The opinion of the court was delivered by
Abbott, J.:
This is a direct appeal by the defendant, Edwin T. Witte, from his conviction of driving while under the influence of alcohol, contrary to K.S.A. 8-1567(a)(1).
Although the defendant raises other issues, the primary and dispositive issue is whether the trial court erred in admitting evidence of the horizontal gaze nystagmus (HGN) test administered by a law enforcement officer and, if so, whether it was harmless error.
In the early morning hours of June 1, 1990, Deputy Ron Goodwyn was traveling his normal patrol route. Shortly before 2 a.m., after noticing a vehicle swerving back and forth, he stopped the vehicle. The deputy informed Witte, the driver of the vehicle, that he had been stopped because of his erratic driving, including crossing the center line and hitting the curb.
Goodwyn testified that Witte said if he had problems driving, it was due to the condition of the car. Witte testified this was the first time he had driven the car. He stated the steering was loose and the car pulled to the left. Sheldon Birmingham, the car s owner and a passenger in the car, testified the car did not hit the curb and he did not notice the car weaving. Birmingham also stated there was “a little play in the steering wheel” and people who had driven his car complained about the loose steering.
Deputy Goodwyn testified he detected a strong odor of alcohol and noticed that Witte’s eyes were bloodshot and watery. Witte admitted he had drunk two beers.
The deputy had Witte perform field sobriety tests, namely, the HGN, the walk and turn, and the one-leg stand. According to Goodwyn, Witte failed the field sobriety tests. Witte testified that after he completed the three tests, Goodwyn told him he had passed. Witte also stated that Goodwyn said he would run a check on Witte’s driver’s license and, if there were no problems, Witte would be free to go. Birmingham testified that when Witte returned to the car, Witte confirmed he had passed the tests. [315]*315Birmingham stated the deputy then came by the car and told them they were free to go after he checked their licenses.
Deputy Goodwyn ran a check of Witte’s driver’s license, which was suspended. Goodwyn arrested the defendant and transported him to the Sedgwick County Jail. Witte then was given a breath test, which registered a .103 alcohol concentration at 3:22 a.m. The defendant was charged with driving with an alcohol concentration in his blood or breath of .10 or more, contrary to K.S.A. 8-1567(a)(l) and (2); one count of driving with a suspended driver’s license, contrary to K.S.A. 8-262(a)(l); one count of lane straddling, contrary to K.S.A. 8-1522(a); and one count of driving left of center, contrary to K.S.A. 8-1514(a).
Witte pled guilty to driving with a suspended driver’s license. The State dismissed the lane-straddling charge on the basis that it merged with the charge of driving left of center. After a jury trial on October 8-11, 1990, the jury acquitted Witte of the charge of driving left of center and convicted him of driving with a concentration of alcohol in his breath of .10 or more, contrary to K.S.A. 8-1567(a)(l).
Witte received a controlling sentence of one year and total fines of $600.
The defendant filed a timely notice of appeal. The case was transferred to this court, pursuant to K.S.A. 20-3018(c).
HORIZONTAL GAZE NYSTAGMUS TEST
In Kansas, it is illegal for an individual with a blood or breath alcohol concentration (BAC) of .10 or higher to operate or attempt to operate a vehicle. K.S.A. 8-1567(a)(l). The National Highway Traffic Safety Administration (NHTSA) has researched and recommended a battery of field sobriety tests to assist in this determination. The NHTSA claims the HGN test is an accurate and effective field sobriety test to determine whether a driver’s alcohol concentration is above .10. National Highway Traffic Safety Administration, DOT-HS-806-512, Improved Sobriety Testing (January 1984) (found at 2 Nichols, Drinking/Driving Litigation § 26 app. A [1991]) (hereinafter 1984 NHTSA Study); National Highway Traffic Safety Administration, DOT-HS-806-475, Field Evaluation of a Behavioral Test Battery for DWI (September 1983) (found at 2 Nichols, Drinking/Driving Litigation § 26 app. B [316]*316[1991]) (hereinafter 1983 NHTSA Study); National Highway Traffic Safety Administration, DOT-HS-805-864, Development and Field Test of Psychophysical Tests for DWI Arrest (March 1981) (found at 2 Nichols, Drinking/Driving Litigation § 26 app. C [1991]) (hereinafter 1981 NHTSA Study); Wichita-Sedgwick County Standardized Field-Sobriety Procedure, Officers Manual (1987); see Carper & McCamey, Gaze Nystagmus: Scientific Proof of DUI?, 77 Ill. B.J. 146 (1988); Seelmeyer, Nystagmus, A Valid DUI Test, Law and Order 29 (July 1985); Rouleau, Unreliability of the Horizontal Gaze Nystagmus Test, 4 Am. Jur. Proof of Facts 3d 439 § 1 (1989).
Nystagmus is “an involuntary rapid movement of the eyeball, which may be horizontal, vertical, rotatory, or mixed.” Dorland’s Illustrated Medical Dictionary 1068 (25th ed. 1974); see 3 Schmidt’s Attorneys’ Dictionary of Medicine N-102 (1991); Stedman’s Medical Dictionary 1074 (25th ed. 1990). HGN is
“a jerking of the eyes as they gaze to the side. Many people .will exhibit some nystagmus, or jerking, as their eyes track to the extreme side. However, as people become intoxicated, the onset of the nystagmus,-or jerking, occurs after fewer degrees of lateral deviation, and the jerking at the more extreme angles becomes more distinct.” 1983 NHTSA Study at .2.
"The theory behind the gaze nystagmus test is that there is a strong correlation between the amount of alcohol a person consumes and the angle of onset of the nystagmus.” Carper & McCamey, 77 Ill. B. J. at 147; see Whitmore, Roadside Sobriety Tests: A Police Officers’ Guide to Making Drunk Driving Arrests Stand Up in Court § 5:05 (1987).
No special equipment is needed to administer the HGN test. The driver is instructed to keep his head stationary. and follow an object, such as a pen, a penlight, or the officer’s finger, with his eyes. The object is held at the driver’s eye level and positioned about 12 to 15 inches away from the driver’s eyes. The NHTSA then instructs officers:
■ “Check the suspect’s right eye by moving the object to the suspect’s right. Have the suspect follow the object until the eyes cannot move further to the side. Make this movement in about two seconds, and observe: 1) whether the suspect was about to follow the object smoothly or whether the motion was jerky; and 2) how distinct the nystagmus is at the maximum deviation.
[317]*317“Move the object a- second time to the 45-degree angle of gaze, taking about four seconds. As the eye follows the object, watch for it to start jerking back and forth. If you think you see nystagmus, stop the movement to see if the jerking continues. If it does, this point is the angle of onset. If it does not, .keep moving the object until the jerking does occur or until you reach the imaginary 45-degree line. Note whether or not the onset occurs before the 45-degree angle of gaze. (The onset point at a BAC. of 0.10 percent is about 40 degrees.)” 1984 NHTSA Study at 3-4.
This procedure is repeated for the left eye.
There are three possible signs of intoxication for each eye:
“Angle of Onset — the more intoxicated a person becomes, the sooner the jerking will occur as the eyes move to the side.
“Maximum Deviation — the greater the alcohol impairment the more distinct the nystagmus is when the eyes are as far to the side as possible.
“Smooth Pursuit — an intoxicated person often cannot follow a slowly moving object smoothly with his eyes.” 2 Nichols, Drinking/Driving Litigation § 26:01, p. 159 (1992 Supp.).
A score of six points is possible, three for each eye. According to the NHTSA, if a driver scores four or more points, the. driver’s BAC is above .10. 1984 NHTSA Study at 4. For a detailed discussion of the administration and scoring of the HGN test, see 1984 NHTSA Study at 3-4; Tenney, The Horizontal Gaze Nystagmus Test and The Admissibility of Scientific Evidence, 27 N.H.B.J. 179, 181 (1986).
In this case, Witte filed a motion in limine to prevent the State from presenting evidence relating to HGN. The defendant claimed the HGN test is scientific evidence and, as such, must meet the Frye foundation requirements for admissibility. See Frye v. United States, 293 F. 1013 (D.C. Cir. 1923) (test enunciated for determining admissibility of scientific evidence); State v. Lowry, 163 Kan. 622, 629, 185 P.2d 147 (1947) {Frye test cited with approval). He also contended that the HGN test is not scientifically reliable and that the officer did not conduct the test properly. A' hearing was heard on Witte’s motion prior to trial. The State claimed the HGN test was not scientific evidence and did not have to satisfy the Frye requirements. The trial court overruled the defendant’s motion, reasoning that “if the other type of tests that are given by. the officers is admissible, the'n the eye test is admissible as an observation by a layperson who [318]*318has been labeled a police officer who has had some training in it.”
On appeal, both Witte and the State make essentially the same arguments. Additionally, the State argues in the alternative that, if the HGN test is scientific evidence, it satisfies the Frye test.
This court has never addressed whether the HGN test is scientific evidence and, if so, whether it meets the Frye admissibility requirements. Three Kansas cases simply mention the use of the HGN test. See State v. McNaught, 238 Kan. 567, 569, 713 P.2d 457 (1986) (state trooper administered the HGN test to the defendant); State v. Maze, 16 Kan. App. 2d 527, 528, 825 P.2d 1169 (1992) (the defendant performed the HGN test and failed it); State v. Kelly, 14 Kan. App. 2d 182, 183, 786 P.2d 623 (1990) (same).
In another case, Angle v. Kansas Dept. of Revenue, 12 Kan. App. 2d 756, 758 P.2d 226, rev. denied 243 Kan. 777 (1988), the appellant claimed the HGN test was not administered correctly. The appellant appealed from the trial court’s affirmance of the hearing officer’s finding that the appellant was driving while under the influence. The Court of Appeals noted that the appellant had failed to raise the issue to the hearing officer and concluded that the trial court erred in considering whether the HGN test was conducted properly. The Court of Appeals also stated that even if it had allowed the appellant’s claim, two out of the three phases of the HGN test were administered properly. Furthermore, there was substantial competent evidence, independent of the HGN test, to support the finding that the appellant was driving while intoxicated. 12 Kan. App. 2d at 758, 764-69.
None of these cases address what foundation must be laid before HGN test results may be admitted at trial. Thus, this is a case of first impression for Kansas. Other jurisdictions have addressed this issue.
Ohio, Iowa, and Texas have held that evidence of HGN testing is admissible and does not require scientific evidence or expert interpretation. The Supreme Court of Ohio, in State v. Bresson, 51 Ohio St. 3d 123, 129, 554 N.E.2d 1330 (1990), reasoned:
“The HGN test cannot be compared to other scientific tests, such as a polygraph examination, since no special equipment is required in its administration. Thus the only requirement prior to admission is the officer’s [319]*319knowledge of the test, his training, and his ability to interpret his observations. The admission of the results of the HGN test is no different from any other field sobriety test, such as finger-to-nose, walk-and-tum, or one-leg stand.”
Although the court concluded that “the HGN test has been shown to be a reliable indicator of BAC levels,” the court stated that HGN test results cannot be used to prove an exact BAC. 51 Ohio St. 3d at 128, 129. See State v. Nagel, 30 Ohio App. 3d 80, 81, 506 N.E.2d 285 (1986).
The Iowa Supreme Court, in State v. Murphy, 451 N.W.2d 154, 156 (Iowa 1990), noted at the outset that “the principal obstacle to the admissibility of the horizontal gaze nystagmus test may be its pretentiously scientific name.” The court followed the Nagel decision’s “liberal approach to the admissibility of technical information.” 451 N.W.2d at 157; see also State v. Edman, 452 N.W.2d 169, 170 (Iowa 1990) (Citing its decision in Murphy, the Supreme Court of Iowa stated: “Because the test may be easily administered and its results objectively recorded by a properly trained officer, it is unnecessary to establish the foundation for such evidence through scientific testimony.”).
Texas courts have held HGN evidence may be admitted to prove intoxication and any witness, lay or expert, may give an opinion on the issue of intoxication. Finley v. State, 809 S.W.2d 909, 913-14 (Tex. App. 1991). HGN evidence, however, may not be used to prove an exact BAC. Richardson v. State, 766 S.W.2d 538, 540 (Tex. App. 1989).
The Supreme Court of Montana, in State v. Clark, 234 Mont. 222, 762 P.2d 853 (1988), concluded that Montana’s liberal foundation requirements for expert testimony permitted the admission of HGN test evidence. The court noted that Montana does not follow the Frye test, reasoning:
“Unless an exaggerated popular opinion of the accuracy of the particular technique makes its use prejudicial or likely to mislead the jury, the better approach is to admit all relevant scientific evidence in the same manner as other expert testimony and allow its weight to be attacked by cross-examination or refutation. [Citation omitted.]” 234 Mont, at 227.
The appellate courts in Illinois have differed in their approach to the issue. The Appellate Court of Illinois, Fourth District, has held that a Frye hearing is not required before the results of [320]*320field sobriety tests, including the HGN , test, can be admitted into evidence. People v. Sides, 199 Ill. App. 3d 203, 206-07, 556 N.E.2d 778 (1990) (“No expert testimony is needed nor is a showing of scientific principles required before a jury can be permitted to conclude that a person who performs badly on the field-sobriety tests may have his mental or physical faculties ‘so impaired as to reduce his ability to think and act with ordinary care.’ ”); see People v. Vega, 145 Ill. App. 3d 996, 496 N.E.2d 501 (1986).
The Appellate Court of Illinois, Second District, never reached the question of whether a Frye hearing is required because it disposed of the issue on other grounds. The court concluded that the statute only authorized analysis of “blood, urine, breath or other bodily substance” to prove intoxication. Therefore, because HGN testing did not fall into any of these categories, HGN test results would not be admissible to prove a statutory violation. The court also noted that “HGN test results . . . are derived from the subjective interpretation of the police officer administering the test.” People v. Dakuras, 172 Ill. App. 3d 865, 869, 527 N.E.2d 163, lv. to appeal denied 123 Ill. 2d 561 (1988).
The Appellate Court of Illinois, Third District, in a case involving whether HGN test results could be used to establish probable cause to make an arrest, stated: “We recognize that the horizontal gaze nystagmus test has been criticized and is not evidence, but the officer’s observation of the defendant’s inability to follow directions is probative and properly admissible.” People v. Jebelian, 204 Ill. App: 3d 11, 14, 561 N.E.2d 1079 (1990). (Emphasis added.) Although the case can be distinguished factually, the court’s comment about the HGN test is worth noting.
The majority of jurisdictions that have considered the issue have held that the HGN test is scientific evidence, most requiring that the Frye foundation requirements for admissibility be satisfied. Ex parte Malone, 575 So. 2d 106, 107 (Ala. 1990); State v. Superior Court, 149 Ariz. 269, 279, 718 P.2d 171 (1986); People v. Williams, 3 Cal. App. 4th 1326, 1334, 5 Cal. Rptr. 2d 130 (1992); State v. Garrett, 119 Idaho 878, 881, 811 P.2d 488 (1991); State v. Armstrong, 561 So. 2d 883, 887 (La. App.), writ denied 568 So. 2d 1077 (La. 1990); State v. Wheeler, 764 S.W.2d 523, 524-25 (Mo. App. 1989); State v. Borchardt, 224 Neb. 47, 58-[321]*32159, 395 N.W.2d 551 (1986); People v. Torrey, 144 A.D.2d 865, 866, 534 N.Y.S.2d 807 (1988); State v. Reed, 83 Or. App. 451, 454-55, 732 P.2d 66 (1987); Com. v. Miller, 367 Pa. Super. 359, 365-66, 532 A.2d 1186 (1987); State v. Barker, 179 W. Va. 194, 197-98, 366 S.E.2d 642 (1988).
These courts have given various reasons for holding that HGN evidence is scientific in nature: The HGN test is distinguished from other field sobriety tests in that science, rather than common knowledge, provides the legitimacy for HGN testing. State v. Superior Court, 149 Ariz. at 276; Garrett, 119 Idaho at 881. Certain reactions to alcohol are so common that judicial notice will be taken of them; however, HGN testing does not fall into this category. Reed, 83 Or. App. at 455. HGN test results are “scientific evidence based on the scientific principle that consumption of alcohol causes the type of nystagmus measured by the HGN test.” Miller, 367 Pa. Super, at 365. HGN evidence could have a disproportionate impact on the jury’s decisionmaking process because of the test’s scientific nature and because the jury may not understand the nature of the test or the methodology of its procedure. Malone, 575 So. 2d at 107; Barker, 179 W. Va. at 197-98.
The California Court of Appeal, Fifth District, ruled that an officer’s testimony concerning the HGN test went beyond common knowledge and experience. The court reasoned the officer’s opinion that the defendant
“was under the influence of alcohol, to the extent it was based on the nystagmus test, rests on scientific premises well beyond [the officer’s] knowledge, training, or education. Without some understanding of the processes by which alcohol ingestion produces nystagmus, how strong the correlation is, how other possible causes might be masked, what margin of error has been shown in statistical surveys, and a host of other relevant factors, [the officer’s] opinion on causation, notwithstanding his ability to recognize the symptom, was unfounded.” Williams, 3 Cal. App. 4th at 1334.
Here, the State maintains that HGN evidence is not scientific in nature. In reaching that conclusion the State urges this court to follow People v. Nagel, 30 Ohio App. 3d 80, and those courts that have followed the Nagel decision; State v. Bresson, 51 Ohio St. 3d 123; State v. Murphy, 451 N.W.2d 154; State v. Edman, 452 N.W.2d 169. Bresson, Murphy, and Edman previously have [322]*322been discussed. The Nagel court concluded that the HGN test “requires only the personal observation of the officer administering it. It is objective in nature and does not require expert inteipretation.” 30 Ohio App. 3d 80.
The Nagel opinion has been criticized. One commentator has observed:
“The Ohio appellate court has apparently ignored the rule that the arresting officer’s personal observations,’ in this instance, constitute opinion testimony. That is, the officer’s opinion (that the jerking and twitching of the suspect’s eyes during the gaze nystagmus test indicated that the suspect had consumed alcohol) is an opinion which is based upon a so-called ’scientific’ interpretation of observed facts which exist outside the common knowledge of the average lay person and therefore would require the testimony of an expert. The court also ignored the significance of the fact that the horizontal gaze nystagmus test draws its convincing force from the supposed scientific principle that alcohol affects the smooth pursuit mechanism of the human eye. It is clear that the horizontal gaze nystagmus test is scientific in nature, as are other numerous reflex response tests (such as Babinsld’s reflex, which does not require the use of a machine to administer, monitor or interpret).” Rouleau, Unreliability of the Horizontal Gaze Nystagmus Test, 4 Am. Jur. Proof of Facts 3d 439 § 10, p. 458 (1989).
Cf. Pangman, Horizontal Gaze Nystagmus: Voodoo Science, 2 DWI Journal 1, 3-4 (1987) (HGN test is subjective, not objective, and based upon scientific principles rather than common knowledge).
The State also contends the HGN test does not differ significantly from other field sobriety tests and, therefore, the admissibility requirements should be the same for all field sobriety tests. The State attempts the following analogy:
“The State is not required to famish an expert to testify as to the effects of alcohol on an individual’s sense of balance prior to the admission of evidence [of] a defendant’s performance on the one-legged stand test. In like manner, the State should not be required to famish an expert to testify as to the effects of alcohol with respect to nystagmus prior to the admission of evidence of a defendant’s performance on the HGN test.”
The State’s analogy fails. Alcohol’s effect on a person’s sense of balance is common knowledge. The same cannot be said for HGN. The HGN test is based upon scientific principles and exceeds common knowledge. We hold that the HGN test results are scientific evidence. As such, the Frye foundation requirements for admissibility must be satisfied.
[323]*323In Smith v. Deppish, 248 Kan. 217, 236, 807 P.2d 144 (1991), this court reviewed the Frye admissibility requirements.
“The Frye test requires that, before expert scientific opinion may be received in evidence, the basis of that opinion must be shown to be generally accepted as reliable within the expert’s particular scientific field. If a new scientific technique’s validity has not been generally accepted as reliable or is only regarded as an experimental technique, then expert testimony based on its results should not be admitted into evidence. [Citation omitted.]”
The rationale behind the adoption of the Frye test is still pertinent.
“ ‘Frye was deliberately intended to interpose a substantial obstacle to the unrestrained admission of evidence based upon new scientific principles. . . . Several reasons founded in logic and common sense support a posture of judicial caution in this area. Lay jurors tend to give considerable weight to “scientific” evidence when presented by “experts” with impressive credentials. We have acknowledged the existence of a “. . . misleading aura of certainty which often envelops a new scientific process, obscuring its currently experimental nature.’ ”
“. . . Courts should be reluctant to resolve the disputes of science. It is not for the law to experiment, but for science to do so. Without the Frye test, juries would be compelled to make determinations regarding the validity of experimental or novel scientific techniques. As a result, one jury might decide that a particular scientific process is reliable, while another jury might find that the identical process is not. Such inconsistency concerning the admissibility of a given scientific technique or process in criminal cases would be intolerable.” State v. Washington, 229 Kan. 47, 54, 622 P.2d 986 (1981).
Witte maintains the State has the burden of satisfying the three prongs of the Frye test, which the defendant describes as follows:
“1. Th[e] reliability of the underlying scientific theory upon which the HGN test is based (i. e., that the nystagmus of the eye is, in fact, an indicator of alcohol consumption to the degree that it influences or impairs the ability to drive).
“2. That the method used to test HGN is a valid test to measure or perceive that phenomenon, particularly if the method actually conducted by the officer in this case varies from the manner in which such test is to be performed according to the scientists familiar with the phenomenon and test.
“3. That this officer has been trained to follow the procedures established by the scientists to test the phenomenon and used those methods properly pursuant to the training.”
[324]*324Witte cites no authority for his three-prong Frye analysis. The State does not dispute or discuss the defendant’s assertion that Frye requires a three-prong analysis.
The first two prongs of Witte’s “three-prong analysis,” whether the scientific theory and the validity of the techniques used to measure that theory generally have been accepted as reliable within the appropriate scientific field, fall within the Deppish court’s description of the Frye test. Other jurisdictions have incorporated the third prong, whether the administering officer was trained properly and administered the test correctly, into its analysis of whether HGN evidence satisfies the Frye test. See, e.g., State v. Superior Court, 149 Ariz. 269, 276, 718 P.2d 171 (1986). Additionally, one commentator discusses the three-prong Frye analysis. See Tenney, The Horizontal Gaze Nystagmus Test and The Admissibility of Scientific Evidence, 27 N.H.B.J. 179, 183-84 (1986).
Witte claims the trial court erred in allowing the State to introduce HGN evidence. He contends that Deputy Ron Goodwyn, the arresting officer and the State’s only witness who potentially could have presented a sufficient foundation to satisfy the Frye admissibility requirements, did not and could not have presented the requisite foundation. At the motion in limine hearing, the trial court ruled there was insufficient foundation to recognize Goodwyn as an expert. The State did not endorse or subpoena any doctors, researchers, or other experts who could have laid the requisite foundation for the admission of HGN evidence under the Frye test.
The State argues that because other jurisdictions have recognized HGN evidence as reliable under the Frye test, the State was not required to establish the reliability of such evidence through expert testimony. The State also contends that because HGN evidence has been established as rehable, an officer, who has been trained properly and who has administered correctly the test to the defendant, can testify about the defendant’s test results. In support of its argument, the State cites State v. Superior Court, 149 Ariz. 269, and State v. Armstrong, 561 So. 2d 883 (La. App. 1990).
[325]*325In State v. Superior Court, the Supreme Court of Arizona determined
“that the HGN test satisfies the Frye standard. The evidence demonstrates the following propositions have gained general acceptance in the relevant scientific community: (1) ■ HGN occurs in conjunction with alcohol consumption; (2) its onset and distinctness are correlated to BAC; (3) BAC in excess of .10 percent can be estimated with reasonable accuracy from the combination of the eyes’ tracking ability, the angle of onset of nystagmus and the degree of nystagmus at maximum deviation; and (4) officers- can be trained to observe these phenomena sufficiently to estimate accurately whether BAC is above or below .10 percent. We therefore hold that, with proper foundation as to the techniques used and the officer’s ability to use it [citation omitted], testimony of defendant’s nystagmus is admissible on the issue of a defendant’s blood alcohol level as would be other field sobriety test results on the question of the accuracy of the chemical analysis.” 149 Ariz. at 279.
The court stated that HGN evidence “may be admitted in evidence to corroborate or attack, but not to quantify, the chemical analysis of the accused’s blood alcohol constent.” .149 Ariz. at 280. One article suggests that even though the Arizona Supreme Court held that HGN evidence satisfied the Frye test, that court doubted the reliability of the HGN test, because it held that a DUI conviction cannot rest solely upon HGN test results. Carper & McCamey, Gaze Nystagmus: Scientific Proof of DUlP, 77 Ill. B.J. 146, 148 (1988).
In reaching its conclusion, the Arizona- Supreme Court discussed what constituted the relevant scientific community.
“No argument has been made why the fields of pharmacology, ophthalmology and criminalistics (beyond those concerned with detecting violators of DUI laws) should be included in the relevant scientific community and no convincing reason occurs to us. We conclude, therefore, that to determine whether the HGN test satisfies the Frye requirement of general acceptance the appropriate disciplines include behavioral psychology, highway safety and, to a lesser extent, neurology and criminalistics.”-149 Ariz. at 278.
In State v. Superior Court, in addition to its testifying witnesses, the State submitted seven publications, including scientific reports and NHTSA studies. The defendant, on the other hand, presented no evidence to counter the State’s expert witnesses or to dispute the State’s publications. The court also conducted its own research. 149 Ariz. at 278, 280-82. The opinion [326]*326concludes with two appendices, one listing the publications submitted by the State and the other listing the court’s research.
In a subsequent case, State ex rel. Hamilton v. City Court of City of Mesa, 165 Ariz. 514, 799 P.2d 855 (1990), the Supreme Court of Arizona reaffirmed the State v. Superior Court decision.
“HGN test results may be admitted only for the purpose of permitting the officer to testify that, based on his training and experience, the results indicated possible neurological dysfunction, one cause of which could be alcohol ingestion. The proper foundation for such testimony, which the State may lay in the presence of the jury, includes a description of the officer’s training, education, and experience in administering the test and a showing that the test was administered properly. The foundation may not include any discussion regarding the accuracy with which HGN test results correlate to, or predict, a BAC of greater or less than .10%.” 165 Ariz. at 519.
The Supreme Court of Louisiana adopted the Arizona court’s holdings and reasoning. Armstrong, 561 So. 2d at 887. The Idaho Supreme Court also followed the Arizona opinion. The Idaho court noted that no evidence or publications had been presented that refuted the Arizona opinion. State v. Garrett, 119 Idaho 878, 881, 811 P.2d 488 (1991).
Our research indicates that the reaction within the scientific community is mixed. Some articles endorse the HGN testing and its accuracy. See, e.g., Good & Augsburger, Use of Horizontal Gaze Nystagmus as a Part of Roadside Sobriety Testing, 63 Am. J. of Optometry & Physiological Optics 467 (1986). Other articles discuss concerns with the HGN test. See, e.g., Carper & McCamey, 77 Ill. B.J. at 149; Halperin & Yolton, Is the Driver Drunk? Ocularmotor Sobriety Testing, 57 J. of the Am. Optometric A. 654, 657 (1986). Several commentators disagree with the Arizona Supreme Court’s conclusions, insisting the HGN test has not been accepted generally within the scientific community and questioning the methodology of the NHTSA’s research. See, e.g., Cowan & Jaffee, Proof and Disproof of Alcohol-Induced Driving Impairment Through Evidence of Observable Intoxication and Coordination Testing, 9 Am. Jur. Proof of Facts 3d 459 § 12 (1990); Pangman, Horizontal Gaze Nystagmus: Voodoo Science, 2 DWI Journal 1, 3-4 (1987); Rouleau, Unreliability of the Horizontal Gaze Nystagmus Test, 4 Am. Jur. Proof of Facts 3d 439 § 7, p. 452 (1989); 1 Erwin, Defense of Drunk Driving Cases [327]*327§ § 8A:06, 8A:08 (3d ed. 1992); 2 Nichols, Drinking/Driving Litigation § 26:01 (1991 & 1992 Supp.). These articles or the particular sections cited are not listed in the Arizona opinion’s appendices. Most of these articles were published after the Arizona opinion was issued April 7, 1986.
The defendant contends the scientific community does not agree about the correlation between the BAC level and the angle of onset at which nystagmus occurs. The NHTSA declares that “the extent of impairment is indicated by the angle at which nystagmus begins.” Officers are instructed to have the suspect move his or her eye sideways to an angle of 45 degrees from the nose and to watch whether nystagmus occurs before the eye reaches the 45-degree angle. 1984 NHTSA Study at 3-4. “The expected angle of onset for a BAC of 0.10% is 40.2 degrees for the right eye and 40.1 degrees for the left eye.” 1981 NHTSA Study at 25. The NHTSA maintains that, if nystagmus is observed at the 45-degree angle, a BAC of .10 can be estimated correctly 78 percent of the time. 1981 NHTSA Study at 25-30. Put another way, 22 percent of the time it is wrong.
Other researchers disagree that 45 degrees is the appropriate angle of onset. According to one authority, 50-60 percent of sober individuals who deviate their eyes more than 40 degrees to the side will exhibit nystagmus, and this nystagmus cannot be distinguished from alcohol gaze nystagmus. Pangman, 2 DWI Journal at 2 (citing Toglia, Electronystagmography: Technical Aspects and Atlas [1976]). Another researcher suggests the threshold appearance of HGN in most individuals is observed at a 40-degree angle with a BAC reading of .06 percent. Pangman, 2 DWI Journal at 2 (citing Aschan, Different Types of Alcohol Nystagmus, Acta Oto-Laryngologica Supp. 140:69 [1957]; Aschan, Bergstedt, Goldberg & Laurell, Positional Nystagmus in Man During and After Alcohol Intoxication, 17 Q.J. of Studies on Alcohol 381 [1956]). Still another researcher contends individuals with a BAC reading of .10 do not exhibit nystagmus until the eye is deviated to a 51-degree angle. Pangman, 2 DWI Journal at 2 (citing Lehti, The Effect of Blood Alcohol Concentration on the Onset of Gaze Nystagmus, 13 Blutalkohol 411 [1976]). See Rouleau, 4 Am. Jur. Proof of Facts 3d 439 § § 7, 8; 2 Nichols, Drinking/Driving Litigation § 26:01.
[328]*328Researchers have expressed concern that the 45-degree angle used by the NHTSA will create false positive readings. The NHTSA Study also has been criticized for “deliberately screening] out people at high risk for being classified as false positives.” 2 Nichols, Drinking/Driving Litigation § 26:01, p. 2.
The NHTSA agrees the angle of lateral deviation is critical. Despite the fact that the NHTSA obtained its research results through the use of mechanical devices that “hold the head in a stable position and precisely measure the angle of lateral deviation of the eye,” the NHTSA instructs officers to estimate the 45-degree angle. A visual estimation of the angle would seem to cause inaccurate and inconsistent results. 2 Nichols, Drinking/ Driving Litigation § 26:01, p. 4. The stability of the suspect’s head, another critical factor, is also questionable when the test is conducted at roadside. Pangman, 2 DWI Journal at 3.
In addition to intoxication, many other factors can cause nystagmus.
“Nystagmus can be cáused by problems in an individual’s inner ear labyrinth. In fact, irrigating the ears with warm or cold water, not a far-fetched scenario under particular weather conditions, is a source of error. Physiological problems such as certain kinds of diseases may also result in gaze nystagmus. Influenza, streptococcus infections, vertigo, measles, syphilis, arteriosclerosis, muscular dystrophy, multiple sclerosis, Korsakoff’s Syndrome, brain hemorrhage, epilepsy, and other psychogenic disorders all have been shown to cause nystagmus. Furthermore, conditions such as hypertension, motion sickness, sunstroke, eyestrain, eye muscle fatigue, glaucoma, and changes in atmospheric pressure may result in gaze nystagmus. The consumption of common substances such as caffeine, nicotine, or aspirin also lead to nystagmus almost identical to that caused by alcohol consumption.” Pangman, 2 DWI Journal at 3.
See Rouleau, 4 Am. Jur. Proof of Facts 3d 439 § 9. Temporary nystagmus can occur when lighting conditions are poor. Rouleau, 4 Am. Jur. Proof of Facts 3d 439 § 9, p. 456.
An individual’s circadian rhythms (biorhythms) can affect nystagmus readings — the body reacts differently to alcohol at different times of the day. One researcher has suggested that because of this, the angle of onset should be decreased five degrees between midnight and 5 a.m. Rouleau, 4 Am. Jur. Proof of Facts 3d 439 § 9, p. 456; Pangman, 2 DWI Journal at 3. A number of driving under the influence arrests occur after midnight, which [329]*329“would seem to indicate that sensitivity of HGN to alcohol is enhanced during the hours of the day when the greatest number of drunk driving arrests occur.” Pangman, 2 DWI Journal at 3.
A prosecution-oriented group in California conducted its own research:
“The study measured the correlation of police officer estimations of the angle of onset of nystagmus against chemical tests involving breath and blood samples. The data in the study revealed that there was virtually no correlation between the actual value of blood alcohol concentration and the predicted value based upon the angle of onset of nystagmus. However, a correlation did develop between the breath alcohol reading and the level predicted by the alcohol gaze nystagmus. Interestingly, the study concluded that this was caused by the very subjective nature of the test itself:
Since the police officers are the ones operating the breath testing equipment, it appears that, at least in some of the cases, an already known breath alcohol value may have influenced the determination of the angle of onset.
“Simply put, the cops fudged the horizontal gaze nystagmus determination to correspond with the already known correct answer determined by the breath test result. However, since they did not know what the correct answer was when the blood sample was tested (since someone else did the analysis), they could not come close to the correct BAC. These were highly trained California police officers, experienced and familiar with the test procedures and aware that their results were being scrutinized for accuracy and crosschecked against actual BAC determinations. This study points out the fact that horizontal gaze nystagmus tests should never be intended as a substitute for actual blood or breath alcohol testing. The purpose of the procedure, if any is strictly a field screening function, like other presumptive tests.” Pangman, 2 DWI Journal at 3.
The group conceded “the use of ‘(HGN) to predict a person’s blood alcohol level does not appear to be warranted.’ ” Rouleau, 4 Am. Jur. Proof of Facts 3d 439 § 8.
If the Arizona Supreme Court had had this evidence before it, it may not have held that HGN evidence satisfies the Frye admissibility requirements. The reliability of the HGN test is not currently a settled proposition in the scientific community. This court holds that HGN evidence requires a Frye foundation for admissibility. If the Frye foundation is established to this court’s satisfaction, HGN evidence will be admitted in other cases without the need to satisfy the Frye test each time. Before this court rules on whether HGN evidence satisfies the Frye admissibility [330]*330requirements, a trial court first should have an opportunity to examine, weigh, and decide disputed facts to determine whether the test is sufficiently reliable to be admissible for any purpose in Kansas.
HARMLESS ERROR
The State argues that because Witte was convicted of violating K.S.A. 8-1567(a) (1) by operating a motor vehicle with a BAC of .10 or more, there is sufficient competent evidence to support the verdict and, thus, any error concerning the HGN test is harmless.
The major problem is the machine used to test Witte’s BAC gave a reading of .103, or .003 above the statutory maximum. The operator testified that he tested the machine with a known percentage (.10) furnished by the State of Kansas and that if the test result is between .09 and .109, the machine is deemed in satisfactory working condition. According to the operator, a machine usually is set to read low and the sample he ran produced a test result of .097. Because a satisfactory range is between .09 and .109, the machine can be off as much as .010. That fact could be important to a jury if, as here, the defendant’s BAC was .103.
An employee of the Kansas Department of Health and Environment testified the department certifies machines and follows national standards for acceptability. The department will certify a machine if the machine is tested with known solutions of .050, .100, and .200, and the accuracy of the readings is “within plus or minus five percent systematic error and a standard deviation not to [exceed] .0042 percent at all three of those levels.” Additionally, that employee stated the machine used to test Witte’s BAC had tested low in the past and discussed a test score of .096 on a .10 solution.
The employee, however, testified she could not say what percentage of error applied to a particular test. Concerning the low readings, she said the .096 was a “systematic error” and within the plus or minus five percent systematic error allowed. The employee described the method used to calculate a systematic error as taking eight readings of a .10 solution and averaging the readings. The average arrived at is added to or subtracted from [331]*331.10 to arrive at a positive or negative “systematic error.” Thus, on the average, the machine in question produces a result lower than the actual alcohol content. Nonetheless, this would not preclude a machine from producing a higher percentage than is average during one or more of the eight tests.
The machine in question had been tested several times by the Kansas Department of Health and Environment and had a “systematic error” average difference of .001, and that was based on four tests of .000, .001, .001, and .002.
Many jurors could find beyond a reasonable doubt that the defendant was guilty of operating a motor vehicle with a BAC of .10 or above. On the other hand, a reasonable juror could give Witte the benefit of the doubt and acquit the defendant.
Here, the admission of the HGN test results and the officer’s testimony that the jerking of defendant’s eyes was with “certainty” because Witte was under the influence of alcohol may have influenced the jury by leading the jury to believe one scientific test supported and gave credibility to the other. Therefore, we conclude the admission of the HGN test evidence was not harmless error. We vacate the conviction and remand for a new trial.
The defendant raised other issues concerning requested instructions and discretionary functions of the trial court, all of which are moot and none of which require our guidance in the event of a retrial.
Reversed and remanded for a new trial.