*401Defendant appeals a judgment convicting him of driving under the influence of intoxicants (DUII). ORS 813.010.1 In his second assignment of error, defendant argues that the trial court erred when it denied his motion for judgment of acquittal. Specifically, defendant argues that, based on the evidence presented at trial, no factfinder could have found beyond a reasonable doubt that defendant's blood alcohol content (BAC) was .08 percent or higher when he was riding or "driving" his motorcycle. On appeal, we conclude that, even though the fact that alcohol in the blood dissipates over time is common knowledge, that knowledge combined with the minimal evidence presented at trial in this case is not sufficient for a reasonable factfinder to find beyond a reasonable doubt that defendant's BAC was above .08 percent at the time he was driving. Accordingly, we reverse.2
The relevant facts are few and undisputed. Oregon State Police Trooper Dunlap stopped defendant for riding his motorcycle without a helmet. Following that stop, Dunlap took defendant into custody for DUII and brought defendant to the North Bend Police Department for a blood alcohol test using an Intoxilyzer. That test began approximately one hour and 45 minutes after defendant was initially stopped. The test indicated that defendant's BAC was .09 percent. Between the time that defendant was stopped and the time that his BAC was tested, Dunlap did not observe defendant consume any alcohol.
Defendant was charged by information with DUII. A short bench trial was held in which the state relied solely on a per se theory of intoxication-i.e. , that defendant was legally intoxicated under the relevant statute because he *402had a .08 percent or higher BAC when he was riding his motorcycle. ORS 813.010(1)(a).3 *1082In defendant's closing argument, he argued that the trial court could not convict him because the state's evidence was legally insufficient to prove that he was intoxicated in violation of ORS 813.010(1)(a) at the time that he was riding. Specifically, defendant noted that the only proof offered by the state was that defendant's BAC was .09 percent when measured one hour and 45 minutes after he had been driving and that defendant had not consumed any alcohol during that time. Defendant argued that there was no basis for a factfinder to determine, without further evidence, whether defendant had a .08 percent or higher BAC at the time that he was driving. The trial court, acting as a factfinder, rejected defendant's argument and convicted defendant, noting:
"Well, in this case, I will find [defendant] guilty because the only evidence before me is what he blew, and I don't have evidence at all that suggests one way or the other what you do with *** that to equate it with time of driving. But that's the evidence I have.
"So, I'll find him guilty of that."
Defendant appealed.
As noted, on appeal defendant argues that the trial court erred in denying his motion for judgment of acquittal because no factfinder could infer beyond a reasonable doubt that defendant had a BAC of .08 percent or higher when he was riding his motorcycle.4 Specifically, defendant contends *403that it is speculative to find that his BAC was above the legal limit at the time he was driving based on the results of a blood alcohol test that he took one hour and 45 minutes later. In response, the state contends that, under prior case law, the trial court was entitled to infer based on the record before it that defendant's BAC when he was riding was at least as high as the BAC measured by defendant's later blood alcohol test. We agree with defendant and reverse.
"The sufficiency of the evidence is a question of law." State v. Reynolds , 250 Or. App. 516, 520, 280 P.3d 1046, rev. den. , 352 Or. 666, 293 P.3d 1045 (2012). We review "questions of the sufficiency of the evidence in a criminal case following a conviction by examining the evidence in the light most favorable to the state to determine whether a rational trier of fact, accepting reasonable inferences and reasonable credibility choices, could have found the essential element of the crime beyond a reasonable doubt." State v. Cunningham , 320 Or. 47, 63, 880 P.2d 431 (1994), cert. den. , 514 US 1005, 115 S.Ct. 1317, 131 L.Ed.2d 198 (1995). However, "[i]f the state has sought to establish an element of a criminal offense by reasonable inference, * * * whether sufficient evidence supports the inference is a question for a court to decide." State v. Guckert , 260 Or. App. 50, 55, 316 P.3d 373 (2013), rev. den. , 354 Or. 840, 326 P.3d 77 (2014).
Here, the state attempted to establish an element of its DUII case by inference. That is, the state contends that, based on defendant's blood alcohol test result of .09 percent obtained one hour and 45 minutes after defendant was driving, the fact that defendant had not consumed any alcohol during that intervening time, and the common knowledge that alcohol dissipates over time, a reasonable factfinder could infer that defendant's BAC was .08 percent or higher at the time that he was riding his motorcycle.
The state is allowed to rely on "circumstantial evidence and reasonable inferences flowing from that evidence" to prove an element of a crime. State v. Bivins , 191 Or. App. 460, 466, 83 P.3d 379 (2004). However, "[t]here is a difference between inferences that may be drawn from circumstantial evidence and mere speculation." Id. at 467, 83 P.3d 379 (internal quotation marks omitted). "Reasonable inferences are permissible;
*1083speculation and guesswork are not." Id. The line *404between reasonable inferences and impermissible speculation " 'is drawn by the laws of logic.' " Id. (quoting Tose v. First Pennsylvania Bank, N.A. , 648 F.2d 879, 895 (3d Cir.), cert. den. , 454 US 893, 102 S.Ct. 390, 70 L.Ed.2d 208 (1981), abrogated on other grounds by Griggs v. Provident Consumer Discount Co. , 459 U.S. 56, 103 S.Ct. 400, 74 L.Ed.2d 225 (1982) ). As a result, the issue before us is whether mere logic renders probable that, when a person's BAC is .09 percent one hour and 45 minutes after he drove and he has not consumed alcohol over that period, that person's BAC was at least .08 at the time that he was driving. See Bivins , 191 Or. App. at 467, 83 P.3d 379 (noting that, if there is a "logical probability that an ultimate fact will follow a stated narrative or historical fact, then the jury is given the opportunity to draw a conclusion because there is a reasonable probability that the conclusion flows from the proven facts" (internal quotation marks omitted; emphasis added) ).
We begin by noting that we agree with the proposition, as does the dissent, that it is common knowledge that alcohol dissipates from the body over time. 290 Or. App. at 409, 415 P.3d at 1085 (Powers, J., dissenting). That proposition has long been recognized, at least in dictum , by Oregon courts. See, e.g. , State v. Eumana-Moranchel , 352 Or. 1, 11, 277 P.3d 549 (2012) (noting, in dictum , that BAC at the time of driving can be proven with a combination of a chemical analysis and "an inference that blood alcohol rates dissipate over time"); State v. Parker , 317 Or. 225, 232 n. 9, 855 P.2d 636 (1993) (noting, in dictum , that "[i]t is common knowledge that the level of alcohol in the blood and alcohol's effect on a person's behavior dissipate over time" (internal quotation marks omitted) ); State v. Conway , 75 Or. App. 430, 435, 707 P.2d 618, rev. den. , 300 Or. 451, 712 P.2d 110 (1985) (noting, in dictum , that it is "common knowledge that alcohol dissipates" over time).
We also agree that a factfinder can rely on that common knowledge in determining whether the state has sustained its burden of proof. "The jury is entitled to draw inferences from matters of common knowledge." Dodge v. Tradewell Stores , 256 Or. 514, 516, 474 P.2d 745 (1970). Thus, although a factfinder may not use personal knowledge as a substitute for evidence not provided by the state, "triers of fact are entitled to draw inferences from facts directly proved based on their common knowledge, experience, or personal *405observation." State v. Clelland , 214 Or. App. 151, 159, 162 P.3d 1081 (2007). However, we do not agree that the factfinder's common knowledge can bear the weight that the state would require it to bear, and that it must bear if we are to conclude that the evidence adduced at trial was sufficient to support defendant's conviction.
The fact that blood alcohol dissipates does not logically lead to any conclusion regarding a specific person's earlier BAC at a specific time. The fact that blood alcohol dissipates is notable for what it does not tell the court. It does not, for example, by itself inform the court whether, at any given time , a person's blood alcohol is dissipating or increasing. After all, it is also a matter of common knowledge that, before a person's blood alcohol can dissipate, alcohol must accumulate in the blood. See generally State v. Trujillo , 271 Or. App. 785, 353 P.3d 609, rev. den. , 358 Or. 146, 363 P.3d 1288 (2015) (discussing scientific evidence regarding blood alcohol accumulation and dissipation); State v. Baucum , 268 Or. App. 649, 343 P.3d 235, rev. den. , 357 Or. 550, 357 P.3d 503 (2015) (same). More significantly-and largely due to the last point-the fact that blood alcohol dissipates says nothing about whether a person's BAC at any given time was higher, lower, or the same as when it was measured sometime later. Thus, without more, a factfinder has no way to determine if defendant's BAC was still rising, but under .08, at the time he was driving and peaked later before falling to .09 at the time of the BAC test.
Further, while it may be common knowledge that alcohol accumulates and dissipates in the blood over time, we can safely conclude that the precise rates of alcohol accumulation or dissipation in the blood are not matters of common knowledge. See Eumana-Moranchel , 352 Or. at 9, 14, 277 P.3d 549 (noting the expert testimony in that case that explained the general rate of alcohol accumulation and elimination in the blood following *1084the time of consumption and holding that the state should have been able to call their expert on retrograde extrapolation); cf . State v. O'Key , 321 Or. 285, 296, 899 P.2d 663 (1995) (holding that the Horizontal Gaze Nystagmus (HGN) test for possible intoxication is a scientific technique distinguished from other field sobriety tests because science, "rather than common knowledge," provides its legitimacy). Nor is it commonly known the extent to which those rates *406could be affected by other factors, such as the intoxicated person's gender and weight, among other factors.
Furthermore, no additional "predicate" facts in this case could help a factfinder determine defendant's BAC at the time he was driving. See Bivins , 191 Or. App. at 469, 83 P.3d 379 (referring to narrative or historical facts that inform what inferences may reasonably be drawn as "predicate facts"). The only facts presented to the court were that defendant's BAC was .09 percent approximately one hour and 45 minutes after he was driving, and that he had not consumed any additional alcohol during that time.
The difficulty in this case is that the state proceeded solely on a per se theory under ORS 813.010(1)(a), which required the state to prove that defendant's BAC was .08 or higher at the time of driving, a minimum calculated amount of alcohol in the bloodstream at a particular time. Without additional evidence guiding the trier of fact on rates of accumulation and dissipation either through an expert's "retrograde extrapolation" or perhaps other admissible evidence that could lead to a permissible inference of likely BAC at the time of driving based on the subsequent Intoxilyzer test, the factfinder cannot, at least on this record, apply the common knowledge that blood alcohol goes up and down over time to make a reasonable inference about when defendant's BAC likely reached .08 or above and whether that occurred while defendant was driving. See State v. Miller , 289 Or. App. 353, 359, 413 P.3d 999 (2017) (holding that a jury's educated guess in determining a particular distance based only on estimates and the use of the Pythagorean Theorem was too speculative where the statute required proof that defendant was within 25 feet of the protected person).
Because the limited evidence presented by the state in this case does not include any evidence bearing on the movement of alcohol through defendant's body or the presence of alcohol in defendant's body at the time or shortly before defendant drove, Eumana-Moranchel , Parker , and Conway do not inform our analysis. Without that additional evidence, it does not follow solely as a matter of probability and logic that a person whose BAC is measured at .09 percent would have necessarily had a BAC of at least *407.08 percent an hour and 45 minutes earlier if he or she consumed no alcohol during that intervening time period.
The dissent asserts that there are "multiple reasonable inferences" that can be drawn by a factfinder regarding defendant's BAC at the time of driving, including that: (1) defendant's BAC could have been lower than the later test result (supporting an acquittal); (2) it could have been higher than the later test result (supporting a conviction); or (3) it could have been the same as the later test result (also supporting a conviction). 290 Or. App. at 414, 415 P.3d at 1088 (Powers, J., dissenting). While it is true that we generally let the factfinder draw the inferences from the evidence, without more information, there is nothing but speculation that guides a factfinder to select from one of those three possible inferences. It could be that defendant's BAC was still rising and had not reached .08 at the time that he was pulled over and, thus the first inference-that defendant's BAC was lower than the later test result-is correct. It could be that defendant's BAC had reached .08 or .09 at the time that he was pulled over, peaked either then or later, and was falling at the time of the test, as we can generally infer that alcohol ultimately dissipates over time. The factfinder is left to speculate.
The trial court erred in denying defendant's motion for a judgment of acquittal.