delivered the opinion of the court:
The defendant, Peppe K. Park, was convicted of one count of possession of “more than 2.5 grams but not more than 10 grams of any substance containing cannabis,” a Class B misdemeanor (Ill. Rev. Stat. 1973, ch. 5614, par. 704(b)). The Appellate Court, Fifth District, reversed because it found that the State had failed to “prove beyond a reasonable doubt that the substance defendant possessed was cannabis” (49 Ill. App. 3d 40, 43), and we granted the State’s petition for leave to appeal. We affirm.
On April 25, 1974, on the basis of a tip from the Lawrence County sheriff’s office, Lawrenceville Police Chief Steve Foster and Patrolman Don Wolfe proceeded to the Lawrenceville High School parking lot to apprehend the defendant, apparently in the belief that he was involved in an illegal drug transaction. Wolfe found the defendant standing among a group of other students next to Barney Bass’ car. Wolfe then approached the defendant and told him to empty his pockets. The defendant complied, but Wolfe found no contraband among the contents of Park’s pockets. Nonetheless, on Chief Foster’s instructions, Wolfe transported the defendant to the Lawrenceville city hall, while Lawrence County Sheriff Richard Hunnicutt escorted Bass to city hall in Bass’ car. Shortly after their arrival, while searching Bass’ car, Hunnicutt observed a yellow envelope on the pavement near Bass’ feet.
At trial, Bass admitted that he had taken the envelope from his car, but claimed .that the defendant had put it there. Wolfe and Foster testified that, after being advised of his Miranda rights, defendant admitted having been in possession of the envelope and admitted that it contained marijuana. Defendant testified that he had not been so advised, that he had not made any such admission, and that he had never seen the yellow envelope before Foster *207showed it to him at city hall.
Harry Hardin, a licensed, registered pharmacist, testified that he had weighed the envelope in question and determined its weight to be six grams. However, the only direct evidence that the envelope contained cannabis was the testimony of Deputy Sheriff Billy Carrico. The day before the trial (four months after defendant’s arrest), Carrico had been summoned to the Lawrenceville city hall to examine the contents of the envelope and determine whether they contained cannabis. The State offered to have Carrico testify as to the results of his testing the substance with his “Narco test kit.” However, because the State was unable to explain the probative value and reliability of the results of the “Narco test kit,” the court excluded all evidence regarding the use of the kit. The State now argues that exclusion of the results of the “Narco test kit” was error. .In light of our disposition of the case, we do not reach that issue. The court did, however, permit Carrico to testify that, as an expert, he could identify marijuana by “feel, smell, texture and looks.”
To establish Carrico’s expertise, the State elicited his testimony that he had been a deputy sheriff for about four years, and that on at least 40 occasions during that period he had handled what he believed to be marijuana. However, because of the manner in which the State framed its next question (“Q. In any of these 40 instances, have these drugs been identified positively by experts as being marijuana? A. That’s correct”), we only know that Carrico’s identification of a substance as being marijuana had been proved correct on as few as 1 of the 40 occasions (2.5%). Carrico further admitted that he was not trained to identify marijuana, but was relying upon having seen what he believed to be marijuana on several occasions. Despite defendant’s arguments to the contrary, the trial court considered this experience sufficient to establish Carrico’s expertise in the identification of marijuana, and permitted Carrico to tell the jury that, in his opinion, the substance *208in the envelope was marijuana.
After an exhaustive review of the question, we feel constrained to disagree with the trial court’s ruling on this issue. As this court observed seven years ago:
“Knowledge of the characteristics and effects of the drug commonly called marijuana, which is obtained from the leaves of the female hemp plant, cannabis sativa, has developed rapidly in the last decade.” (People v. McCabe (1971), 49 Ill. 2d 338, 342.)
The pace of that knowledge explosion has not slowed during the past seven years, and one of the areas of knowledge affected has been that of the reliability of certain means of identifying marijuana. Statistics from the State Crime Laboratory in our neighboring State of Wisconsin are quite startling in this regard. “During the period March 1970 to March 1971, 1674 samples of marihuana, morphologically identified as such, were submitted to the Wisconsin Crime Laboratory for confirmatory testing. Only 85.6 percent of these were in fact marihuana. Therefore, 14.4 percent, or one in every seven samples, turned in as suspected marihuana were not marihuana.” (Stein, Laessig, & Indriksons, An Evaluation of Drug Testing Procedures Used by Forensic Laboratories and the Qualifications of Their Analysts, 1973 Wis. L. Rev. 727, 770 (hereinafter Drug Testing Procedures).) At the very least, these statistics demonstrate that even if it is possible, as Carrico claimed, to reliably identify cannabis in the manner he claimed to have used (feel, smell, sight and touch), such means are highly prone to error in the hands of anyone but an expert, because of the number of plants whose gross morphological characteristics closely resemble Cannabis sativa L. See Drug Testing Procedures, at 770-71.
Thus, the admissiblity of Carrico’s opinion that the substance in question was marijuana hinges upon whether, *209despite his complete lack of training on the subject, Carrico’s limited experience as a deputy sheriff qualified him to reliably identify the substance in question as cannabis through the means he allegedly used to reach that conclusion. As a threshold matter, we note that the burden of establishing the qualifications of an alleged expert witness is that of the proponent of his testimony — in this case the State. As Professor Wigmore has so emphatically stated:
“The possession of the required qualifications by a particular person offered as a witness, must be expressly shown by the party offering him. This follows from the nature of the situation and is universally conceded.” (Emphasis in original.) (2 J. Wigmore, Evidence 640-41 (3d ed. 1940).)
That the defendant had an opportunity to cross-examine Carrico as to his qualifications therefore did not eliminate or in any way reduce the State’s burden of establishing those qualifications. Also, while the question of whether the State had met that burden was a matter within the discretion of the trial court, its decision is subject to reversal if it constituted an “abuse of discretion.” (People v. Fiorita (1930), 339 Ill. 78, 89.) Here, we conclude that while the trial court diligently pursued this question with counsel in chambers, its admission of Officer Carrico’s “expert” testimony was an abuse of discretion.
Although the resolution of this issue requires us to venture into fields of human knowledge as to which we, as judges, possess no special expertise, we are not without benchmarks in our journey. By 1942, the qualifications of expert witnesses had been reviewed in over 200 reported cases in Illinois. (King & Pillinger, A Study of the Law of Opinion Evidence in Illinois 58 (1942).) Examination of this wealth of judicial experience is central to our analysis of the question presented by this case. The rule which we derive from those cases is that the degree and manner of *210knowledge and experience required of an alleged expert is directly related to the complexity of the subject matter and the corresponding likelihood of error by one insufficiently familiar therewith. Thus, in People v. Palmer (1932), 351 Ill. 319, 324, prior ownership and marketing of hogs was sufficient practical experience and special knowledge to render an opinion as to value of a pregnant sow; in People v. Hayes (1960), 20 Ill. 2d 319, a police officer with 15 years’ experience with stolen auto parts was qualified to testify that certain stolen auto parts were worth more than $50; in State v. Ajster (1925), 318 Ill. 230, 234, prior drinking of liquor qualified a witness to testify that a certain beverage contained more than .5% alcohol on the basis of its taste; in Webster Manufacturing Co. v. Mulvanny (1897), 168 Ill. 311, 314, steam fitters with some experience in running engines “but not to any great extent” were competent to testify to the cause of an explosion in a steam pipe connecting a steam engine with a boiler; and in Keating v. People (1896), 160 Ill. 480, 487, a paying teller of a bank was qualified to render an opinion of the genuineness of a bank note. On the other hand, in People v. Jennings (1911), 252 Ill. 534, 547, fingerprint experts were qualified to render opinions where one had 6 years of study and over 2,000 prior identifications; another had 6 years’ study, extensive study on the subject under a Scotland Yard inspector, and 2,500 prior identifications; another had 6 years’ experience, studied at Scotland Yard, passed an examination on the subject, and started the first Bureau of Identification in Washington, D.C.; and the final expert had 27 years in the Bureau of Identification, 6 years’ study of prints, and over 6,000 prior identifications. However, significantly, in People v. Fiorita (1930), 339 Ill. 78, 89, a police employee was not qualified as a ballistics expert merely by engaging in identification of guns for several months where he had no other experience or education; but in People v. Green *211(1963), 28 Ill. 2d 286, 288, a microanalysis expert was qualified by his college education, extensive study in the field, and over 2,000 prior analyses. And we note also that police officers may not be presumed to possess the requisite expertise to identify a narcotic substance. See People v. Sanford (1962), 24 Ill. 2d 365 (identification of heroin).
Thus, while the value of pregnant sows and stolen auto parts might be readily assimilated with only limited experience, the ability to draw fine distinctions between fingerprints and ballistic markings requires more thorough and systematic study. The positive identification of cannabis without the aid of chemical and microscopic analysis falls into the latter category, because it simply is far too likely that a nonexpert would err in his conclusions on this matter, and taint the entire fact-finding process. Cross-examination often is a clumsy, counterproductive and therefore ineffective means of purging that taint. Accordingly, we hold that Carrico’s statement that the substance was in fact marijuana should not have been admitted and did not help to establish this element of the State’s burden. See People v. Williams (1962), 25 Ill. 2d 562.
We next consider whether the remainder of the evidence was sufficient to establish beyond a reasonable doubt that the substance in question contained cannabis.
We agree with the appellate court’s finding that the . evidence was not sufficient to meet the State’s burden of proof on this question. While we have no doubt that the State produced evidence from which the jury could have concluded that the defendant believed the substance contained cannabis, the defendant was not charged with possession of a substance which he believed to contain cannabis. Rather, he was charged with possession of a substance which in fact contained cannabis. The State therefore was required to prove the actual identity of the substance in question beyond a reasonable doubt. People *212 v. Sanford (1962), 24 Ill. 2d 365, 368 (State must prove substance was heroin).
While we agree with the general proposition put forward by the State (and conceded by the defendant) that circumstantial evidence can be used to establish that a substance contains cannabis (see, e.g., People v. Robinson (1958), 14 Ill. 2d 325, 332 (identification of narcotic substance by circumstantial evidence)), we do not agree that the particular circumstantial evidence adduced in this case was sufficient.
Park’s own alleged admission was insufficient to convict him of the offense charged. Even assuming that Park’s alleged statement was admissible (an issue which neither we nor the appellate court reach), its probative value or weight is limited, in the absence of substantial evidence that this 17-year-old defendant had some means of knowing that the substance in question contained cannabis. Although neither the State nor the defendant has cited it, the well-established rule governing this situation is that requiring substantial, independent corroboration of the defendant’s alleged admission of the elements of the corpus delicti. (See People v. Hubbard (1967), 38 Ill. 2d 104, 110.) The reasons for this ancient rule are many, but underlying them all is the general problem of the trustworthiness of such alleged admissions and a reluctance to indulge in the fiction that the jury can accurately determine the weight to be given them. (See generally Bergen v. People (1856), 17 Ill. 425, 427.) Here, the “corroborating circumstances” (as well as the circumstantial evidence taken as a whole) were plainly insufficient.
This becomes readily apparent upon comparison of this case to the cases cited by the State. It is undisputed that there was no proof of any sort of “sale” of drugs at a high price suggesting that the principals had taken steps to assure themselves of the identity of the substance. (Cf. *213 People v. Harrison (1962), 26 Ill. 2d 377, cert denied (1963), 373 U.S. 928, 10 L. Ed. 2d 426, 83 S. Ct. 1531; United States v. Lawson (7th Cir. 1974), 507 F.2d 433, cert. denied (1975), 420 U.S. 1004, 43 L. Ed. 2d 762, 95 S. Ct. 1446; United States v. Fiotto (2d Cir. 1972), 454 F.2d 252; United States v. Agueci (2d Cir. 1962), 310 F.2d 817.) Neither is the quantity involved so large as to support a similar inference (cf. People v. Rege (1976), 64 Ill. 2d 473; United States v. Bentvena (2d Cir. 1963), 319 F.2d 916); nor is there any evidence of the effect of the substance upon a human subject (cf. People v. Robinson (1958), 14 Ill. 2d 325, 332; United States v. Dolan (4th Cir. 1976), 544 F.2d 1219; Ewing v. United States (9th Cir. 1967), 386 F.2d 10). In short, the State would have us cure its lack of diligence in preparing its case by overlooking the substantial possibility that, on the basis of the evidence presented, the substance in question was not cannabis. This we cannot do.
One of the chief safeguards of our liberty is the requirement that, before punishing an individual as a criminal, the executive branch of government must prove to the satisfaction of the judicial branch of government that the individual has violated the laws enacted by the legislative branch of government. Any relaxation of this standard poses the gravest possible threat to our most basic institutions. While we must also take care not to unnecessarily impede the State from dealing effectively with the vexatious problems of illegal drug traffic which plague our society (see generally People v. McCabe (1971), 49 Ill. 2d 338, 354 (Ryan, J., dissenting); see also section 1 of the Cannabis Control Act (Ill. Rev. Stat. 1973, ch. 5614, par. 701)), the requirement that the State provide more substantial evidence than it did here is but a minor burden.
To determine accurately that a particular substance contains cannabis, all that is necessary is microscopic examination combined with the Duquenois-Levine test. *214When observed at 60x magnification through an ordinary-compound microscope, cystolith “hairs” are visible on leafy material from Cannabis sativa L. While certain other plants also have such cystolith hairs, apparently none of them has been demonstrated to test positive under the Duquenois-Levine test. That test requires only readily available prepackaged materials and is easy to perform. The substance in question is dissolved in hydrochloric acid and combined with the Duquenois reagent (vanillin dissolved in a solution of acetaldehyde and alcohol). If the mixture produces a violet color, chloroform is added to the mixture and if the violet color is then confined to the bottom two layers of the resulting mixture, the test is positive for cannabis. Although substances other than cannabis may produce a “false positive” with this test, apparently none of them has been demonstrated to have cystolith hairs. Thus, in combination, the two tests are highly reliable. (See Nakamura, Forensic Aspects of Cystolith Hairs of Cannabis and Other Plants, 52 J.A. Off’l Anal. Chem. 5 (1969). See also Drug Testing Procedures, at 746-71; State v. Wind (1973), 60 Wis. 2d 267, 270 n.3, 208 N.W.2d 357, 360 n.3.) The existence and widespread use of such tests (see generally Annot., 75 A.L.R.3d 717 (1977)) make all the more troublesome the manner in which the State sought to meet its burden of proof in the instant case.
For the foregoing reasons, the judgment of the appellate court is affirmed.
Judgment affirmed.
MR. JUSTICE KLUCZYNSKI took no part in the consideration or decision of this case.