Judge Throws Out 49 Breath Tests

Posted by Lawrence Taylor on May 20th, 2008

An Arizona judge threw out the breath tests in 49 drunk driving cases yesterday….because the manufacturer refuses to let the defense look into its machines.  Thank god for a few honest and gutsy judges still left…. 

I have posted in the past about a battle being fought in courts across the country — notably in Florida, Minnesota, New Jersey and Arizona — concerning the right of an accused to take a closer look at the all-important breath  machine.  Under our current legal system, that machine has essentially become the judge, jury and executioner.

Trial by machine.  But how can this be?

Let’s review the laws, which are essentially similar across the country now due to increased federal pressure and MADD’s influence on state legislatures.  Let’s assume you’ve been arrested for “drunk driving” and you take a breath test which gives a .09% blood-alcohol concentration (BAC) reading.

First, your license will immediately be seized by the officer and he will serve you with a notice of suspension from the DMV: the so-called .08% per se suspension.  That’s it.  And the only evidence to support this immediate suspension at the police station is…the machine.  In most cases, unless you appeal the suspension and are able to prove the machine was wrong, your license remains forfeited.

Second, you will be criminally charged with driving with .08% or higher BAC.  The only evidence for this in most cases will be…the machine.  The machine is rebuttably presumed by law to be reliable and accurate.  In other words, unless you are able to show that the machine is wrong, the jury is told you are presumed to be guilty

Third, you will be charged with “driving under the influence” (DUI), “driving while intoxicated”  (DWI) or “operating a vehicle while intoxicated” (OWI); each state uses different terminology.  The primary evidence for this charge will be…the machine.  Most states now have laws which impose a legal presumption of intoxication if the BAC was .08% or higher.

In other words, unless you are able to prove that the machine was wrong, the jury will be instructed that you are presumed to be guilty.

But it takes time to investigate the suspect in the field, transport him to the station and prepare the breath  test.  Commonly, this can take an hour or two, often more.  So…if your BAC was .10% at the station, what was it two hours later when you were actually driving?  We know that alcohol can take 1/2 to 4 hours to absorb, so the BAC may well have been lower when driving.  No problem.  Again, our state legislatures have relieved prosecutors of that awkward scientific fact, as the following statute from California exemplifies:

In any prosecution…it is a rebuttable presumption that the person had 0.08 percent or more, by weight, of alcohol in his or her blood at the time of driving the vehicle if the person had 0.08 percent or more, by weight, of alcohol in his or her breath at the time of the performance of a chemical test within three hours after the driving.  California Vehicle Code sec. 23152(b)

Again, presumed guilty…unless you can prove the machine was wrong

Ok, so everything depends upon proving the machine was wrong.  How do we do that?

Well…there are two ways, really.  You can have your breath sample re-analyzed by an independent laboratory,  just as can be done with a blood or urine test. Or you can look inside one of these machines and inspect its brain — the software programs that run everything.

Fine.  Let’s try the first method: get the breath sample from the police and have it reanalyzed.  Except that the breath sample was thrown away.  In all police agencies today, the all-important breath sample is dispersed into the air — despite the easy availability of cheap capture devices.  When this constitutional denial of the right to access evidence was raised on appeal to the California Supreme Court, the conviction was reversed:

Due process simply demands that where evidence is collected by the state, as it is with the Intoxilyzer, or any other breath testing device, law enforcement agencies must establish and follow rigorous and sytematic procedures to preserve the captured evidence or its equivalent for the use of the defendant. People v. Trombetta, 142 CalApp.3d 138 (1983).

Not so fast, said the U.S. Supreme Court, which reversed the California Supreme Court and decided that to require saving the breath sample “the evidence must both possess an exculpatory value that was apparent before the evidence was destroyed, and also be of such a nature that the defendant would be unable to obtain comparable evidence by other reasonable means.”  California v. Trombetta, 467 U.S. 479 (1984).  A classic Catch-22: you have to save the breath sample only if it clearly shows him to be innocent.  Why would you then be charged?

Result: no police agency in the U.S. today bothers to save the breath samples.

Hmmm….Ok, but at least we can take a look inside, can’t we?  I mean, we can see if the software running it is accurate and reliable, right?  After all, the U.S. Supreme Court attempted to justify its dubious decision by pointing out that the accused could always inpsect the machine itself:

To protect against faulty calibration, California gives drunken driving defendants the opportunity to inspect the machine used to test their breath as well as that machine’s weekly calibration results and the breath samples used in the calibrations.

Now, I’ve written ad nauseum in the past about all of the reasons why these devices are inherently unreliable and inaccurate.  See, for example, How Breathalyzers Work and Why They Don’t, Close Enough for Government Work, Why Breathalyzers Don’t measure Alcohol, The Mouth Alcohol Problem and Breathalyzer Inaccuracy: Testing During the Absorptive State.  Yet, they are cloaked with an aura of infallibility — to the extent of imposing presumptions of guilt.

Well, at least the Supreme Court says the accused citizen has the right to “inspect the machine”, right?


This is the battle that is being fought around the country today…and, as usual when DUI is involved, largely being lost.  Problem:  the prosecutors don’t have the software code to turn over, nor do the police or the police laboratories.  The code inside of the machine is the property of the manufacturers.  Problem: the manufacturers have all refused to provide their codes, hiding behind the “trade secrets” theory.

Unfortunately for the manufacturers, the Supreme Court of New Jersey recently ordered one maker to turn their “secret” code over to the defense or lose its exclusive contract with the state.  The code was then turned over to defense attorneys, who turned it over to a specialized computer laboratory, which then turned over its findings.  As I layed out in an earlier post (Secret Software Code Finally Revealed),  the results indicated that the software was…a mess.  Nor was there nothing unique about the codes to justify characterizing as “trade secret”.  After pointing out numerous technical problems, the report concluded:

It is clear that, as submitted, the Alcotest software would not pass development standards and testing for the U.S. Government or Military. It would fail software standards for the Federal Aviation Administration (FAA) and Food and Drug Administration (FDA), as well as commercial standards used in devices for public safety…If the FAA imposed mandatory alcohol testing for all commercial pilots, the Alcotest would be rejected based upon the FAA safety and software standards…

No wonder the manufacturers didn’t want to let anyone look at what was inside their profitable little machines.

They shouldn’t worry.  Their machines will continue to be kept secret, breath samples will continue to be safe from re-analysis, test results will continue to be accepted by courts as proof beyond a reasonable doubt, and citizens will continue to be presumed guilty based upon them.

  • jim



    The Alcotest was recently approved as acceptable by the NJ Supreme Court in March of 2008, following a 3 year challenge, give or take, to its’ scientific reliability.

    This case, commonly referred to as CHUN, because of the lead defendants last name, explored many issues, but did not explore the issue I am about to raise.

    I am an expert analytical chemist with 20 years experience. I have used hundreds of software driven analytical instruments, from a variety of manufacturers, for thousands upon thousands of methods, in a FDA driven environment.

    The Alcotest is a device, an analytical instrument,  which one breaths into at a police station, following a DUI arrest in NJ ( in 17 of the 21 counties).

    The Alcotest is an analytical instrument that uses an ethanol (alcohol) standard solution to calibrate the instrument, then determines the amount of alcohol in a persons blood, employing a 2100-1 ratio for alcohol in breath to blood.

    OK. Now we can start.

    Analytical instruments that produce a result of significance in NJ,  100% of the time supply the raw data along with the final result.( except for the Alcotest)

    A result of significance, importance, can be categorized as a result for potency. That is, for example, a result from a HPLC analysis for a batch of acetaminophen tablets manufactured in NJ, or shipped to a NJ distributor, would be of the utmost importance to determine if the product can be sold to the public.

    This type of result would apply to any analytical method employing HPLC, GC, AA, UV, IR , and EC technology ( as well as others).

    The Alcotest employs 2 types of technology ( detectors) IR and EC.

    The result produced by the Alcotest is significant, because if the result is 0.08% BAC or greater, a defendant can be subject to a lot of penalties, including loss of drivers license, jail time, fines, insurance fees, loss of employment for a lifetime, and mandatory rehab as directed by the IRDC (which can be cost prohibitive and thus lead to jail time).

    The significance of the Alcotest result is not in question.

    The question is whether or not the Alcotest produces a valid result.

    Validity in the analytical environment is documented in a number of ways that includes many aspects. That being said, what I am looking for is nothing beyond the basic minimum requirement (even below the minimum requirement).

    One of these minimum requirements is the ability to show the result produced by the instrument is correct, in as far as the math.

    When a standard of ethanol is used to calibrate, or verify calibration, in the Alcotest, it is  ran through an infrared detector and  produces an electric signal response. This response is not displayed or recorded as a number called a “count”.  However, every analytical instrument in NJ that produces a result like this, of significance or importance, has the raw data “count” number in the documentation.

    Now, after the Alcotest is calibrated, it is used to analyze breath samples. The response from a persons breath sample is compared to the response from a standard alcohol solution, and a final BAC is determined, employing the 2100-1 ratio.

    My argument is this….

    Without knowing the signal response from the standard solution, and the signal response from the breath sample, there is absolutely no way anyone in the world can prove the result is correct. The raw data, the “count” , must be available for standards and samples, in order to determine the correct result was produced. This is fundamental basic analytical chemistry for a result of importance.

    There are other adjacent issues to this in the fundamental validation of an analytical result.

    For an IR (infrared detector), there is an acceptable range for an electronic response. This range is typically, referred to as being between 0-1. That is the signal response must fall within the signal range capable of the detector.

    The signal response also must be shown to be linear within the results for a breath test.

    Minimum instrument detection levels and minimum method detection levels must also be determined using standard industry practices.

    None of these things can be determined without the raw data electronic “count” result!

    Now, whether or not the instrument is able to take a 0.12% ethanol solution calibration, vaporize it to 2100 times is not the real starting question.

    The question is whether or not it can be proven. The only way to start to prove this is with RAW DATA, the electronic signal in digital count form.

  • joe

    Here, here. Everyone accused has the right to face their accuser, but if they’re accuser is a machine, how does that equal justice? The companies that make these machines know that increased scrutiny will lead to them losing profits, and God forbid that happens. The state can provide its law enforcement officials with pricey machines that equal convictions, and it leaves the citizens with no recourse. Take away the citizen’s right to inspect the machine and what do you have? Machine produced guilty verdicts. It’s as if “The Matrix” is the goal for some of these people.

  • jim

    listen, it is very simple solution as far as i am concerned, simply write the software to report the integrated signal count, like all the other analytical instruments do, like hplc, gc, gcms, ir, uv, aa, icp, …ect..

    results of significance must have the raw data, … BAC shouldnt be treated like the ph of you pool water.

    it is impossible for drager to have made the alcotest without mweasuring the electronic signal response.

    see my point?

  • koivisto

    As an engineer I know how flawed machinery is and can be. That goes for software too. Software “crashes” all the time. This is another exceptional piece of work by Mr. Taylor. What needs to be taken into account is not a arbitrary number, but the actual condition of the driver.

  • jim

    as an analytical chemist, the number can be taken into account, if the raw data was with it, and the thing was validated as eveery analytical instrument is validated, not just the technology, EACH INSTRUMENT

  • koivisto

    Raw data? Garbage in garbage out. How many gage repeatability studies have you seen for breathalyzers? How about Bias and Linearity? What master sample is used for calibration?Validated by whom? What about stability studies? No the number is not even close to validated with these machines, and peoples futures sometimes ride on it. I eat chemists like you for lunch.

  • jim

    now, you do not. when i use the word validation, i know what i am talking about. all i’m saying is, if you are going to use the alcotest breath tester, and the legislature agrees on the 2100-1 ration, and the instrument can be VALIDATED, then the number, in of itself, would be the number the legislature wants. if it is not v alidated, which it isnt, then the number shouldnt be used.

    kov- i am an expert analytical chemist with 20 years experience in validation of analytical instruments and methods. i have seen and done so much in so many companies. i know what i am talking about.

  • jim

    the point about raw data was and is, without the raw data, there is no way to begin to validate the instrument, to see if it works. it is literaly impossible.

  • koivisto

    Jim, you do have good arguments. I like the fact that we can agree to disagree. I’ve been in manufacturing for 28 years and have seen the impossible happen time and time again. Nothing is exact as much as we’d like it to be. Keep up the good fight my friend!

  • jim

    i do not think we are disagreeing. no one said the result needed to be exact down to 0,0000000… ect..

    the result from the alcotest however, should be defined as such a result that can be verified, from a validated instrument, with complete understanding what the result entails and is about. and if this is what we as a people want, to decided whethere or not a person is guilty of the perse 0,08 bac dui offense, then so be it. we have the right to run for office and try to change the statute.

    that being said, there is little hope in futher defining the statute, but more hope, although also very little, in even having the alcotrest, each instrument validated.

    vaildation is much more than running standards and xseeing a final result, as you alluded to.

    i have written almost 2000 pages on this subject concerning the alcotest, and have including things such as saftey validation in the factory, ISO, and audits of the factory.

    my basic poiint is, even if they had everything in place, did everything to comply with every other validation requirment, the instrument can still not be validated, because it does not provide raw data signal count.

  • jim

    in other words, do you not have to be as smart as justice scelia to know that when the nj statute says a person is guilty of dui if he has a blood alcohol content of 0,08%, with no other defining criteris, that the result must be correct.
    the only way to have a correct result, even if you wanted to use the alcotest to provide the result, we be to prove is through validation of each instrument on at least an annual basis.

    we can stop right here, because this is an impossible task, unless they rewrite the software to provide the raw data signal count.

    or– the legislature can rewrite the statute to say a bac result from the alcotest as it is today, without validation.

    the problem with this whole area is that the legislature dumped the problem on the courts, for the courts to be the bad guys, and do their dirty work, under the guise of legal language, hidden political deals, and hidden strupidity.

    in nj, the landmark case, state v chun, recently decided, concerning the scientific aceptability of the alcotest, although long proceedings, 3 years, was a mockery and a sham put upon the citizens of new jersey.

    i still think, that if some one wanted to argue the result from the alcotest can not comply with the statute as written, they would win in the us supreme court.

    i have a strong feeling justice scelia would agree that the number must be proven as fact, in order to comply with the statute as written.