Of DataMasters, Discovery and DUIs

By Stephen Hayne

Shortly after midnight on December 23, 1985, a somewhat bewildered police officer stood before a strange machine in a small room in the Bellevue Police Department. Manual in hand, he tentatively pushed a button on the unfamiliar keyboard. Suddenly, the machine came to life; sucking, wheezing and whirring through its preliminary cycles. A few moments later, history was made as a DWI suspect blew the world’s first official breath sample into the BAC Verifier DataMaster II. Thus began a new era in the prosecution of drunk driving cases in Washington.

Is the widely-touted DataMaster the answer to prosecutors’ prayers, as claimed by the state toxicologist and its fledgling manufacturer? Or is it an unreliable, enormously complicated, technological boondoggle, as many defense attorneys maintain? Only time will tell, but the state’s choice of the DataMaster over its obviously more qualified competitors raises serious questions for future judges and juries. How and why the DataMaster was selected, how it does (and doesn’t) work, and what its use means to attorneys, judges, and juries in DWI cases is the subject of this article.

I do not claim a lack of bias on the subject. I defend DWIs , plain and simple. On the other hand, the factual information in this article is accurate and based on the state toxicologist and State Patrol’s own documents, studies, statements, and testimony. The materials were collected during a four-month-long discovery process defending the first cases involving the DataMaster, and represents a review of 20,000 + documents, six days of depositions, three days of hearings and untold hours of research. The conclusion is inescapable: the state may have made a very big mistake.

I. The DataMaster: How? Why?

In August 1983, the State Patrol convinced an ad hoc budget committee that it was time to replace its 30-year-old Breathalyzers with new “infrared” alcohol breath testing machines. Four manufacturers were invited to submit machines for testing: the Intoxilizer 5000, Intoximeter 3000, Breathalyzer 2000, and the BAC Verifier (the DataMaster’s predecessor). The initial evaluations were designed to test each machine for precision (how close together repeated tests of the same sample are), accuracy (how close the test result is to a known value), and blood/breath correlation (how close the test result is to a contemporaneously drawn blood sample). No tests involving quality control, manufacturing history, or reliability were involved.

The tests, performed by the Washington State Patrol Crime Laboratory from February to March 1984, reveal some disturbing results. The Verifier failed over half the tests for precision and accuracy, misidentified acetone as ethyl alcohol, and completely broke down during other important test procedures. This disappointing performance led to a State patrol memorandum, “Conclusions and Opinions on Each Instrument,” which hailed the Intoxilizer 5000 as the Patrol’s clear “instrument of choice.” The memo described the Verifier’s performance:

Several functional problems were experienced with the instrument during testing. The printout was not of an acceptable format. The company is new in the breath test field and has only recently demonstrated an interest in data collection capability. No data collection has been demonstrated . . .

Despite the Patrol’s misgivings and the Verifier’s poor performance on the evaluation tests, the State Toxicologist, Dr. Vidmantas Raisys, testified:

The initial evaluations of the four instruments, the results as far as I was concerned, were acceptable; therefore we could have gone with any of the instruments (Raisys’ dep., p. 23)

Dr. Raisys’ conclusions were at odds with those of the people who did the actual testing, but all four manufacturers were invited to bid on the state’s contract. Three bids were received in August 1984. Verax Systems, Inc., manufacturer of the Verifier, was the clear winner, underbidding the Intoxilizer by almost 30%. In November 1984, the contract was signed, but subsequent events would more than confirm the Patrol’s fears about the machine’s performance and the fledgling manufacturer’s ability to meet the demanding specifications and delivery dates.

While awaiting production and delivery of the first 25 machines (due March 1985), the state toxicologist signed into law new WAC provisions in 448.12 officially “approving” the BAC Verifier DataMaster for use in DWI prosecutions in the state of Washington. The name change from “Verifier” to “DataMaster” was more than cosmetic; the DataMaster was, in many respects, an entirely different machine from the Verifier evaluated by the Patrol in 1984. A completely untested, much more complicated device than its predecessor, the DataMaster had added a “data collection capacity.” This required numerous fundamental design changes, none for the purpose of improving the machine’s accuracy, precision, or reliability. In fact, Dr. Raisys “approved” the machine without ever having seen one:

Q: Then specifically did you [prior to approving the DataMaster] review any tests, the results of any tests that were performed on the DataMaster rather than the Verifier?

A: I don’t think so.

Q: Had you received any reports, studies, test results on precision, accuracy, blood/breath correlation on the BAC Verifier DataMaster before you certified it?

A: Before the Washington Administrative Code, no, I don’t think so. Or at least I don’t remember.

Dr. Raisys’ testimony was confirmed by Sergeant Rod Gullberg, the chemist in charge of the testing program for the State Patrol:

Q: . . .at the time that the [DataMaster] was approved for service in the state of Washington, had you relied on any tests performed by any other laboratory, state agency, research facility?

A: No.

No other state or federal agency tested the DataMaster, either, simply because the machine was still on the drawing board. Building one proved an extremely difficult task for the previously unknown and financially strapped manufacturer. The delivery date for the first 25 machines came and went, and the State Patrol’s anxieties increased. When four prototype machines were finally delivered, they didn’t work. At this point, termination of the contract was considered:

Q: . . . had any consideration been given at that point in time to terminating the contract with Verax?

A: Yes.

Q: When were those first concerns or discussions held?

A: I would—to the best of my recollection, about the time the 120 days [for delivery] was expiring and we had those first four [machines] and they were locking up. At some point in there we had, I guess, to make a determination as to—at that point I think we could have legally terminated the contract or agreement and we had to make a decision as to whether or not it was in our best interest to do that or to continue with the company. . . [w]e sent George Ishii, who was at one time the director of the Seattle Crime Lab, to the factory to meet the people and look at the factory and make a judgment as to whether or not he felt they [could provide us with an instrument that would do what it was supposed to do].

Q: His conclusion was?

A: His conclusion basically was that they could, but that it would be a tough, tough chore for them, that they were a relatively small, relatively new company. He felt that they had the personnel and the desire and the ability to deliver us the instrument that would perform properly and that because of their size and their growing pains that they could fulfill the contract, deliver the appropriate number of instruments, but it would be a tough, hard battle for them. (Program Mgr. Lloyd Danielson’s dep., p. 33)

Thus did the state reluctantly decide to gamble on Verax and its DataMaster, a bet which may or may not pay off. Considering the machine’s record under controlled lab conditions, the Stat Patrol was justifiably concerned over its potential under the stresses and strains unique to actual “field” conditions.

Defense counsel and prosecutor alike must become knowledgeable enough about the design and function of the machine to recognize the often subtle hints of a malfunction. Due to space limitations, what follows is an overly simplified explanation of how the machine is supposed to work, mechanically, scientifically and electronically.

II. The Testing Sequence

Since the machine is always “warmed up,” the testing sequence can begin any time after completion of the 15-minute observation periods required under WAC 448.12.230.

The officer begins by typing in the time the 15-minute observation period began. All entries are made on a standard typewriter-type computer keyboard connected to the machine by an electrical cord. He types in his name, the code number for his police department, the subject’s name, date of birth, sex and ethnic origin.

The officer then enters coded numbers for the arrest location, crime arrested for, whether an accident was involved (and what type), where the subject had been drinking, and the subject’s occupation. He has the option of reviewing the data to check if for accuracy, and then he enters all of the information into the computer memory of the machine. After the data is stored, the actual breath test sequence is begun and the officer’s role, barring difficulty or a refusal, is essentially over, since the remaining steps are largely automatic.

After insertion of the “evidence ticket,” the machine “purges” itself by drawing room air into the “sample chamber.” This purging process, which occurs several times during the test sequence, is designed to clear the chamber of any residual alcohol and to test the “ambient” (room) air to ensure the absence of alcohol or other interferents in the immediate environment.

The next step is “ambient zeroing”: the machine establishes the level of any alcohol remaining in the machine and subtracts it from the subsequent test results. The result of the ambient air test is then recorded on the evidence ticket as the first “blank test” (see evidence ticket, figure 1), along with the time. In the author’s experience, all “blank test” results have been .00.

Next is the “internal standard” check, in which a quartz filter is automatically inserted into the sample chamber between an infrared light source and a “detector board.” The quartz filter absorbs infrared energy in a known, predetermined amount and is used to verify that the machine is working properly by correctly “recognizing” the presence of the filter. If it does, the word “verified” is printed on the evidence ticket.

The officer then attaches a mouth piece to the breath tube. The machine allows one minute for the officer to type in “y” or “n” to the displayed question: “subject refused?” If no answer is typed in within the 60 seconds, the test sequence automatically terminates, erases the data previously entered, and goes back to step one. If “y” is typed in, the “refusal” is printed on the evidence ticket and the test ends.

If the “n” is entered, the subject is given two minutes to provide an “acceptable” breath sample by “inhal[ing] a full breath and then provid[ing] a slow, consistent, continuous breath sample through the mouthpiece” until the officer says to stop. The subject must keep blowing until the machine flashes the words “test results, alcohol ._______” (to two digits).

If the subject blows, but the machine does not “accept” the sample, after two minutes it will again flash “subject refused?” At this point, the officer must decide whether the subject has made “an earnest attempt” to provide an acceptable sample. If not it is considered a refusal, the test is terminated, the defendant is usually charged based on the remaining evidence and will face a difficult battle with the Department of Licensing over the one year loss of license for “refusing” the test.

In most cases, however, the breath sample is accepted and the alcohol (or any other substance which the machine “sees” as alcohol) will be measured via a complicated mathematical-electronic process (described later). A two-digit result is then printed on the evidence ticket following “subject sample.” The purging or blank test process is then repeated with another .00 result.

The next step is to test a “simulator solution” (shown on the evidence ticket as the “external standard”). Attached to the back of the machine is a jar filled with a mixture of alcohol and water with a small space of trapped air above the solution. This mixture is supposed to approximate a .10 alcohol concentration in the trapped air which the machine must “read between .090 and .110 (WAC 448.12.230). If the result is outside these parameters, the test results are invalid. (Note, however, that an “external standard” of less than .10 does not mean the machine is “reading low,” since the potency of the solution deteriorates rapidly and is usually replaced every 30 days.)

After this test of the simulator solution, the subject must give a second sample (using a new mouth piece). The results for the subject’s two tests must agree “within plus or minus 10% of the average of the two” to be considered valid under WAC 448.12.220.

The State Patrol reports that 11% to 20% of completed tests on DWI subjects are outside the acceptable range, despite the fact that Washington’s standard is less stringent than most other states’. So far, the Patrol has been unable to solve the mystery of so many inconsistent results, a phenomenon which is theoretically impossible according to the manufacturer’s claims and the machine’s “failsafe” design.

Once the second sample has been measured, the machine does a final room air “blank test,” which is printed as a .00; the time is recorded, and the evidence ticket is removed from the machine. The whole testing sequence takes approximately three to four minutes.

Most of the entered data and test results are then stored in the memory of the machine for later gathering by the Patrol’s centralized “host” computer in the crime laboratory in Seattle. Regularly, the “host” computer automatically telephones each DataMaster and “polls” it for the accumulated data, freeing the limited memory capacity of the machine for the next tests. This accumulated “database” is permanently stored in the host computer’s memory for later review by interested persons.

Next, we will discuss the role of individual mechanical and electronic components of the machine, how it attempts to measure alcohol in the breath and, finally, where to look for relevant discovery materials in DWI cases.

III. What is a DataMaster and How Does it Work?

The BAC Verifier DataMaster II is the still-evolving descendant of a relative newcomer to alcohol breath testing: the BAC Verifier. It is a machine which was designed and produced specifically for Washington by a young company headquartered in Fairport, New York, Verax Systems, Inc.

As far as breath testing machines go, the DataMaster is very appealing in theory and appearance. It differs from other machines in that it has the capacity to gather and transmit information from individual police departments to a host computer located in Seattle. This feature proved irresistible to Washington officials, even though most of the data gathered is, at best, of passing interest in the prosecution of a DWI case. However, this “database” is a potential gold mine to the defense attorney willing to put the time, money, and effort into careful research on the track record of a particular machine. This topic will be discussed more thoroughly below.

First, how does the machine actually measure a breath sample for alcohol?

As a suspect blows into the heated DataMaster breath tube, two components are ensuring the highest possible test results: the thermistor and the slope detector. The thermistor is the first component the breath sample passes through inside the machine and is a device which measures “resistance” electrically. It is designed to ensure that the subject is blowing hard and long enough to deliver a sample from deep in the lungs, where the highest concentration of alcohol is found. The machine is programmed to require the subject to blow hard enough to establish a sufficient resistance level for at least five seconds in order for the sample to be accepted. Once the air sample passes the thermistor, it travels into the sample chamber via the three-way valve (the same valve which directs samples of room air and simulator solution into the sample chamber during other phases of the test sequence).

As the breath passes into the sample chamber, its alcohol content is “read” four times a second, and the results are plotted electronically by the slope detector. The slope detector is programmed to expect a steadily rising curve, indicative of deeper and deeper lung air. When the curve or slope reaches the highest reading and begins to fall off or level out, the machine captures that reading and flashes the result on the control panel screen. The result is then printed on the evidence ticket and stored in memory for eventual transfer to the host computer in the crime lab.

The slope detector is claimed by the manufacturer to “protect” suspects against false high readings by detecting mouth alcohol in the breath sample. Theoretically, it will detect mouth alcohol by seeing a reverse slop (high-to-low) and flashing “invalid sample.” However, no serious testing of this aspect of its function has been performed, and such devices have a poor reliability record on other breath machines.

Thus, while the slope detector and thermistor ensure an “adequate” sample, they usually guarantee the highest result, something the old Breathalyzer couldn’t do.

How does the DataMaster detect alcohol in the breath sample? It employs a method known as “infrared spectroscopy.” This simply means that it shines light through the breath sample and measures the loss of energy between the light source and the detector. The method is based on the Beer-Lambert Law (molecules “absorb”) infrared energy at predictable wave lengths, and if one knows at which wave lengths alcohol is absorbed, one can tell how much alcohol is in the sample by measuring absorption only at that wave length). The DataMaster measures alcohol from 3.37 to 3.44 microns.

The DataMaster’s sample chamber comprises three tubes, approximately one meter (three feet) long. It uses mirrors to reflect the light from one tube to another and on to the detector. The breaks in the tube are necessary in order to lengthen the light path and provide better “resolution” in what the detector sees.

However, despite the relatively long light path, the actual volume of air sampled is only 50 cc (milliliters), approximately the size of a golf ball. Thus, the diameter of the 3-foot-long sample chamber tubes is very small—only about one-third of an inch.

In order for the machine to read a breath sample for the presence of alcohol, the detector transmits data to its internal computer for analysis. The computer applies a complex and as yet undisclosed mathematical/electronic process to interpret the detector’s findings into blood/breath terms (e.g., “subject sample: .l5”). To illustrate the complexity of the process, eight pages of Sergeant Gullberg’s deposition were required to describe just one of the several mathematical formulae employed in the DataMaster.

The manufacturer of the machine, seeking protection under trade secrets and proprietary interest claims, has gone to great lengths to keep the process, mathematical formula and electronic schematics secret. The schematic description of electronics alone reportedly encompasses several hundred pages and weights over twenty pounds.

The actual breath-testing-to-alcohol result process thus remains a secret at this time. But its very complexity allows significant opportunities for error. For instance, despite the guarantee of accepting only consistent samples provided by the thermistor and slope detector, the machine continues to be plagued by frequent test results outside expected or acceptable limits. In theory it is impossible for the machine to register.10 and .14 readings within three minutes on the same subject. In reality, such test results are not uncommon.

However it occurs, once the breath sample has been accepted and analyzed by the machine, it exists through a one-way valve back into the room. Significantly, the original purpose of the one-way valve was to allow a “sample trap” to preserve a sample of the breath for retesting. However, the State Patrol elected to eliminate this feature in their bid specifications for the new machines. Upon being questioned, Dr. Raisys revealed that the decision was based upon Washington law, which does not require preservation of a sample. State v. Canaday, 90 Wash.2d 808 (1978).

The final stage of the testing sequence involves the printing of the test results on the evidence ticket and storage of the data for later retrieval by the host computer at the crime lab.

The data collected is permanently stored and may be retrieved by interested persons. This “database” provides the defense attorney with a wealth of discovery material useful at hearing or trial. At the touch of a button, the entire history of a specific machine, including all test results and recorded malfunctions, can be retrieved. The database can reveal how many times the machine gave results outside the requirements of the Washington Administrative Code, how many times the machine has experienced an error code, how many “incomplete” tests may have preceded a client’s “refusal,” how many times the machine has “misdiagnosed” an interferent, etc.

In preparing for cross-examination of the state’s expert in a DataMaster case, keep in mind the newness of the machine and breath test procedure. Cross examination is limited only by your imagination, knowledge of the machine’s theory, electronic and mechanical features, and its performance history.

IV. Cross-Examination of the DataMaster Technician

The DataMaster is an incredibly complex piece of machinery. Most attorneys will initially be intimidated by its apparent sophistication. However, it is this very complexity which provides many of the fruitful areas for cross-examination. For the relative newcomer, however, it is advisable to keep it simple and direct. The following is a brief outline of suggested areas of cross-examination of the DataMaster technician.

A. History of the Machine

To begin, review the cross-examination of State Patrol Sergeant Rod Gullberg in City of Bellevue vs. Van Noort, (Bellevue District Court Cause BE 92686). The examination reveals the many changes in individual components in the machine since its approval for use by the state toxicologist. Remember, the DataMaster was designed and built specifically for the state of Washington and was literally “approved” before the state had even received the first prototype. But continuous malfunctions and consistent unreliability required literally dozens of changes in individual components. The DataMaster of today bears little resemblance to the DataMaster approved in March of 1985. In fact, individual components continue to be changed, altered, reprogrammed and redesigned. Stress the unproven tack record of the machine and continual alterations. The unfortunate truth is that the machine was rushed into service well before the bugs had been worked out, and is still experiencing unexplained difficulties.

B. Maintenance Records

A record of maintenance and repair worksheets is kept by the State Patrol for each machine. Every time a malfunction or other complaint is received regarding a machine, a maintenance technician is supposed to identify the problem, repair it, recalibrate the machine if necessary, and record the incident. In many cases, the maintenance records will reveal a recurrence of the same problem or malfunctions close in time to the date of the defendant’s breath test. The records also document the frequent inability of the technicians to recreate the problems complained of. Thus, the cause of the malfunction often remains a mystery. The Patrol also keeps a log of telephone complaints form officers in the field identifying problems which are never recorded by the host computer.

C. The Database

The information contained in the database of the host computer can reveal much, but not all, about an individual DataMaster’s past performance. For instance, an interesting development with the advent of the DataMaster is its ability to record incidents of “radio frequency interference.” For years, the State Patrol has maintained that RFI was simply not a problem in breath tests in the state of Washington, despite contrary findings in other states. In many instances, a DataMaster located where Breathalyzer had been—on the same table, in the same room, in the same police station—has shut down repeatedly due to radio frequency interference.

The database also provides ”error codes” which record instances of the machine shutting itself down due to malfunctions: “system won’t zero,” calibration error, printer error, etc., as well as radio frequency interference. A review of a typical database will almost certainly reveal numerous incidents of error codes which may prove useful to defense counsel.

When cross-examining the DataMaster technician, use great caution to avoid losing the judge or jury by over-complicating the process. It is important to reduce these complex principles to simple explanations and easily understood examples.

Use considerable care preparing cross-examination so that the questions and answers will be easily understood by a layperson. Set simple, straight-forward goals. Develop questions requiring yes or no answers whenever possible. Careful preparation will allow the attorney to control the examination by requiring the technician to answer the question asked and only the question asked. As in all cross-examination, remember: Never ask a question when you don’t know the answer.

In summary, the BAC Verifier DataMaster infrared breath testing device signals a new era in prosecuting and defending drunk driving cases in Washington. Its sophistication and complexity offer great challenges to attorneys, especially those on the defense side. However, it should also be remembered that with challenge comes opportunity. To those willing to put forth the effort, the result will ultimately be fairer trials for those clients accused of this increasingly unpopular offense.

1 Data Master Operator’s Manual, p. 25

2 DataMaster Operator’s Manual, p. 25

3 Much of the data gathered is simply demographic.

Stephen W. Hayne currently limits his practice to defending DUIs. Co-author with Douglas Cowan of Defending DUIs In Washington (Butterworths, Legal Publishers), he is former chairman of the criminal law sections of WSTLA, SKCBA, the state chapter of NACDL, and is a founding member of WACDL. Currently on the Executive Board of the WSBA Criminal Law Section, Hayne is a frequent lecturer on criminal law topics. He gratefully acknowledges the assistance of attorneys Doug Cowan and Jon Scott Fox in the research and preparation of this article.