In this original proceeding under C.A.R. 21, we issued a rule directing the Arapahoe County District Court to show cause why it should not be prohibited from reducing the charge against the defendant from first degree murder1 to second degree murder2 as a sanction for failing to preserve the victim’s hands in a condition suitable for trace metal testing by the defendant. We now discharge the rule.
The defendant, Thomas Reynolds, was charged with first degree murder based on the shooting of his wife, Elaine Reynolds. The defendant’s theory appears to be3 that the victim held the pistol from which the fatal bullet was fired and that the gun discharged during a struggle in which the defendant was attempting to defend himself against an attack by his wife.
Elaine Reynolds was shot at the family home during the early morning hours of February 28, 1982. Officer Garbett of the Aurora Police Department was summoned to the scene and was in charge of the investigation. While the initial on-scene investigation was in progress, a deputy district attorney asked Garbett to obtain a trace metal test of the victim’s hands. Defense counsel also requested that the police perform such a test and asked that the victim’s hands be covered with bags until the analysis could be accomplished. The purpose of the requested trace metal test was to attempt to determine whether the victim had held the gun. After receiving these requests, on the morning of the same day Elaine Reynolds was shot Garbett asked the supervisor of the police department crime laboratory to perform the trace metal test. The supervisor stated that his laboratory would not conduct the examination, but also said that if Garbett wished to ask one of the laboratory workers to accomplish such an analysis as a personal favor the supervisor would not interfere. Garbett advised defense counsel on that same morning that the police department crime laboratory would not perform the test.
Bags were placed around the victim’s hands at the death scene at the request of the police crime laboratory supervisor. The body was then taken to a mortuary, where an autopsy was performed.4 No defense representative was present at the autopsy; the record does not indicate that defense counsel was notified and given an opportunity to appear. After the autopsy, the body was transported to another mortuary for embalming. A mortician began preparing the body for burial shortly after midnight on March 2, 1982. When the mortician began his work there were no bags on the victim’s hands. As part of the usual procedures to prepare a body for burial, the mortician massaged the victim’s hands with a solution of surgical soap and Clorox *1285bleach to obtain better circulation of the embalming fluid and later washed and scrubbed the hands with the same solution to remove dried blood resulting from the postmortem examination. On the evening of March 2,1982, at the request of Garbett, an Aurora Police Department crime laboratory technician performed a trace metal test on the victim’s hands.
The defendant moved to dismiss the murder charge on the ground that by failing to preserve the victim’s hands in a condition suitable for trace metal testing the prosecution had violated his right to due process of law. The trial court held a hearing at which Garbett, the crime laboratory supervisor, the mortician, and a Colorado Bureau of Investigation (CBI) expert on trace metal tests gave testimony. The CBI expert explained that the trace metal test was devised to determine whether a person has held a metal object in the recent past. It is a chemical test and is based upon the scientific fact that when a human hand touches certain metals perspiration reacts with the metal and a minute layer of metal is deposited on the tissue of the hand. After a hand has been sprayed with the appropriate chemical compound and is exposed to a particular kind of light, it will fluoresce in colors that vary depending on the type of metal present.
Although in the most successful analyses it is possible to discern on the hand an image of a pattern appearing on the metal, only in approximately four percent of tests conducted under laboratory conditions on a subject who has held metal is it possible to identify the particular object that was grasped. The absence of a positive reaction does not necessarily mean that the subject has not held a metal object,5 nor does a positive reaction necessarily mean that the object that produced it was a weapon. Where a positive reaction is obtained it is not possible to tell at what time in the recent past the subject touched the metal object. Notwithstanding these shortcomings, the CBI expert was of the opinion that such a test should be conducted in all cases where there is a question whether the subject held a metal weapon. His opinion was based principally on the fact that the four percent possibility of definite evidence that the subject held a particular weapon is too significant to ignore.
The CBI expert also testified that, although he could not state definitely the manner in which the washing of hands with the Clorox bleach and surgical soap solution would affect the trace metal test, he would expect some reaction, especially because Clorox is a strong chemical. He also noted that many soaps contain zinc, which produces its own characteristic color on trace metal testing.
The police crime laboratory supervisor’s testimony, although not so detailed as that of the CBI expert, was not significantly in conflict with it. The supervisor was self-taught in trace metal testing, had engaged in periodic testing for eight to ten years for the Aurora Police Department, and had done his last testing six months earlier. He stated that over the years he had conducted about seventy-five to one hundred tests, about twenty-five percent of them in field situations. The laboratory supervisor was of the opinion that because a negative test result is not conclusive that the subject held a metal object and because a positive reaction does not necessarily establish that the source of the metal was a weapon, the test is too inconclusive to be useful. That is the reason he assigned for deciding not to conduct the trace metal test notwithstanding the requests by the investigating officer, deputy district attorney, and defense counsel.
The trial court found the prosecution’s position at the motion hearing that the trace metal test has no value to be implausible, particularly in light of the long history of use of the test by the Aurora Police Department and the testimony of the CBI expert as to the validity and usefulness of *1286the analysis. The court ruled that under the circumstances here, including the fact that the defendant relies on the theory of self-defense, the prosecution had an affirmative duty to preserve the victim’s hands in a condition suitable for testing on behalf of the defendant. The court concluded that the failure to perform that duty deprived the defendant of due process of law.
As a remedy for the due process violation, the trial court ordered the charge reduced from first degree murder to second degree murder. In.addition, it vacated its earlier determination that a capital offense was involved and that the proof was evident or the presumption great,6 and set bail. The trial court also suppressed from use as evidence in the prosecution’s case the results of the trace metal test conducted after the victim’s hands had been cleansed by the mortician. In this original proceeding the People challenge only the order reducing the charge to second degree murder.
Although we have considered on several occasions whether due process of law was violated by prosecutorial conduct depriving a defendant of relevant evidence and the appropriate remedies for such an infringement of a defendant’s constitutional rights, this case presents a novel fact pattern for application of familiar legal principles. We first address the question of whether the defendant’s due process rights were violated and then consider whether the relief granted by the trial court was appropriate.
I.
In determining whether the defendant’s due process rights were violated by the failure of the prosecution to conduct a trace metal test or to make the victim’s body available to the defendant for testing we begin with the principles established in the seminal case of Brady v. Maryland, 373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963), where the United States Supreme Court held
that the suppression by the prosecution of evidence favorable to an accused upon request violates due process where the evidence is material either to guilt or to punishment, irrespective of the good faith or bad faith of the prosecution.
Id. at 87, 83 S.Ct. at 1196-97, 10 L.Ed.2d at 218; accord, e.g., Moore v. Illinois, 408 U.S. 786, 92 S.Ct. 2562, 33 L.Ed.2d 706 (1972); People v. Garries, 645 P.2d 1306 (Colo.1982); Garcia v. District Court, 197 Colo. 38, 589 P.2d 924 (1979); People v. Hedrick, 192 Colo. 37, 557 P.2d 378 (1976). For convenience in analysis, the Brady v. Maryland criteria commonly have been broken down into three parts: (1) suppression by the prosecution after a request by the defense, (2) the evidence’s favorable character for the defense, and (3) the materiality of the evidence. Id.7 Only if all three parts of the test are met is there a due process violation. People v. Hedrick, supra. We address the application of the second, third and first parts of the Brady v. Maryland test in that order.
A.
The desired evidence — a trace metal test of the victim’s hands — was not obtained before the mortician scrubbed them, so it can never be known whether such a test would have provided evidence favorable to the defendant. In such circumstances “it is not necessary that the defendant prove the exculpatory value of the evidence so long as that evidence is not merely incidental to the prosecution’s case or the defendant’s de*1287fense.” People v. Morgan, 199 Colo. 237, 241, 606 P.2d 1296, 1299 (1980); accord, Garcia v. District Court, supra; People v. Harmes, 38 Colo.App. 378, 560 P.2d 470 (1976). We have required, however, that the defendant establish “the reasonable possibility that the evidence could have been of assistance to the defense.” People v. Morgan, 199 Colo. at 241, 606 P.2d at 1299.
In order to support Thomas Reynolds’ theory that Elaine Reynolds’ death was caused when he struggled with her to obtain possession of the gun with which she was assaulting him, it would be valuable to the defendant to establish that his wife had held the gun. Although the evidence shows that there is only a four percent chance of a positive finding that Elaine Reynolds held the death weapon even if she did, and notwithstanding that proof of that fact would not be conclusive that the events occurred as the defendant’s theory would portray them, we believe that a reasonable possibility that the evidence could have been of assistance to the defense has been established. As the CBI expert indicated, a four percent chance of determining a fact centrally significant to the defendant’s theory of the case in this first degree murder prosecution is simply too significant to ignore.
B.
We also hold that the materiality criterion has been satisfied. It will never be known whether the test would have shown that Elaine Reynolds held the gun. If she did, the evidence reflects that there is a four percent chance that the trace metal test would have supported the defense claim. The defendant specifically requested performance of the test. Had the test revealed that the victim had held the pistol, this would have tended to prove or disprove a matter at issue — whether Elaine Reynolds was the aggressor — and we conclude that is sufficient to meet the materiality test under the circumstances here. See People v. Holloway, 649 P.2d 318 (Colo.1982); People v. Morgan, supra; but see United States v. Agurs, 427 U.S. 97, 96 S.Ct. 2392, 49 L.Ed.2d 342 (1976).8
C.
Finally, we consider whether the trace metal test was suppressed by the prosecution.
This is not a case where the prosecution has failed or refused to furnish existing exculpatory material to the defense. Rather, the police were asked to perform an additional test. We have recognized that police investigators have no general duty to search out possible exculpatory evidence or to perform tests to determine marginally relevant facts that, with the benefit of hindsight, a defendant might speculate would have been of possible value to support his defense against a criminal charge. People v. Roark, 643 P.2d 756 (Colo.1982); People v. Hedrick, supra; People v. Culp, 189 Colo. 76, 537 P.2d 746 (1975); People v. Sasson, 628 P.2d 120 (Colo.App.1980). We have cautioned, however, that when evidence can be collected and preserved in the performance of routine procedures by state agents, failure to do so is tantamount to suppression of the evidence and that the state must 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. Gomez, 198 Colo. 105, 111, 596 P.2d 1192 (1979), cert. denied, 455 U.S. 943, 102 S.Ct. 1439, 71 L.Ed.2d 655; Garcia v. District Court, supra; see also People v. Shaw, supra; People v. Poole, 192 Colo. 56, 555 P.2d 980 (1976); People v. Harmes, supra; cf. People v. Hedrick, su*1288pra.9 We have also recognized that, when the police conduct scientific tests, they must preserve samples to permit the defendant to accomplish independent testing, permit the defendant’s experts to monitor the police testing, or provide some other suitable means to allow the defendant to verify independently the appropriateness of the procedures and the accuracy of the results of the testing. People v. Garries, supra; People v. Morgan, supra; People v. Gomez, supra; Garcia v. District Court, supra.
Applying these criteria to the facts here, we conclude that the police conduct was tantamount to suppression of evidence. The police laboratory had conducted trace metal tests in field situations on numerous occasions. It had the equipment and personnel to accomplish the testing. Both the prosecution and defense counsel specifically requested the test very shortly after the victim’s death. See People v. Thatcher, 638 P.2d 760 (Colo.1981) (the court recognized that a timely request by defense counsel may be relevant to the question of whether the prosecution suppressed the evidence).10 The only reason assigned by the police laboratory supervisor for his refusal to conduct the test was his opinion that the test was inconclusive. The trial court, while stopping just short of a finding of police bad faith,11 found implausible the prosecution’s argument that the test had no value. Although the police bagged the hands of the victim to protect them for future testing, they took the position that they had no responsibility for the body after it was taken from the crime scene. After the hands were bagged, no care was taken to preserve them in a condition suitable for testing.12 Moreover, knowing of the defendant’s desire for a trace metal test, no opportunity was given to his counsel to obtain an independent test before the mortician chemically cleansed the hands. The request by the defendant for a test was explicit and timely. It was never withdrawn. What we said in Garcia v. District Court, supra, is apt in the present case:
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. [Citation omitted.] 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.
197 Colo, at 47, 589 P.2d at 930.
Under the circumstances here we conclude that at a minimum the prosecution had the duty to preserve the hands in a condition suitable for testing and to make available to the defendant an opportunity for independent trace metal testing of the victim’s hands before burial. It failed to do so, with the result that the defendant has been deprived of due process of law.
II.
We next consider the appropriateness of the reduction of the charge from first degree to second degree murder as a remedy for the due process violation.
*1289The imposition of sanctions serves the dual purposes of protecting the integrity of the truth-finding process and deterring the prosecutor and the police from destroying material evidence. People v. Garries, supra; People v. Morgan, supra; see People v. Poole, supra. In furtherance of these purposes, the trial court has broad discretion in fashioning an appropriate remedy to protect a defendant’s rights where a due process violation has denied him access to evidence. People v. Holloway, supra; People v. Morgan, supra; People v. Poole, supra. That remedy, however, should be “no more restrictive than necessary to protect the defendant’s right to due process.” People v. Morgan, supra, 199 Colo, at 242, 606 P.2d at 1299; accord, People v. Holloway, supra. See Comment, Judicial Response to Governmental Loss or Destruction of Evidence, 39 U.Chi.L.Rev. 542 (1972). In selecting the remedy “the court must weigh the significance of the evidence lost or destroyed and the conduct of the prosecution which led to its loss or destruction.” People v. Poole, supra, 192 Colo, at 60, 555 P.2d at 983. In appropriate cases, dismissal of charges is a proper sanction. People v. Poole, supra; People v. Harmes, supra; cf. People v. Gillett, 629 P.2d 613 (Colo.1981) (dismissal of driving under the influence charges was an appropriate remedy for denial of the defendant’s statutory right to a blood-alcohol test).
In evaluating the appropriateness of the trial court’s order we note that, although the likelihood that the trace metal test would have shown that the victim held the gun is slight, such a showing would be highly material to the defendant’s claim that the gun was held by the victim and discharged during a struggle. In this connection we think it important that the present case is not one where, with the benefit of hindsight and after the possibility of testing has been foreclosed, the police are faulted for failure to conduct a test that they might reasonably have rejected as having only a negligible prospect for production of useful evidence. Cf. People v. Roark, supra (failure to test the alcoholic content of blood in absence of a defense request did not violate due process where the issue of intoxication was not central to the offense charged and was not shown to be relevant or material to guilt or innocence); People v. Hedrick, supra (it was relevant to the issue of whether the police suppressed evidence that the defendant’s first request for a breath sample was made nearly three months after the incident that served as a basis for the charge of driving under the influence of intoxicating liquor). Instead, not only the defense but also the prosecution requested the testing at a time when a simple testing procedure, commonly performed by this very police department in the past, could have been conducted. We find the opinion of the CBI expert to be compelling: in a case involving the serious charge of first degree murder, even a slight chance that a simple test would support the claim that the victim held the pistol is too significant to ignore. The prosecution has suggested no other way, except for the defendant’s own possible testimony, that this disputed fact can be determined. Compare this case with People v. Holloway, supra, and People v. Alonzi, 40 Colo.App. 507, 580 P.2d 1263 (1978), aff’d, 198 Colo. 160, 597 P.2d 560 (1979) (in each case alternative ways were available to establish the content of conversations recorded on a destroyed tape). Moreover, the trial court heard the evidence at the preliminary hearing, the record of which is not before us, and most probably is better informed than we as to the details of the defendant’s proposed defense and the importance of the lost evidence to his defense. See People v. Morgan, supra. Certainly, if the victim held the gun, that fact would be relevant to the question of whether or not she was shot “after deliberation,” an essential element of first degree murder but not second degree murder. See sections 18-3-102(l)(a) and -103(l)(a), C.R.S.1973 (1978 Repl.Vol. 8). Under these circumstances we cannot say that the trial court abused its discretion in reducing the charge to second degree murder as a reasonable remedy for the loss of material evidence as the result of the due process violation.
We discharge the rule.
. Section 18-3-102, C.R.S.1973 (1978 Repl.Vol. 8) (1982 Supp.).
. Section 18-3-103, C.R.S.1973 (1978 Repl.Vol. 8).
. Our understanding of the defendant’s proposed theory of defense is taken from the briefs filed in this case and inferences from the transcript of the hearing on the defendant’s motion to dismiss or suppress evidence.
.Officer Garbett testified that he could not remember the exact date of the autopsy but that it was either February 28, 1982, or March 1, 1982.
. The CBI expert testified that in only forty percent of tests conducted under laboratory conditions will a trace metal test yield positive results where the subjects have handled weapons.
. “All persons shall be bailable by sufficient sureties except for capital offenses, when the proof is evident or the presumption great.” Colo. Const. Art. II, § 19.
. The three part analysis is described as set forth in Moore v. Illinois, supra. Although our own descriptions of the three criteria have been essentially the same, the precise language of our formulations has varied somewhat from case to case. For example, we have frequently referred to the “favorable nature of the evidence” inquiry as a test of whether the evidence is “exculpatory.” People v. Garries, supra; People v. Hedrick, supra. Also, in our later cases we have not required a request by the defendant as an essential part of a showing of a due process violation, for there may be no defendant or no defense counsel at the time evidence is destroyed. People v. Morgan, 199 Colo. 237, 606 P.2d 1296 (1980).
. Because the results of a properly conducted trace metal test in this case are unknown, this case is distinguishable from United States v. Agurs, supra, and other cases where the evidence suppressed by the prosecution is available for review or the probable content of such evidence is clear to the reviewing court. In such cases, the standard for determining materiality is whether the evidence could have affected the outcome at trial in light of the other evidence presented. See People v. Shaw, 646 P.2d 375 (Colo.1982); People v. Gallegos, 644 P.2d 920 (Colo.1982); People v. Thatcher, 638 P.2d 760 (Colo.1981).
. In People v. Hedrick, supra, the absence of a showing that a breath sample could be simply preserved was important to a determination that no due process violation occurred when no sample was retained for testing by the defendant.
. Compare the present case with People v. Hedrick, supra (it was relevant to the determination of whether evidence had been suppressed that the defendant’s request for a breath sample was first made nearly three months after the breath test was conducted).
. The good faith or bad faith of the police in suppressing evidence is not determinative in the decision whether a due process violation has occurred. Brady v. Maryland, supra. “If the suppression of evidence results in constitutional error, it is because of the character of the evidence, not the character of the prosecutor.” United States v. Agurs, supra, 427 U.S. at 110, 96 S.Ct. at 2401, 49 L.Ed.2d at 353; accord, People v. Shaw, supra; People v. Morgan, supra. As discussed in part II of the present opinion, however, prosecutorial good faith is significant in selection of the appropriate sanction for a due process violation.
.The evidence established that, while bagging was important to preserve the hands in a condition suitable for testing for the presence of gunpowder, it was not critical to trace metal testing. The same cannot be said of the mortician’s scrubbing of the hands with the Clorox and surgical soap solution.