Evaluation of New Evidential Breath Alcohol Instrumentation

Preliminary Results

R . G . Gullberg

Washington State Patrol

Breath Test Program

December 2007

Introduction

The purpose of this review was to evaluate the most recently

developed technology in forensic breath alcohol instrumentation .

The effort was largely investigational in order to learn what

features were now available and to determine what might be relevant

for the evidential breath test program in Washington State . Part of

the objective was also to consider revisions to Washington ' s breath

test program regarding both instrumentation and protocols. Since

Washington ' s current program was largely developed approximately

twenty years ago , many features offered by new instrumentation were

considered as possible useful additions to program development .

Four instruments were provided by vendors for evaluation .

These included : ( 1) Datamaster DMT - National Patent Analytical

Systems, Inc. , ( 2 Intoxilyzer 8000 - CMI Inc. , Intoximeter.

ECIR II - Intoximeters, Inc . , and ( 4 ) Alcotest 9510 - Drager Safety ,

Inc . All of the experimental work was performed at the Roanoke

office of the Breath Test Program in Seattle. None of the

instruments were re- calibrated in the Seattle Lab prior to beginning

the following analyses. The instruments were evaluated with the

calibration established by the manufacturer.

The following summary will present the principle results from

this preliminary evaluation .

Accuracy and Precision

Accuracy and precision were evaluated by performing n = 10

replicate simulator measurements on each of the four instruments at

eight concentrations. The solutions were obtained from the State

Toxicology Laboratory where reference values were determined by

headspace gas chromatography Figures 5 plot the bias

along with one coefficient of variation ( ) error bars at each of

the eight reference concentrations. Lines representing bias are

also shown for reference. Figures 6 through 9 are linearity plots

137

for each instrument . Each point represents the mean of n = 10

simulator replicates along with standard deviation ( SD ) error

bars . The linear regression equation is shown as well for each

instrument .

All instruments revealed very acceptable accuracy with bias

estimates being less than 4 . 7 % All instruments, except the

Alcotest 9510 , revealed acceptable precision with estimates being

less than 2 . 5 % . The Alcotest did not have heated simulator

tubing and showed some very large estimates at both the IR and EC

detectors. Figure 2 shows the bias plot for the Intoxilyzer 8000

when calibrated at the factor with the four point calibration

algorithm . Figure 3 shows the bias plots for the Intoxilyzer 8000

when calibrated with the one point calibration routine. The one

point calibration appears to yield very satisfactory results. The

plot in figure 5 shows the results on the Alcotest 9510 when the

simulator tubing was purged between runs of n = 10 . When doing so ,

the results were very acceptable. Heated simulator tubing was a

feature that Drager added at a later time.

Figure 10 plots the total error for each instrument determined

at each of the eight reference concentrations. Total error, as

shown by the equation in the plot , combines both bias and random

error ( CV) The Intoximeter ECIR performed the best overall by this

criteria .

Effect of Loose Simulator Jar

Figure shows the results of employing a tight and loose

simulator jar . set of n = 10 simulator measurements were performed

on the instrument initially with a tight simulator jar . Then the

jar was loosened slightly and another set of n = 10 measurements were

performed. instruments provided acceptable results with both

the tight and loose simulator jar except the Intoxilyzer 8000

shown in the figure, this instrument was very sensitive to the loose

jar condition . Both accuracy and precision suffered in this

condition . Very likely the vacuum sampling mechanism of the

Intoxilyzer 8000 was the cause of this . 11 also shows the

effect of shorter and simulator tubing in the Intoxilyzer

8000 . The effect of having a loose simulator jar would be an

important consideration where simulators are used as control

standards for every subject tested .

Day- to -Day Simulator Differences

Employing a new instrument in Washington State will likely

involve a change in the evidential test control standard

138

measurement. There will most likely be two simulator or gas

standards performed with each subject tested. In order to evaluate

the day - to - day variation, duplicate simulator tests were performed

on each instrument on different days . Figures 12 to 15 show these

results for each instrument The difference was determined by

subtracting the second result from the first Figures 12 and 14

( for the Datamaster DMT and Intoxilyzer 8000, respectively) a

tendency for the difference to be negative. This results from the

second result being slightly higher than the first This can result

in simulator tests where the tubing is not heated, as in the case of

the Intoxilyzer 8000 . The Datamaster DMT, on the other hand, had

heated simulator tubing of the differences along with the

actual results themselves, were very acceptable and would comply

with the current statutory requirements in Washington State.

Figure 16 shows another evaluation of day - to - day stability , in

this example for the Intoxilyzer 8000. Each point represents the

mean of n = 10 measurements of either the wet bath simulator or dry

gas standard . Also shown are error bars. Each run of

measurements were performed on different days . From the plot we see

that the wet bath simulator results tended to decrease over time at

a greater rate than the dry gas standard . This results from the

depletion effect which is more pronounced in use of the wet bath

simulator. We also see that the variability (magnitude of the 2SD

error bars) is generally greater with the simulator. This results

largely from the lack of heated simulator tubing in the Intoxilyzer

8000 . Figure 17 shows the same plot for the Datamaster DMT. Here

we see less depletion effect when compared to the Intoxilyzer 8000

in figure 16 . We also note that the variability is generally more

for the simulator than the dry gas standard. Figure 18 shows the

same results for the Intoximeter ECIR II. From this plot we see the

superior day - to - day precision compared to either the Intoxilyzer

8000 or the Datamaster DMT .

Carryover Effect

One important characteristic of an analytical system is that

each separate sample introduced for analysis must be considered

totally analytically independent. Each sample must be measured

without influence from prior samples introduced. This is not really

expected to be a problem with any of the instruments considered here

since they have purging routines along with zeroing requirements

before a subsequent sample can be introduced. In order the evaluate

the instruments for any carryover effect, alternatinghigh ( 0 .40

g /210L) and low ( 0 . 04 g /210L) were introduced. Some of the

instruments received the sample through the breath tube and some

through the simulator ports. Figures 19 to 22 reveal the results

for each instrument. Both the Datamaster DMT and the Drager 9510

produced very stable and reproducible results as seen in figures 19

and 20 . The Drager 9510 actually did not yield stable lower results

when the samples were introduced through the simulator ports . This

resulted from not having heated simulator tubing. Later , heated

simulator was installed which improved precision of the

lower results. The Intoxilyzer 8000 was very sensitive to the use

of non -heated simulator tubing. Figure 21 reveals the tendency for

the results to be increasing with sequential analysis until

measurement number when the tubing was cleared and the results

stabilized. The response of the Intoximeter ECIR in figure 22 was

very unusual Both the high and low sequential results tended

towards each other . These samples were also introduced through the

heated breath tube .

Wet Bath Simulators and Dry Gas Comparisons

One important feature of the instrument must be the ability to

employ either wet bath simulator solutions or dry gas as external

control standards . have such capability . Figure 23

shows the plot of means along with 2 SD error bars for both simulator

and dry gas standard results. Each point is the mean of n = 10

measurements . All sequential results were performed on different

days . From the figure we see that the dry gas standards yield

slightly better precision . This probably results from condensation

in the simulator headspace and tubing along with non - constant

temperature control. We also see a slight trend down for the

simulator results. Although the instrument employs recycle tubing

this represents some slight depletion over time . The total

measurements were well over 150 . Figure 24 shows the of means

with 2SD error bars for the Intoximeter ECIR . This plot also

reveals slightly better precision for the dry gas standards . We

also see a more significant depletion rate in the simulator results

compared to that of figure 23 for the Datamaster DMT. Figure 25

shows the results for the same analysis in the Intoxilyzer 8000 .

This reveals an even greater rate of depletion in the simulator

results . Again , the dry gas standard results reveal slightly better

precision than the simulator results. While figure 25 is a plot of

percent bias rather than actual mean concentrations , the trend and

relative precision are still observable . Figure 26 is another

illustration of similar analysis on the Intoxilyzer 8000 . The means

of n = 10 measurements for five sequential sets of runs are plotted

along with one coefficient of variation error bars . The three

configurations were simulator without clearing the tubing between

runs of n = 10, simulator with cleared tubing and dry gas standards .

The figure reveals that when the tubing is not cleared on the

instrument there is significant increase in variability . This

results from the simulator tubing not being heated as the instrument

140

was configured Clearing the tubing greatly improved precision ,

even better than the dry gas standard runs.

Interference Test Results

All of the instruments employed features designed to detect

the presence of an interfering substance . Figures 27 through 29

show the results of testing these features . Figure 27 shows the

results of first measuring the sample from a simulator containing on

ethanol. The simulator was arranged with four -way tubing so that

the same sample was received by all four instruments at the same

time. A common pressure source (portable compressor ) was employed

as the positive pressure sampling source . All four instruments

yielded very similar results for the ethanol only sample . Next, five

drops of acetone was added to the simulator to represent a low

concentration . The Drager 9510 and the Intoxilyzer 8000 yielded no

significant change in the result The Datamaster DMT and the

Datamaster CDM showed slightly increased measured results . Next , a

total of ten drops of acetone were in the simulator and finally a

total of fifteen drops were in the simulator . On these last two

occasions the Intoxilyzer 8000, Datamaster DMT and Datamaster CDM

all detected the presence of an interferents. In all three

conditions with added acetone , the Drager 9510 did not show any

significant change in the measured result . This results from the

use of 9 . 5 um in the IR region and fuel cell technology , both of

which do not respond to acetone . Figure 28 shows the results of

similar testing employing isopropanol. Again the four instruments

respond very comparably to the ethanol only simulator. Adding

200 of isopropanol slightly increased the result for the Drager

9510 instrument while it was detected as an interfering for the

other three instruments. Finally, adding a total of 400

isopropanol yielded an interfering result on the Drager 9510 . The

differing response of the Drager 9510 after adding 200

probably due to the use again of 9 . 5 um in the IR region along with

the fuel cell Finally, a separate set of tests were done on just

the Drager 9510 with the results shown in figure 29 . A set of ten

measurements were first performed using a simulator with ethanol

only . Then 200 of acetone was added and another measurement was

performed . additional acetone is added we see from the figure

than the IR result tends to increase while the fuel cell result

tends to decline. Eventually at 600 and 800 acetone, one of

the measurements for each run yielded interference on one of the

analytical methods. At 1000 of acetone both methods yielded

interference 200 of acetone, the vapor concentration is

approximately 926 ug/ is an extreme concentration only

likely to be present in uncontrolled diabetics .

141

Additional Analyses

There were a few additional analyses that were done with

different instruments. One evaluation done on the Intoxilyzer 8000

involved the calibration procedure. The instrument as initially

received was designed to be re - calibrated at four different

concentrations . This process would take about 30 minutes to

accomplish . This did not seem an acceptable process so the

manufacturer provided another instrument that was designed to employ

one-point calibration 30 and 31 show the results of this

analysis . Figure 30 plots the means and one coefficient of

variation error bars as percent bias for n = 10 measurements

throughout the concentration range where the instrument had received

the manufacturer four point calibration We see that all results

remained remain within 5 % of the reference value . 31 plots

the same analysis for the instrument calibrated with one point

calibration . Again we see that all results remain within 5 % of the

reference value. This supports the sufficiency of one -point

calibration in the Intoxilyzer 8000 which would be a more acceptable

procedure as part of an annual re- calibration program .

Another evaluation completed was the comparison of a different

simulator . A Guth model 590 simulator was compared to the Guth

model 2100 simulator. A total of n 20 measurements were performed

on each simulator using the same solution batch and the same

Intoxilyzer 8000 instrument Figure 32 shows the results While

both simulators yielded very similar precision estimates, the Guth

590 simulator results were slightly higher . A runs test on both

sets of data showed random results with no trend.

Desirable Instrument Features

Part of this evaluation was to learn of the existing technology

and identify desirable features for the Breath Test Program in

Washington State. The following is a partial list of desirable

features for the instrument:

1 . The ability employ either wet bath simulators or dry gas as

control standard devices.

2 . The ability to employ operator and technician card readers

using 2D bar code technology While all of the manufacturers

claimed to offer this feature, none actually fully demonstrated it

on their instrument.

3 . The ability of the instrument to provide a page Quality

Assurance Procedure printout of the final results following a QAP

procedure .

142

. Windows based operating system technology with touch screen

capability .

5 . The ability to store and produce when requested , the breath

alcohol concentration and breath flow curves .

6 . The ability to recall a complete breath test printout record

from a prior date when needed. This can be provided at the host

computer end.

7 . The ability to monitor and record the simulator temperature on

the actual printout document when a simulator control standard is

employed

8 . The ability during a QAP to monitor and record the accuracy ,

precision and test for outliers .

9 . The ability during a QAP to monitor and record critical voltage

settings in the instrument.

10 . Part of the contract should state that the State of Washington

will not have any ownership of or any access to the source code or

any copyrights thereof within the instrument .

Conclusions and Opinions

All of the instruments evaluated represent the technology ,

features and capabilities for state -of- the-art forensic breath test

instruments that are fully fit - for- purpose. After working with the

instruments it is my opinion that the Datamaster DMT or the Drager

9510 would be the instruments best suited to meet the needs for the

State of Washington these two instruments, I believe that the

Datamaster DMT is best suited for immediate use and implementation .

There are, however , some features still in need of revision on the

Datamaster DMT ( i . e . , the gas standard configuration ) . While the

Drager 9510 has many desirable features it is not yet ready for

implementation in my opinion . If time and program needs were to

allow it , I would like to continue the evaluation of these two

instruments for the next year . At that time both technologies would

be advanced and better suited for a more informed decision.

R G Gullberg

12 / 13/ 2007

143

Figure 1

PercentBias and One CoefficientofVariation Error Bars ForDatamaster DMT

PercentBias

8

- -

0 0 0 . 1 0 . 4 0 . 5

144

0 .2 0 . 3

Reference Value( g/ 210L )

Figure 2

Percent Bias and One Coefficient ofVariation Error Bars For Intoxilyzer 8000

PercentBias

- -

- 8

0 .0 0. 1 0 . 4 0 . 50.2 0.3

Reference Value (g /210L )145

Figure 3

PercentBias and One Coefficientof Variation Error Bars For Intoxilyzer 8000

Using One PointCalibration

PercentBias

87One PointCalibration

n = 10 at each point

6

E - - - - - -- -

- -

- 8

0. 1 0.2 0 . 40 3

Reference Value (g/210L )

Figure 4

Percent Bias and One Coefficient of Variation Error Bars For Intoximeter ECIR

PercentBias

- - - - - - - - - - - - - - - - - - - - - - -

- - - -

0 . 0 0. 10 . 1 0. 2 0.3 0.40 . 4 0.50 . 5

147

0 . 2 0 . 3

Reference Value ( g /210L )

Figure 5Percent Bias and One Coefficient of Variation Error Bars For Alcotest 9510

PercentBias

n = 10 at each point2 27 /2007

- - -

-- - - -

- 6

0 .00 0. 2 0.40. 3

Reference Value( g/210L)148

Figure 6

Linearity PlotFor Datamaster DMT

Measured Concentration (g / 210L )

0 . 5

0 . 4

0 . 3

Y = 0. 9873 X + 0 . 000048

0 .

0 . 0

0 . 0 0 : 1 0 . 2 0 . 3 0 . 4 0 . 5

149 Reference Value ( g /210L)

Figure 7Linearity PlotForIntoxilyzer 8000

MeasuredConcentration ( g /210L )

0 . 5

0 . 4

Y = 0 . 9806 X + 0.00089

0 . 0 +0 . 1 0 . 2 0 . 3 0 . 4 05

150

Reference Value ( g/210L )

Figure 8

Linearity Plot For Intoximeter ECIR

Measured Concentration ( g/ 210L )

0 .5

0 . 4

0 . 3

Y = 0 . 9810 X + 0. 00428

0 .0

0 . 0 0 . 1 0. 2 0 . 3 0 . 4 0 . 5

ReferenceValue ( g/210L )

Figure 9PlotFor Alcotest 9510

Measured Concentration ( g / 210L )

0 . 5

0 . 4

0 . 3

0 .

Y = 0 . 992 X + 0 . 00026

0 . 1

0 . 0

0.00 . 0 0 1 0 . 21 0 . 3 0. 4 0 . 5

152

Reference Value (g/210L )

Figure 10

Plots of TotalError For Each InstrumentDetermined atEightConcentrations

Percent TotalError

Datamaster DMT

.... .... . Intoxilyzer 8000

- - - Intoximeter ECIR

Datamaster CDM

= bias + 2 CV-

4

---

0 . 0 0153

0 . 2 0. 3

Reference Concentration (g/210L)

Figure 11

Percent Bias and One CV Error Bars Comparing Tight and Loose Simulator

Percent Bias

2 .5 inch tubing

- 20Loose SimulatorTightSimulator

Seven inch tubing

-30

-40

Intoxilyzer 8000 DatamasterDMT Intoximeter ECIR Alcotest9510

Instrumentand SimulatorArrangement154

Figure 12PlotofDifferences Between Duplicate Simulator Tests Performed on Different

Days For the Datamaster DMT

Difference Sim 1-Sim ( g /210L)

0 . 010DatamasterDMT

Day- to -day precision: 0 .0015 g/ 210L

0.005

0 . 000

- . 005

- 0 .010

3020

Test Sequence155

Figure 13

Plot of Differences Between Duplicate Simulator Tests Performed on Different

Days For the Intoximeter ECIR

Difference ( g / 210L )

0.Intoximeter ECIR

Day to -day precision : 0 .0008 g/ 210L

0 .005

0.000

- 0 . 005

-0 . 010

10 20 30156

Sequence

Figure 14Plot of Differences Between Duplicate Simulator Tests Performed on Different

Days Forthe Intoxilyzer8000

Difference( g/ 210L)

0. Intoxilyzer 8000

Day- to -day precision : 0 .0028 g / 210L

0 .005

0 .000

- 0 . 005

- 0 .010 +

10 30157

20

TestSequence

Figure 15Plot of Differences Between Duplicate Simulator Tests Performed on Different

For the Drager 9510 IR Detector

Difference( g/210L)

0 . 010

Drager 9510 IR result

Day- to -day precision : 0.0008 g/210L

0 . 005

0 .000

- 0 .005

-0.01010

158

Test Sequence

Figure 16PlotofMean Along With 2SD Error Bars For Simulator and Dry Gas Results on

Intoxilyzer 8000

Percent Bias

10n = 10 at each pointGas Standard Reference : 0 . 100 g /210L

Simulator Reference : 0 .0827 g / 210L

5 + - - - - - - - -

Simulator results

- 5 -!

Gas standard results

1015

Day159

Figure 17

Plot of Mean Along With 2SD Error Bars For Simulator and Dry Gas Results onDatamaster DMT

Standard Results (g/ 210L )

0. 11Gas standard results

n = 10 at each point

Gas Standard Reference: 0 100 210L

Simulator Reference: 0 .0820 g /210L

0 . 100 g /210L

. 10

0 . 10

. 09

Ref. 0. 0820 g /210L

0.08

Simulator results

0 5 10 1515160

Day

Figure 18PlotofMean Along With 2SD Error Bars For Simulator and Dry Gas Results on

Intoximeter ECIR

StandardResults ( g/ 210L )

0. 100n = 10 at each point

Simulator Reference: 0. 0814 g/ 210L

Gas Standard Reference: 0 .0820 g /210L

0 . 095

Simulator results

0. 090

0.085Dry Gas Reference = 0. 0820 g/210L

HOH

0.080Simulator Reference = 0 .0814 g /210L

Gas standard results

0 .075

0 .070

10 15

Day161

Figure 19Sequential Plotof Alternating High and Low Alcohol Concentrations

Alcohol Concentration ( g / 210L)

0 .5 DatamasterDMT

Samples through simulatorports

mean = 0 .3940 g /210LSD = 0 . 0039 g /210L

mean = 0 .0442 g /210L

SD = 0 .0017 g/210L

8 10 126SequentialTest

162

Figure 20

Sequential Plot of Alternating High and Low AlcoholConcentrations

Alcohol Concentration (g /210L )

0 .57

Alcotest 9510 IR Resultthrough breath tube

0. 4

mean = 0 .4339 g /210L

SD = 0 .0021 g/ 210L

mean = 0 . 0383 g /210L

SD = 0 .0007g/ 210L

0 . 1

163Sequential Test

Figure 21

Sequential Plotof Alternating High and Low Alcohol Concentrations

Alcohol Concentration ( g /210L ) Intoxilyzer 8000

Samples through simulator ports0.5

0 . 4

mean = 0 .4086 g /210L

SD = 0 .0062 g / 210L

Cleared tubingmean = 0 .0537 g /210L

SD = 0 . 0088g/210L

0 . 1

0. 0 +

164 Sequential Test

Figure 22Sequential Plot of Alternating High and Low Alcohol Concentrations

AlcoholConcentration (g/210L) Intoximeter ECIR

samples through breath tube

mean = 0 .3561 g / 210LSD = 0 .024 g /210LCV = 6 . 7 %

mean = 0 .0774 g/210LSD = 0 .024g

CV = 31 %

0 . 0

4 6 10 12165

Sequential Test

Figure 23

PlotofMeans and ErrorBars ForBoth Simulator and Dry Gas Results

Standard Results ( g/210L )0 . 117

Gas standard results

Datamaster DMT

n = 10 at each point

Gas Standard Reference: 0 . 100 g /210LSimulator Reference : 0 .0820 g /210L

I I Ref. = 0. 100 g/210L

0. 10

0 .09

Ref. = 0 .0820 g /210L

0 . 08

Simulator results

0. 0710 15

Day

Figure 24

PlotofMeans and Error Bars For Both Simulator and Dry Gas Results

Standard Results ( g /210L )0 . 100

Intoximeter ECIRn = 10 ateach point

Simulator Reference : 0 .0814 g / 210L

Gas Standard Reference: 0. 0820 g / 210L0 .095

Simulator results0. 090

0.085Dry GasReference = 0 . 0820 g /210L

0 .080Simulator Reference = 0.0814 g / 210L

Gas standard results

0.075

0. 070 +

10 15

167 Day

Figure 25

PlotofMeans and 2SD Error Bars For Both Simulator and Dry Gas Results

PercentBias

Intoxilyzer 8000

n = 10 at each point

Gas Standard Reference: 0. 100 g /210L

Simulator Reference: 0 .0827 g/210L

5

Simulator results

101

:

-5 + - - - - -

Gas standard results

1010 20 2515

Day168

Figure 26Percent Bias and One Coefficient ofVariation Error Bars Comparing

Simulator Tubing Configurations and Dry Gas Standards on the Intoxilyzer 8000

Percent Bias

- -

n = 10 ateach point

3 /19/ 2007

- 2

Simulator Without Cleared TubingDry Gas Standard

Simulator With Cleared TuabingA

169Testing Sequence

Figure 27Single MeasurementResults Where Increasing Amounts of Acetone

Are Added to a Simulator Initially Containing Only Ethanol

Result ( g /210L ).

Ethanolonly5 dropsacetone added

10 dropsacetone added

15 drops acetoneadded Interference Detected

0 . 12

0. 10

0 .08

0. 06Drager 9510 Intoxilyzer 8000 Datamaster Datamaster CDM

170Instrument

Figure 28

Single MeasurementResults Where Increasing Amounts of

Are Added to a Simulator Initially Containing Only Ethanol

Result ( g/ 210L )

0 . 14

Ethanol only

Isopropanol

400ul Isopropanol

Interference Detected0.

0 . 10

0 .08

0 . 06

Drager 9510 Intoxilyzer8000 DatamasterDMT DatamasterCDM

Instrument

171

Figure 29Results Where Increasing Amounts of Acetone Are Added to a Simulator InitiallyContaining Only EthanolFor the Drager 9510 to Compare the Response of the

IR and EC

Result (g / 210L )All samples introduced through breath tube

200ul acetoneyields approx. 926 ug/L vapor concentration0. 117

Interference DetectedIR Result

EC Result

0. 10

These two series had one

interferentand one quantified result0 . 09

meanof ten results

alongwith bars

0 .08

0.07

0. 06EthanolOnly

Acetone Acetone Acetone Acetone

1000uL

Acetone

172

Simulator Solution

Figure 30

PlotofBias and One Coefficientof Variation Error Bars ForFour-Point

Factory Calibration on the Intoxilyzer 9000

PercentBias

FactoryCalibrationn = 10 at each point

- - - - - - -

-

0.0 0 1 0 . 4 0 . 50. 2

Reference Value (g /210L)173

Figure 31Plotof Bias and One Coefficient ofVariation Error Bars ForFour-Point

Factory Calibration on the Intoxilyzer 9000

Percent BiasOne PointCalibration

n = 10 at each point

-

-

0 . 0 0. 1 0. 2 0 . 4 0 . 50 3

Reference Value ( g/210L )174

Figure 32SequentialResults From Two DifferentSimulators Tested on the Intoxilyzer8000

Concentration ( g/ 210L )

0.090

0 . 088Guth 590 Simulatormean 0843 g /210LSD = 0 0005 g /210L

Runs test: Z 0 .49 p = 0 .62

0. 086

0 . 084 bo boooo

0 . 082 + - - - - - - - -

Reference Value 0.0820 g /210L

0 .080Guth 2100 Simulator

mean = 0 .0830 g /210LSD = 0.0003 g /

Runstest: Z = 0 p = 1. 0

0 .078

10 15 20175

SequentialTest

Proposed Protocol For New Instrument Evaluation

Washington State Patrol

Breath Test Program

November 2006

A . Introduction

1. Purposea to evaluate new breath alcohol instrumentation and

technology for establishing type approval and possibleimplementation within Washington State

b . to develop a forensically sound breath alcohol test

program in Washington State centered around an

instrument that has been found to be fit - for -purpose

to develop documented evidence having a high degree of

assurance that a particular instrument type with perform

according to pre - determined specifications and

analyticalattributes

to provide documented for the validity and

subsequent selection of a particular instrument type

that has been shown to fit - for - purpose with a high

degree of confidence

2 . Manufacturers will be asked to come to Seattle and present

an overview of their instrument and leave one for evaluation

Manufacturers will be provided with a copy of this proposed

evaluation protocolThe evaluation results will be available to all interested

individuals5 . The evaluation is expected to take approximately six months

to completeThe instruments will be tested together as much as possible

in order to minimize any differences due to experimental

design , procedure or the use of other equipment .

A Datamaster CDM will also be included in the testing

protocol.

3 .

7 .

Desirable Instrumental Features

1 .

3.

Infrared and / or Electrochemical analytical method

Data collectionRe - cycle tubing for simulator analysis

Ability to continue using Guth 2100 digital simulators

a minimize length of simulator tubing

b ability to heat simulator tubing

c future purchase of new instruments must include a Guth

176