OPINION
SINGLETON, Judge.The state has petitioned for review of an order suppressing Intoximeter results in forty-four cases of driving while intoxicated, AS 28.35.033. We find that we must remand these cases to the district court.
In Anchorage v. Serrano, 649 P.2d 256, 258 (Alaska App.1982), we concluded that due process “requires the prosecution to make reasonable efforts to preserve a breath sample or to take other steps to allow a defendant to verify the results of the breathalyzer test” in those cases in which it wishes to admit the results of a breath test.
The state subsequently began using the Intoximeter 3000 breath testing machine. In People v. Jones, 118 Misc.2d 687, 461 N.Y.S.2d 962 (Albany Cty.1983), the machine was described as follows:
The Intoximeter 3000 is a breath-testing device which utilizes infrared energy and electrical current to detect the presence of ethyl alcohol (ethanol) in the breath of a driver. The device, as are all breath-testing devices, is based upon Henry’s Law and, unlike the more common breath-testers such as the breathalyzer, upon the Beer-Lambert Law of Absorption.
The Intoximeter 3000 is equipped with an infrared energy source of nichrome (an alloy of nickel, chromium and iron) surrounding a ceramic core. An electric current passing through this source *402causes two beams of infrared energy to be emitted in the direction of a two-chambered gas cell after reflecting off a colli-mating mirror. After traversing the gas cell, the infrared beams pass through a narrow band interference filter. This filter permits infrared energy with a wavelength ranging from 3.3 to 3.48 microns to pass through and strike a detector while at the same time blocking energy with wavelengths greater or lesser than that range. Thus the beam of energy striking the detector is modified so that it corresponds to one of the major absorption bands of ethyl alcohol.
The cell through which the infrared beams pass has two chambers. One, the reference cell, contains only room air. The other, the sample cell, contains during a test 900 cubic centimeters of the subject’s alveolar air. The device gives a reading of the amount of interfering substance in the subject’s breath by comparing the amount of infrared energy striking the detector after the two beams pass through the chambers, one through the sample cell, the other through the reference cell. Otherwise stated, the device compares the amount of infrared energy absorbed by the air from the lungs of the subject with the amount absorbed by the air from the room. If the ratio resulting from that reading (energy absorbed by the alveolar air/energy absorbed by the room air) is greater than one, there is present in the subject’s breath some substance which absorbs infrared radiation at 3.39 microns. The amount of interfering substance present is determined by the amount of infrared energy it absorbs. At this point it is impossible to conclude that the absorbing substance is ethyl alcohol because in addition to ethyl alcohol there are other substances which absorb radiation at 3.39 microns.
In order to enable the Intoximeter 3000 to give a specific reading for ethyl alcohol, another device is necessary. Thus the Intoximeter 3000, in addition to its infrared analysis of breath based on the Beer-Lambert Law of Absorption, also contains a semi-conductor (a Taguci sensor) by which it is able to distinguish ethyl alcohol from other substances which absorb infrared radiation in the area of 3.39 microns.
The Taguci sensor is a semi-conductor device the conductivity of which is influenced by the ambient air in the sample chamber. The conductivity of the semiconductor varies where there is present in that ambient air an oxidizible vapor such as ethyl alcohol or other hydrocarbon. Programmed into the memory of the computer that is mated to the Intox-imeter 3000 are the specific conductivity readings (in amperes) of the sensor when ethyl alcohol is present in the sample chamber at varying levels which correspond to the various blood-alcohol levels. These conductivity readings are predetermined empirically in the laboratory, and are specific for ethyl alcohol. Every other substance that absorbs infrared radiation at a wavelength of 3.39 microns produces a different current in the semiconductor. The computer incorporated into the device compares the outputs from the infrared sensor (amount of infrared energy absorbed) and the semiconductor (amperes of electrical current). If the semi-conductor reading does not correspond to that stored in the computer’s memory for the blood-alcohol level reported by the infrared sensor, an inter-ferent other than or in addition to ethyl alcohol is present in the subject’s breath. The computer performs a calculation to determine the difference between the reading for the blood-alcohol level reported by the infrared sensor.and the reading from the semi-conductor sensor, which difference automatically reduces the infrared reading by a corresponding amount. That adjusted amount is reported as the subject’s blood-alcohol level.
Operation of the Intoximeter 3000 is relatively simple and requires minimal operator intervention compared with more common breath-testing devices.
Each test commences with a twenty minute waiting period during which the *403subject is observed to insure that he does not ingest any alcohol, regurgitate or vomit. The operator presses a “start” button and then follows the machine’s commands to enter his name and identification number and the subject’s name. Following the last entry, the machine automatically blanks and purges to remove any residual alcohol fumes and to take a baseline reading, which ought to read “.00.” The machine then commands the subject to blow into the breath tube until the machine indicates that a sufficient sample has been entered. The In-toximeter then reports the subject’s blood-alcohol content on its display and proceeds to purge and blank itself. The machine then automatically runs a test with a reference solution, the ethyl alcohol content of which has been previously certified by the State Police Laboratory. Following the test with the reference sample, the machine again purges itself. The operator then presses the “print” key, whereupon the information previously entered and the results of the subject’s test, the reference sample test and the blank tests, the latter two being control tests to assure that the machine is functioning properly, are printed along with the times that the tests took place. The operator thus is required merely to type certain minimal information on the machine’s keyboard; he does not have to handle ampules of chemicals nor turn dials and levers as with more common breath testing devices.
The machine automatically safeguards against any contaminants in the room air and contains fail-safe devices to abort if there are room temperature or electrical or voltage problems, which rarely occur.
Id. at 963-65 (footnotes omitted).
The Intoximeter was certified for use by Alaska law enforcement agencies by the state Department of Health and Social Services in 1983, after extensive testing by a chemist employed by the department, Mr. Everett Clary. Kerr and the other respondents in this case do not challenge the state’s certification of the Intoximeter 3000.
/
We may quote the district court’s written order in this case for other relevant facts:
The Intoximeter 3000, a technologically advanced and approved instrument for breath testing, has been certified for use in Alaska. The Intoximeter is capable of preserving for external collection, the identical alcohol from the same breath sample which was analyzed by the Intox-imeter.
The external Collection Devices (hardware) which allow for the recapture of the analyzed breath from the Intoxime-ter are:
a. A metal L-shaped device, hereinafter referred to as an “adaptor”.
b. An 8 inch glass tube, hereinafter referred to as “Perchlorate Tube”. It can be roughly described as hourglass shaped. The center of the hourglass tube is filled with white powdery substance, magnesium perchlorate which is packed tightly into place with wads of cotton like fiber. The fiber plug is held in place with a plastic loop at each end. Each end of the glass tube is threaded. A conical polyethylene seal or clear plastic insert is inserted into each end of the tube and a black plastic cap is screwed onto each end of the Perchlorate Tube.
In order for the Intoximeter operator to recapture the alcohol from the analyzed breath sample, the operator must unscrew and remove both of the black plastic caps on the ends of the Perchlorate Tube. He must then remove both of the clear plastic inserts. He then screws the Perchlorate Tube snugly into the female threaded portion of the Adaptor. He then snugly plugs the flanged male portion of the Adaptor into the Intoxime-ter, so that the Perchlorate Tube will be vertical to the base of the Intoximeter and parallel to the side of the Intoxime-ter. When the permanent magnet on the Adaptor then makes contact inside the Intoximeter, the operator then presses a key on the Intoximeter which forces the heretofor tested breath sample through the inlet ports of the Adaptor and into *404and through the Perchlorate Tube. The Magnesium Perchlorate traps all the alcohol from the breath sample as the breath sample is forced through the Perchlorate Tube into the room air.
The operator then must unplug the Adaptor, unscrew the Perchlorate Tube from the Adaptor, reinsert the clear plastic inserts in each end of the Tube, and recap both ends of the Tube with the threaded black plastic caps. This should be done promptly to reduce any potential issue of contamination.
Before the magnesium perchlorate tube (MPT) system was certified, law enforcement officers were complying with our decision in Serrano by informing arrestees of their right to an independent blood test. See AS 28.35.033(e). In mid-1983, law enforcement agencies began using the MPT system.
In early 1984, DWI defendants began to file motions to suppress Intoximeter results based upon the argument that when their MPTs were tested, the results differed so greatly from Intoximeter results that they were of little use in verifying Intoximeter results. On February 8, 1984, a three-judge district court panel in Fairbanks heard evidence in support of a motion to suppress in State v. Eddie Logue, 4FA-S83-3771 Cr. On April 5 and April 9, more evidence was presented to the panel in the cases of State v. Bruce Hanson, et al., 4FA-S83-3398 Cr. The panel announced its decision, suppressing all Intoximeter results in the cases before it, on April 11, 1984. On May 31, 1984, written findings of fact and conclusions of law were issued by the panel.1 The panel’s holding was as follows:
Perchlorate sample preservation is capable of becoming a technologically sound, efficient and effective method of breath preservation. It has not reached that point. The perchlorate method of preservation and testing may yield accurate results in the future. Up to this point we found that the method as used in the field does not consistently produce scientifically acceptable results. For that reason we HOLD that the Intoxime-ter and Perchlorate Tube test results will be suppressed in the above captioned cases, as the defendants have been deprived of due process as set forth in the Serrano opinion.
The Perchlorate Tube method will be considered scientifically reliable when results from field samples that are tested, in a laboratory available to an accused, are shown ninety percent of the time to be consistently accurate within + 10% or — 15% of either the actual Intoximeter test results or actual blood test results. Until that time, we HOLD if only the Perchlorate Tube method of preservation of breath is used, the Intoximeter results are Suppressable upon a timely motion being made prior to trial UNLESS there is clear compliance with AS 28.35.033(e). The only other alternative open to the State is to select one of the other methods of breath preservation which the Court of Appeals found to be reasonable in cost and accurate in results.
On the same day the panel issued its decision, a hearing was held to determine the status of another group of cases arising after the Hanson cases. Essentially, the state was attempting to show that the procedures it had adopted since discovering the problems with the MPT system were sufficient to correct most errors. The cases in question arose in Fairbanks between March 1 and April 11, 1984. The panel eventually ruled that the state was not yet in compliance with its order in the Hanson cases, so that the Intoximeter results in the newer cases must be suppressed as well.2
*405The state petitioned for review. We granted the petition and asked the parties to discuss in their briefing the applicability of our decision in Anchorage v. Flack, 685 P.2d 108 (Alaska App.1984). As the case now stands, the state is seeking review only of the order relating to the second group of defendants.
In Flack, the arrestee was given an In-toximeter test, but no sample was preserved because no perchlorate tubes were available. The trial judge suppressed the Intoximeter results, but his written conclusions were ambiguous with respect to the standard used. Id. at 110. We held that the Serrano standard “is essentially one of negligence, not strict liability.”
If the governmental unit was free of fault in failing to provide the defendant a means of verifying the breathalyzer result, then suppression should not follow. We stress, however, that the duty is owed by the governmental unit, not its individual agents. Thus, a finding that an individual police officer or other person administering a breathalyzer examination was free from fault, would not satisfy Serrano if it was established that the governmental entity, through antecedent negligence, had created a situation in which individual officers administering the test could not provide the defendant a means of verification.
Generally, a person asserting another person’s negligence has the burden of proving it. We believe that there are some good reasons for departing from the general rule in this kind of case. First, as we pointed out in Serrano, in the typical case, the governmental agency will be in a position to furnish the defendant a basis for verification. 649 P.2d at 259. It is not unreasonable to require the governmental agency to establish that a particular case is not typical and justifies relieving it of the duty. Second, the circumstances preventing a governmental entity from complying with Serrano would be peculiarly within the knowledge of its agents. It is therefore not unreasonable to require it to establish its freedom from fault. We therefore conclude that a governmental entity seeking to excuse its failure to preserve a breath sample, or otherwise failing to enable a defendant to verify the results of a breathalyzer examination, has the burden of proving by a *406preponderance of the evidence its freedom from negligence.
Id. Since it was unclear whether the lower court had applied this standard we remanded for further proceedings. Id. at 111.
Flack governs the instant case. To the extent that the panel found that the system was not yielding accurate results, it may be said these defendants were placed in the same position as Flack: they had no independent means of verifying the Intox-imeter result. While the panel made detailed findings with regard to possible and actual sources of error in the MPT system, there was no explicit finding that the state’s negligence was the cause.3 In fact, the formulation of the holding suggests that once the panel found that the system was not yielding scientifically acceptable results, suppression was automatic.4 This amounts to a strict liability standard. Because the standard applied does not comport with our decision in Flack, we must remand these cases to the trial court.
In order to aid the trial court and the parties on remand, a few additional remarks about the parties’ respective burdens of proof are in order. Under Flack, the defendant has the burden of showing that by virtue of some action or inaction on the part of the prosecuting authority, he was not furnished a reasonable means of verifying an adverse breath test result. He may show this by establishing either (1) that the means furnished were so generally and systematically defective that the results of an independent test should be inadmissible in evidence as a matter of law and without regard to the facts of his particular case; or (2) that the results of an independent test in his particular case were so inaccurate that they were worthless for the purpose of impeaching or verifying the original breath test results, regardless of whether the means furnished were generally yielding independent test results which were admissible in evidence. If the defendant succeeds in showing that the means furnished were systematically defective, he might not need to show, as a prerequisite to seeking suppression, that he attempted an independent test in his own case.
Once the defendant has sustained his burden of showing that he was not furnished a reasonable means of verification, he has established a prima facie ease that the breath test results should be suppressed. In order to avoid suppression, the governmental agency in question must then prove by a preponderance of the evidence that its failure to provide the defendant an independent means of verifying the result was free of fault.
We may restate these respective burdens in other terms. Negligence entails four elements: (1) duty; (2) breach of duty; (3) causation; and (4) damage. Under our analysis, the defendant has the burden of proving causation and damage, at which point, the prosecution must disprove breach of duty. In Serrano, we established the duty: to exercise reasonable care under all the circumstances to provide a means of independent verification. Causation and damage exist when the defendant is unable to independently verify the original result because of the governmental unit’s acts or omissions. Proof that the defendant, in fact, obtained a timely blood test, or was furnished the means of independently veri*407fying the result but failed to do so through his own fault or that of his agent, would disprove causation. See Klepzig v. Anchorage, 661 P.2d 1096 (Alaska App.1983). In most cases, however, the trial court’s attention will focus on the alleged breach of duty which will require an analysis of the steps taken by the prosecution to furnish a means of independently verifying the breath test.
Since we cannot conclude that the panel applied the correct standard, we REMAND
these cases to the district court for further proceedings. The court is free to consider further evidence, or to make a ruling based on the evidence already presented, as to whether (1) the defendants in these forty-four cases were denied an opportunity to verify their Intoximeter results, and (2) the state was free from fault in this matter. On the question of fault, the state must bear the burden of proof, in accordance with our decision in Flack.
APPENDIX A
IN THE DISTRICT COURT FOR THE STATE OF ALASKA FOURTH JUDICIAL DISTRICT AT FAIRBANKS, ALASKA
STATE OF ALASKA, ) vs. BRUCE HANSON 4FA-S83-3398 Cr.
EDDIE LOGUE 4FA-S83-3771 Cr.
JOHN EBRON 4FA-S83-4262 Cr.
KAREN PASKE 4FA-S84-295 Cr.
DAVID MILLER 4FA-S84-104 Cr.
PAUL VASSEUR 4FA-S84-290 Cr.
KRIS KENYON 4FA-S84-204 Cr.
SHIRLEY SELVIDGE 4FA-S83-3932 Cr.
DANIEL MORRIS 4FA-S83-4019 Cr.
RICHARD L. GUILLAUME 4FA-S83-4197 Cr.
TONY MONTOYA 4FA-S83-3469 Cr.
SONJIA COOPER 4FA-S84-135 Cr.
NILS HVIID 4FA-S83-3818 Cr.
CHARLES CULVER 4FA-S83-4184 Cr.
MICHAEL COOPER 4FA-S83-3997 Cr.
CYNTHIA STONE 4FA-S83-4024 Cr.
JAMES R. CONNORS 4FA-S84-77 Cr.
BRADLEY FINNESETH 4FA-S83-4320 Cr.
LESTER H. THOMPSON 4FA-S84-136 Cr.
RUSSELL JOSWICIK 4FA-S84-322 Cr.
ROBERT DUYCK 4FA-S83-4266 Cr.
ROBERT TIERNEY 4FA-S83-3825 Cr.
RUSSELL KIRKENDALL 4FA-S84-434 Cr.
KENNETH SVOBODA 4FA-S83-4002 Cr.
MALANIE LAFLEY 4FA-S84-103 Cr.
*408CHRISTOPHER SENUNGETUK 4FA-S83-3958 Cr.
DANIEL McHENRY_ 4FA-S83-4206 Cr.
JOSEPH PHÍLBIN_ 4FA-S84-39 Cr.
TIMOTHY TIERNEY_ 4FA-S84-396 Cr.
ROBIN VEER 4FA-S84-8 Cr.
Motions to Suppress in the above-captioned cases having been heard before this Court sitting En Banc on April 5, 1984 and April 9, 1984, and the Court, at the request of counsel, incorporating into those hearings the testimony taken in State of Alaska vs. Eddie Logue, Case No. 4FA-S83-3771 Cr. which was heard on February 8, 1984, and the Court having considered the written briefs, as well as incorporated matters of record, finds as follows:
FINDINGS OF FACT *
1. Prior to August, 1982, DWI arrestees were given a breathalyzer test to determine blood alcohol content by means of chemical analysis of breath.
2. As a result of the decision of the Court of Appeals in Municipality vs. Serrano, et al., 649 P2d, 256 (Alaska 1982), DWI arrests were offered a blood test, at public expense, to verify or contest the accuracy of the breath test.
3. Subsequent to August 6, 1982, and prior to approximately July of 1983, the arresting officers continued to give arres-tees a breathalyzer test and offered a blood test at public expense.
3.1 This blood test could hardly be called “Independent” since the facility where it was drawn and the person who drew the blood were usually chosen by the officer and the results of the test were available to the prosecution.
4.During the course of the Serrano litigation, “after hearing extensive evidence, District Court Judge Cutler concluded that there are acceptable, inexpensive methods which could be used to preserve a breath sample for the defendant.” Serrano at 260. The Appellate Judges confirmed stating: “We have independently reviewed the record of the extensive testimony which was presented below, and concluded that the technology is available to preserve breath samples at reasonable cost.” Serrano at 260. The three District Court Judges at Fairbanks heard contradictory evidence, but we are bound by the Appellate Judges holdings.
4.1 While we find that $3.00 for each Magnesium Perchlorate tube to be a reasonable cost to collect and preserve a breath sample, the cost of testing (or cross-examining) that sample is $115.00 which is expensive (See Exhibit # 3 in Logue). We do not know what other methods the Trial and Appellate Courts considered in the Serrano hearings to be inexpensive, efficient and accurate. We understand that they may be the Indium Tube or Silica Gel methods. But whatever they were, they were not chosen nor approved by the State Department of Health and Social Services. The only external collection device which has been approved by the Department of Health and Social Services is the Magnesium Perchlorate Tube (See Exhibit # 4 from this hearing).
*4094.2 The Appellate Court noted however, that other methods of providing opportunities for independent testing might also satisfy the requirements of due process. Serrano at 258.
5. Subsequent to August 6, 1982, and prior to July, 1983, State and Municipal prosecutors, in conjunction with the Department of Health and Social Services, began to focus on certifying a technique for the collection, preservation, and testing of preserved breath samples. •
The primary motivation which prompted the hasty development of the perchlorate tube method and its use appears to have been to avoid the inconvenience of informing, transporting and paying for the blood tests of the accused persons, although the blood test was “tried and true”.
6. In January of 1983, Mr. Everett J. Clary, a highly respected chemist, was hired by Dr. Harry Colvin of the Department of Health and Social Services. He immediately began investigating the magnesium perchlorate tube and ultimately developed and obtained approval for that method.
Although he was a chemist for the Department of Health and Social Services, his work site was the Alaska State Trooper Crime Lab.
He performed numerous tests with the Intoximeter 3000, the L shaped device called the Adaptor, and the glass tube containing magnesium perchlorate which is called the Perchlorate Tube.
7. Ultimately these tests under laboratory conditions resulted in a high correlation between known alcohol samples and the amount of alcohol found in the Perchlorate Tubes.
8. The Intoximeter 3000, a technologically advanced and approved instrument for breath testing, has been certified for use in Alaska. The Intoximeter is capable of preserving for external collection, the identical alcohol from the same breath sample which was analyzed by the Intoximeter.
9. The External Collection Devices (hardware) which allow for the recapture of the analyzed breath from the Intoximeter are:
a. A metal L-shaped deviee, hereinafter referred to as an “Adaptor”. See figure 2 on Attachment A. (Also Exhibit # 6 in Logue)
b. An 8 inch glass tube, hereinafter referred to as “Perchlorate Tube”. It can be roughly described as hourglass shaped. The center of the hourglass tube is filled with white powdery substance, magnesium perchlorate which is packed tightly into place with wads of cotton like fiber. The fiber plug is held in place with a plastic loop at each end. Each end of the glass tube is threaded. A conical polethylene seal or clear plastic insert is inserted into each end of the tube and a black plastic cap is screwed onto each end of the Perchlorate Tube. See Figures 1 and 3 on Attachment A. (Also see Exhibit # 5 in Logue).
10. In order for the Intoximeter operator to recapture the alcohol from the analyzed breath sample, the operator must unscrew and remove both of the black plastic caps on the ends of the Perchlorate Tube. He must then remove both of the clear plastic inserts. He then screws the Perchlorate Tube snugly into the female threaded portion of the Adaptor. He then snugly plugs the flanged male portion of the Adaptor into the Intoximeter, so that the Perchlorate Tube will be vertical to the base of the Intoximeter and parallel to the side of the Intoximeter. When the permanent magnet on the Adaptor then makes contact inside the Intoximeter, the operator then' presses a key on the Intoximeter which forces the heretofor tested breath sample through the inlet ports of the Adaptor and into and through the Perchlorate Tube. The Magnesium Perchlorate traps all the alcohol from the breath sample as the breath sample is forced through the Perchlorate Tube into the room air.
11. The operator then must unplug the Adaptor, unscrew the Perchlorate Tube from the Adaptor, re-insert the clear plastic inserts in each end of the Tube, and recap both ends of the Tube with the threaded *410black plastic caps. This should be done promptly to reduce any potential issue of contamination. The Perchlorate Tube is then placed in the cardboard tube in which it was originally received. The cardboard tube is then resealed, marked and preserved as evidence.
12. In the spring of 1983, training sessions were held to educate peace officers and certify them as Intoximeter operators. During that period of education and certification, there were very few Perchlorate Tubes available as the initial lot which had been received was contaminated and had to be returned.
13. During the course of their Intoxime-ter 3000 training, peace officers were required to operate the Intoximeter in a manner which proved that they could correctly obtain a known target reading on the In-toximeter. However, the officers were not given an opportunity to learn to collect the breath sample from the Intoximeter and capture and preserve the alcohol therefrom in a Perchlorate Tube. There were insufficient numbers of Perchlorate Tubes available to provide for “hands-on” training.
13.1 There was no independent certification of peace officers as Perchlorate Tube and Adaptor operators, although proper collection and preservation of the breath sample which contained the alcohol to be captured in the Perchlorate Tube are critical to an accused’s right to verify or discredit (cross-examine) the Intoximeter Test results.
13.2 Both Mr. Clary and the police Supervisor-Instructors believed that the Perchlorate Tube method of preserving breath samples was so simple as to be virtually error free. Therefore a laboratory was never given an opportunity to test one, let alone several Perchlorate Tubes containing samples taken during training to see if officers who were certified as Intoximeter Operators were able to follow proper procedure and obtain reasonably accurate test results from known samples.
14. There was a concerted effort to get the Perchlorate Tubes and the Adaptors “on line” in early July, 1983. Within a month the Perchlorate Tubes were modified to include the clear plastic insert at each end under the black plastic caps. Lo-gue page 212.
14.1 In August, 1983, new Adaptors were put on line. Logue page 212.
15. No particular method of breath preservation was mandated by either the Trial Court or the Appellate Court in the Serrano case. Neither was there any time deadline set by the Court of Appeals to put a breath saving technique into operation because police agencies had the option of offering an accused the independent chemical test performed by someone of the accused own choosing as provided in AS 28.-35.033(e). “effective compliance with AS 28.35.033(e) would constitute an acceptable alternative to routine preservation of breath samples”. Serrano page 258, footnote 5.
16. No field testing, whatsoever, was done with regard to the independent breath saving technique of Perchlorate Tube collection before it was put into operation in early July of 1983.
17. When the Perchlorate Tubes and Adaptors were put on line, both the Fairbanks City Police and the Alaska State Troopers ceased providing and advising accused persons of their right to an independent chemical test of their blood.
18. There was no quality control program, no recorded regular method of inspection, no method devised to review or monitor the external collection components which comprised the Perchlorate technique of preserving a breath sample from an Intoximeter.
19. It was not until January 25, 1984, that Officer John Kairis, the Supervisor-Instructor for the Fairbanks Police Department became aware that there was supposed to be an “O” ring in the female base of the Adaptor. He knew that the Fairbanks Police Department’s Adaptor did not have one, so he sliced a clear plastic insert (from a Perchlorate Tube) and sealed it in with glue. Logue page 214-215.
*41120. Officer Kairis tested the Jury-Rigged “O” ring and said “it seems to be working.” Logue 215.
21. This “Jury Rigging” of a plastic 0-shaped ring was not tested, authorized nor directed by the Department of Health and Social Services.
22. The Adaptors were not an inventory controlled item at the Fairbanks City Police Headquarters. They did not have identifying marks nor serial numbers.
23. The Adaptor at the Fairbanks City Police Station had no “0” ring in its female base until after January 25, 1984. Even at that time it was a jury rigged plastic insert until the new generation of Adaptors went on line about March 1, 1984.
24. If any problems arose with the Perchlorate Tubes or further information was received by Kairis, he would advise the officers of it at their briefing sessions by “grabbing an officer and walking up to the (Intoximeter) instrument and saying “Here’s how it’s done”. Logue page 212.
25. The original generation of Adaptors were replaced with a new 2nd generation Adaptor in August of 1983. Logue 212.
26. That 2nd generation Adaptor has been changed with a 3rd generation Adaptor with nylon threads in the female end. The 3rd generation nylon modified Adaptor went on line about March 1, 1984.
27. Forty-five Fairbanks City Police Officers received In-Service Training between February 6, 1984 and March 5, 1984. (The undated certificate of John Kairis was filed April 12, 1984 at 3:38 PM.)
28. Mr. Clary developed a procedure to allow for the extraction of alcohol from the Perchlorate Tube in such a manner that the amount of alcohol trapped within the magnesium perchlorate relates back to the In-toximeter results within a 5% correlation. Everett Clary, in lab testing, has been able to obtain consistent and reliable results on perchlorate tube re-testing by exposing the perchlorate to a known sample of alcohol and obtaining the same result from the perchlorate.
29. In Exhibit # 8 received in this hearing, Mr. Clary tested 11 field-collected perchlorate samples which were taken between February 11, 1984 and February 24, 1984 — All were 45% or more lower than the Intoximeter result. In the same Exhibit # 8, Mr. Clary tested 22 field-collected perchlorate samples which were taken between February 26, 1984 and March 11, 1984. In that series 5 out of 22 of the tests showed results of 14% to 52% lower readings than the Intoximeter result.
29.1 The improved results of the tests taken after February 26, 1984, have been attributed to Police Officer Re-training. Nevertheless, 20% of those readings are unreliable.-
30. In another sample of approximately 30 cases obtained from a controlled field study from Juneau, Mr. Clary was able to obtain, on the average, results within 6% of target values.
31. In the opinion of Mr. Clary, the perchlorate method of breath sample preservation is at a workable stage, and the inability to achieve consistent correlation between the preserved sample and the Intox-imeter is the end product of “operator mishandling” or “hardware” problems.
32. Mr. Clary found that the first two or three generations of Adaptors which he used in his lab have not proven reliable in the field when used by numerous police officer operators.
32.1 He believes and hopes that the new generation of Adaptors with nylon threads will prove reliable and durable in the field.
33. Mr. Clary stated that the original education and certification of the officers was done with perchlorate tubes that had a different cap and no plastic inserts. The perchlorate tubes which were actually put in use contained plastic inserts and different plastic caps.
When it came to his attention that the officers were not taking out the plastic inserts and in some cases not removing the plastic caps he immediately notified all of *412the law enforcement stations of the problem.
34. None of the changes in hardware described above have been reviewed for any form of recertification for acceptability by the Department of Health and Social Services. Mr. Clary has encouraged Health and Social Services to require each operator to perform a perchlorate test against a known standard as a type of certification of the adaptor, the instrument and the officer. Mr. Clary testified that the work which has been done to date is “developmental” as to methodology.
35. With regard to operator error, Mr. Clary testified that he was “flabbergasted” with the sloppy collection methods. He indicated that no checklist was provided to the officers as it appeared to be a simple straight-forward procedure. He became aware of the collection problems in November, 1983.
He just developed a proposed checklist for the Perchlorate Tube and Adaptor (External Collection Devices) about February 1, 1984. Logue page 143.
36. Dr. Rogers, a pathologist from Hu-mana Hospital, and one of the co-directors of the private lab which performs the retest on perchlorate tubes, testified that the perchlorate breath sample method of collection has potential for working but it has not been working up to now. Logue page 44.
37. Dr. Rogers testified that either officers must be vigorously trained to be meticulous about the method of collection or a lab technician hired to assure that the collection method more closely adheres to the necessary laboratory standards.
38. Dr. Propst, a pathologist from Hu-mana Hospital, testified to essentially the same conclusion reached by Dr. Rogers— that the present method, with its attendant field collection problems, is not a scientifically reliable method of preserving an independent breath sample.
39. Dr. Rogers testified at the Logue hearing on February 8, 1984, and Dr. Propst testified on April 5, 1984.
40. On April 5,1984, both Mr. Clary and Dr. Propst testified that Jill Booth, the lab technician who actually ran the gas chro-matograph testing of the Perchlorate Tubes at Humana Lab, used an erroneous number in calculating the results of the re-test. Instead of a .10 standard against the results tested, the known standard was probably between .169 and .175, resulting in a 40%-42% erroneously low reading on all the test results obtained between November 29, 1983 and February 9, 1984.
41. The results were re-calculated using the .169 standard. Ms. Booth and Mr. Clary agreed that the re-calculated number would bring one 40 and 42 percent closer to the true value, but that there could be no greater refinement of the results due to the inability to determine the source of any other error.
42. in addition to the erroneous calculation performed by Jill Booth, Mr. Clary determined that there were two other possible sources of error in the retesting procedures used by Humana Hospital: First, inadequate salination of the perchlorate solution resulting in less gas or head space to analyze. Second, the lab technician’s failure to use a glass syringe and failure to heat that syringe each time after the gas was extracted from the perchlorate capsule and injected into the gas chromatograph may have resulted in some “carry over” alcohol-tainted gas between each test.
42.1 This “carry over” effect became more obvious when (in controlled tests of nine Perchlorate Tubes at Humana Lab) a reading of .065% blood alcohol was obtained which should have been .000%. This is a glaring error because if the proper “known standard” were used the erroneous result was over .10%. (See Exhibit #9)
43. Mr. Clary stated that because of the erroneous procedures used to obtain the “recalculated” results, there is no ability to recalculate a true result.
44. The procedure used by Humana Hospital Lab to test perchlorate tubes was developed by Mr. Clary while he was an employee of the Department of Health and *413Social Services. He is presently an employee of the Alaska State Troopers. The procedures were not specific enough in the precise methods and equipment to be used in testing, particularly in salinization, use of glass syringes and reheating of glass syringes.
45. Human Lab is the only private lab in Alaska presently performing Perchlorate Tube analysis. It has not strictly followed the method of analysis or the order of the steps in the analysis developed by Mr. Clary which led to erroneous and uncorrectable results.
46. Fourt out of five tests conducted by Alaska State Troopers after they had been retrained were 72.5% below the Intoximeter target reading. Exhibit # 2.
47. Only one out of five of the tests conducted by the Fairbanks City Police was off the Intoximeter reading by more than 10%. Exhibit # 2.
48. Out of all sixty-seven known tests, prior to the use of the new generation of Adaptor with nylon threads, only nineteen were within 15% of the Intoximeter Reading and clearly unacceptable as being scientifically reliable at that stage of development and experience. Exhibit # 2.
49. In manufacturing glass perchlorate tubes, the threads on one end are machine made and on the other end they are hand blown. They are not individually checked to see if they will properly fit into an Adaptor, until the officer uses one in the field.
A defective fit will give an unreliable perchlorate tube test result. Twenty to forty percent of the tubes have had a defective end when inspected by Mr. Clary.
The new generation Adaptor may cure some of the glass thread problem but no controlled tests have been conducted. At the hearing, a perchlorate tube was broken while putting it in a new Adaptor.
50. Mr. Clary requested the opportunity to field test the Adaptors and perchlorate tubes before putting them “on line” in 1983, but his request was turned down by his supervisors in the Department of Public Safety.
51. There is no systematic checking of the “lots” of perchlorate tubes when received to determine if glass threads on the tubes are free of defects.
52. The $115.00 cost of testing a perchlorate tube is not inexpensive. Exhibit # 2 in Logue.
53. When the results of a test of the perchlorate tube are favorable to an accused, the cost of round trip airplane transportation from Humana Hospital in Anchorage to Fairbanks and expert witness fees is not inexpensive.
54. Raymond Funk, Esquire, of the Fairbanks Public Defender’s Office was unable to obtain the necessary $115.00 fee for testing a perchlorate tube on behalf of a defendant he represented through the Public Defender’s Office.
It is obvious that the Public Defender’s Office would be hard pressed to obtain funding to check all perchlorate tubes.
55. As of March 30,1984, there is now a set of instructions from Health and Social Services to the Supervisor-Instructors on preserving breath samples and approved checklists and procedures for breath alcohol collection. See Attachment “B”.
CONCLUSIONS
1. The Intoximeter 3000 is an accurate breath alcohol testing device.
2. In a controlled laboratory setting the perchlorate method of separate breath sample preservation has been found to be an accurate method of collection and preservation when performed by a skilled, trained professional such as Mr. Clary.
3. The gas chromatograph testing procedure developed and employed by Mr. Everett Clary has established that breath samples properly collected and preserved using the perchlorate tubes can potentially be accurately tested to furnish results within acceptable scientific standards of accuracy.
*4144. Less than a third of the field collected perchlorate tubes tested have yielded results within 15% of known Intoximeter values which were collected prior to the use of the new generation nylon threaded Adaptor on March 1, 1984. Exhibit #2.
5. The most significant errors appear to come from the following sources:
First, inadequate Adaptors.
Second, defectively threaded perchlorate tubes.
Third, improper collection techniques by inadequately trained police officers.
Fourth, improper hardware, the use of erroneous known standards, and inadequate test procedures at the Humana Lab.
6. We do not conclude, as Judge Andrews does in Municipality vs. Hernandez et al., that the written protocol for testing the perchlorate tubes appears to be adequate and will yield results within tolerance if the separate breath sample is properly collected in the field by the officer, because Humana Hospital Lab has made too many mistakes prior to March 1, 1984. That is the only lab in the State of Alaska with any extensive testing experience where the work isn’t done by Mr. Clary. Mr. Andrews in Juneau has had some limited experience which appears favorable. Andrews is a chemist with Health and Social Services.
7. Because of the numerous potential sources of error and the actual sources of error which have been identified in the past, it is impossible to pinpoint with any exactitude the source of the errors in any single perchlorate test result. Because of these errors then, it is also impossible to tell whether a correct result was reached in perchlorate tube testing. An apparently satisfactory result may be the product of good procedure and testing or an accidental coincidence.
8. Humana Lab test results through at least February 9, 1984, cannot be “rehabili-, tated” due to numerous errors in testing. We are unable to conclude that the protocol for testing supplied to that lab by the State was adequate or inadequate.
9. If we were to operate from a premise that the perchlorate tube test results were valid rather than the Intoximeter results being valid, 29 of the people out of 67 listed in Exhibit #2 of this hearing could not have been charged under the .10% theory.
10. Everyone is aware of the fact that a blood test is generally the most reliable test to determine alcohol consumption. When the State prosecution or an officer avoids advising an accused of his statutory right to another chemical test being administered by a person of the accused’s own choosing and offers a less reliable test which the State has selected, the bias of the prosecution or officer becomes obvious.
This is especially true when in Fairbanks, blood tests are readily available at the hospital and from other qualified persons who are available to testify locally without the necessity of air transportation, high witness fees and unbelievable scheduling conflicts.
11. We do not agree with that portion of Judge Andrew’s opinion which infers that anytime a Perchlorate test is fortuitously within ± 15% of the Intoximeter test, an accused has been afforded his due process right to a cross check of the Intox-imeter. The State must first establish the validity of the Perchlorate Tube testing method of field samples as being consistently scientifically accurate within acceptable limits before an accused’s due process right has been met.
DISCUSSION
As a result of the decision in the matter of Municipality vs. Gilbert Serrano, 649 P2d 256, the Appellate Court concluded that the due process clause of the Alaska Constitution requires the prosecution to make a reasonable effort to preserve the breath sample or to take steps to allow a defendant to verify or dispute the results of the breath test. The Court found that because state statute places great emphasis on the breath test by making it an offense to drive a car with a .10% or more breath alcohol, the ability of the defendant *415to check (cross-examine) these test results is critical to his case and the integrity of the criminal justice system. The wisdom of the Appellate Court’s decision is obvious when one looks at the practical application. A suspected intoxicated driver whose test result is a .10 on the Intoximeter can be convicted upon sufficient proof of that fact. If that same Intoximeter reading can be proven to be erroneous by one-one hun-dreth of a percent such that the Intoxime-ter should properly have read .09, then the defendant cannot be convicted of driving with .10 or more breath alcohol. He may be convicted of driving under the influence with that level of blood alcohol. (Although our collective experience indicates that indeed such convictions are rare). The potential to check and actual ability of the defendant to establish a reasonable doubt or prove that the Intoximeter was not operating within acceptable limits of accuracy is important and cannot be discounted on the theory that the Intoximeter is almost always accurate.
Prior to the introduction of the perchlorate preservation method the State in compliance with Serrano, offered defendants the opportunity to preserve an adequate test by providing them with the opportunity to have a blood test to determine the alcohol content of their blood. The blood alcohol test is a routine laboratory test which yields results, usually higher than a breath alcohol test, and has been accepted as reliable by the Courts. The new perchlorate tube technique introduced by the State in July of 1983 appears to be a potentially reasonably accurate substitute for the blood test. However, it was never field tested to determine that it would provide, in fact, a reasonably accurate check of the Intoximeter result.
The State claims in its argument and through Mr. Clary that it could not perform field testing of the technique on actual arrestees as the perchlorate sample was the defendant’s evidence. AS 28.35.031, the Implied Consent statute, states that a person who drives has impliedly consented to a “chemical test or tests” of his breath. It is fairly obvious to this Court that the State could have continued to offer the independent blood samples while collecting Perchlorate Tubes and testing them for accuracy and providing a cross check against both the Intoximeter and Perchlorate Tube testing. The State failed to take any precautionary measure whatsoever to assure that the perchlorate method would work in the field. The Court, in Serrano, set no requirement that perchlorate tubes be used. It is clear to this Court that the State, in its attempts to provide for a less expensive, more efficient method of preserving breath alcohol, introduced a method which had not been properly tested to assure its reliability and provided no backup testing while the method’s reliability could be determined.
Perchlorate sample preservation is capable of becoming a technologically sound, efficient and effective method of breath preservation. It has not reached that point. The perchlorate method of preservation and testing may yield accurate results in the future. Up to . this point we found that the method as used in the field does not consistently produce scientifically acceptable results. For that reason we HOLD that the Intoximeter and Perchlorate Tube test results will be suppressed in the above captioned eases, as the defendants have been deprived of due process as set forth in the Serrano opinion.
The Perchlorate Tube method will be considered scientifically reliable when results from field samples that are tested, in a laboratory available to an accused, are shown ninety percent of the time to be consistently accurate within +10% or —15% of either the actual Intoximeter test results or actual blood test results. Until that time, we HOLD if only the Perchlorate Tube method of preservation of breath is used, the Intoximeter results are Suppress-able upon a timely motion being made prior to trial UNLESS there is clear compliance with AS 28.35.033(e). The only other alternative open to the State is to. select one of the other methods of breath preservation which the Court of Appeals found to be reasonable in cost and accurate in results.
*416Modifications in the Adaptor and Perchlorate Tube should be approved by the Department of Health and Social Services. Modifications of procedures or equipment used in laboratory testing of Perchlorate samples should be approved by the Department of Health and Social Services.
We further find that the Adaptor is a critical, essential “External Collection Device” within the meaning of 7AAC 30.055. It should be tested and approved by the Department of Health and Social Services. We HOLD 7AAC 30.055 requires that the Adaptor be placed on the “List of Remote Collection Devices Approved for Use With Breath • Test Instruments" BEFORE the Perchlorate Tube or Adaptor is used with any Intoximeter.
We further’HOLD that pursuant to AS 28.35.033(d) each officer must have been individually certified as to the use of the Adaptor and the Perchlorate Tube before the officer uses them in the field. That certification must be filed with the Department of Health and Social Services in the same manner as the Breathalyzer and In-toximeter certifications filed and retained by the department.
DATED at Fairbanks, Alaska.this 31st day of May, 1984.
/s/ Hugh H. Connelly Hugh H. Connelly District Court Judge /s/ H.E. Crutchfield Herschel E. Crutchfield District Court Judge /s/ Jane F. Kauvar Jane F. Kauvar District Court Judge
. We have included the entire written order as an appendix to this opinion. The description of the MPT system above is taken from this order.
. Evidence was presented that officers in the Fairbanks area were retrained in MPT collection procedures in February and March. The panel also heard testimony concerning improved results obtained when two chemists employed by the state tested MPTs that had been collected statewide during this period and made available because the defendant had entered a plea of guilty or no contest. Finally, the panel *405was told that a new model adaptor went into use in Fairbanks on March 1, and that in laboratory tests using the new model adaptor the two chemists had achieved a high correlation between MPTs and the known samples used to generate MPTs.
On June 25, 1984, the parties argued. The prosecutor asked the court to limit its holding in Hanson to those defendants who had taken the trouble to have their MPTs tested. The court issued its findings orally. The court found no reason to change its original Hanson holding, except that the deviation to be allowed would be plus or minus 15% rather than plus 10% and minus 15%. The panel left the actual decision on whether the Intoximeter results on the new cases should be suppressed for the next day. On June 26, 1984, the court held that the defendants need not have had their tubes tested. The court relied in part on the fact that the state had not arranged a system of certifying officers in the collection of MPTs, so a given defendant had no way of knowing if the officer administering the Intoximeter to him had been retrained during the recent effort. (The court apparently did not consider the defendant’s ability to call the examiner as a witness and ask him.) A written decision covering the conclusions reached on both days was issued on June 27, 1984. The court stated that once it was established that "results of local tests" were within ± 15% 90% of the time, all Intoximeter results would be admitted regardless of whether an individual defendant could show that his MPT result was substantially outside this range.
The state argues that the panel erred in concluding that Kerr and the other defendants were entitled to suppression regardless of whether they had had their MPTs tested. See Best v. State, 712 P.2d 892 (Alaska App.1985). We have no information on how many of the 44 respondents have had their MPTs tested, or what the results were. This issue should be addressed on remand. If the court finds negligence in the design or selection of the MPT, then a finding that governmental negligence caused a given defendant damage might be sustainable even without retesting. Conversely, if it is negligence in the implementation of the MPT collection system which is in issue, a retest to establish a causal link between the negligence of the state, if any, and damage to the individual defendant might be required. See Best, 712 P.2d at 896 n. 6.
. The word "negligence” does not appear in the Hanson order, nor was it used in discussion during the hearings, which occurred prior to our decision in Flack. While the panel listed the four “most significant" sources of error, it did not directly assign responsibility for these sources to the state. See Appendix A. And, while some of the sources identified might be easily attributable to the state, we are reluctant to make such a finding at the appellate level, in part because of the great number of separately identified possible sources of error. Some of these sources are probably not attributable to the state. Rather than speculate as to which sources were most important to the panel in reaching its conclusion, we return the matter to the district court.
. The panel stated:
Up to this point we found that the method as used in the field does not consistently produce scientifically acceptable results. For that reason, we HOLD that the Intoximeter and Perchlorate Tube test results will be suppressed in the above captioned cases....
See Appendix A, p. 407.
While there is no clear procedure for an En Banc Hearing in the District Court, such procedure was agreed to by counsel and Judges Hugh H. Connelly, Herschel E. Crutchfield and Jane Kauvar. It has been done in District Court in Fairbanks before for the following reasons:
1. The desireability of hearing the evidence only once.
2. To keep expenses of transportation and expert witness fees down for both the State and various defendants.
3. The desireability of reaching a consensus at this level of Court, if possible, so as to avoid judge shopping, although each Judge is completely free to decide his or her cases.
As much as was reasonably possible, the same format, the same order of Findings of Fact, and even the same language as used by the Honorable Elaine M. Andrews, District Court Judge, in her opinion Municipality of Anchorage vs. Ramiro Hernandez, Case No. 3ANM-S83-4629, et al. was used. This was done in order to assist counsel statewide and the Appellate Courts on Appeal or Review.