MR. JUSTICE ERICKSON delivered the opinion of the Court.

The defendant Andrew Joe Garcia and the defendant Kenneth Ray Feland have both asserted that the present procedures used in conducting breath tests to determine blood alcohol percentage in a drunk driving prosecution deprives them of the opportunity to test the accuracy and validity of the procedure. Determination that a person had a blood alcohol content of .10 percent creates a presumption that he was operating a vehicle while under the influence of intoxicating liquors. Section 42-4-1202, C.R.S. 1973 (1976 Supp.).

We granted certiorari to review the Mesa County District Court decision in the Garcia case. We also granted certiorari to review the Jefferson County District Court decision in the case wherein Kenneth Ray Feland was charged with driving while intoxicated.

The issues in both cases are similar, and for that reason we have consolidated the cases for the purpose of this opinion. Our determination to review these two cases is based upon issues that were unanswered in People v. Hedrick, 192 Colo. 37, 557 P.2d 378 (1976). We reverse the convictions of both Garcia and Feland and return the cases to the respective district courts with directions to remand to the respective county courts for new trials. Evidence apart from the blood alcohol tests may in and of itself be sufficient to establish guilt.

[1] The procedures which must be followed and which will be discussed in this opinion have prospective effect and are not to be considered retroactively, except as to the parties to these proceedings and as to those cases wherein motions for production of breath samples or breathalyzer ampoules have already been made. Lauderdale v. State, 548 P.2d 376 (Alaska 1976).

Both defendants were tried in the county court and found guilty of driving under the influence of intoxicating liquors. Section 42-4-1202, C.R.S. 1973 (1976 Supp.). Both defendants appealed to the district court where their convictions were affirmed. Both defendants sought to suppress evidence of the blood alcohol tests which were conducted with breath analysis equipment shortly after their arrest.

[2,3] Samples of blood and urine which are taken in blood alcohol tests are customarily preserved for the use of the defense and to insure that the test is accurate. See note 1, infra. Preservation of the blood, urine, or breath which formed the basis for the conclusion that a person was operating a vehicle while under the influence of intoxicating liquors is essential in view of the presumption that arises from the test. Section 42-4-1202 provides, in pertinent part:

" Driving under the influence — driving while impaired — implied consent to chemical tests — penalties. (1)(a) It is a misdemeanor for any person who is under the influence of intoxicating liquor to drive any vehicle in this state . . . .

"(2) In any prosecution for a violation of subsection (1)(a) or (1)(b) of this section, the amount of alcohol in the defendant's blood at the time of the commission of the alleged offense or within a reasonable time thereafter, as shown by chemical analysis of the defendant's blood, urine, or breath, shall give rise to the following presumptions:

. . . .

"(c) If there was at such time 0.10 percent or more by weight of alcohol in the defendant's blood, it shall be presumed that the defendant was under the influence of alcohol."

Different equipment was used in each of the cases before us, but the record before us in the combined cases establishes that breathalyzer procedures exist which permit the preservation of a breathalyzer sample for use by the defense.

[4] In People v. Hedrick, supra, there was no evidence to show that the products of the breath test could have been preserved for later use and for separate testing by the defense. In both the Garcia and Feland cases, there was ample evidence to reflect that the state of the art is such that samples of a defendant's breath, taken at the time a blood alcohol breath test is given, can be preserved at a cost far less than that which is required to preserve urine or blood samples. Because of the procedures which are available to insure fundamental fairness in the prosecution of drunk driving cases, we hold that procedures must be utilized which will cause a sample of the defendant's breath, taken at the time that a breath test is administered, be preserved and available to the defendant for separate testing as to blood alcohol content. The Facts People v. Garcia

An officer of the Grand Junction police department observed Andrew Joe Garcia operating a motor vehicle at an excessive speed. Garcia was stopped and given a roadside sobriety test. He was thereafter arrested for driving under the influence of alcohol and was advised of his rights under Colorado's Implied Consent Law, section 42-4-1202(3), C.R.S. 1973 (1976 Supp.). He elected to take the breath test. The breathalyzer reflected a blood alcohol content of .18%, and as a result, Garcia was charged with driving under the influence of alcohol.

The breathalyzer machine which was used in the Garcia case was manufactured by the Smith and Wesson Company. The manner in which the machine was used and the method for determining blood alcohol content of breath is described in this manner:

"The breathalyzer is a machine designed to measure the amount of alcohol in the alveolar breath and is based upon the principle that the ratio between the amount of alcohol in the blood and the amount in the alveolar breath from the lungs is a constant 2100 to 1. In other words, the machine analyzes a sample of breath to determine the alcoholic content of the blood . . . .

"To operate the machine, the subject blows into the machine through a mouthpiece until he has emptied his lungs in one breath. The machine is so designed that it traps only the last 52-1/2 cubic centimeters of air that has been blown into it. This air is then forced, by weight of a piston, through a test ampoule containing a solution of sulphuric acid and potassium dichromate. This test solution has a yellow hue to it. As the breath sample bubbles through the test solution, the sulphuric acid extracts the alcohol, if any, therefrom, and the potassium dichromate then changes the alcohol to acetic acid, thereby causing the solution to lose some of its original yellow color. The greater the alcoholic content of the breath sample, the greater will be the loss in color of the test solution. By causing a light to pass through the test ampoule and through a standard [reference] ampoule containing the same chemical solution as the test ampoule (but through which no breath sample has passed), the amount of the change in color can be measured by photoelectric cells which are connected to a galvanometer. By balancing the galvanometer, a reading can be obtained from a gauge which has been calibrated in terms of percentage of alcohol in the blood." State v. Baker, 56 Wash.2d 846, 355 P.2d 806, 809 (1960).

When arraigned, Garcia requested that the test and reference ampoules used in his breathalyzer test be produced. He was informed that in accordance with the standard police department procedure, both ampoules had been destroyed. Garcia then moved to suppress the results of the breathalyzer test as evidence. His motion was predicated on the contention that had the test and reference ampoules, or a separate sample of his breath, been preserved, he would have been able to submit those items for an independent test of his blood alcohol level. He contended that the test could have been conducted by experts of his own choosing and would have been available to impeach or verify the result indicated by the breathalyzer.

A stipulation appears in the record to establish that the breathalyzer machine which was used to test Garcia's breath was certified by the Colorado State Board of Health, was in proper working order, and was operated by a qualified person. Expert testimony was produced in the trial court to establish the manner in which the breathalyzer machine was operated and the feasibility of maintaining the test and reference ampoules and a separate breath sample. There was conflicting testimony, but it was conceded that a sample of Garcia's breath could have been preserved at a minor expense by a number of methods. Experts called by the defense testified that the preservation of the test and reference ampoules would have enabled them to determine whether the proper amount of solution was in each vial and whether the proper amount of potassium dichromate was present in the solution, as well as whether the glass ampoules were of the correct diameter and free of imperfections. The defense expert testified that the standard solution (potassium dichromate), the glass ampoule's diameter, and the other factors which were outlined could have affected the accuracy of the test. It was also the opinion of the defense experts that the preservation of a breath sample would have enabled an independent laboratory test.

The trial court concluded that the test and reference ampoules could not easily be preserved and that, thus, the state was under no obligation to preserve the test and reference ampoules for Garcia, but it did not make any findings regarding the feasibility of preserving a separate sample of Garcia's breath for use by the defense.

People v. Feland

A police officer stopped Kenneth Ray Feland for driving a motor vehicle at an excessive speed. The officer administered roadside sobriety tests, and Feland was then taken to the Arvada police department for a blood alcohol test. Feland agreed to take a breath test, and that test was performed with a "Luckey Alco-Analyzer" gas chromatograph. No evidence was offered which creates an issue as to either the accuracy of that particular machine or the manner in which the test was administered.

The gas chromatograph measures the percent of alcohol in the blood. It first mixes the suspect's breath with any one of several inert gases. The resulting mixture is then burned in a hydrogen flame which is surrounded by an electrical field. The ions formed by combustion contribute to the existing electrical field, and variations in that field are measured by an electrometer. Those measurements are recorded on a card, and the various readings obtained are then compared with a standardized form to obtain a measurement of the suspect's blood alcohol level. See R. Erwin, Defense of Drunk Driving Cases § 23.01 (3d ed. 1978). Before the test is administered, the machine is calibrated by passing a standard solution containing .10 percent alcohol through it.

The gas chromatograph destroys the sample of the suspect's breath which it analyzes. Thus, unlike the breathalyzer machine, the gas chromatograph does not preserve any substance, such as the test ampoule, which could be made available for independent testing by a defendant.

Prior to trial, Feland asked that he be provided with a sample of his breath taken at the time of the blood alcohol test and a sample of the standard solution used to calibrate the machine at that time. He was informed that, in accordance with the standard policy of the police department, no separate sample of his breath had been preserved, and that the standard solution was no longer available.

Feland then moved to suppress his blood alcohol test as evidence. At the suppression hearing, Feland's expert witness testified that at the time of the test several inexpensive and practical methods existed to preserve a separate sample of Feland's breath. The district attorney in Feland's case has conceded that it is possible to preserve such a sample.

Feland's expert witness further testified that, had a separate sample of Feland's breath been preserved, it would have been possible to retest his blood alcohol level with greater precision at a later date and under laboratory conditions. He also expressed the opinion that if a sample of the standard solution had been retained, that solution could have been tested to determine whether it had been properly prepared. The trial court denied the motion to suppress.

The trial courts in both of these cases justified their denial of the defendants' motions to suppress on the basis of our decision in People v. Hedrick, supra. People v. Hedrick, however, was not grounded on this court's conclusion that the state could never be required to preserve a sample of defendant's breath for independent testing. Rather, People v. Hedrick was predicated on a record that demonstrated "a complete lack of evidence . . . regarding the type of machine used in the test on the defendant," and the fact that a breath sample could be preserved. Id. at 381. In People v. Hedrick, we said:

"Neither defendant not the prosecution presented any evidence of the type of machine used in administering the test. The only testimony was from the defendant, who stated he had freely chosen the breath test and testified, inter alia: 'Well, I don't feel 1.28 [sic] was under the influence. I don't feel like I was under the influence, so I don't think the test was right.' Id. at 379.

"In summary, where there is a failure, as here, to prove that the evidence is preservable or that there was any prejudice to defendant by failure to have available to him a breath sample, we must hold that the wider interests of society favor the admissibility of the test results at trial." Id. at 382.

In both cases which are now before us, the record is replete with evidence that a sample of the defendant's breath could have been preserved inexpensively and expediently. At defendant Garcia's trial, the evidence was uncontradicted. In defendant Feland's case, the district attorney had conceded that methods exist to preserve a defendant's breath. Moreover, the defendant in each case presented competent evidence that a separate breath sample could have been independently tested and the results of that test used to "impeach" the "testimony" of the machine which tested his breath. Lauderdale v. State, supra, at 381; see also, Davis v. Alaska, 415 U.S. 308, 316, 318, 94 S.Ct. 1105, 39 L.Ed.2d 347 (1974); Alford v. United States, 282 U.S. 687, 692, 51 S.Ct. 218, 75 L.Ed. 624 (1931); U.S. Const., amend. VI; Colo. Const., Art. II, Sec. 16. Thus, the vacuum in the record which dictated our decision in People v. Hedrick, supra, has been filled by the evidence produced by the defendants in these cases.

[5] In People v. Hedrick, supra, we set forth three tests to determine whether a denial of discovery infringed upon the due process right of a defendant to have the state make available to him potentially exculpatory information: (1) whether the evidence was suppressed or destroyed by the prosecution; (2) whether the evidence is exculpatory; and (3) whether the evidence is material to the defendant's case. Moore v. Illinois, 408 U.S. 786, 794-5, 92 S.Ct. 2562, 33 L.Ed.2d 706 (1972); Brady v. Maryland, 373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963).

[6] The breath samples requested by the defendants are obviously material to the proof of the drunk driving charges. Section 42-4-1202(2)(c), C.R.S. 1973 (1976 Supp.) provides that a blood alcohol level of .10 percent gives rise to a presumption that the defendant was driving under the influence of alcohol. A defendant's ability to attack the validity of that presumption is, thus, dependent upon his ability to attack the accuracy of the machine which tested his blood alcohol level. State v. Lauderdale, supra.

[7] It is, of course, impossible for us to determine whether the breath samples requested by the defendants would have exculpated them. It is not necessary for a defendant to demonstrate that the evidence he seeks to discover, but which is no longer available for examination by the court, would have been favorable to him, People v. Harmes, 38 Colo. App. 378, 560 P.2d 470 (1976), so long as that evidence is not merely "incidental" to the prosecution's case or to the defendant's affirmative defense. People v. Bynum, 192 Colo. 60, 556 P.2d 469 (1976). It is sufficient that the material requested "'might' be 'favorable' to the accused." United States v. Bryant, 439 F.2d 642, 652, n. 21 (D.C. Cir. 1971).

[8,9] The final question to be answered is whether the evidence requested by the defendant has been suppressed or destroyed by the prosecution. In these cases, the separate samples of the defendant's breath have been neither suppressed nor destroyed — they were never collected and preserved. But this fact does not mean that the defendants' case must fall. The failure of the state to collect and preserve evidence, when those acts can be accomplished as a mere incident to the procedure routinely performed by state agents, is tantamount to suppression of that evidence. It is incumbent upon the state to employ regular procedures to preserve evidence which a state agent, in the regular performance of his duties, could reasonably foresee "'might' be 'favorable' to the accused." People v. Bryant, supra; People v. Poole, 192 Colo. 56, 555 P.2d 980 (1976); State v. Maloney, 105 Ariz. 348, 352, 464 P.2d 793, cert. denied, sub. nom. Maloney v. Arizona, 400 U.S. 841, 91 S.Ct. 82, 27 L.Ed.2d 75 (1970); Comment, Judicial Response to Governmental Loss or Destruction of Evidence, 39 U. Chi. L. Rev. 542 (1972).

The fact that a .10 percent blood alcohol reading from the devices used in the present cases is sufficient to bring about a defendant's conviction militates against a conclusion that independent testing of the accuracy of those machines is unnecessary.

"The prosecution has the duty to prevent the loss or destruction of evidence that is favorable to the accused . . . . Trial by ambush is no longer acceptable as a means for ascertaining the truth. In the past, we have endorsed liberal discovery procedures in criminal cases . . . ." People v. Roblas, 193 Colo. 496, 568 P.2d 57, 60 (1977).

[10] The trial of a criminal case is not a game of fox and hounds in which the state attempts to outwit and trap a quarry. Giles v. Maryland, 386 U.S. 66, 87 S.Ct. 793, 17 L.Ed.2d 737 (1967) (Fortas, J., concurring). It is, instead, a sober search for truth, in which not only the resources of the defendant, but those readily available to the state must be put to work in aid of that search.

[11] We hold, therefore, that in all cases where a defendant elects to submit to a breath test to determine his blood alcohol level, he must be given a separate sample of his breath at the time of the test, or the alcoholic content of his breath in a manner which will permit scientifically reliable independent testing by the defendant, if that test is to be used as evidence. United States v. Bryant, supra, 439 F.2d at 652; People v. Hitch, 113 Cal. Rptr. 158, 11 Cal.3d 159, 520 P.2d 974 (1974); 111 Cal. Rptr. 9, 12 Cal.3d 641, 527 P.2d 361 (1974).

Accordingly, the evidence relating to the breath tests and any derivative evidence is suppressed in both cases, a new trial is ordered for both defendants, and the cases are remanded for further proceedings not inconsistent with the directives contained in this opinion.