Massachusetts breath testing for alcohol: a computer science perspective.

• Author: Workman, Thomas E., Jr.

1. Introduction

   Massachusetts relies primarily on the technology of breath testing (1) in the prosecution of criminal complaints for driving under the influence of alcohol. (2) Breath testing is performed on Draeger Alcotest 7110 MK III-C machines, which are purchased statewide as the machine of choice for state police and city police departments. The Draeger Alcotest 7110 is used statewide in two other states: New Jersey and Alabama. Other states, such as New York and California, manage their breath testing programs on a county by county basis, using different machines throughout the state.

   Prosecutions for Operating Under the Influence have decreased in recent years, as reflected in statistics collected by the Massachusetts District Courts, where most prosecutions are adjudicated. (4)

   [GRAPHIC OMITTED]

   Prosecutions also vary from Court to Court, again as reported by the Clerks of the District Court through the District Court's webpage, whose data is presented here. (5)

   [GRAPHIC OMITTED]

   Cases are prosecuted on the theory that the citizen was impaired, and the evidence of impairment is proved through the testimony of witnesses as to the manifestations of that impairment. (6) A second method is now available, as the "per se" provisions of Mass. Gen. Laws ch 90 [section] 24 provide that the presence of .08 % alcohol in the blood, as exhibited by that percentage present in the breath, is per se a crime, provided that the other elements of the offense are present. (7)

   The advent of the "per se" statutory provisions placed an increased burden on the science of breath testing, because the testing of breath in the previous statutory scheme permitted a defendant to argue that despite an alcohol level that exceeded .08, that in fact they were not impaired. Under the new statutory scheme, that defense is no longer available. (8) One can argue that the machine was not operating properly, and that the reported result is incorrect. (9) This places a new emphasis on the science of breath testing, for prosecutors, defense attorneys, and Judges.

   In the United States, four companies manufacture nearly all of the evidentiary machines used by law enforcement to evaluate human breath for ethanol. CMI has the largest market share, and manufactured the Intoxilyzer 5000 (the machine that was employed statewide in Massachusetts prior to the statewide deployment of the Draeger Alcotest 7110 MK III-Q. (10) The Datamaster models, from National Patent Analytical Systems, are the second most popular machines. (11 The Draeger and EC/IR machines share a small minority market share.

2. The Appearance of Simplicity

   The User Interface for breath testing devices is designed with simplicity in mind. Basically, the power is turned on, the start button is pressed, and the subject blows into the machine until the operator tells them to Stop. (12) The sequencing of the test is controlled by the software in the machine, and the operator's role is to follow the steps as directed by the display. (13) To the casual observer, the machine is like a bathroom scale: blow into it and it reports ethanol content.

   What goes on inside the machine is anything but simple and straightforward. Statistics of breath tests are not available for Massachusetts (the data has not been provided in response to requests), however data is available from other states. Breath test data is available on the internet for the state of Washington. (14) Florida data has been analyzed and presented here, and it is expected that the data in Massachusetts is not significantly different from data in other states. (15)

   One in four tests in Florida do not report a numeric result. (16) In Arizona, which uses the same machine that is used in Florida (the Intoxilyzer 8000), one in three tests does not report a result. (17) The distribution of test "Results" for Florida's data set of 90,000 breath tests is as follows:

   Value Reported 74% Refusals 19% Malfunctions 7% Note: Table made from pie chart. III. Breath Tests that Report a Value

   Of the tests reporting a value for the amount of ethanol in the subject's system, one in five reported a result less than the statutory .08 threshold. (18) A subject who "passes" the test is equally likely to have a zero Breath alcohol content, as they are to have a value below .08 but above zero. (19)

   The Florida tests are tainted by a software defect which reported results when the volume of the breath test sample was less than the required 1.1 liter sample. (20) In addition to this problem, some breath tests, which measured a sufficient sample of breath, were incorrectly rejected and reported as a "Refusal." (21)

4. Breath Tests that Result in a "Refusal"

   Most breath testing machines report different explanations for tests that result in a "Refusal". Some subject samples are rejected due to improper processing by the machine. Other tests are rejected because of improper instructions (e.g. the Operator tells the subject to stop blowing before an adequate volume of air is supplied). The distribution of "Refusals" is provided graphically here:

   Subject Refused 85% Aborted by Officer 10% Volume Not Met 4% Improper Sample 1% Note: Table made from pie chart. Not all "Refusals" are caused by the subject. Of the Refusals in Florida where a valid first sample was measured, 92% of the subjects supplied a first sample that was sufficient for the machine to measure the ethanol, and that first sample was alcohol free. (22)

   Breath testing machines, including the Draeger Alcotest 7110 MK III-C used in Massachusetts, permit the operator to abort a test, generating a refusal, even after the subject has supplied a sample that has been evaluated by the machine and has generated a valid measurement of subject alcohol. On these machines, operators are trained that at any time they can press the "Start" button and the machine will "Reset." They also learn that if a test is in progress, pressing the "Start" button test will record a "Refusal." All known breath testing machines discard the data associated with the aborted sample. This "feature" has no known legitimate purpose, and abuse of this "feature" permits a police officer to generate a "refusal", with all of the administrative penalties associated with a refusal, at will.

   In addition to the problems of operator-induced refusals, courts in Florida have ordered the suppression of breath tests because the conduct of the operator can manipulate the result. In State v. Briggs, the Court held an evidentiary hearing to evaluate the degree of operator manipulation that is possible and, at the conclusion of that hearing, ordered the suppression of breath test results. (23)

5. Self-Diagnosed and Reported Malfunctions

   An examination of the "malfunctions" that are self diagnosed and self reported, for the Florida dataset, (24) discloses the following distribution:

   Tolerance Failure 23.3% RFI Detected 37.6% Purge Fail 23.0% Other 16.2% Note: Table made from pie chart. No. 020 Agreement 3.3% Slope 2.2% Diagnostic Failure 1.9% Range Exceeded 0.1% Interference Detected 5.0% Note: Table made from pie chart. * RFI Detected: Self reporting of RFI indicates that Radio Frequency Interference has been detected by the machine. RFI can be self generated by components that are failing within the machine, or they can be the result of electronic equipment, which includes cell phones, police radios, air conditioners, refrigerators, computer printers, fluorescent lights, or virtually anything that either plugs in or has a battery.

   * Tolerance Failure: The calibration sample supplied to the machine with a known amount of alcohol does not compute a value that is within the accepted tolerance.

   * Purge Fail: The machine was unable to remove all of the ethanol from the chamber by blowing fresh air through the machine.

   * Interference Detected: The machine has detected the presence of some material that distorts the ability to measure the subject's breath.

   * Ambient Failure: The machine is unable to calibrate itself based on the initial evaluation of the chamber of air before the test begins.

   * No. 02 Agreement: The samples supplied by the subject produce results that are more than .02 apart, which indicates that the machine is not able to measure the samples correctly.

   * Slope: The machine is unable to calculate a result because the sensors vary more than a predetermined amount during the course of measuring 120 values per second over the course of a breath sample.

   * Diagnostic Failure: The self diagnostic that is performed at the beginning or end of the test has determined that some aspect of the machine is not functioning correctly.

   * Range Exceeded: The amount of measured alcohol exceeds .500, and it is presumed that the machine is not working correctly.

   When errors are encountered, the common practice is to have the subject submit to an additional testing procedure, in an effort to get a test result that is admissible in Court. The statute in Massachusetts does not explicitly require a citizen to submit to multiple attempts to test their breath, and in fact, includes language that is singular in syntax, inferring a single test. (25) Notwithstanding the requirement for citizens to submit to "a test," multiple testing on the same breath testing instrument is commonplace. In Arizona, one in every three administered tests is a repeat test. The practice of requiring an additional test when the machine self diagnoses an error is the norm. In states where there are enhanced penalties for breath levels that are higher than .08 (referred to as "aggravated" DUI in some states), it is not uncommon for a citizen with a test result just below the statutory aggravated limit to be re-tested in an attempt to obtain a test result that carries a more severe penalty. (26)

   In Florida, one citizen was tested 13 times on one machine, by one officer, in one hour. (27) These instances occur because in some situations, a machine that registers an error or multiple...