Brachtenbach, J.
In these consolidated cases two primary issues are presented: (1) Is the state toxicologist required to approve and publish in the Washington Administrative Code procedures for evaluating, certifying, and maintaining DataMaster breath testing machines, and procedures for formulating, testing, and using the simulator solution used in the machines, before DataMaster breath test results may be admitted in prosecutions for driving while intoxicated? (2) Invalid sample code messages appear on the DataMaster's screen when the machine is unable to complete a breath test under certain circumstances. Must breath test results be suppressed on the ground that a record of the invalid sample code messages is not preserved by the State and made available to defendants?
As to the first question, we conclude that the state toxicologist has carried out the statutory directive in RCW 46.61.506(3) that he approve methods for breath alcohol [863]*863analysis, that in any event he need not promulgate further formal rules pursuant to the administrative procedure act (Act), RCW 34.05 (formerly RCW 34.04), that certain protocols approved by the state toxicologist are not "rules" within the meaning of the Act, and that defendants' due process rights are protected under existing law. As to the second question, we hold that due process does not require that the State preserve a record of the invalid sample code messages.
For ease of analysis and the reader's convenience, the facts and resolution of State v. Straka follows in full, with the facts and resolution of State ex rel. Dawson v. South District Court addressed thereafter.
State v. Straka
In the first case before us defendant Donald Straka was arrested for driving while intoxicated and then taken to the Monroe Police Department where he underwent physical sobriety tests. Straka then agreed to submit to breath alcohol analysis performed on a DataMaster machine. The test results showed readings of .18 percent and .17 percent breath alcohol content.
Straka moved to suppress the breath testing results, arguing, among other things,1 that the breath test results should have been suppressed on the grounds that the state toxicologist failed to properly promulgate written procedures and protocols for evaluation and certification of the DataMaster machines, and failed to properly promulgate written procedures for the formulation, testing and use of the simulator solution used in the machines, contrary to applicable court rules and the Washington Administrative Code (Code). A suppression hearing was held in Evergreen District Court on January 31, 1990. The court was presented with testimony of Dr. Vidmantas Raisys, who was then the state toxicologist, and testimony of employees of the state toxicology lab and the Washington State Patrol, [864]*864including Sergeant Rod Gullberg of the Washington State Patrol, who is attached to the breath testing section of the Washington State Patrol Crime Lab and directs the Data-Master testing program.
Essentially this is a representative case. The parties submitted the relevant issues to the trial court on a stipulated record from another case. The aim of the parties has been to present a complete record for purposes of deciding the issues, which have been the subject of considerable litigation.
Breath alcohol analysis is currently tested using BAC Verifier DataMaster II machines. According to the testimony in the stipulated record, when the State Patrol receives a DataMaster machine from the manufacturer, a written protocol approved by the state toxicologist must be followed before a DataMaster is put into the field. This evaluation process, developed largely by the Washington State Patrol Crime Lab, involves testing all the major functions of the DataMaster machines, including, among numerous tests, tests of the acetone detector, the radio-frequency interference detector, and the mouth alcohol detector. Results of the evaluation are recorded on a 4-page form. The evaluation procedure has been approved by the state toxicologist.
After evaluation of the machines, and before they are placed in the field, the DataMaster machines undergo a certification procedure. The state toxicologist has approved the certification procedure, which is also in written form, although it was not prior to 1987. The first certification procedure was developed by the breath testing section of the Washington State Patrol Crime Lab, and was approved by the state toxicologist. A second and third version do not contain any substantive changes in the procedure. The certification procedure involves tests of the DataMaster conducted using solutions of water and alcohol at known values—if the known strengths of the solutions are reproduced within acceptable limits, the machine is properly [865]*865functioning. DataMaster machines in the field are recertified after repair or replacement of certain parts of the machines.
In addition to the evaluation and certification procedures, the state toxicologist has approved three successive versions of a written protocol for preparing the simulator solution used in breath testing. The DataMaster breath alcohol analysis includes running a simulator test between two breath measurements, in order to determine if the machine is in proper working order and correctly analyzing a subject's breath samples. In simple terms, the subject blows into a mouthpiece on the machine, then vapor from the simulator solution is tested, then the subject blows a second time.2 The machine measures the alcohol content of these three samples. Accuracy and reliability are thus tested in at least two ways—the correlation between the two breath analyses, and the calibration check which the simulator test provides.
The simulator solution is prepared and tested by employees at the state toxicology laboratory. The solution is made by mixing ethyl alcohol and water in known concentration. The sole procedure used at the lab to check the ethanol concentration of the simulator solution is gas chromatography. While the first two written protocols simply directed that the ethanol concentration be determined after mixing, the third directs that this procedure be accomplished through gas chromatography.
After mixing and testing, the solution is placed into individual jars, sealed, and sent to the field for use in the DataMaster machines.
The protocols for evaluation and certification of the machines, and preparation and testing of the simulator solution were not promulgated through formal rulemaking under the administrative procedure act, and they are not set out in the Washington Administrative Code.
[866]*866Additional testimony is recounted in the analysis below. The trial court denied Straka's motion to suppress the breath test results, adhering to its prior decision (in the case in which the stipulated record was made) where it concluded that the state toxicologist has met the statutory requirement that he approve methods for testing breath for alcohol content. The court also concluded that adherence to the protocols for evaluation and certification of the Data-Master machines, and adherence to the protocol for preparing and testing the simulator solution, when coupled with strict compliance with applicable WAC's, produces scientifically reliable results in the field.
Straka was convicted, and appealed. The RALJ appeal was certified by a judge of the Snohomish County Superior Court to this court, which accepted direct review.
In this case we must decide whether RCW 46.61.506(3) requires the state toxicologist to promulgate through formal administrative rulemaking methods for evaluating and certifying DataMaster machines, and for preparing and testing the simulator solution used in breath testing; and whether the written protocols approved by the state toxicologist are "rules" within the meaning of the administrative procedure act and therefore must be published in the Washington Administrative Code.
RCW 46.61.506(3) provides:
Analysis of the person's blood or breath to be considered valid under the provisions of this section or RCW 46.61.502 or 46.61.504 shall have been performed according to methods approved by the state toxicologist and by an individual possessing a valid permit issued by the state toxicologist for this purpose. The state toxicologist is directed to approve satisfactory techniques or methods, to supervise the examination of individuals to ascertain their qualifications and competence to conduct such analyses, and to issue permits which shall be subject to termination or revocation at the discretion of the state toxicologist.
The state toxicologist has approved the DataMaster as a device for the measurement of a person's breath for [867]*867alcohol concentration. WAC 448-12-210.3 In State v. Ford, 110 Wn.2d 827, 755 P.2d 806 (1988), this court held that the state toxicologist's approval of the DataMaster was not arbitrary or capricious, and therefore was valid.
Defendant maintains that the approval upheld in Ford relates only to the technique for measuring the breath, i.e., the DataMaster, but that the methods by which the technique is to be utilized must be provided through formal rulemaking pursuant to the administrative procedure act and must be set out in the Washington Administrative Code. Defendant reasons that evaluation and certification procedures and procedures for mixing and testing the simulator solution are "methods" which must be published in the Code. Defendant contends that lack of written standards promulgated through formal rulemaking and set out in the Code makes it impossible to determine what procedures must be followed to ensure accuracy and integrity of breath alcohol analysis, and to determine whether those procedures have been followed.
We do not agree that the statute mandates formal rule-making as defendant contends. The statute does require the state toxicologist's approval of "satisfactory techniques or methods" of breath analysis, but does not direct that approval of procedures for evaluation and certification of the machines and preparation and testing of the simulator solution occur through formal administrative rulemaking.
Moreover, the approved procedures do not constitute "rules" within the meaning of the Washington administrative procedure act, and therefore they need not be published in the Code. RCW 34.05.210 (former RCW 34.04.050) [868]*868requires that rules be published in the Washington Administrative Code. RCW 34.05.010(15) (in relevant part the same as former RCW 34.04.010(2)) defines rule:
"Rule" means any agency order, directive, or regulation of general applicability (a) the violation of which subjects a person to a penalty or administrative sanction; (b) which establishes, alters, or revokes any procedure, practice, or requirement relating to agency hearings; (c) which establishes, alters, or revokes any qualification or requirement relating to the enjoyment of benefits or privileges conferred by law; (d) which establishes, alters, or revokes any qualifications or standards for the issuance, suspension, or revocation of license to pursue any commercial activity, trade, or profession; or (e) which establishes, alters, or revokes any mandatory standards for any product or material which must be met before distribution or sale. . . .
Two of the categories of rules in RCW 34.05.010(15) are argued by defendant, (c) and (e). Clearly (a), (b), and (d) are not implicated.
Initially defendant argues that "rule" be broadly defined in light of the expression of legislative intent that the Act be interpreted consistently with the federal administrative procedure act and decisions in other states, that the Washington Act be interpreted to provide public and legislative access to agency decisionmaking, and that the Act be construed consistently with prior Washington court decisions. RCW 34.05.001 (Laws of 1988, ch. 288, § 18 effective July 1, 1989). These principles do not dictate, however, that the definition of "rule" set out in RCW 34.05.010(15) be ignored in favor of some other definition, such as that set out in the federal act (5 U.S.C. § 551(4)) (federal definition of "rule" differing markedly from state definition).
Also, defendant's reliance on the reference to "directive" found in our state's definition of "rule" as meaning "anything directive in nature" fails to consider the fact that one of the five categories in the definition must be involved, regardless of whether a "directive" is at issue.
Defendant explains that the argument relating to subsection (c) of the definition of "rule" found in Washington's [869]*869Act is that a person's privilege to drive in this state is conditioned upon consenting to a breath or blood test upon request by a law enforcement officer, RCW 46.20.308(1), and this consent is in turn conditioned upon the state's compliance with RCW 46.61.506(3). Therefore, the argument goes, suspension of a driver's license (the privilege to drive) relates back to the operation of the breath test machine, i.e., the methods for testing are thus related to a privilege conferred by law as contemplated by RCW 34.05-.010(15)(c).
We do not agree with this reasoning. The methods by which the breath testing is accomplished do not establish, alter, or revoke the "requirement" that a person consents to a breath test by exercising the privilege to drive.4
As to RCW 34.05.010(15)(e), defendant argues that the simulator solution must meet a mandatory standard established by the "recipe" protocol before the solution is "distributed" to the police throughout the state, and therefore that protocol establishes a mandatory standard for a product or material which must be met before distribution, as contemplated by subsection (e).
Again, we do not agree with this reasoning. As the State notes, if defendant's argument were accepted, an agency which adopted a certain letterhead for agency stationery which was sent out to branch offices would have to publish the standards for the stationery in the Code. This is clearly not within the intended scope of the subsection.
[870]*870The "standards" in the protocol must be met to comply with the state toxicologist's approved methods for breath testing, but they are not mandatory standards in the sense of subsection (e). Subsection (e) is concerned with ensuring that certain action is taken before distribution of a product or material. In marked contrast, the simulator protocol relates to accuracy of breath testing and admissibility of test results in a DWI prosecution, i.e., evidentiary concerns.
We conclude that the protocols are not rules which must be published in the Washington Administrative Code.
We also agree with the trial court that adherence to the protocols for evaluation and certification and the protocol for preparing and testing the simulator solution, when coupled with compliance with applicable WAC's, produces scientifically reliable results. In State v. Ford, 110 Wn.2d 827, 833, 755 P.2d 806 (1988), we said that " [t]he ultimate concern of the judiciary is that the methods approved result in an accurate test, competently administered, so that a defendant is assured that the test results do in fact reflect a reliable and accurate measure of his or her breath content."
When the protocols at issue here and existing Code provisions are followed, there is sufficient assurance of accuracy and reliability of the test results to allow for general admissibility of test results. These results are still subject to challenge in any particular case as explained later in this opinion.
The state toxicologist has directed that a simulator be attached to each DataMaster, providing an external standard which tests to ensure that the DataMasters operate correctly and are calibrated correctly. WAC 448-12-210. The simulator test must be run in each case between two breath measurements. WAC 448-12-210.
In WAC 448-12-220 the breath test is defined. It requires that the subject blow twice into the instrument, with the two breath samples constituting one test. An accurate test is presumed if the results of each measurement are [871]*871within plus or minus 10 percent of the average of the two measurements. WAC 448-12-230 sets out the procedure approved by the state toxicologist for performing the breath test. The subject must have had nothing to eat or drink for at least 15 minutes before the test, and must not have any foreign substances (other than dental work) in the mouth at the beginning of the 15-minute period. The operator is then to follow the instructions displayed on the DataMaster.
WAC 448-12-230 further provides that "[t]he temperature of the solution in the simulator must be 34 Centigrade, plus or minus .2 Centigrade, prior to the time the test is given. The reading from the simulator test must be between .090 and .110 inclusive." Defendant argues that this description does not satisfy the statement in WAC 448-12-210 that the simulator to be attached to each DataMaster "will provide a known external standard as defined in WAC 448-12-230."
The phrase "as defined in" means whatever definition is actually set out in WAC 448-12-230. The WAC defines the "external standard" in terms of the reading it is to produce on the DataMaster. Because the recipe for the simulator solution is set out in a protocol approved by the state toxicologist, a manner of "approval" wholly consistent with the statutory directive in RCW 46.61.506(3), the requirement that the reading from the simulator test be between .090 and .110 inclusive does in fact define the known external standard which must be satisfied to assure accuracy of the DataMaster when the simulator solution is prepared as approved by the state toxicologist.
Therefore, contrary to defendant's contention, there is no necessary definition lacking from the Code which is declared to be in the Code.
Other WAC's detail qualification, training and review of instructors and operators. See WAC 448-12-240 through 448-12-330.
Correlating procedures under these WAC's and the approved protocols, the testimony in the record establishes [872]*872that accuracy and reliability in general are assured if both the WAC's and the protocols are followed.
The state toxicologist testified that if two breath measurements are taken with the simulator solution run between them, as required, the results should be good.I ***5 Sergeant Gullberg emphasized the importance of the breath testing procedure:
I think that accuracy and proper working order of the instrument is best evaluated at the time of the test in question. Not at some prior date. And I think that the breath test protocol that we have that the instrument goes through, and it conducts testing on a particular individual, is adequate to insure accuracy and precision and proper working order. . . . Blank tests, duplicate breath tests, external standard tests, internal standard tests, 15 minute observation. That gives me confidence that at that time, that's an accurate result of their breath alcohol concentration.
Verbatim Report of Proceedings, at 105.
. The simulator test is of particular significance in certification of the DataMaster machine, and in the machine's self-testing of calibration which it goes through each time a breath alcohol analysis is performed in accordance with procedures in the WAC's.
[T]he simulator is a device that contains a glass jar and the top portion has a thermometer, a motor and heating elements and ports. The purpose of it is to simulate a breath alcohol sample. And it contains a solution of alcohol and water that has been prepared. The solution is heated to a certain temperature and thermostatically regulated to stay at that certain temperature. And then it can produce a known vapor alcohol concentration and it can be used as a calibrating device and as a testing [873]*873device when you are testing a breath test, any type of breath test instrument. Simulators are typically the standard for testing the calibration. The solutions are received from the state toxicology laboratory, and are used for certification purposes, they prepare and test [the] solutions then provide [them] to us [the State Patrol].
Verbatim Report of Proceedings, at 89.
Obviously, the simulator solution is key to simulator testing. As noted, the three successive protocols identified in the record have all been approved by the state toxicologist. A scientist working at the state toxicology lab, Matthew Friel, testified that mixing the simulator solution was the easiest part of his job, requiring the educational level of a high school chemistry student. Friel testified that the lack of more detailed instructions as to the manner of mixing the water and ethanol, and the type of water used in the lab in Seattle (distilled or tap) did not result in any problems so far as mixing a proper solution is concerned. He testified that the sole procedure used at the lab to check the ethanol concentration of the simulator solution is gas chromatography, which is "a very standard technique that has been used in many laboratories throughout the country." Verbatim Report of Proceedings, at 15. While the first two written protocols simply directed that the ethanol concentration be determined after mixing, the third directs that this procedure be accomplished through gas chromatography. Exhibits 20, 21, 22. The record does not contain any evidence that the methods approved by the state toxicologist for preparing the simulator solution, if followed by those working at the toxicology lab, will yield inaccurate testing results. Friel, after working at the lab for IV2 years, testified that the analysis of the simulator solution was the "most rigidly controlled procedure" that he had worked with in 20 years. Verbatim Report of Proceedings, at 15.
Dr. Raisys testified that there were certain practices, such as using clean jars, that he expected his employees, the forensic toxicologists, to do because they are trained and experienced, which are not in and of themselves included in the protocols.
[874]*874! The record tends to show that the state toxicologist has sometimes approved methods after their implementation by the State Patrol and has sometimes lacked detailed knowledge of each step in each of the evaluation and certification procedures. While the former state toxicologist (who has since resigned) may have attempted to initially delegate authority to others and may have relied upon the scientific expertise of those in the State Patrol, this record establishes that the methods required to be approved by the state toxicologist have in fact been approved by the state toxicologist. RCW 46.61.506(3) requires that the state toxicologist approve the methods of breath analysis, but it does not require that he design the procedures.
The record also supports the conclusion that breath testing is not less reliable because the procedures are not set out in the Code. In this regard, we disagree with defendant's claim that it is impossible to tell what procedures must be followed and whether there has been compliance with those procedures if they are not set out in the Code. While detailed WAC's might provide a standard for comparing procedures actually employed with those required, ability to ascertain compliance or noncompliance does not require that they be there. As evidenced by detailed defense questioning contained in this record about the certification and evaluation procedures used with regard to specific DataMaster machines, defendant has been able to obtain the protocols and records necessary to ascertain compliance.6
Defendant is not denied due process as a result of the manner in which the breath testing methods have been approved. This follows from this court's long-standing requirement that when relying on results of machine tests for breath alcohol, the State must establish that the machine was in proper working order, that if chemicals [875]*875were used in testing they were correct and correctly used, that the operator was qualified and performed the test correctly, and the results are accurate. This foundational requirement was first set out in State v. Baker, 56 Wn.2d 846, 355 P.2d 806 (1960). See State v. Brayman, 110 Wn.2d 183, 191-92, 751 P.2d 294 (1988) (reaffirming foundational requirement and describing WAC's where these requirements have been set forth); State v. Franco, 96 Wn.2d 816, 828, 639 P.2d 1320 (1982). Breath test evidence alone is not conclusive evidence of the per se offense. Brayman, at 191.
The defendant believes that unless the approved methods for evaluation, certification, and preparation of the simulator solution are published in the Code the State cannot possibly meet this burden. We do not agree. The State may meet its burden regardless of whether the approved methods are set out in the Code. The evidentiary burden is the same, regardless.
If the State satisfies its initial burden, the test results are admissible. A defendant still has the opportunity to attack the test results. Defendant may introduce evidence refuting the accuracy and reliability of the test reading. Brayman, at 192; Franco, at 828. As noted above, the procedures for evaluating and certifying the DataMasters approved by the state toxicologist are known to (or discoverable by) the defense, as is apparent from the record in this case. The same is true of the protocol setting out the procedures for mixing the simulator solution. If defendant presents refuting evidence, resolution of the issue of the reliability and accuracy of the test results is for the trier of fact. Cf. Bremerton v. Osborne, 66 Wn.2d 281, 282, 401 P.2d 973 (1965) (challenges to qualifications of operator of Breathalyzer and the sufficiency of the checking and testing procedures go to the weight rather than the admissibility of the evidence).
Finally, we address defendant's reliance on State v. Watson, 51 Wn. App. 947, 756 P.2d 177 (1988). There, the Court of Appeals held that breath test results were not admissible due to noncompliance with WAC 448-12-015, requiring checking and calibrating Breathalyzer machines [876]*876at least once every 3 months. Breathalyzer machines were used to test breath for many years before DataMaster machines were approved for this purpose. The court's conclusion was rooted in State v. Peterson, 100 Wn.2d 788, 791, 674 P.2d 1251 (1984), where this court held that the Legislature, in enacting a statute defining the DWI crime in terms of a Breathalyzer reading, created a presumption that Breathalyzer machines would function properly for 3 months if the maintenance requirements in the WAC were followed. The court in Watson also relied upon State v. Ryan, 43 Wn. App. 488, 490, 717 P.2d 1390 (1986), where the court held in reliance on Peterson that for Baker's foundational requirement to be satisfied the directives of the WAC had to be followed.
In Watson, the question was not whether the maintenance procedure had to be set out in the Code, but instead was whether the State could meet the Baker requirement absent compliance with the existing WAC, which the Court of Appeals reasoned was the only way to meet the foundational requirement. Watson is thus clearly distinguishable.
For the reasons set out above, the trial court's ruling denying Strain's motion to suppress results of the breath test is affirmed.
State ex rel. Dawson v. South District Court
The second case before us involves the consolidated cases of several defendants whose breath test results were suppressed on the ground that defendants' due process rights were violated by the State's failure to preserve a record of code messages ("invalid sample" messages) which appeared on the DataMaster screen and prompted the operator to reset the machine and attempt another breath analysis.
Each of the four defendants, Theresa Garrison, Robert Slater, Patrick Selfridge, and Orville Luder, was arrested and charged with driving while intoxicated. Each defendant agreed to submit to a breath test from a DataMaster machine located in Snohomish County. The breath analysis [877]*877showed breath alcohol content for Garrison at .18 percent and .17 percent; for Slater at .15 percent and .15 percent; for Selfridge at .19 percent and .19 percent; and for Luder at .14 percent and .14 percent. Each defendant pleaded not guilty. In each case, counsel for the defendant moved for dismissal of the charges or for suppression of the breath test results, arguing that defendant's due process rights were violated by the State's failure to preserve (or destruction of) evidence, i.e., a record of the code messages.
A hearing was held August 7, 1989, and October 19, 1989, during which testimony was given by Sergeant Gullberg of the Washington State Patrol and by a DataMaster technician.
According to the testimony given by Sergeant Gullberg, who was identified as the most qualified person to testify as a witness regarding the technical aspects of the Data-Masters: The DataMaster consists of essentially two devices—infrared analysis equipment and a computer which calculates test results and stores some information. When a breath sample is tested, the machine performs a test reading each one-fourth, for the duration of the breath sample. While the machine performs these test readings, the computer calculates the average of the three most recent analyses. Through this process, the machine can detect the slope of the breath alcohol curve. The alcohol in a breath sample should increase until reaching a fairly level plateau. If the machine detects a decrease in the slope over a certain time, a negative slope, it is programmed to indicate that an "invalid sample" exists. If this happens, an "invalid sample" code message, known as message code 14 (formerly known as error code 14), appears on the Data-Master screen, and the particular test is aborted. The operator is instructed to reset the machine and begin a new test. The "invalid sample" messages are not preserved in the machine's permanent memory.
Sergeant Gullberg testified that there are two major causes of the invalid sample error messages—the machine "believes" it has detected mouth alcohol in the sample, or [878]*878the machine is out of adjustment. As to the latter, Sergeant Gullberg testified that the electrical settings may be out of adjustment and adjusting electrical settings back into their ranges and proper tolerance seems to affect the frequency of invalid sample message occurrences.
Sergeant Gullberg explained that identification information about the operator and the subject, and the final alcohol reading results, are sent to the permanent memory. That memory is later polled by a host computer located at the State Patrol Crime Lab, and the stored material is sent to the host computer. That material can be printed and used by interested persons; the printed form is known as a database. Sergeant Gullberg testified that the primary purpose of maintaining this information is for statistical analysis of the statewide DWI enforcement program, and for educational and research purposes. When the Data-Masters were received from the manufacturer, they were not programmed to preserve a record of the invalid sample messages and thus did not have that capacity.
Sergeant Gullberg testified that the State Patrol did not originally know that the invalid sample messages were not being recorded in the permanent memory. About 6 months after the new machines were deployed the State Patrol learned that invalid sample code messages were occurring in the field but message code 14 occurrences were not showing up in the records. Gullberg said that in September 1986 he requested from the DataMaster manufacturer a variation of the software then used, to provide for recording the invalid sample messages. This change was one of several requested. Thereafter the State Patrol experimented with variations of the software. In late 1987, the State Patrol sought permission from the state toxicologist to install new software with a number of changes, including the recording of the occurrence of the invalid sample messages, on all DataMaster machines. The state toxicologist approved the changes in late 1987.
The new software was installed on one machine, but removed after a short time because of a significant number [879]*879of invalid sample messages being reported by the machine. Sergeant Gullberg reported the problem to the manufacturer, requesting that the problem be corrected, but leaving the manner of correction up to the manufacturer. A second, revised version of the new software was installed in two machines. The software was again revised in April and June 1988. Each time, the revisions were requested by the State Patrol because the machines were reporting large numbers of invalid sample code messages. Sergeant Gullberg testified that the fourth revision corrected the problem. The first three versions showed the rates of invalid sample messages to be, respectively, about 30 percent (31.4 percent), 22.8 percent, and 40.5 percent. The fourth version showed the error rate to be 4.3 percent.7 No version of the revised software was ever installed in more than a few machines, and none was ever installed in Snohomish County. Sergeant Gullberg testified that he would expect the same rate of recorded messages to result if the same software was installed in other machines.
While the invalid sample messages generally indicate the presence of mouth alcohol or an electrical problem needing adjustment, Sergeant Gullberg testified that he believed the first three versions of the software were too sensitive in their calculations as a result of programming errors (and this is why the recorded rates were so high). He said that the software would show invalid samples on experiments when mouth alcohol was not actually present. Further, Sergeant Gullberg testified, the results recorded with the first three versions of the software were inconsistent with anecdotal information received from operators of the machines about the frequency of the invalid sample error code messages. Gullberg testified that he could not give an opinion as to the actual incidence of invalid samples occurrences before any versions of the new software were installed. He [880]*880said there were no statistics based on any of the "old" software.
The DataMaster technician who testified said that one of the DataMaster machines, located in Lynnwood, is used 1,500 times a year. She stated that if the machine was giving invalid sample messages one-fourth of the time she would probably have been getting phone calls from the operators every night; however, she was not. She testified that she had experienced instances where electrical problems caused invalid sample messages.
The last version of the software was removed from : machines in King County after some courts ruled that the software could not be used because it was not approved by the state toxicologist. Sergeant Gullberg testified that none of the DataMaster machines in the state currently have any version of the new (amended) software.
Following argument by counsel, the trial court made oral ‘findings which are generally supported by the evidence, including a finding that technology exists to gather and preserve a record of the occurrences of the invalid sample messages. The trial court also found that the State Patrol elected to make the DataMaster program changes based on a number of factors, including drivers' license examination statistics, and to gather statistics about the error code situation for the database.
The court said, however, that the motives for the requested changes were immaterial as a matter of law, because the effect of the changes "was to suppress evidence that was available—to reduce the evidence that was available in the first program, the second, the third, and so on— that information was available in the database for discovery to Counsel for Defense." Verbatim Transcript, at 56 (Dawson) (Oct. 19, 1989). The court said that evidence "might cast doubt in the mind of the jury as to whether the State had proven that the machine was working beyond a reasonable doubt." Verbatim Transcript, at 57 (Dawson) (Oct. [881]*88119, 1989). The trial court suppressed the results of defendants' breath tests, reasoning that culpability may not be based on a machine test if the test result is in doubt.
The State sought and was granted permission to obtain interlocutory review of the suppression ruling. Snohomish County Superior Court then requested that this court accept direct review of the four consolidated cases, which this court did.
In seeking discretionary review of the trial court's ruling, the State argues that the United States Supreme Court's holding in Arizona v. Youngblood, 488 U.S. 51, 102 L. Ed. 2d 281, 109 S. Ct. 333 (1988) should be applied, and that under that case, there is no due process violation here. For the reasons explained below, we agree, and conclude that the trial court erred in suppressing the test results.
We first examine defendants' claim that failure to provide information about the occurrence of invalid sample messages violates RCW 46.61.506(6). The statute provides: "Upon the request of the person who shall submit to a test or tests at the request of a law enforcement officer, full information concerning the test or tests shall be made available to him or his attorney." By its terms, the statute requires information be disclosed about a defendant's own test. The statute does not require that other information about the machine's operation be made available.
Defendants argue, however, that in amending the statute in 1986, the Legislature intended to expand the information available to defendants. They note that the amended statute requires defense access to "all information" about "the test or tests” (defendants' emphasis). There is no merit to this contention. Prior to amendment, the statutory scheme generally provided for chemical analysis of blood or breath, and RCW 46.61.506(6) then provided that one submitting to "a chemical test or tests” must have access to "full information concerning the test or tests . . .." (Italics ours.) The only change made in 1986 [882]*882was the deletion of the word "chemical" in the provision. See Laws of 1986, ch. 153, § 4; compare Laws of 1979, 1st Ex. Sess., ch. 176, § 5. Defendants also seem to imply that the term "all information" in the statute is new. As noted above, this is simply not true. Because the language in the statute did not change as defendants contend, we do not perceive legislative intent that information such as sought by defendants is now encompassed within it.
We next examine the State's claim that because the invalid sample error message does not actually exist in any permanent form capable of storage and retrieval, the evidence does not exist at all. The State argues that defendants are actually asking the State to create evidence by installing the software approved by the state toxicologist which will record the error code messages. The State maintains there is no violation of due process based upon the State's failure to create evidence.
We disagree with this characterization of the issue. We have recognized that where preservation of evidence is at issue, that evidence may either be a "tangible object or a sense impression." State v. Judge, 100 Wn.2d 706, 717, 675 P.2d 219 (1984) (quoting State v. Jones, 26 Wn. App. 551, 554, 614 P.2d 190 (1980)). The invalid sample message appears on the screen; the operator responds to that message by resetting the machine in order to attempt another breath test. However briefly the message appears on the screen, it does exist and is the subject of the operator's sensory impression. The issue is not one involving a question of creating evidence.
Instead, the issue concerns either destruction of evidence or failure to preserve evidence.
The prosecution has a duty to disclose material exculpatory evidence to the defense. Brady v. Maryland, 373 U.S. 83, 10 L. Ed. 2d 215, 83 S. Ct. 1194 (1963); United States v. Agurs, 427 U.S. 97, 49 L. Ed. 2d 342, 96 S. Ct. 2392 (1976). This court found a related duty to preserve evidence under certain circumstances. State v. Wright, 87 Wn.2d 783, 557 P.2d 1 (1976). Then, in State v. Vaster, 99 Wn.2d 44, 659 [883]*883P.2d 528 (1983), the court adopted a "reasonable balance" analysis in cases involving an inadvertent or good faith loss or destruction of evidence.
Since these decisions were handed down, the United States Supreme Court has explained the constitutional due process analysis applicable in cases involving lost or destroyed evidence. In Wright we relied upon federal cases for our analysis, and Vaster depends heavily upon Wright. To the extent those decisions explicate federal constitutional principles inconsistent with recent United States Supreme Court decisions they are, of course, supplanted. It is thus significant that while defendants maintain that Vaster is still controlling, defendants do not argue that the due process analysis under the state constitution varies in this area from that set out in the more recent United States Supreme Court opinions. See State v. Gun-wall, 106 Wn.2d 54, 720 P.2d 808, 76 A.L.R.4th 517 (1986); State v. Wethered, 110 Wn.2d 466, 755 P.2d 797 (1988). That argument has been raised in another case, State v. Ortiz, cause 57551-7 (argued Jan. 15, 1991).
Looking to the recent Court decisions then, in California v. Trombetta, 467 U.S. 479, 81 L. Ed. 2d 413, 104 S. Ct. 2528 (1984), the Court held that the Fourth Amendment's due process clause does not require law enforcement agencies to preserve breath samples in order to introduce breath test results at trial. The Court observed that in destroying defendant's breath samples after testing, the officers acted "'in good faith and in accord with their normal practice.'" Trombetta, at 488 (quoting Killian v. United States, 368 U.S. 231, 242, 7 L. Ed. 2d 256, 82 S. Ct. 302, reh'g denied, 368 U.S. 979, 7 L. Ed. 2d 441, 82 S. Ct. 476 (1961)). The Court then stated:
Whatever duty the Constitution imposes on the States to preserve evidence, that duty must be limited to evidence that might be expected to play a significant role in the suspect's defense. To meet this standard of constitutional materiality, see United States v. Agurs, 427 U. S. [97,] 109-110, [49 L. Ed. 2d 342, 96 S. Ct. 2392 (1976)], evidence must both possess an exculpatory value that was apparent before the evidence was [884]*884destroyed, and be of such a nature that the defendant would be unable to obtain comparable evidence by other reasonably available means.
(Footnote omitted.) Trombetta, at 488-89.
Following Trombetta, the Court decided Arizona v. Youngblood, supra. In Youngblood, defendant was convicted of child molestation, sexual assault, and kidnapping. Defendant claimed the State failed to preserve semen samples from the victim's body and clothing, in violation of his due process rights. The Court characterized the issue before it as concerning "the extent to which the Due Process Clause of the Federal Constitution requires the State to preserve evidentiary material that might be useful to a criminal defendant." Youngblood, at 52.
The Court reasoned:
The Due Process Clause of the Fourteenth Amendment, as interpreted in Brady [v. Maryland, 373 U.S. 83, 10 L. Ed. 2d 215, 83 S. Ct. 1194 (1963)] makes the good or bad faith of the State irrelevant when the State fails to disclose to the defendant material exculpatory evidence. But we think the Due Process Clause requires a different result when we deal with the failure of the State to preserve evidentiary material of which no more can be said than that it could have been subjected to tests, the results of which might have exonerated the defendant.
Youngblood, at 57. The distinction drawn by the Court was based in part upon the difficulty encountered by courts facing the "treacherous task" of determining the materiality of lost evidence and partly upon the Court's unwillingness to read the fundamental fairness requirement as "imposing on the police an undifferentiated and absolute duty to retain and to preserve all material that might be of conceivable evidentiary significance in a particular prosecution. " Youngblood, at 58.
The Court held that "unless a criminal defendant can show bad faith on the part of the police, failure to preserve potentially useful evidence does not constitute a denial of due process of law." Youngblood, at 58.
[885]*885Defendants maintain that the "evidence" at issue here is not unknown or speculative evidence that might conceivably be of significance, but is instead "actual evidence that the prosecution's only evidence [of the] breath alcohol level was inaccurate." Brief of Respondent Dawson, at 14. Defendants argue that the "bad faith" standard of Young-blood does not apply because this is a case where the prosecution has destroyed material exculpatory evidence.
We disagree. The invalid sample messages are not directly related to guilt or innocence. Their relationship to the guilt determination is tangential at best in that, as defendants acknowledge, any record of occurrences would be used in an attempt to discredit the reliability of breath test results. Any record of invalid sample occurrences could not directly serve to confirm or deny particular breath test results.
Moreover, as we have explained in our discussion of Straka's case, the state toxicologist has approved the Data-Master, and the evaluation and certification procedures used by the State Patrol. The state toxicologist has also approved the protocol providing for mixing and testing the simulator solution. The manner in which each breath test is conducted on the DataMaster machines provides machine self-testing each time a defendant's breath is tested. The simulator test provides assurance that a particular breath test is accurate.
According to the testimony, the invalid sample error code messages appear on the computer screen to tell the operator when the machines are unable to run a reliable test. Thus, even a high error rate does not indicate that breath test results from completed tests are unreliable. The State maintains, and there is no controverting evidence on this point, that the invalid sample messages show that the machine is working, i.e., that it refuses to analyze what it perceives as flawed samples. Sergeant Gullberg testified that the "old" software currently in use accurately and reliably tests breath, samples. The testimony in this case is consistent with that in State v. Straka, discussed above.
[886]*886We conclude the evidence which defendants claim should have been preserved is not material in the constitutional sense. Therefore, contrary to defendants' view, this case falls within Youngblood's analysis. At most, a record of invalid sample error messages can only be termed "material that might be of conceivable evidentiary significance in a particular prosecution," Youngblood, at 58, or "evidentiary material that might be useful to a criminal defendant," Youngblood, at 52.
Under Youngblood, the destruction of or failure to preserve such evidence does not violate defendants' due process rights if the police did not act in bad faith.
The trial court did not reach the question whether the police acted in bad faith, terming the State Patrol's motives for its policy decisions immaterial. Defendants claim that bad faith is shown because the State violated RCW 46.61-.506(6) while knowing of the invalid sample message codes since 1985 or 1986. Because we have held RCW 46.61.506(6) does not require defense access to this information, this contention fails.
Defendants also claim bad faith is shown because the State Patrol stopped trying to find an acceptable version of software that will record the messages, and has instead reverted to the old software rather than use the software approved by the state toxicologist in 1987. Put another way, the defendants claim that the State Patrol has chosen to use software that destroys the invalid sample messages.
However, the record establishes that there was sound reason for the State Patrol's efforts to resolve what was perceived as a programming error. Sergeant Gullberg testified that he believed there were software programming errors which resulted in the high rate of recorded invalid sample error code messages in the first three amended Versions of the software. Sergeant Gullberg is the best qualified person in the state to testify about the technical aspects of the machines. The recorded invalid sample rates for the first three amended versions of the software vary Considerably, going from 31.4 percent to 22.8 percent to [887]*88740.5 percent. The widely varying rates appear to support Sergeant Gullberg's conclusion that software programming errors occurred.
Further, the testimony was that the invalid sample message rates recorded by the first three versions of amended software were not consistent with anecdotal reports from the field. The version approved by the state toxicologist is in this group. Sergeant Gullberg testified that he did not recall the same problem occurring with the variations experimented with before the 1987 version.
The State Patrol chose not to use the last three versions (including the one reporting a 4.3 percent invalid sample message rate) because courts had rejected test results on the basis that those versions were not approved by the state toxicologist.
We fail to perceive any bad faith on the part of the State Patrol. When received from the manufacturer, the machines did not have capacity to record the invalid sample messages. The State Patrol sought to record the message rate for statistical, research, and educational purposes, and in endeavoring to do so encountered considerable difficulty obtaining and using software which accurately reflected truly invalid samples. There is no showing of any bad faith, much less of any bad faith directed toward any defendants.
Under Youngblood, we conclude that defendants' due process rights have not been violated by the destruction of or failure to preserve a record of the invalid sample error code messages. We note that defendants can still challenge the breath test results, as explained above in State v. Straka.
The State's motion to strike portions of defendants' brief is denied.
The trial court's order suppressing the breath test results is reversed, and these cases are remanded for trial.
Dore, C.J., Dolliver, Andersen, and Guy, JJ., and Callow, J. Pro Tern., concur.