delivered the opinion of the court:
Defendant James D. Kirk was charged with driving under the influence of alcohol (DUI) (625 ILCS 5/11—501(a)(2) (West 1994)) and improper lane usage for driving the wrong direction on a one-way road (625 ILCS 5/11—708 (West 1994)). Following a jury trial, defendant was convicted on both counts. Defendant appeals, arguing that it was error for the trial court to allow certain scientific testimony without conducting a Frye hearing. See Frye v. United States, 293 F. 1013 (D.C. Cir. 1923). We agree a Frye hearing should have been conducted but hold that any error was harmless.
The evidence established that defendant watched the second half of the Super Bowl at his son’s Bloomington home. Defendant arrived at the house during halftime and did not appear to be impaired. He remained at the home for a couple of hours, through the second half. During this time, defendant and his son drank some beer from a 12-pack and ate some snacks. Defendant testified he consumed three or four beers "at the most.” Defendant’s daughter-in-law, Joyce Kirk, testified there was still beer left in the 12-pack when defendant left her home, but she did not know how may beers actually remained. She did not think defendant was impaired when he left and testified his walk and speech were no different than usual. Joyce also stated that defendant walks with a limp because he has bad knees.
Defendant, who does not live in Bloomington, testified that when visiting his son he will generally only travel the Bloomington roads he was driving at the time of his arrest. Defendant testified that af*328ter leaving his son’s house, he decided to visit his other son. He then decided it was too late to do so and turned on Oakland Avenue to head east. As he approached Lee Street and saw the traffic lights, defendant realized he was driving the wrong way on a one-way street. He made a right turn when he got to MacArthur and was stopped shortly thereafter.
Officer Darrin Woodin saw defendant driving the wrong way on Oakland Avenue. He watched defendant drive 300 to 400 feet before he turned off Oakland Avenue and onto Madison Street. Woodin did not see defendant violate any other traffic laws prior to the stop. Woodin smelled alcohol as he approached defendant’s vehicle. Woodin believed the smell of alcohol emanated from defendant and not defendant’s car because the smell became stronger when defendant spoke. Woodin described defendant’s speech as being sometimes clear and. sometimes "drift[ing] off into a mumble that was almost unintelligible.” Woodin asked defendant to take the field-sobriety tests.
Woodin first administered the horizontal gaze nystagmus (HGN) test. This test involves testing a suspect’s eye movement in an effort to determine intoxication. Based upon the HGN test, Woodin believed defendant to be intoxicated. Woodin then asked defendant to "recite the alphabet from E through N.” According to Woodin, defendant skipped the letters "H” and "I” and continued through the letter "Z.” Woodin said defendant drifted "in and out of the slurred speech.”
Woodin demonstrated, then asked defendant to perform, the finger-to-nose test. Defendant reportedly refused to take the test, stating that it was "impossible.” Woodin did not ask defendant to perform either the walk-and-turn test or the one-leg-stand test. Woodin explained that, in light of defendant’s bad knees, he did not think the tests would be fair. Based upon the tests and his observations of defendant’s movements, Woodin was of the opinion that defendant was impaired. Woodin also testified that, in response to questions, defendant said he had been at a bar and that he had consumed a few drinks. Woodin’s police report made no mention of defendant’s statement that he had been at a bar. Defendant denied stating that he had been at a bar. Woodin noticed three to five empty beer cans in the backseat of defendant’s car. Defendant testified he had picked these cans up for recycling.
Defendant was arrested for DUI. After being read the motorist warning, defendant refused to take the breathalyzer. Woodin testified defendant said he would not pass the test anyway. At trial, defendant explained that he did not trust the machine, that he had heard several people talk about the test and he did not believe anyone ever passed the test. Defendant was found guilty and appeals.
*329Defendant raises only one argument upon appeal, that it was improper for the trial court to allow Woodin’s testimony concerning the HGN test. Defendant argues the HGN test is based upon scientific principles and that an Illinois court has yet to properly determine whether the HGN test is generally accepted within the scientific community.
Nystagmus, a physiological phenomenon, is a term used to describe an involuntary jerking of the eyeball. People v. Buening, 229 Ill. App. 3d 538, 539, 592 N.E.2d 1222, 1223 (1992); see also Webster’s Tenth New Collegiate Dictionary 800 (1996) ("a rapid involuntary oscillation of the eyeballs”). Nystagmus can be congenital or it may be caused by " 'a variety of conditions affecting the brain, including ingestion of drugs such as alcohol and barbiturates, palsy of lateral or vertical gaze, disorders of the vestibular apparatus and brainstem and cerebellar dysfunction.’ ” (Emphasis omitted.) Buening, 229 Ill. App. 3d at 539, 592 N.E.2d at 1223, quoting The Merck Manual of Diagnosis and Therapy 1980 (14th ed. 1982); see also Schultz v. State, 106 Md. App. 145, 180-81, 664 A.2d 60, 77 (1995) Qisting 38 possible causes of nystagmus in addition to alcohol consumption). The HGN test, as routinely performed by law enforcement officers, consists of:
" 'the driver [being] asked to cover one eye and focus the other on an object (usually a pen) held by the officer at the driver’s eye level. As the officer moves the object gradually out of the driver’s field of vision toward his ear, he watches the driver’s eyeball to detect involuntary jerking. The test is repeated with the other eye. By observing (1) the inability of each eye to track movement smoothly, (2) pronounced nystagmus at maximum deviation[,] and (3) onset of the nystagmus at an angle less than 45 degrees in relation to the center point, the officer can estimate whether the driver’s blood[-]alcohol content (BAG) exceeds the legal limit of [0.10].’ ” Buening, 229 Ill. App. 3d at 539-40, 592 N.E.2d at 1223, quoting State v. Superior Court, 149 Ariz. 269, 271, 718 P.2d 171, 173 (1986) (en banc) (hereinafter Blake).
In Buening, the defendant filed a motion in limine seeking to exclude the results of his HGN test. The trial court granted the motion, and the State, after filing a certificate of impairment, appealed. The Buening court reviewed Illinois case law concerning the admissibility of HGN test results as well as the case law of other states. Relying upon Blake as "one of the more extensively researched and well-reasoned decisions on the subject” (Buening, 229 Ill. App. 3d at 541, 592 N.E.2d at 1225), the court concluded that HGN testing meets the Frye standard (see Frye, 293 F. 1013; People v. Baynes, 88 Ill. 2d 225, 430 N.E.2d 1070 (1981)) and that "HGN test results are admissible, *330as is any other evidence of a defendant’s behavior, to prove that the defendant is under the influence of alcohol, provided a proper foundation has been laid” (Buening, 229 Ill. App. 3d at 546, 592 N.E.2d at 1227-28). Buening, a fifth district opinion, has been followed by the third district in People v. Wiebler, 266 Ill. App. 3d 336, 640 N.E.2d 24 (1994).
This court addressed HGN testing in People v. Vega, 145 Ill. App. 3d 996, 496 N.E.2d 501 (1986), a DUI case in which the trial court admitted evidence that the defendant had failed the test. The entire foundation for the test’s admission was the testimony of the police officer explaining the procedure of the test. There was no testimony about either the officer’s training in the administration of the test or the scientific validity of the test. This court noted that when evidence "beyond the general knowledge of the average individual is sought to be introduced, a proper foundation by way of expert testimony is required. This becomes especially true of technological evidence. It is a natural inclination of jurors to regard such evidence as extremely trustworthy.” Vega, 145 Ill. App. 3d at 1000, 496 N.E.2d at 504. Perhaps recognizing that a Frye hearing should have been conducted in the trial court, both parties in Vega submitted materials to this court in support of their position as to the validity (or invalidity) of the HGN test. Those materials had not been presented to the trial court. Determining that the foundation presented at trial was insufficient to admit the HGN test evidence, we declined to follow Blake and declined the opportunity to rule upon the validity and admissibility of the HGN test: "That decision must await another day and another case. Exhibits or attachments to appellate briefs, not seen by the trial court, are improper.” Vega, 145 Ill. App. 3d at 1001, 496 N.E.2d at 505. When read as a whole, Vega stands for the proposition that the HGN test evidence should not have been admitted in the absence of a finding that it met the Frye standard. Vega was followed by the second district in People v. Smith, 182 Ill. App. 3d 1062, 538 N.E.2d 1268 (1989).
In People v. Hood, 265 Ill. App. 3d 232, 638 N.E.2d 264 (1994), this court was faced with the question of whether the results of an HGN test were admissible in an implied-consent proceeding. This court first noted that neither Vega nor Buening dealt with the issue then before the court, statutory summary suspension of drivers’ licenses. This court then said:
"As the fifth district determined the HGN test was sufficiently reliable to meet the Frye standard for admissibility in criminal proceedings, we are persuaded it is sufficiently reliable to be admitted in implied-consent proceedings; thus, where evidence *331involving the HGN test is sought to be admitted in implied-consent proceedings, the State need not call an expert witness to attest to its reliability. Accordingly, the circuit court properly overruled Hood’s objection based on lack of scientific reliability.” Hood, 265 Ill. App. 3d at 245-46, 638 N.E.2d at 274.
See also People v. Rose, 268 Ill. App. 3d 174, 181, 643 N.E.2d 865, 870 (1994) ("Generally, field-sobriety tests are admissible not only in proceedings to determine whether probable cause existed, but also in criminal proceedings to prove intoxication. (See People v. Buening (1992), 229 Ill. App. 3d 538, 592 N.E.2d 1222 ([HGN] test).”) This case presents the specific issue, not involved in Hood and Rose, of whether it is necessary to conduct a Frye hearing prior to the admission of the result of a HGN test in a criminal trial for DUI. We conclude that a Frye hearing is necessary.
The Buening court relied primarily upon the Blake decision in reaching its conclusion. In Blake, the defendant made pretrial motions (1) to dismiss the prosecution for lack of probable cause to arrest, and (2) to preclude the admission of HGN evidence at her upcoming DUI trial. The trial court conducted an evidentiary hearing, at which the prosecution presented four witnesses. The first was Dr. Marcelline Burns, a research psychologist who studied the effect of alcohol on behavior. Burns testified the HGN test, when used in conjunction with walk-and-turn and one-leg-stand tests, resulted in 83% accuracy in determining BAG above and below 0.10. Blake, 149 Ariz. at 271, 718 P.2d at 173. Burns also testified "the HGN test had been accepted as valid by the highway safety field, including the [National Highway Traffic Safety Administration (NHTSA)], Finnish researchers, state agencies such as the California Highway Patrol, Arizona Highway Patrol, Washington State Police, and numerous city agencies.” Blake, 149 Ariz. at 272, 718 P.2d at 174. The other three witnesses were police officers. Their testimony suggested that the HGN test is between 80% and 90% accurate and that it is particularly "useful in detecting violations where a driver with [a] BAG over 0.10 is able to pull himself together sufficiently to pass the traditional field[-]sobriety tests and thus avoid arrest and subsequent chemical testing.” Blake, 149 Ariz. at 272, 718 P.2d at 174. The trial court concluded that the HGN test represented a new scientific principle subject to the Frye standard of admissibility, that the test did not satisfy Frye, and therefore could not form the basis of probable cause. Blake, 149 Ariz. at 272, 718 P.2d at 174. The appellate court reversed, noting that the Frye standard applied only to the admissibility of evidence at trial, not to probable cause for arrest, and that the HGN test was sufficiently reliable to form the basis of *332probable cause. Alternately, the appellate court found the HGN test did satisfy the Frye standard. Blake, 149 Ariz. at 272-73, 718 P.2d at 174-75.
The Arizona Supreme Court noted the "HGN test is a different type of test from balancing on one leg or walking a straight line because it rests almost entirely upon an assertion of scientific legitimacy rather than a basis of common knowledge.” Blake, 149 Ariz. at 276, 718 P.2d at 178. The court concluded the Frye standard applied. Blake, 149 Ariz. at 277, 718 P.2d at 179. The court then determined that, before the HGN test could be found to satisfy the Frye rule, it would have to be shown that the HGN test was generally accepted in four different disciplines: "behavioral psychology, highway safety and, to a lesser extent, neurology and criminalistics.” Blake, 149 Ariz. at 278, 718 P.2d at 180. (Unlike other tests employed in court, such as deoxyribonucleic acid (DNA), it appears that HGN is not used by anyone except police officers.) Based upon its own research as well as articles submitted by the prosecution, the Arizona Supreme Court concluded the HGN test satisfied the Frye standard. Accordingly, the court held that, "with [the] proper foundation as to the techniques used and the officer’s ability to use it [citation], 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.” Blake, 149 Ariz. at 279, 718 P.2d at 181.
There are several ways a proponent of evidence subject to Frye can prove the "general acceptance” of the proffered evidence. The proponent may use scientific publications, prior judicial decisions, practical applications, as well as the testimony of scientists as to the attitudes of their fellow scientists. See 1 J. Strong, McCormick on Evidence § 203, at 870 (4th ed. 1992) (and cases cited therein) (hereinafter McCormick). At least a couple of courts have taken judicial notice of general acceptance where the published indications of general acceptance are unequivocal and undisputed. McCormick § 203, at 870 n.22. Blake appears to rely upon this method, at least to some extent. While it may be proper for a trial court to take judicial notice of numerous articles, we believe it is improper for this court to decide the validity or acceptance of a scientific test on such a basis.
We agree with Buening and Blake in at least one respect: HGN evidence is scientific evidence that must meet the Frye standard before it is admissible. This seems to be the majority view. See Schultz, 106 Md. App. at 158-60, 664 A.2d at 66-67 (and cases cited therein). In the absence of a Frye hearing in the instant case, however, we decline to follow the Buening conclusion that HGN evidence is admissible. The Buening court relied heavily upon the *333opinions of other courts, including Blake. Perhaps the Buening court relied upon other evidence, but we cannot tell whether it did so from the opinion.
The other cases relied upon in Buening in turn relied upon Blake. Four of the cases were from courts that did not deem Frye applicable to the HGN test. See Howard v. State, 744 S.W.2d 640 (Tex. Crim. App. 1987); State v. Bresson, 51 Ohio St. 3d 123, 554 N.E.2d 1330 (1990); State v. Nagel, 30 Ohio App. 3d 80, 506 N.E.2d 285 (1986); State v. Murphy, 451 N.W.2d 154 (Iowa 1990). Of the remaining cases, none of the appellate courts had the benefit of a Frye hearing. Rather, those courts concluded that the Frye standard had been met due to the Blake court’s conclusion. See State v. Garrett, 119 Idaho 878, 881, 811 P.2d 488, 491 (1991) (finding that HGN evidence satisfies Frye standard because it had not been provided with authority that "refutes the reasoned decision of [Blake]”); State v. Armstrong, 561 So. 2d 883 (La. App. 1990); State v. Clark, 234 Mont. 222, 762 P.2d 853 (1988).
Reliance upon other courts’ opinions can be problematic: "Unless the question of general acceptance has been thoroughly and thoughtfully litigated in the previous cases, *** reliance on judicial practice is a hollow ritual.” McCormick § 203, at 870 n.20. For example, McCormick cites Glover v. State, 787 S.W.2d 544 (Tex. Crim. App. 1990), as a case where the court held that DNA fingerprinting enjoys general acceptance following a hearing in which defendant produced no expert testimony. In reaching its decision, the Glover court relied upon other cases in which no defense experts were available. McCormick § 203, at 870 n.20. " '[Jjudicial notice could become a yellow brick road for judicial acceptance of bogus or at least unvalidated scientific theories or techniques.’ ” McCormick § 203, at 870 n.20, quoting J. Starrs, Frye v. United States Restructured and Revitalized: A Proposal to Amend Federal Evidence Rule 702, 115 F.R.D. 92, 97 (1987). The State’s evidence in Blake consisted of four witnesses: one research psychologist and three police officers. The defendant did not present any evidence. The Blake court relied upon its own research of relevant articles. While the Blake defendant "won” the Frye hearing at the trial court level, that decision was reversed by the appellate court. Blake is questionable authority for the proposition that the HGN test meets the Frye standard in Illinois courts.
The expert retained by the prosecution in Blake, Dr. Burns, was the individual who conducted the study that led to the NHTSA’s adoption of the HGN test. Police departments, in turn, have adopted the NHTSA’s recommendations. In Blake, Dr. Burns supported the proposition that the HGN test is accepted and reliable, in part, by *334relying upon the NHTSA’s manual and the fact that the test is used by different police departments. By doing so, however, she in essence referred back to her own conclusions, magnifying the opportunity for error. We do not say that Dr. Burns’ conclusions on the subject are flawed, only that the issue has not been fully and thoroughly litigated. The proper place for this litigation is in the trial court, and it was error to admit the HGN test evidence without a proper Frye hearing.
The concurrence cites a recent decision, Zimmerman v. State, No. 130 1996 (Del. April 2, 1997) (1997 WL 158121). In Zimmerman, the Delaware Supreme Court reversed a decision admitting HGN evidence because the trial court had not conducted Delaware’s equivalent of a Frye hearing. That reversal is consistent with our holding, that because of the conflict in the decisions the trial court should not simply choose a decision of some other jurisdiction to follow, but should actually review the scientific testimony on both sides in a Frye hearing. Zimmerman noted that State v. Ruthardt, 680 A.2d 349 (Del. Super. 1996), decided after the case sub judice, conducted a Frye hearing, and Zimmerman agreed with Ruthardt but the subsequent decision did not justify affirmance in the case before it. Zimmerman, slip op. at 3, 4 n.18. Ruthardt recognized there is some opposition to the HGN test both within the scientific community and among the courts, there are several scientific articles published subsequent to the Blake case that insist that the HGN test has not gained general acceptance within the scientific community, and several articles maintain that the research procedures in the NHTSA studies are suspect. Ruthardt, 680 A.2d at 359. The dissent cites State v. Taylor, No. CUM—95—706 (Me. April 18, 1997) (1997 WL 200304), but that decision also relies heavily on the Blake case and the questioned NHTSA studies.
The State argues that any error was harmless. We agree. While defendant may have had valid objections to both the walk-and-turn and one-leg-stand tests, Woodin did not require defendant to take these tests. Defendant has offered no valid explanation for his failure to take the finger-to-nose test. He simply stated that the test was impossible, even after Woodin demonstrated the test. Defendant’s only explanation for his refusal to submit to the breath test was because he had heard that no one ever passed it. Defendant’s explanation that hay fever and dentures caused his slurred speech were self-serving at best. Defendant failed the alphabet test. While we believe the admission of the HGN test was error, we believe it was harmless error in light of the other evidence of defendant’s guilt.
*335For the foregoing reasons, the judgment of the trial court is affirmed.
Affirmed.
McCullough, j., concurs.