the basis that the BAC of his blood plasma exceeded the per se limit. We conclude that this error resulted in some harm because it completely obviated the jury’s need to examine appellant’s alternative defense, and there was affirmative evidence in support of this defense showing that appellant was not impaired. See Anderson v. State, 774 S.W.2d 733, 735 (Tex.App.-Houston [14th Dist.] 1989, no pet.) (erroneous instruction on the per se theory of intoxication was harmful, even though there was sufficient evidence to convict under the impairment theory of intoxication); cf. Bagheri, 119 S.W.3d at 763-64 (erroneous admission of retrograde extrapolation testimony was harmful, even though there was sufficient evidence to convict under the impairment theory of intoxication).
CONCLUSION
The evidence is legally insufficient to support a conviction for a Class A misdemeanor for driving while intoxicated. We render a judgment of acquittal as to this offense, meaning that the State is barred from retrying it. ' See Ex parte Granger, 850 S.W.2d 513, 518 (Tex.Crim.App.1993).
There is sufficient evidence to support a conviction for a Class B misdemeanor, but because we conclude that the trial court committed harmful errors with respect to this offense, we reverse the court’s judgment and remand the case for a new trial.9
(Frost, C.J., dissenting).. In this opinion, all numerical references to BAC are expressed in the same units: grams of alcohol per 100 milliliters of blood. We denote in each instance whether the blood at issue is whole blood or blood plasma.
. We explain, infra, at note 6, that the expert's calculations were wrong. However, even if the expert had correctly converted the results of appellant’s blood plasma testing, there would still be no showing that the BAC of appellant’s whole blood was at least 0.15 at the time the analysis was performed.
. Although alcohol concentration is a separate element of the Class A misdemeanor offense, appellant has not argued that the trial court violated his due process rights by not submitting that element to the jury. Cf. Ex parte Boyd, 58 S.W.3d 134, 136 (Tex.Crim.App.2001) (applying Apprendi v. New Jersey, 530 U.S. 466, 490, 120 S.Ct. 2348, 147 L.Ed.2d 435 (2000)).. In any event, we need not address this issue because we conclude that appellant is entitled to an acquittal of the Class A misdemeanor on other grounds.
. The State contends that its trial argument was "plainly true” and an accurate description of the jury charge. Because the charge did not specify that appellant’s alcohol concentration must be measured in units of whole blood, the State believes it could not be improper to say that the charge was silent. We disagree. By emphasizing that the charge was silent, the State's manifest objective was to convey to the jury that it could base a finding of intoxication on either the evidence of whole blood or the evidence of blood plasma. This effect made the argument improper, even though it was otherwise grounded by an accurate statement of fact. Cf. Myers v. State, 573 S.W.2d 19, 20-21 (Tex.Crim.App. [Panel Op.] 1978) (prosecutor made an improper closing argument when a comment about an accurate statement of fact created an implied or indirect allusion to the defendant’s failure to testify).
. The expert did not provide a ratio as it specifically relates to a person's blood "plasma,” but we recognize that the differences between blood serum and blood plasma are small. Studies have shown that alcohol concentrations are often very close between the two samples, if not the same, with an average variance of approximately 1%. See Charles L. Winek & Mark Carfagna, Comparison of Plasma, Serum, and Whole Blood Ethanol Concentrations, 11 J. Analytical Toxicology 267, 267-68 (1987) (cited approvingly in Bigon, 252 S.W.3d at 368 n. 6).
. It appears that the expert miscalculated here. He subtracted 16% of 0.158 from 0.158 (which is the same as multiplying 0.158 by 0.84), when he should have divided 0.158 by 1.16. See Jessup v. State, No. 13-02-00024-CR, 2004 WL 2612958, at *3 (Tex.App.-Corpus Christi Nov. 10, 2004) (mem. op., not designated for publication) (reciting expert testimony that the BAC of whole blood is derived by "dividfing] the serum concentration by 1.16”). The difference between the expert's miscalculation and the correct calculation is de minimis.
.Our dissenting colleague points out that appellant’s hematologic history was checked as normal, according to a report attached to his medical records. However, this report was never discussed at trial, and there is no indication that the jury ever saw it. At the end of closing arguments, the trial court advised the jury that the exhibits, if requested, would be delivered to the deliberation room: "I’ll remind you again that if you want to see the evidence, you can ask for any or all — any or all of that, and we'll send it into the jury room with you.” There is no record that the jury made a request, and the trial court did not note a request in any of its docket sheets.
Even if the medical records had been requested, there is no showing that the jury would have known how to interpret them. The meaning of “hematologic” is unlikely to *703be within the common understanding of a lay juror. Furthermore, the records clearly denoted that appellant had a "High” count of "WBC” — white blood cells, we presume. A juror could have seen the records and concluded that appellant’s blood cell count was higher than average.
. There is no concrete evidence that the accident actually happened one and a half hours before the blood draw. The expert was simply asked to conduct his analysis assuming that time lapse. From the live testimony, we know that the eyewitness who saw the accident was traveling on the interstate between 11:00 p.m. and midnight. The eyewitness did not testify about the exact time that he saw the accident, but a first responder testified that the accident happened "just before midnight.” As for the other time point, the evidence is much clearer: a nurse testified that she drew appellant's blood at the hospital at 1:05 a.m.
. This disposition makes it unnecessary to address appellant’s fourth issue, in which he argues that the trial court imposed an unconstitutional condition of community supervision.