At Raul Mata's trial for driving while intoxicated, an expert witness testified that Mata's blood alcohol content two hours after his arrest showed that he was intoxicated at the time of his arrest. We must decide whether this expert's testimony met the requirements of Daubert v. Merrell Dow Pharmaceuticals, Inc., ⁽¹⁾ and Kelly v. State. ⁽²⁾ We conclude that it did not.

Officer Kenneth Thompson pulled Mata over at 3:05 a.m. on May 25, 1992, for a traffic violation. He smelled alcohol on Mata's breath, so he asked Mata to perform some field sobriety tests. After the tests, Officer Thompson arrested Mata for driving while intoxicated. Over two hours later, at 5:14 a.m., Mata took two breath tests in quick succession and registered a blood alcohol content (BAC) of .19.

The State charged Mata with driving while intoxicated both by driving with a BAC in excess of .10 and by driving without the normal use of his mental and physical faculties.

Mata filed a motion to suppress the results of his breath test because, among other reasons, they were obtained "from scientific techniques which have not been shown by clear and convincing evidence to be reliable and relevant." He relied on Rules 403 and 702 of the Texas Rules of Criminal Evidence - the applicable rules at the time of Mata's 1993 trial.

At the suppression hearing, George Allen McDougall, Jr., testified that he was the breath test technical supervisor for Bexar County. He explained that an individual's BAC is reflected by an alcohol concentration curve which begins at the low point when there is no alcohol in the person's blood and rises as the alcohol is absorbed into the bloodstream until it reaches the peak, that is, the maximum alcohol concentration. Then the curve falls as the alcohol dissipates from the bloodstream to the low point when there is no alcohol left. McDougall testified that he had "done some calculations like this on his own" and that to draw such a curve for an individual, he would need "more than a couple" of reference points. He would need "a lot of them." McDougall conceded that, although Mata blew into the breathalyzer twice, the readings were only two minutes apart and, for purposes of drawing an alcohol concentration curve, constituted only a single reading. That single reading did not give McDougall enough information to determine whether Mata was in the absorption phase or the elimination phase at the time of the breath test.

McDougall testified that, in order to determine an individual's BAC at the time the individual was driving, he "assume[s] several variables." He testified to several possibilities: the individual could have had a higher BAC at the time he was driving than at the time of the breath test; the individual could have had a lower BAC at the time he was driving than at the time of the breath test; or the individual could have the same BAC at both times. McDougall testified that he determines how high the BAC could have possibly been and how low it could have possibly been. Then "between those two somewhere is where the actual value will be."

McDougall agreed that the alcohol concentration curve of an individual who drank on an empty stomach would rise higher and faster than the curve of an individual who drank on a full stomach. He conceded that without a number of tests, and without knowing an individual's weight and whether he had eaten anything, he could not determine the steepness of that individual's alcohol concentration curve. He testified that, regardless of what the person had in his stomach, "an hour and a half is adequate for a complete absorption of the drinks." ⁽³⁾ He testified that alcohol eliminates at a rate of .02 grams per 210 liters. ⁽⁴⁾

When given a hypothetical case based on the facts of this case, McDougall testified that if an individual's breath test registered .19 two hours and nine minutes after that person had been driving, McDougall would say that the person had a higher BAC at the time of the test than at the time of the driving. He also testified that, nevertheless, based on those facts, he would believe the person was intoxicated at the time of the driving.

McDougall testified that, given Mata's BAC of .19 two hours after his arrest, he believed Mata was impaired at the time of driving and had lost the normal use of his mental or physical faculties at the time he was driving. But he emphasized that he was not saying that Mata's BAC at the time of driving was .10 - only that it was at least .08. He said Mata's BAC range was between .12 or .13 all the way up to .25. He testified that these figures were based on Mata having an empty stomach. If he had a full stomach, the range would be more narrow -- that is, from a low end of .16 up to a high end of .21 or .22. Then, when asked if Mata's BAC could have been less than a .10, McDougall said "there was an "extreme situation" in which that was possible. ⁽⁵⁾

McDougall conceded that he did not know how much Mata weighed, how much Mata had to eat or drink before taking the breath test, or when Mata's last drink was.

Defense counsel argued that McDougall's testimony was unreliable insofar as it attempted to extrapolate Mata's .19 breath test result back to the time of his arrest. The trial court denied the motion to suppress.

At trial, McDougall testified to his background again. When the prosecutor asked McDougall to give the jury a range for a person's BAC at the time of driving if two hours later his breath test showed a BAC of .19, defense counsel examined McDougall outside the presence of the jury "concerning the underlying facts or data upon which he relies to render this opinion, so that the Court may determine that the expert has a sufficient basis for his opinion."

During voir dire examination, McDougall testified that, while elimination rates are standard, absorption rates vary from person to person. He testified that the absorption phase does not last more than one hour. He then testified that it could last up to an hour and a half if the person had a full stomach. But he maintained that it was not possible for a person's BAC to peak two hours after he quit drinking. When presented with an exhibit showing a person's BAC peaking two hours after drinking stopped, McDougall testified that that case involved a very small amount of alcohol, which was not the norm. He later testified that he had "tested thousands of individuals" and, "with all of the studies that [he had] read," it was his experience that the BAC will not rise for more than one and a half hours after the first drink, whether the person has a full or empty stomach. ⁽⁶⁾

McDougall acknowledged that he did not know when Mata began drinking or when he stopped or when Mata reached his peak BAC. He explained that, without knowing when Mata peaked, he considers all the possible peak points to calculate the BAC range for Mata at the time of the arrest. He testified that the normal or average person eliminates alcohol at a rate of .02 per hour, that he uses the .02 elimination rate, ⁽⁷⁾ and that he bases his calculations on a "normal drinking pattern." According to McDougall, an individual who "chug-a-lugs" shots of hard liquor in a short period of time is not engaging in a "normal" drinking pattern. He admitted that, if he were to include the chug-a-lug situation in his calculations, there would be "a lot of variability." He conceded that a study by the National Highway Traffic Safety Administration (NHTSA) reflected that a person's BAC can rise 10 points in one hour in the chug-a-lug situation, but he repeated that this was not a "normal" drinking pattern. ⁽⁸⁾

At the conclusion of the voir dire, defense counsel argued that McDougall was only qualified to testify about the "average" or "normal" person, but he could not apply his calculations to Mata, because he did not know if Mata engaged in "normal" drinking patterns, and he did not know if Mata had been eating, how much he had been eating, how much he had had to drink, or his weight. The trial court allowed McDougall to offer his opinion before the jury.

Before the jury, McDougall testified that, given a .19 BAC two hours after driving, he would estimate a person's BAC while driving to be between .13 and .23. ⁽⁹⁾ He also testified that the person's BAC could be lower than .10 if he chug-a-lugged a lot of hard liquor just before being stopped. ⁽¹⁰⁾ He first testified that it would take five shots to increase from a .10 to a .19 in two hours. He later testified that it would take seven to twelve shots. ⁽¹¹⁾

In performing a hypothetical calculation, McDougall utilized a .02 or .03 elimination rate. He later testified that the elimination rate could range from .018 to .025. ⁽¹²⁾

McDougall again testified that he based his calculations and estimates on "normal drinking patterns" and he did not know if Mata followed such patterns. ⁽¹³⁾ He also testified that he did not believe a person's peak BAC could occur more than an hour and a half after the arrest. ⁽¹⁴⁾

The jury convicted Mata of driving while intoxicated and the court sentenced him to 90 days in jail, probated for two years, and a fine of $400. Mata filed a motion for new trial arguing that the trial court erred in overruling the motion to suppress and in allowing into evidence "the opinion of . . . McDougall . . . because that opinion was non-scientific and unreliable." After hearing argument, the trial court denied Mata's motion.

In his first two points of error on appeal, Mata argued that the trial court erred in overruling his suppression motion because McDougall's testimony was unreliable and because the probative value of the testimony was outweighed by the unfair prejudice. Mata relied in part on Daubert and Kelly. The court of appeals summarily rejected these claims, quoting a portion of its prior opinion in Hartman v. State ⁽¹⁵⁾:

Given McDougall's impeccable qualifications, including extensive personal observations of the alcohol absorption and elimination process, and the limits which McDougall placed on his opinion, we find the trial court did not abuse its discretion in admitting his testimony. Once the trial court so found, any further doubts as to the veracity of McDougall's opinions were for the trier of fact to weigh in its deliberations. ⁽¹⁶⁾

The appellate court did not discuss the requirements of Daubert or Kelly and whether those requirements were met in this case. In a lengthy dissent, Justice Cadena concluded that McDougall's testimony was not sufficiently reliable under Kelly and that the trial court abused its discretion in admitting it. ⁽¹⁷⁾

We granted Mata's petition for discretionary review to decide whether McDougall's testimony was admissible pursuant to the requirements of Daubert and Kelly.

The State argues that Mata did not preserve error because his written motion to suppress did not attempt to suppress McDougall's testimony. Instead, it attempted to suppress the results of the breath test. The court of appeals found the error preserved. ⁽¹⁸⁾

Appellate Rule 33.1(a)(1)(A) provides that error is preserved if the "complaint was made to the trial court by a timely request, objection, or motion" that states the grounds "with sufficient specificity to make the trial court aware of the complaint." At the suppression hearing, Mata argued for the suppression of McDougall's testimony on the grounds that it was not reliable. The trial court denied the motion to suppress after hearing Mata's argument. We conclude that Mata timely informed the trial court of his grounds with sufficient specificity and, therefore, that he preserved error.

Evidence Rule 702 provides that an expert witness may testify as to his opinion based on scientific knowledge if it will help the trier of fact understand the evidence or determine a fact in issue. A trial court's responsibility under Rule 702 is to determine whether proffered scientific evidence is sufficiently reliable and relevant to assist the jury. ⁽¹⁹⁾ The proponent of the scientific evidence must demonstrate by clear and convincing evidence that the evidence is reliable. ⁽²⁰⁾ This is accomplished by showing the validity of the underlying scientific theory, the validity of the technique applying the theory, and proper application of the technique on the occasion in question. ⁽²¹⁾ Factors that may affect reliability include, but are not limited to, the following: (1) the extent to which the underlying scientific theory and technique are accepted as valid by the relevant scientific community, if such a community can be ascertained; (2) the testifying expert's qualifications; (3) the existence of literature supporting or rejecting the underlying scientific theory and technique; (4) the technique's potential rate of error; (5) the availability of other experts to test and evaluate the technique; (6) the clarity with which the underlying scientific theory and technique can be explained to the court; and (7) the experience and skill of the person who applied the technique on the occasion in question. ⁽²²⁾

While Rule 702 involves the "dual inquiry of relevance and reliability," the "overarching subject of Rule 702 is the scientific validity of the evidence at issue." ⁽²³⁾ In weeding out the so-called "junk" science, trial judges are called upon to serve as "gatekeepers." ⁽²⁴⁾ The proffered testimony must be sufficiently tied to the facts of the case that it will aid the jury in resolving a factual dispute. ⁽²⁵⁾ We will not disturb the trial court's decision to admit scientific evidence absent an abuse of discretion. ⁽²⁶⁾

The science which we analyze today is known as "retrograde extrapolation." Retrograde extrapolation is the computation back in time of the blood-alcohol level - that is, the estimation of the level at the time of driving based on a test result from some later time. ⁽²⁷⁾

As alcohol is consumed, it passes from the stomach and intestines into the blood, a process referred to as absorption. ⁽²⁸⁾ When the alcohol reaches the brain and nervous system, the characteristic signs of intoxication begin to show. ⁽²⁹⁾ The length of time necessary for the alcohol to be absorbed depends on a variety of factors, including the presence and type of food in the stomach, ⁽³⁰⁾ the person's gender, ⁽³¹⁾ the person's weight, ⁽³²⁾ the person's age, ⁽³³⁾ the person's mental state, ⁽³⁴⁾ the drinking pattern, ⁽³⁵⁾ the type of beverage consumed, ⁽³⁶⁾ the amount consumed, ⁽³⁷⁾ and the time period of alcohol consumption. ⁽³⁸⁾ At some point after drinking has ceased, the person's BAC will reach a peak. After the peak, the BAC will begin to fall as alcohol is eliminated from the person's body. The body eliminates alcohol through the liver at a slow but consistent rate. ⁽³⁹⁾

In 1932, Swedish chemist E.M.P. Widmark first calculated absorption and elimination rates in the body, and his work still represents the benchmark for other scientists' studies today. Widmark created what we know today as the "BAC curve," which represents the rise and fall of an individual's BAC as his body absorbs and eliminates alcohol. A reading from a single breath test will not reflect where the person is on his BAC curve. In other words, it will not indicate whether the person is in the absorption phase, at his peak, or in the elimination phase. ⁽⁴⁰⁾

So if a driver is tested while in the absorption phase, his BAC at the time of the test will be higher than his BAC while driving. If tested while in the elimination phase, his BAC at the time of the test could be lower than while driving, depending on whether he had reached his peak before or after he was stopped. Obviously, the greater the length of time between the driving and the test, the greater the potential variation between the two BACs.

To clarify, we are not addressing whether retrograde extrapolation is necessary in order for the State to prove a defendant guilty in a DWI case. ⁽⁴¹⁾ Nor do we address whether test results showing a defendant's BAC at some time after the alleged offense are admissible at trial in the absence of retrograde extrapolation. ⁽⁴²⁾ Our only concern today is whether McDougall reliably applied the science of retrograde extrapolation in Mata's trial.

We may take judicial notice of scientific literature not presented by either party at trial or on appeal. ⁽⁴³⁾ Our research revealed a number of articles on the subject. Mark Montgomery of the College of Public Health at the University of Florida and Mark Reasor of the Department of Pharmacology and Toxicology in West Virginia contend that retrograde extrapolation is reliable. ⁽⁴⁴⁾ They concede that "the scientific community is divided" on this issue. ⁽⁴⁵⁾ But they believe that extrapolation to a range of BACs can be accomplished reliably, as long as "justifiable assumptions are made that are based on sound principles of pharmacology, toxicology, and physiology." ⁽⁴⁶⁾ The authors set forth several hypothetical cases in which, given several known factors, they are able to calculate the subject's expected BAC at the time of driving. In each hypothetical case, the known factors include the subject's weight, the length of time in which drinking occurred, and the time at which the drinking stopped.

Other scientists are more cautious in their examination of reliability. Richard Watkins, Assistant Director of the Phoenix Crime Detection Lab, and Eugene Adler, a toxicologist for the Arizona Department of Public Safety Crime Lab, write that retrograde extrapolation is somewhat reliable. ⁽⁴⁷⁾ In their study on the effect of drinking on a full versus an empty stomach, they discovered that "the alcohol elimination rate was lower in the full stomach condition compared to the empty stomach condition. The difference was statistically significant." ⁽⁴⁸⁾ They concluded that it was possible to make estimates of BACs at some time removed from the breath test that were "sufficiently reliable." ⁽⁴⁹⁾ But they nevertheless cautioned that, "[h]owever useful such estimates may be in [DWI] cases, it should be remembered that the process of alcohol absorption is highly variable. The limitations and pitfalls associated with retrograde extrapolations are often not appreciated by laymen and the courts." ⁽⁵⁰⁾ The authors concluded that "[a]ny attempt at retrograde extrapolation should be made with caution, and performed by persons able to assess and discuss the applicability of a retrograde extrapolation to a particular situation." ⁽⁵¹⁾

Still others are even more wary of the science. Alan Jones, Ph.D., Associate Professor in the Department of Alcohol Toxicology at the University Hospital in Sweden, Kjell-Ake Jonsson, M.D. and Ph.D., Clinical Associate at the Department of Clinical Pharmacology and Internal Medicine at the University Hospital, and Aldo Neri, M.D., Research Technician at the Department of Alcohol and Drug Addiction Research at the Karolinska Institute in Stockholm, Sweden, point out the potential for error in the science. ⁽⁵²⁾ They write that retrograde extrapolation is a "dubious practice" and that expert testimony on the issue "requires careful consideration of the absorption kinetics of ethanol and the factors influencing this process." ⁽⁵³⁾ They explain that "[t]he absorption profile of ethanol differs widely among individuals, and the peak [BAC] and the time of its occurrence depend on numerous factors. Among other factors, the drinking pattern, the type of beverage consumed, the fed or fasted state, the nature and composition of foodstuff in the stomach, the anatomy of the gastroinstestinal canal, and the mental state of the subject are considered to play a role." ⁽⁵⁴⁾

These authors point out the limitations of the Widmark equations. In their study, they found that "rapid consumption of a moderate dose of ethanol on an empty stomach" resulted in a peak BAC that was "higher than would be expected . . according to Widmark calculations." ⁽⁵⁵⁾ They refer to this as the "overshoot effect." They explain that after this "early peak" there would be a "diffusion plunge." They state that "[i]f the apparent rate of elimination is calculated from the change in [BAC] between two time points immediately after an overshoot, that is, on a diffusion plunge, the results are abnormally high." ⁽⁵⁶⁾ Professor Jones and his colleagues conclude that "[t]he status of ethanol absorption in drunk drivers at the time of the offense is a more difficult question to tackle. In practice, this will depend on such circumstances as the previous drinking spree - the duration and quantities consumed - and the time lapse from the end of drinking to the time of arrest or the time an accident occurred." ⁽⁵⁷⁾ The authors "highly recommend" that such "speculation" be avoided by states defining the offense of driving drunk as having a certain BAC at the time of the breath test rather than at the time of the driving. ⁽⁵⁸⁾

P.R. Jackson, G.T. Tucker, and H.F. Woods, all of the University of Medicine and Pharmacology at Royal Hallmshire Hospital in Great Britain, agree that retrograde extrapolation should be used with caution. ⁽⁵⁹⁾ They state that the method used should "account for possible variations from several sources." ⁽⁶⁰⁾ The "major possible source of error," the authors state, is when "continuing absorption is ignored, as this can make a large difference to the shape and position of" the BAC curve. ⁽⁶¹⁾ Another source of error "arises from the unique status of the observed subject" because there is "generally no information as to his or her position in the population distribution of the parameters describing ethanol elimination." ⁽⁶²⁾ These authors conclude that, "in the absence of continuing absorption," reasonably reliable BAC estimates can be made. But "when absorption continues after drinking, especially when at a slow rate, backtracking calculations become markedly inaccurate." ⁽⁶³⁾

Jones and Neri, in another paper, agree that Widmark's method, while it can be "practical and reliable" in some instances, is not reliable "if alcohol is consumed together with a meal" because "the resulting blood-ethanol concentrations are underestimated." ⁽⁶⁴⁾

The complicating factors were discussed by Y. Al-Lanqawi of the Department of Pharmacology and Clinical Pharmacology at the University of Dundee Medical School, and his colleagues. ⁽⁶⁵⁾ He writes that "[i]ndividuals vary with respect to age, sex, body weight and lean body mass, all of which may effect the disposition of ethanol. In addition, the absorption of ethanol into the body may be variable and may continue over a long period of time." ⁽⁶⁶⁾ He found that retrograde extrapolations using the "mean elimination rate"observed in their study resulted in a rate of error of "essentially zero." ⁽⁶⁷⁾ He nevertheless noted that "the variability in the extrapolation error obtained in individual suspects with the mean elimination rate was considerable." ⁽⁶⁸⁾

In particular, Al-Lanqawi discovered that the potential rate of error increased as time went on. Indeed, "this variability was particularly large" when extrapolation back one hour or more was attempted. ⁽⁶⁹⁾ Finally, Al-Lanqawi and his colleagues found that the mean elimination rate in their study was 25% greater than the "frequently cited population mean rate" of .015. "When the latter slope was used in the back extrapolation procedure, a consistent underestimation in the plasma ethanol concentration was observed in the majority of the subjects." ⁽⁷⁰⁾

Finally, Edward Fitzgerald, a lawyer in Massachusetts, and Dr. David Hume, a professor of analytical chemistry at M.I.T., write that there are "serious questions about the reliability of any later single test as an indicator of an earlier BAC." ⁽⁷¹⁾ They explain that most extrapolations are based upon certain assumptions which simply are not true, including the assumption that a person's BAC will decline once drinking has stopped. In fact, they say, in most DWI cases, the person is arrested shortly after his last drink. ⁽⁷²⁾ His stomach has not yet absorbed all the alcohol, and his BAC will continue to rise for some time after arrest, resulting in a tested BAC which is higher than what it was at the time of the driving. Another assumption these authors criticize is the assumption that all people have an alcohol elimination rate of .015. While that is the average rate of elimination, the authors point out that the actual individual range is generally between .01 and .02, with values seen as high as .04 and as low as .006. ⁽⁷³⁾ They conclude that the attempt to engage in retrograde extrapolation is "fraught with difficulties." ⁽⁷⁴⁾

Few jurisdictions have considered the reliability of retrograde extrapolations. One reason for this is that many states have eliminated any need for retrograde extrapolation as a matter of law. Their statutes provide for a rebuttable presumption that an individual is guilty of DWI if his BAC at the time of the test is over the legal limit, assuming the test was conducted within a specified or reasonable time from the driving. ⁽⁷⁵⁾ In these states, a BAC of .10 or more at the time of the breath test would essentially prove the State's case; it is then the defendant's burden to prove that his BAC was lower than the legal limit at the time he was driving. In contrast, at the time of Mata's offense, our statute required the State to prove that the defendant had a BAC of .10 "while operating a motor vehicle." ⁽⁷⁶⁾

We found only two courts that have touched upon the reliability of retrograde extrapolation. An Arizona appellate court has stated in a footnote, without citation to authority, that the science of retrograde extrapolation "has achieved general acceptance in the scientific field." ⁽⁷⁷⁾ The Alabama Court of Criminal Appeals, on the other hand, appears to disagree. That court, while finding the issue was not preserved for appeal, cautioned that its opinion should not be interpreted as "tacit approval" for the use of retrograde extrapolations. ⁽⁷⁸⁾ Citing a law review article, the Court stated that

[c]areful analysis of these studies indicates that retrograde extrapolation is an unreliable method of determining a defendant's condition at the time of operation. The inadequacies of retrograde extrapolation extend beyond mere technical inaccuracies to problems which are inherent in the basic premises and calculations of this technique. These inadequacies render retrograde extrapolation inherently untrustworthy and therefore inappropriate for use as evidence to convict drunk drivers." ⁽⁷⁹⁾

Under Kelly we may also examine the clarity with which the expert explained the underlying scientific theory and technique. Initially, we note that McDougall is a popular expert in Bexar County DWI prosecutions. ⁽⁸⁰⁾ The court of appeals found that he has "impeccable qualifications, including extensive personal observations of the alcohol absorption and elimination process." But we find nothing in the record supporting any "extensive personal observations of the alcohol absorption and elimination process" or any ability to perform retrograde extrapolations. It was inappropriate for the court of appeals to rely on its opinion in Hartman as evidence of McDougall's qualifications in this case. Nevertheless, Mata conceded at trial that McDougall was qualified to perform retrograde extrapolations generally; he complained only that McDougall could not reliably apply the technique to Mata.

In reviewing the testimony, we find that, while McDougall's testimony indicates a general understanding of the BAC curve and the concepts of absorption and elimination, it reflects failure to explain the theory to the court with any clarity. First, McDougall contradicted himself within his testimony. He testified, at various times, that the absorption phase could not last longer than an hour, or an hour and a half, or two hours. ⁽⁸¹⁾ He testified that Mata's BAC at the time of driving would have been at least .08, then .12 or .13. He testified that Mata's high-end BAC would have been .25, then .23. ⁽⁸²⁾

In addition, McDougall seemed unaware of the subtleties inherent in any retrograde extrapolation calculation. He initially testified that the average elimination rate was approximately .02, but most of the literature suggests it is .015, ⁽⁸³⁾ and McDougall did not acknowledge the difference (a difference of 33%) or explain it. He later testified that elimination rates could range from .018 to .025, but he nevertheless stated that he used a .02 or .03 elimination rate in his calculations. McDougall's .03 elimination rate is not only outside his own stated range, it is twice the generally-accepted elimination rate. The dissent believes McDougall "clearly misspoke" ⁽⁸⁴⁾ in referring to elimination rates, but we find nothing in the record to indicate this. The dissent also suggests that McDougall's erroneous elimination rate is irrelevant since scientific literature also suggests that the .015 rate is too slow, and "recent literature indicates that anything from .015 to .02 is . . . normal." ⁽⁸⁵⁾ But our focus today is not on what the proper elimination rate is or should be. We merely observe that McDougall's testimony was inconsistent and failed to acknowledge scientific evidence to the contrary. This is just another factor we consider in determining whether McDougall explained the science to the court with clarity.

Moreover, McDougall's math appeared to be in error. At the suppression hearing, McDougall calculated a high-end BAC for Mata as being .25 on an empty stomach and .22 on a full stomach. He testified that he used a .02 elimination rate. But if Mata's BAC was .19 at the time of the test, and McDougall was using the average elimination rate of .02, McDougall should have calculated a high-end BAC of .23 at the time of the offense, which was two hours before the test. McDougall did not explain the difference in the numbers. At trial, he corrected his math and calculated a high-end BAC for Mata as being .23. McDougall's changing his calculations from the pretrial hearing to the trial is a factor we can consider in determining the reliability of his testimony.

McDougall's inconsistent statements did not stop here. Regarding the number of drinks required to increase from below a .10 to a .19 in two hours, McDougall first testified that five drinks would be sufficient, then changed his testimony to seven to twelve drinks. ⁽⁸⁶⁾

Finally, McDougall conceded that his calculations were based on a "normal drinking pattern." He acknowledged that the "chug-a-lug" situation was not a normal drinking pattern, and he also acknowledged that he had no idea how much Mata had had to drink, what he had had to drink, or how long Mata had been drinking. As a result, McDougall essentially conceded that all of his calculations as to Mata were speculative. ⁽⁸⁷⁾

Our study of retrograde extrapolation leads us to several conclusions. Initially, we recognize that even those who believe retrograde extrapolation is a reliable technique have utilized it only if certain factors are known, such as the length of the drinking spree, the time of the last drink, and the person's weight. ⁽⁸⁸⁾ McDougall knew none of those factors, or any other individual characteristics of Mata, for that matter. In addition, there appears to be general disagreement on some of the fundamental aspects of the theory, such as the accuracy of Widmark's formulas, ⁽⁸⁹⁾ whether a standard elimination rate can be reliably applied to a given subject, ⁽⁹⁰⁾ and the effect that food in the stomach has on alcohol absorption. ⁽⁹¹⁾ Nevertheless, given the studies, other concepts seem indisputable, including that multiple tests will increase the ability to plot a subject's BAC curve, a test nearer in time to the time of the alleged offense increases the ability to determine the subject's offense-time BAC, and the more personal information known about the subject increases the reliability of an extrapolation.

We believe that the science of retrograde extrapolation can be reliable in a given case. The expert's ability to apply the science and explain it with clarity to the court is a paramount consideration. In addition, the expert must demonstrate some understanding of the difficulties associated with a retrograde extrapolation. He must demonstrate an awareness of the subtleties of the science and the risks inherent in any extrapolation. Finally, he must be able to clearly and consistently apply the science.

The court evaluating the reliability of a retrograde extrapolation should also consider (a) the length of time between the offense and the test(s) administered; (b) the number of tests given and the length of time between each test; and (c) whether, and if so, to what extent, any individual characteristics of the defendant were known to the expert in providing his extrapolation. These characteristics and behaviors might include, but are not limited to, the person's weight and gender, the person's typical drinking pattern and tolerance for alcohol, how much the person had to drink on the day or night in question, what the person drank, the duration of the drinking spree, the time of the last drink, and how much and what the person had to eat either before, during, or after the drinking.

Obviously, not every single personal fact about the defendant must be known to the expert in order to produce an extrapolation with the appropriate level of reliability. As the Kentucky Supreme Court has recognized, if this were the case, no valid extrapolation could ever occur without the defendant's cooperation, since a number of facts known only to the defendant are essential to the process. ⁽⁹²⁾ If the State had more than one test, each test a reasonable length of time apart, and the first test were conducted within a reasonable time from the time of the offense, then an expert could potentially create a reliable estimate of the defendant's BAC with limited knowledge of personal characteristics and behaviors. In contrast, a single test conducted some time after the offense could result in a reliable extrapolation only if the expert had knowledge of many personal characteristics and behaviors of the defendant. Somewhere in the middle might fall a case in which there was a single test a reasonable length of time from the driving, and two or three personal characteristics of the defendant were known to the expert. We cannot and should not determine today the exact blueprint for reliability in every case. Suffice it to say that the factors must be balanced.

In this case, the inconsistencies in McDougall's testimony prevented him from explaining the science to the court with any clarity. Regarding the other factors, there was only one test of Mata's BAC, and it occurred over two hours after the alleged offense. This is a significant length of time and seriously affects the reliability of any extrapolation. Add to that the fact that McDougall did not know one single personal characteristic of Mata - he did not know whether Mata had eaten anything that night and if so, how much; how much Mata had had to drink; what Mata had been drinking; when Mata's last drink was; the length of Mata's drinking spree; or even Mata's weight.

We find that the factors in this case weigh against a finding of reliability. Even the scientists who find retrograde extrapolations reliable would require more known quantities than what McDougall had in this case. Given McDougall's inconsistent testimony, along with the single breath test conducted over two hours after the driving, we conclude that, in this case, the State failed to prove by clear and convincing evidence that McDougall's retrograde extrapolation was reliable.

We conclude the trial court abused its discretion in admitting that part of McDougall's testimony pertaining to the extrapolation of Mata's BAC. We remand this cause to the court of appeals for a determination of harm under Appellate Rule 44.2.

DATE DELIVERED:

PUBLISH

Suppression Hearing

Q: And is it conceivable then that over a period of two hours, if a person had a very full stomach, that it is going to take awhile for them to reach their peak of alcohol concentration?

A: Yes.

Q: Regardless of how much they have had to drink, if they have a full stomach, you know, lots of meat and potatoes, it is just going to take some time for it to get up there?

A: Yes.

Q: Two hours maybe?

A: I think an hour and a half is adequate for a complete absorption of the drinks. It doesn't take two hours to digest a stomach full of food. So all of the alcohol would be absorbed in the hour and a half.

Q: An hour and a half?

A: Yes, and it would be less than that if it were not a full stomach.

Voir Dire

Q: Consequently, if all you have is the test result and you have the arrest time, the only angle that [you] can draw, nothing [sic] that the peak occurred after the test, is the angle from zero, at the time of arrest, to the point of the chart on the test?

A: No, I don't agree with that.

Q: Okay. Tell me why you disagree with that?

A: I don't know of any situation where a person would be in the absorption phase after an hour. So let's assume that he was arrested and he drank just before he was arrested. He could be rising for an hour and then he would peak. I don't know of any instance where he would peak two hours later.

Q: Okay. Are you saying that an individual who has a full stomach on him, a meal of potatoes, would not peak more than two hours?

A: If the person had an empty stomach he would peak in about half an hour to 45 minutes. And if he is on a full stomach it would be somewhere around an hour to an hour and a half at the very most.

Q: So are you saying that it is not possible?

A: For him the peak two hours later, no. He has to peak before two hours. He has to peak in about an hour.

Q: Well, I am just going to show you this chart that we have gone over. I know you have seen it. This is that chart based upon the U.S. Department of Transportation, Alcohol and Safety study and we referred to this. And you will note that the second or the lower of the two was based upon, you know, the same amount of alcohol with an individual immediately after eating a meal of potatoes and it shows that this person is peaking right at about two hours, doesn't it?

A: Yes, it does.

Q: Now, would you say you would disagree with that?

A: Here is the problem. On this particular chart our peak is at a .04. That is a very limited alcohol concentration.

Q: It is lower.

A: So it is a very small amount of alcohol in the amount of potatoes. It varies greatly with more alcohol. It would not be delayed as much for a greater amount of alcohol. Potatoes can't dilute the alcohol any more than potatoes can dilute the alcohol.

Q: I agree, but what it does is it slows down the system and it slows down the alcohol getting into the gut, basically?

A: Yes, but that is also dependent upon the concentration of the alcohol and the amount of the alcohol. This is a limited amount of alcohol. This is just two ounces.

Q: Well, here it says two ounces of pure alcohol and 8 ounces of water, equivalent to 5 ounces of 80 proof Vodka. Would you agree with that?

A: Yes, sir.

Q: This is an extreme. Okay. I agree with you that that is extreme on one end. But would you agree with me that there is also an extreme on the other end that in some individuals, that they could have been drinking for a long period of time and had achieved a high level of alcohol in their system, perhaps several hours before this, and then continued to rise for a period of time?

A: For an hour to an hour and a half. That is the most.

Q: Yes.

A: Anything more than that I don't believe. It has not been my experience and I have tested thousands of individuals and it has not been my experience that their alcohol concentration rises more than an hour and a half after their last drink.

Q: Even though this survey might suggest otherwise?

A: That is not a survey. That is an experiment done with a gulp of five ounces on potatoes and then you measure the alcohol concentration.

Q: Right. And one was done on a full stomach and one was done on an empty stomach and there was a variability. You will agree with that?

A: I am not going to accept the two hours as an absolute, no. In my experience with all of the studies that I have read it is between half an hour to an hour and a half to reach a peak, after drinking on a full or empty stomach. Within the limitations of that graph they could have been [talking] about an hour and a half and it looks like two hours. It is not very significant.

Trial

Q: What would [it be] if the peak in this situation came after the test? How would that change your range?

A: I don't believe it would have occurred more than an hour and a half after the arrest. I don't know of an instance, in my own experience, I don't know of anybody that has reached a peak more than an hour and a half after the last drink.

Q: You have not come across that situation?

A: No, but I must admit that I have not tested a chug-a-lugger.

Suppression Hearing

Q: And for the alcohol to be metabolized and processed out of the system, we are talking about - that is when we talk about this absorption/elimination. In other words, it is going to rise fairly rapidly in terms of absorption and then it will dissipate or eliminate at .02 percent grams per 210 liters, is that correct?

A: Yes.

Q: Approximately?

A: Yes, sir.

Voir Dire

Q: Now, alcohol . . . eliminates at about a .02 per hour, something like that?

A: Yes, sir.

Q: What you are saying then as I understand it is, that you take the time of the test, and following the .02 elimination rate, you would draw a chart up to, say the time of the arrest and past it to a potential peak, is that right?

A: Yes, sir.

Q: Now, the reason we can testify about this range right here, or you can, is that you have two knowns. You have the known breath test and you have the known standard elimination rate which is pretty standard among all persons. Will you agree with me?

A: Yes, sir.

Trial

Q: Could you draw for the jury a graph in the form of an XY axis, showing us that range?

A: By XY, do you mean alcohol concentration versus time?

Q: Right.

A: I guess I can.

Q: Your Honor, may the witness step down to the diagram?

Court: Yes.

A: This is time and this would be alcohol concentration. The only time that I really have is the time of the test. And this is the only real test and I will put that at .19 up here, a .193 and the time of the test. Alcohol is eliminated at the rate of one drink per hour. Two hours before it could have been as high as a .23. And then, let's say it was at a .02 or .03 elimination per hour, it would make it at .06 less than this - a .13. His alcohol concentration would have been somewhere in that range (indicating on the chart).

Q: Now, let's talk about the elimination rates. Is that a fairly standard number in all persons' elimination rate?

A: .02 per hour of alcohol concentration for a 150 pound person is pretty stable. It ranges from [.018 to a .025]. A .02 is very close to normal.

Q: So we are not going to see any large shift in elimination rates over a period of time. It is going to be right around a .02?

A: If all he is doing is eliminating and if he has completed absorption.

Q: And that will probably be pretty close to the same for the average person throughout?

A: Yes, sir.

Q: So given an elimination rate of .02, if we assume, based upon your chart, that the individual peaked prior to the test then you can draw sort of a line back showing, you know, on this chart a rise of .02 per hour?

A: If he peaked two hours before or if he peaked prior to that.

Q: Right. And that is what this line would represent because these are known things, right? We know that we have a breath test score of a .193 or whatever it was and we know about a .02 elimination rate, over time. Now, what about absorption rates?

A: They vary more because of the amount of food in the stomach and also the amount of the concentration of the alcohol being consumed.

Suppression Hearing

Q: You are not testifying, today, about what his conditions were while he was in actual, physical control of a motor vehicle, are you?

A: Yes, I believe I was.

Q: What are you saying about his condition when he was in actual, physical control of a motor vehicle?

A: That he was impaired and that he had lost the normal use of his mental or physical faculties at the time that he was in control of a motor vehicle.

Q: You are not saying that he had a breath alcohol content of .10 or higher, are you?

A: No, I am saying that he had an alcohol concentration of at least a .08 or higher.

Q: So you will state that in your opinion, at a .08, it means an individual has lost the normal use of their mental or physical faculties?

A: Yes.

Q: If that is the result of ingesting alcohol?

A: Yes.

Q: Now, however, you will agree that the State, in terms of the breath alcohol testing program, the State minimum is a .10?

A: Yes, I will.

Q: All right. But based upon the information that you have, you just really can't give us a figure for this individual based upon that test taken two hours later, can you?

A: I can't give you an individual number. I can give you a range, but not a number.

Q: And they will ask - so I am going to ask you, what is that range?

A: I think it is possible for the person to be as much as a .06 less than a .19, two hours and ten minutes earlier. It could have been as low as a .12 or a .13.

Q: I am sorry, a what?

A: A .12 to a .13 at the low end, and somewhere about a .25 at the high end, somewhere in between there.

Q: So we are talking about a range here of .13?

A: To a .25, yes, two hours previous.

Q: A range then including 12 points?

A: Yes.

Q: So it is a 12 point range that you will testify to?

A: Yes.

Q: Are you saying that there is not a possibility that it could be lower than 10, a .10?

A: No. There is an extreme situation where it is possible that it could be less than a .10.

A: He could have been as low as maybe a .16 on a full stomach and not quite as high [as] a .25, maybe a .21 or .22.

Trial

Q: A male blows a .193 two hours after they were stopped, two hours after we know at least that they had stopped drinking two hours prior. Can you give us a range as to what his alcohol concentration would have been at the time he was stopped?

A: I believe the high side would have been somewhere around a .04 higher than that, which would be a .23 and then his body would have eliminated the equivalent of two drinks during those two hours. On the low side, his body could have absorbed two or three drinks and as much as a .04 or a .06, reached a peak and either maintained that level or decreased a little bit until the time of arrest, which would make him, I guess, at the low side of a .13. So I would say somewhere between a .13 and a .23.

Q: Mr. McDougall, can you conceive of any factuall [sic] facts that would support an alcohol concentration below a .10 at the time that he was stopped?

A: Yes.

Q: Okay. What would that be?

A: If just before he was stopped if he chug-a-lugged a lot of hard liquor, hard alcohol.

Q: Now, in this particular case you have testified that you thought there was a range of .13 to .23?

A: Yes, sir.

Q: However, you previously testified that the range could have been from a point .12 to a point .25. Is that correct?

A: That is very possible. I don't know if we used the same assumptions.

Q: I don't want to put words in your mouth and that is why I marked this and I want to make sure that you are the same George Allen McDougall that testified in this case back in June of 1993. There is no other George Allen McDougall, is there, testifying?

A: Well, my father, but he is not testifying.

Q: Okay. Then I will submit to you that this was your testimony in response to my questions.

A: It could have been as low as a .12 or a .13. That is what I testified to here, and to a .25, yes. Yes, what did I say previously, a .23? It is possible I could have testified to that.

Voir Dire

Q: Let's assume that we do have a .19, and we have that .19 an hour and 55 minutes, tested, after the arrest. And that is all the information that you have?

A: That is correct.

Q: Now, not nothing [sic] when that peak occurs, before or after, why is that figure any different in factoring your theoretical analysis of range than the figure that we are dealing with here?

A: Because you are assuming that he chug-a-lugged twice as much alcohol to get to that level.

Q: Perhaps. See, I am not assuming that. I think that you have to assume that. You have to assume many different variables is what I am saying?

A: I don't think that I have to assume that. I think that I can assume a normal drinking pattern. And if you want to raise the issue of the possibility of a chug-a-lug, then we can talk about it. But I don't assume that as normal. I am sorry, but I don't assume that chug-a-lugging alcohol is normal. In my drinking experience that is very unusual.

Q: Well, that is true?

A: I am not going to give you an example based on a very unusual situation.

Q: But you will agree with me, won't you, that there are persons that drink that much?

A: Who drink that much, yes, but not chug-a-lug. But yes, there are people that drink that much. Yes, I will agree with you.

Q: Mr. McDougall, the point I am trying to make is that you don't know anything about Mr. Mata or his drinking habits and what he had to drink on that night, so you cannot factor in any of these variables to determine the angle, you know, back which we can recreate this range.

A: No, but I can state the parameters under which I am making my hypothetical answer and you can live with that or you can illustrate the limitations of my parameters. But I think my parameters are very well within the bounds of what you are going to find.

Q: But what you are saying is you are basing your parameters on some hypothetical individual?

A: Yes, on a normal drinking individual.

Q: But that may mean something totally different to this jury than it does to you because you are a breath test supervisor?

A: But I am not adverse to explaining what my parameters are and they can assume those parameters and they can understand the parameters I am using. I am not using any magic.

Q: No, and I agree with that, but wouldn't you agree that your experience with drinking and with the capacities of individuals and the vari[e]ties within the population, that your experience is far different than these people on this jury. You have a much broader range of information.

A: And with that knowledge a chug-a-lug is even more exceptional. To state that a chug-a-lug - which is the only possibility that we are talking about here, is that he actually gulped down a quart of liquor just before the arrest and then he went driving.

Q: Right, that is one of the possibilities that you have to consider?

A: That is one that would not be considered normal. I am sorry.

Q: Well, you say normal but you have to apply that opinion to Mr. Mata. Are you saying to this court that you would limit your testimony not to Mr. Mata but to a hypothetical individual who is not him?

A: No, I would explain what is normal and if you were to bring up other parameters then I will tell you what those other parameters will do to an individual. I will explain it and lay it all out.

Trial

Q: You say when he could have peaked within a normal drinking pattern - is that what you just said?

A: Yes.

Q: But you don't know what the drinking pattern is of Mr. Mata?

A: That is correct.

Q: So that range is based upon an assumption that doesn't apply to Mr. Mata, is that correct?

A: It may not, yes, sir.

Q: So that range that we have there may not be his range?

A: That is possible, yes.

Q: This is a hypothetical based upon your person, the average person?

A: Yes.

Q: Mr. McDougall, can you conceive of any factuall [sic] facts that would support an alcohol concentration below a .10 at the time that he was stopped?

A: Yes.

Q: Okay. What would that be?

A: If just before he was stopped if he chug-a-lugged a lot of hard liquor, hard alcohol.

Q: About how much? Can you tell us at this point about how much he would have to chug-a-lug to blow a .10 and up to a .19?

A: Again I need to know his body weight. If we are talking about a 150 pound person then that is five extra drinks, from a .10 to a .19. He would have to consume at least five shots of alcohol for 150 pounds and proportionally more if he weighs more than 150 pounds.

Q: So a 150 pound person, it would be about five shots?

A: Five shots of hard liquor.

Q: And how many ounces constitute the shot?

A: An ounce and a quarter.

Q: And that would be right before the person was stopped?

A: That would be so that the person had consumed the alcohol but it hadn't had a chance to get into his blood stream yet, which means very close to the time that he was arrested. He would have to very quickly be drinking and be very quickly arrested thereafter.

Q: Are you assuming in this fact situation as in a .09, or are you assuming he is a .00 at the time that he had the five shots?

A: No, he would be at five shots in addition to a .10. Let's assume that he is at .10 and to reach a .19 he would have to consume at least five shots if he weighed 150 pounds.

Q: If the person was stone, cold sober and had not a drop to drink at the time that they were stopped, they would have to drink significantly more than the five shots?

A: He would have to consume about 12 shots, from .00 to .19. Yes, 12 shots of hard liquor.

Q: Based upon the factual hypothetical that we talked about earlier, what are you considering the abnormal drinking pattern?

A: Chug-a-lugging a bottle of hard liquor and gulping it down as fast and as much as you can in a short period of time.

Q: And this hypothetical that we talked about, him going from a [.10 to a .19] ⁽⁹³⁾, that it had to take him five shots just before he got arrested, do you agree with that?

A: No. Those aren't the same. Chug-a-lugging is drinking much more than that. It would be seven, eight, nine, ten, or eleven or twelve drinks consumed very quickly.

Q: For him to come from below a .10 up to a .19?

A: Yes, he would have to have much more than that.

Q: And an individual from that fact situation, this person that blows a .19, two hours after arrest, they have above a .10 at the time that they are stopped?

A: I believe so, yes.

1. 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993).

2. 824 S.W.2d 568 (Tex. Crim. App. 1992).

3. See Appendix A.

4. See Appendix B.

5. See Appendix C.

6. See Appendix A.

7. See Appendix B.

8. See Appendix D.

9. See Appendix C.

10. See Appendix C.

11. See Appendix E.

12. See Appendix B.

13. See Appendix D.

14. See Appendix A.

15. 2 S.W.3d 490 (Tex. App. - San Antonio 1999, pet. ref'd) (en banc).

16. Mata v. State, 13 S.W.3d 1, 2 (Tex. App. - San Antonio 1999) (quoting Hartman, 2 S.W.3d at 493-94).

17. Id. at 4-37 (Cadena, J., dissenting).

18. Id. at 2 n. 1.

19. Jackson v. State, 17 S.W.3d 664, 670 (Tex. Crim. App. 2000); Jordan v. State, 928 S.W.2d 550, 554-55 (Tex. Crim. App. 1996).

20. Jackson, 17 S.W.3d at 670.

21. Kelly, 824 S.W.2d at 573.

22. Ibid.

23. Jordan, 928 S.W.2d at 554-55, quoting Daubert, 509 U.S. at 593-95, 113 S.Ct. at 2797.

24. Id. at 555.

25. Ibid.

26. Hinojosa v. State, 4 S.W.3d 240, 250-51(Tex. Crim. App. 1999); Griffith v. State, 983 S.W.2d 282, 287 (Tex. Crim. App. 1998).

27. See Lawrence Taylor, Drunk Driving Defense § 5.2 (5^th ed.) (2000).

28. Nat'l Inst. on Alcohol Abuse & Alcoholism, Alcohol Alert, "Alcohol Metabolism," No. 35 (Jan. 1997).

29. Jennifer Pariser, Note: In Vino Veritas: The Truth About Blood Alcohol Presumptions in State Drunk Driving Law, 64 N.Y.U.L. Rev. 141, 146 (1989), citing S. Brent & S. Stiller, Handling Drunk Driving Cases, § 4:2 (1985); Edward Fitzgerald & Dr. David Hume, The Single Chemical Test for Intoxication: A Challenge to Admissibility, 66 Mass. L. Rev. 23, 28 (1981).

30. Pariser, 64 N.Y.U.L. Rev. at 147, citing 2 R. Erwin, Defense of Drunk Driving Cases Criminal/Civil § 15.04[1][b][i], at 15 (3d ed. 1988); Rodney Gullberg, Variation in Blood Alcohol Concentration Following the Last Drink, 10 J. of Police Science & Admin. 289 (1982); Alan Jones et al., Peak Blood-Ethanol Concentration and the Time of Its Occurrence After Rapid Drinking on an Empty Stomach, 36 J. of Forensic Science 376, 381 (1991).

31. NIAAA Alcohol Alert, "Alcohol Metabolism," No. 35 (Jan. 1997).

32. Y. Al-Lanqawi et al., Ethanol Kinetics: Extent of Error in Back Extrapolation Procedures, 34 British J. of Clinical Pharmacology 316, 320 (1992).

33. Ibid.

34. Jones et al., 36 J. of Forensic Science at 381.

35. Ibid.

36. Ibid.

37. Gullberg, 10 J. of Police Science & Admin. at 289.

38. Ibid

39. Pariser, 64 N.Y.U.L. Rev. at 149.

40. Pariser, 64 N.Y.U.L. Rev. at 152, citing W. Frajola, Defending Drinking Drivers 16.1 (1985).

41. Compare, e.g., McCafferty v. State, 748 S.W.2d 489 (Tex. App. - Houston [1^st Dist.] 1988, no pet.) (evidence insufficient without retrograde extrapolation); State v. Ladwig, 434 N.W.2d 594 (S.D. 1989) (same); Desmond v. Superior Court, 161 Ariz. 522, 779 P.2d 1261 (Ariz. 1989) (same); Allman v. State, 728 N.E.2d 230 (Ind. Ct. App. 2000) (same); State v. Rollins, 141 Vt. 105, 444 A.2d 884 (Vt. 1982) (same); State v. Geisler, 22 Conn. App. 142, 576 A.2d 1283 (Conn. App. Ct.) (same), appeal denied, 215 Conn. 819, 576 A.2d 547 (Conn. 1990), vacated on other grounds, 498 U.S. 1019, 111 S.Ct. 663, 112 L. Ed. 2d 657 (1991); with Haas v. State, 597 So.2d 770 (Fla. 1992) (evidence sufficient without retrograde extrapolation); Sullivan v. State, 517 N.E.2d 1251 (Ind. Ct. App. 1988) (same); People v. Mertz, 68 N.Y.2d 136, 497 N.E.2d 657, 506 N.Y.S.2d 290 (N.Y. 1986) (same); Commonwealth v. Slingerland, 358 Pa. Super. 531, 518 A.2d 266 (Pa. Super. Ct. 1986) (same); Comm. v. Wirth, 936 S.W.2d 78, 84 (Ky. 1996) (same); State v. Kubik, 235 Neb. 612, 456 N.W.2d 487 (1990) (same).

42. See, e.g., Mireles v. Dep't of Public Safety, 9 S.W.3d 128 (Tex. 1999) (test results admissible without retrograde extrapolation); State v. Barber, 42 Conn. App. 589, 681 A.2d 348 (Conn. Ct. App. 1996) (same); Desmond, supra (same).

43. Emerson v. State, 880 S.W.2d 759, 764-65 and 765 n.1. (Tex. Crim. App. 1994).

44. Mark Montgomery & Mark Reasor, Retrograde Extrapolation of Blood Alcohol Data: An Applied Approach, 36 J. of Toxicology and Envtl. Health 281-292 (1992).

45. Id. at 282.

46. Id. at 288.

47. Richard Watkins & Eugene Adler, The Effect of Food on Alcohol Absorption and Elimination Patterns, 38 J. of Forensic Science 285-291 (1993).

48. Id. at 288.

49. Id. at 290.

50. Ibid.

51. Ibid.

52. Jones et al., 36 J. of Forensic Science 376-385 (1991).

53. Id. at 377.

54. Id. at 381.

55. Ibid.

56. Id. at 381-83.

57. Id. at 384.

58. Ibid.

59. P.R. Jackson et al., Back-tracking Booze with Bayes - the Retrospective Interpretation of Blood Alcohol Data, 31 British J. of Clinical Pharmacology 55-63 (1991).

60. Id. at 61.

61. Id. at 61-62.

62. Id. at 62.

63. Ibid.

64. A.W. Jones & A. Neri, Reinvestigation of Widmark's Method for Quantitative Evaluation of Blood-Ethanol Profiles: Influence of Alcohol Dose and Mode of Drinking, 33 Clinical Chemistry 1469 (1987).

65. Al-Lanqawi et al., 34 British J. of Clinical Pharmacology at 321.

66. Id. at 320.

67. Ibid.

68. Ibid.

69. Ibid.

70. Ibid.

71. Fitzgerald & Hume, 66 Mass. L. Rev. at 28 (1981).

72. Id. at 31.

73. Id. at 31.

74. Id. at 32.

75. See, e.g., Cal. Veh. Code § 23152(b); Nev. Rev. Stat. § 484.381(1); Ind. Code Ann. § 9-11-4-15; Iowa Code Ann. § 321J.2(7); Minn. Stat. §§ 169.121 subd. 2.

76. Former Tex. Rev. Civ. Stat. Art. 6701l-1 (repealed 1995), now located at Tex. Penal Code § 49.04.

77. Ring v. Taylor, 141 Ariz. 56, 69 n.6, 685 P.2d 121, 134 n.6 (Ariz. Ct. App. 1984).

78. Smith v. Tuscaloosa, 601 So.2d 1136, 1140 (Ala. Crim. App. 1992).

79. Ibid., quoting E. Abbott, "One for the Road" - The Reliability of Retrograde Extrapolation and the Implications for Vermont Statutes, 16 Vt.L. Rev. 395, 397 (1991).

80. See Coward v. State, 993 S.W.2d 307 (Tex. App. - San Antonio 1999, no pet.); Mireles v. Texas Department of Public Safety, 993 S.W.2d 426 (Tex. App. - San Antonio 1999), aff'd, 9 S.W.3d 128 (Tex. 1999); State v. Reed, 888 S.W.2d 117 (Tex. App. - San Antonio 1994, no pet.); Kapuscinski v. State, 878 S.W.2d 248 (Tex. App. - San Antonio 1994, pet. ref'd); State v. Krager, 810 S.W.2d 450 (Tex. App. - San Antonio 1991, pet. ref'd); State v. Kost, 785 S.W.2d 936 (Tex. App. - San Antonio 1990, pet. ref'd); Douthit v. State, 739 S.W.2d 94 (Tex. App. - San Antonio 1987, no pet.); Harrell v. State, 725 S.W.2d 208 (Tex. Crim. App. 1986).

81. See Appendix A.

82. See Appendix C.

83. See Fitzgerald & Hume, 66 Mass. L. Rev. at 31 (average elimination rate is .015); Pariser, 64 N.Y.U.L. Rev. at 152 n. 76, citing Watson & Batt, Prediction of Blood Alcohol Concentrations in Human Subjects: Updating the Widmark Equation, 42 J. Stud. Alchol 547 (1981); Taylor, Drunk Driving Defense § 5.2.3 (Widmark's research indicates average elimination rate of .015).

84. Post, slip op. at 4 (Keller, P.J., dissenting).

85. Id., slip op. at 3.

86. See Appendix E.

87. See Appendix D.

88. Montgomery & Reasor, 36 J. of Toxicology and Envtl. Health at 287-88.

89. Compare Jones et al., 36 J. of Forensic Science at 381 (consumption of alcohol on empty stomach results in rates different from Widmark equations); Jones & Neri, 33 Clinical Chemistry at 1469 (consumption of alcohol on full stomach results in rates different from Widmark equations); with Montgomery & Reasor, 36 J. of Toxicology and Envtl. Health at 282 (finding retrograde extrapolations generally reliable).

90. Compare Watkins & Adler, 38 J. of Forensic Science at 288 (standard elimination rate cannot apply because elimination rates vary depending on full or empty stomach, and variation is statistically significant); Pariser, 64 N.Y.U.L. Rev. at 152 n. 76 (individuals will have higher or lower elimination rates than standard elimination rates); with Montgomery & Reasor, 36 J. of Toxicology and Environmental Health at 282 (finding retrograde extrapolations generally reliable).

91. Compare Watkins & Adler, 38 J. of Forensic Science at 288 (food does not affect the time it takes to reach peak BAC); with Jones et al., 36 J. of Forensic Science at 381 (food does affect the time it takes to reach peak BAC).

92. Wirth, 936 S.W.2d at 84.

93. The reporter's record actually states ".01 to a .09" at this point. But nowhere in the "hypothetical . . . talked about earlier" did these numbers occur. Instead, the "hypothetical talked about earlier" concerned increasing from a .10, the legal limit, to a .19, Mata's BAC. Moreover, in the prosecutor's very next two questions, the reporter's record reflects the numbers .10 and .19. The reference to ".01 to a .09" is either a typographical error or a misstatement by the prosecutor. Either way, the record indicates that both the prosecutor and McDougall understood the hypothetical as referring to the numbers .10 and .19.