Facts
This matter is before the court on the defendant’s motion to suppress evidence. The defendant, Vincent Workman, Jr., was arrested for a violation of R.C. 4511.19 and took a breath test on May 7,1995 on a BAC Datamaster. On May 3, 1995, four days before defendant’s test, the BAC Datamaster was checked for calibration and registered a test result of 0.102. The machine was checked again for calibration on May 10, 1995, three days after defendant’s arrest and registered a test result of 0.100. Both of the calibration checks were performed with calibration solution from RepCo. Marketing, Inc.’s (“RepCo.”) Batch No. 93002, bottle 621.
The defendant is charged under R.C. 4511.19(A)(3). Pursuant to this section, the threshold test for admissibility of alcohol test results is (1) whether the bodily substance was withdrawn within two hours of the alleged violation, (2) whether the bodily substance was analyzed in accordance with methods approved by the Director of Health, and (3) whether the test was analyzed by an individual possessing a valid permit issued by the Director of Health pursuant to R.C. 3701.143.
The issue presently before the court is whether the bodily substance was analyzed in accordance with methods approved by the Director of the Ohio *29Department of Health. Specifically, the motion to suppress the breath test is based on defendant’s contention that the calibration batch solution (1) was not properly approved by the Director of Health, (2) was used more than nine months after the manufacturer’s expiration date, and (3) should not have been approved by the Ohio Department of Health.
This hearing was the first time this court has thoroughly examined the process by which batch solutions are certified by the Ohio Department of Health. Attorneys for the prosecution and the defense adduced exhaustive evidence upon the issues herein.
1993 Calibration Batch Solution Approval Process
Pursuant to the regulations contained in Ohio Adm.Code 3701-53-04, the- Ohio Department of Health (“ODH”) has the duty to issue certificates of approval for calibration solutions. The court has heard ten days of testimony by the Ohio Department of Health and other experts regarding the approval process which leads to the issuance of these certificates.
As required by the regulations, breath-testing equipment must be checked for calibration. The calibration process involves the use of an alcohol and water “Calibration Solution,” which is heated in a simulator jar to 34° C ± 2° C to simulate the temperature of human breath. The gas generated by the heated solution is then pumped into the breath-testing device to simulate an actual breath test. The result obtained when the test is complete must be within ± .005 of the optimum test solution reading. Ohio Adm.Code 3701-53-04(A)(l). The calibration process in essence verifies that a breath-testing instrument is in proper working condition.
The Ohio Department of Health approves the process for calibrating solutions. ODH certified RepCo. Batch No. 93002 on December 8, 1993. The purpose of approval is for quality assurance and not to establish an independent concentration of calibration solution. (Testimony of Leonard Porter, pg. 46, lines 8-12.) There was no written procedure or protocol outlining the approval process in 1993. The earliest written protocol, as was stipulated by the parties, was not written until September 1994. Some of the testimony regarding the process appears to be in conflict with those later written standards.1 There is no evidence of a consistent mandated procedure for certifying batch solutions.
*30The calibration process in 1993 began with manufacturers of calibration solutions preparing solutions for sale to local entities. The manufacturers would submit three to four sealed bottles from a batch to the Ohio Department of Health for review, analysis, and approval. These randomly selected bottles would come from the beginning, middle, and end of the batch. The sealed solutions were received by the organic chemistry lab, where a series of tests and analyses were performed with a gas chromatograph. The laboratory analyst would forward the lab results on a summary sheet to Leonard Porter, Chief of the Alcohol Testing program. Porter would review the results and determine if the analytical findings of the laboratory were within a plus or minus five thousandths in breath alcohol of the target value as specified in the information submitted by the manufacturer.2 Leonard Porter would then determine whether or not the batch met his criteria and his administrative assistant or secretary would prepare the approval certificate. At no point in the process did Dr. Peter Somani, Director of the Ohio Department of Health, actually review the summary or sign the approval certificate. The administrative assistant or secretary would fill in the blanks on a certificate that already had a photocopy of the Director of Health’s signature. (Defendant’s Exhibit C.) Porter would sign the certificate and the certificate would be sent back to the manufacturer. (Testimony of Leonard Porter, pg. 32.) The manufacturer then provided that approval certificate with other purchasing inserts as they became necessary to the local end users.
1993 Approval of RepCo. Batch No. 93002
On November 15, 1993, the Ohio Department of Health received four randomly selected bottles of Certified Calibration Solution, Lot No. 93002, for analysis from RepCo. Marketing, Inc. The certificate of analysis provided by the manufacturer, RepCo., stated that the solution was analyzed by an independent laboratory utilizing a gas chromatograph and was found to contain 0.1217 gms/dl wt./vol. ethyl alcohol. When used in an alcohol-breath test instrument it should produce a reading of .10 percent BAG.3 The certificate of analysis also listed an expiration date of July 31, 1994 at 11:59 p.m. (See Defendant’s Exhibit B-25.)
The independent laboratory which tested Batch No. 93002 was Northwest Toxicology, Inc. It performed its analysis on two sample bottles provided by RepCo. According to RepCo.’s President, Cecil Garner, Northwest Toxicology was supposed to analyze five samples from each of the two bottles. Northwest performed an analysis on August 30, 1993 and found that the solution contained 0.1217 gms/dl wt./vol. ethyl alcohol. No other evidence was introduced regarding *31the laboratory procedures for Northwest Toxicology. (Defendant’s Exhibit B-25.)
After receipt of the solutions by ODH, RepCo. Batch No. 93002 and Guth Batch No. 93390 were assayed together on a gas chromatograph4 on December 3, 1993 and again on December 6, 1993.5 The summary data sheet for Batch No. 93002 was prepared by Andrew Smith (an employee of ODH) on December 7, 1993 and forwarded to Porter. It lists a “grand mean” of 1.2236 using five standard calibration reference concentrations. (Defendant’s Exhibit B-25.) The grand mean represents a mean of the determined bottle concentrations. Raw data was not included with the summary sheet. The summary data also does not reference whether the data came from the December 3 or the December 6 assay. However, the evidence is substantial that the batch failed the December 3 assay.
Porter determined that the test met his criterion. This criterion was not written, and there has been conflicting testimony as to what it actually was. For example, Porter testified that he would ascertain that the solution was within ± 5% of the manufacturer’s target value. Other ODH employees testified that a 1994 written protocol was actually in use during 1993. (Testimony of Steve Wagner, pg. 75 and see Defendant’s Exhibit DD1.) According to that protocol, the solution must be within ±2.5% of the manufacturer’s target value.
On December 8, 1993, the Ohio Department of Health issued the approval certificate for Batch No. 93002. The certificate stated in part: “This calibration solution contains 1.21 mg/mL ethyl alcohol in distilled water. When used according to calibration checklists, it will produce a reading of 0.100 g/210L ± .005 in the approved breath testing instrument, when that instrument is in proper working condition. This solution shall not be used more than three months after its date of first use.” (Defendant’s Exhibit B-25.) The concentration determined by ODH’s quality assurance tests, 1.2236, is not listed on the certificate. The certificate does not list the expiration date established by the manufacturer, but pursuant to the regulations the solution shall not be used more than three months after its date of first use. Ohio Adm.Code 3701-53-04(A)(l).
Subsequent Actions by the Ohio Department of Health
Subsequent to the approval of RepCo. Batch No. 93002, a number of events have occurred within the Department of Health. The first significant event *32occurred in September 1994, when a protocol titled “Quantitative Determination of Ethanol in Alcohol Calibration Solutions by Flame-Ionization Gas Chromatography” was put into effect. See Defendant’s Exhibit DD1. Based on conversations with laboratory employees in 1996, Ferguson and Steve Wagner believed that a similar protocol was being followed in 1993. There is no other evidence that indicates whether or not the protocol was actually followed or consistently followed. Yet, this 1994 protocol is still important for several reasons. First, under Section 7.1, Sample Handling, only solutions prepared by Guth Laboratories and Miami Valley Regional Crime Laboratory are listed. RepCo. is not listed as an approved source. Second, Section 7.1 does not specifically address approval or rejection criteria. The closest language pertaining to approval is contained in Section 9.2.4. It states that “the mean recovery should be within 97.5-102.5%.” It does not state that a batch is accepted or rejected on this criterion. Third, Section 11.1 .states: “prepare a five point calibration curve6 using either an internal standard or an external standard procedure.” (Footnote added.) The external and internal standards are two different approaches7 to determining what the concentration is in a solution based upon a set of standard solutions. (Testimony of Dr. Craig Sutheimer, pp. 32-33.)
The next event occurred on January 29,1996, when Leonard Porter of the Ohio Department of Health Laboratory issued a memorandum that proposed different standards for the information contained on the approval certificates issued by ODH. See Defendant’s Exhibit B-29. This memorandum created confusion among law enforcement agencies and the community at large. On February 22, 1996, Dr. Peter Somani, Director of the Ohio Department of Health, issued a memorandum to courts, prosecutors, and law enforcement, which disavowed the earlier Porter memo. See Defendant’s Exhibit B-l. It also restated the existing policy that the approval certificates would continue to be based on the manufacturer’s stated values. Sometime in late January or early February 1996, Dr. Somani spoke with Dr. Craig Sutheimer, Chief Toxicologist and Director of the Toxicology Laboratories for the Cuyahoga County Coroner’s Office, and requested that he review the ODH lab. After his preliminary review of the lab, Dr. Sutheimer suggested a complete data audit. Dr. Somani agreed and asked Dr. Sutheimer to perform a complete audit to ensure that the Ohio Department of *33Health was correct in approving the nineteen solutions originally approved between January 1993 and January 29, 1996. (Testimony of Dr. Craig Sutheimer, pg. 14.)
Dr. Sutheimer and James Ferguson, Chief Toxicologist and Director of the Forensic Toxicology Laboratory for the Franklin County Coroner’s Office, then conducted a data audit. In late February, Dr. Sutheimer and Ferguson extracted raw data from the original chromatograms generated for the batches. Steve Wagner, Interim Chief of Environmental and Health Toxicology for the Ohio Department of Health, created a model to analyze the results. Using the raw chromatograms from the original assays, Dr. Sutheimer and Ferguson extracted the areas from each chromatogram for each analyte. Dr. Sutheimer and Ferguson would each extract the data from the chromatograms for half of the batches and then switch, and the other person would double-check the results of the first person. (Testimony of James Ferguson, pp. 17-18.) Steve Wagner would then enter the data into the databases. The summary data produced would then be analyzed. At the end of the audit, Ferguson and Dr. Sutheimer both submitted letters and a summary report to Dr. Somani. (See Defendant’s Exhibit H and EE1.) The summary report contained only their final conclusions. It did not contain worksheets, raw data, or comments. As will be discussed below, the worksheets contained information suggesting that the batch was rejected. The summary report, however, gave only summaries and stated the conclusion that the Ohio Department of Health was correct in approving the nineteen batch solutions. The significance of this evidence is that it appears that nothing except the sanitized final conclusions were intended for public dissemination. (Testimony of James Ferguson, pg. 59.)
1996 Data Audit of RepCo. Batch No. 93002 by Dr. Sutheimer and James Ferguson
When RepCo. Batch No. 93002 was reevaluated in 1996, Dr. Sutheimer determined that the December 3, 1993 assay was so flawed that it could not be used as a basis to accept or reject Batch No. 93002. See Defendant’s Exhibit I, pg. X9. The December 6, 1993 assay also was flawed in many respects. First, Ferguson testified that the assay “wasn’t pretty and there was more slop in it than he would like.” (Testimony of James Ferguson, pg. 37, line 16.) Second, there were split peaks and tailing problems that resulted in several select individual sample results being thrown out of the analysis. According to the experts, split peaks can occur as a result of inaccurate injection; the results, therefore, are not necessarily indicative of a problem with the solution, but could indicate an equipment or operator problem. Yet, the problems caused by split peaks can make the data obtained from the assay unreliable. Third, Ferguson *34also indicated' that the assay contained some very poor chromatography. He stated in his report:
“[S]erious split peak problems, # 1 calibration verification standard not found; RSI coincides with reported cal. ver std # 2. RS2 = cal ver Std # 3 etc. Original calculations add split peaks areas to area of peak of interest, v poor chromatography.” (Defendant Exhibit I, pg X8.)
The grand mean for Batch No. 93002 was 1.319728 or 108.4 percent8 of the target value. Additionally, the worksheets noted both in handwriting, “batch rejected,” and on the top of the worksheet as part of the printout that assay failed, “REJECTED.” (Defendant’s Exhibit I, pp. X5-X6.)
The next step was to do an analysis based on an internal standard. The internal standard would measure the ratio of a marker (n-propanol) put in along with the target analyte. The internal standard, however, was missing. (Defendant’s Exhibit I, pg. X7.) It appears then that Batch No. 93002 should have ultimately failed. However, RepCo. Batch No. 93002 did not fail in the final summary reported to Dr. Somani. The reasons for this are of considerable concern.
The data for Batch No. 93002 was then used and compared to a calibration curve constructed with calibration reference concentrations in the last part of the assay, which corresponded to Guth9 Batch No. 93390. Using a standard five-point calibration curve, the correlation coefficient used to determine the reliability of the data would have been only .990. (Testimony of Dr. Robert J. Bellotto, Jr.) This .990 would have warranted rejection under the standards used by Dr. Sutheimer and Ferguson for their 1996 audit. (Defendant’s Exhibit H, pg. 4).10 The standard used in 1993 is unclear. Yet, the 1994 and 1996 written protocols indicate that a five-point calibration curve is to be constructed. (Defendant’s Exhibit DD1, Section 11.1; Defendant’s Exhibit K, Section 11.1.) Dr. Sutheimer explained that they looked at the R2, and it was lower than they would accept. They then tried to determine if there was a reason for it. They felt that the standard 1183.5 was too different from the other calibrators and therefore simply dropped it. They did not perform any statistical studies to determine if it was an outlier 11; although it was possible to do the study. (Testimony of Dr. Craig *35Sutheimer, pp. 76-78.) The data was then fitted to a four-point calibration curve constructed from the reference standards run with the Guth Batch No. 93390.
Even using only four calibrators, the batch was approved with one of the method validation samples still outside ± 5%. Method validation sample “QC1” was calculated at 94.7%. Additionally, this “94.7%” was later determined in the course of the hearing to be a miscalculation. Instead of 94.7% it should have been 93.4%. This is also clearly outside either the ± 2.5% or ± 5% ranges. The next step taken by Dr. Sutheimer and Ferguson was to drop this data point because it was assumed to be an outlier and suffered from split peaks. Other data containing similar split peaks were not dropped. So based on 40%12 of the calibrators and dropping QC1 from the assay the batch finally was passed.13 A brief summary of the outcomes based on the 1993 chromatograms is as follows:
*361993 12/3/93 Assay. REJECTED Based on the assumption that it failed originally since a second assay was performed.
1993 12/6/93 Assay14. APPROVED Grand mean of 1.2236. However, later questioned by Director of Health and subjected to a re-evaluation at his request.
1996 12/3/93 Assay15 REJECTED
1996 12/6/93 Assay16 w/o Internal. Using a five point calibration curve constructed with . reference standards corresponding to RepCo 93002. REJECTED Grand mean of 1.319728. Mean Calibration Verification Recovery 124.6%. Mean of method validation samples recovery 108.7%.
1996 12/6/93 Assay17 w/o internal. Using a five-point calibration curve constructed with reference standards corresponding to Guth 93390 REJECTED Correlation coefficient 99020 which is less than required .999.
1996 12/6/93 Assay18 w/o internal. Using a four point calibration curve constructed with calibration reference standards corresponding to Guth 93390 QCI 93.4%.
1996 12/6/93 Assay19 w/o Internal. Using a four point calibration curve constructed with calibration reference standard' corresponding to Guth 93390 and dropping QC1. Grand Mean 1.195209.
*37 Application of the Law
The legislature by enacting R.C. 4511.19(D) has made the threshold requirement for admissibility of alcohol test results dependent on substantial compliance with the applicable regulations. Therefore, to substantially comply with Ohio Adm.Code 3701-53-04, the state need only show that the solution in question was approved by the Director of Health. Once the state introduced into evidence a certificate of approval for Repco. Batch No. 93002 (State’s Exhibit 1), the burden shifted to the defendant to show prejudice by anything less than complete technical compliance with the challenged regulation. State v. Plummer (1986), 22 Ohio St.3d 292, 22 OBR 461,490 N.E.2d 902.
Technical Compliance with Regulations
The defendant has attempted to show that because Dr. Peter Somani did not personally review or approve RepCo. Batch No. 93002 there was not technical compliance with the regulations. This argument is not well taken. The legislature has given a broad power to the Director of Health to determine methods for alcohol testing. R.C. 3701.143 states:
“For purposes of section 4511.19 of the Revised Code, the director of health shall determine, or cause to be determined, techniques or methods for chemically analyzing a person’s blood, urine, breath, or other bodily substance in order to ascertain the amount of alcohol, a drug of abuse, or alcohol and a drug of abuse in the person’s blood, urine, breath, or other bodily substance. The director shall approve satisfactory techniques or methods, ascertain the qualifications of individuals to conduct such analyses, and issue permits to qualified persons authorizing them to perform such analyses. Such permits shall be subject to termination or revocation at the discretion of the director.” (Emphasis added.)
Pursuant to this grant of power from the legislature, the Director of Health, not the legislature, promulgated administrative regulations pertaining to alcohol testing, approval, and permits. The regulations are contained in the Ohio Adm.Code Chapter 3701-53.
Ohio Adm.Code 3701-53-04 specifically addresses calibration. It states in part:
“(A) Approved evidential breath testing instruments shall be checked for calibration no less frequently than once every seven days by a senior operator using a solution of ethyl alcohol approved by the director of health and using the *38calibration checklist for the instrument being cheeked, as set forth in appendices A to G to this rule.” (Emphasis added.)
The legislature has also given the Director of Health comprehensive powers to administer the Department of Health in R.C. 3701.04(A), which states:
“(A) The director of health shall:
* *
“(2) Provide such methods of administration, appoint such personnel, make such reports, and take such other action as may be necessary to comply with the requirements of the federal act and the regulations thereunder[.]”
Thus, the Director of Health may delegate statutory responsibility to personnel under his control. Notwithstanding the fact that approval of calibration solutions is contained in the administrative rules promulgated by the Director of Health, it is a well-recognized principle of administrative law that “subdelegation of authority, impliedly or expressly, exists — and must exist to some degree.” Bell v. Lawrence Cty. Gen. Hosp. Bd. of Trustees (1973), 34 Ohio St.2d 70, 74, 63 O.O.2d 115, 117, 296 N.E.2d 276, 278. Dr. Somani and Leonard Porter both testified that the power to approve calibration batch solutions was delegated to Porter some twenty years ago. (Testimony of Leonard Porter, pg. 45.) In 1992 when Dr. Somani became the director, he continued this delegation to Porter. Therefore, the approval issued by Porter and a photocopy of Dr. Somani’s signature on the certificate was in technical compliance with the law and the regulations.
The defendant herein also argues that the Director of Health abused his discretion on a variety of issues. A decision by the Director of Health in the determination of matters entrusted to him by the legislature by virtue of R.C. 3701.143 should not be disturbed by the court in the absence of fraud, collusion, bad faith, or gross abuse of discretion. “Gross abuse of discretion” can best be defined as an “unreasonable, arbitrary or unconscionable attitude.” State ex rel. Fogle v. Steiner (1995), 74 Ohio St.3d 158, 656 N.E.2d 1288. There is a difference between actions which constitute an abuse of discretion and actions which are against the weight of evidence. No matter how persuasive the defendant’s arguments, this court cannot substitute its judgment for that of the Director of Health unless defendant shows that the director’s actions are either not in accordance with the law or are not supported by substantial evidence. 2 Ohio Jurisprudence 3d (1977), Administrative Law, Section 176, at 387; Sterling Drug, Inc. v. Wickham (1980), 63 Ohio St.2d 16, 17 O.O.3d 10, 406 N.E.2d 1363.
Expiration Dates
The defendant attacked the approval of Batch No. 93002 on the basis that the Director of Health abused his discretion in not indicating expiration data on *39ODH certificates and bottles for the calibration batch solutions and allowing for the use of the solution after the manufacturer’s stated expiration date. The regulations do not use the term “expiration date”; instead, the regulations state, “A calibration solution shall not be used more than three months after its date of first use.” Ohio Adm.Code 3701-53-04(A)(1). In State v. Brockway (1981), 2 Ohio App.3d 227, 231-232, 2 OBR 247, 252, 441 N.E.2d 602, 607, the Fourth District Court of Appeals ruled:
“[I]t is evident that the General Assembly intended, because of the numerous types of equipment and procedures available for such alcohol tests, to achieve uniformity in testing in Ohio by delegating to the Director of Health authority to choose and approve of the specific tests and procedures to be used in Ohio.”
The defendant’s argument rests on the premise that the certificate of approval does not contain an expiration date and that federal law mandates that the Director of Health cannot ignore an expiration date. The prosecution, however, has clearly established that breath and saliva tests for alcohol are not medical devices when used for law enforcement purposes. (State’s Exhibit 11.) As a result, calibration solutions are not drugs and are not governed by the drug laws as defendant has argued.
The defendant also urges the court to find that the ODH should adopt the National Safety Council’s recommendations. However, since there is no applicable federal law that the director must follow with regard to expiration dates, his decision to adopt a policy of using solutions within three months after their first use is supported by substantial evidence. The court cannot substitute its judgment with regard to which rule is better, when the legislature has specifically given this power to the Director of Health. R.C. 3701.143. Therefore, the defendant’s assertion that the Director of Health abused his discretion by failing to promulgate regulations relating to use of a solution after the stated expiration date fails.
Abuse of Discretion in Approving RepCo. Batch No. 93002
The defendant then set forth a number issues relating to an abuse of discretion in approving RepCo. Batch No. 93002 for use as a calibration solution in the state of Ohio. As previously discussed, an action by the Director of Health to approve calibration batch solutions is in accordance with the law as set forth in R.C. 3701.143 and Ohio Adm.Code 3701-53-04. The question then is whether the decision to approve RepCo. Batch No. 93002 in 1993 is supported by substantial evidence.
The reliability and accuracy of breath-testing equipment, including calibration solution, are critical in cases charging a violation of R.C. 4511.19(A)(3), the so-called per se violation. A violation of this statute occurs when an individual *40operates a motor vehicle with a concentration of alcohol in his bodily substances equal to or exceeding the percentage designated in the statute, regardless of whether the accused was under the influence of alcohol. Serious consequences are incurred by the defendant solely as a result of this chemical test. The results of this test must be reliable. The certificate of approval issued by ODH needs to be based on reliable data obtained using scientifically valid methodology subject to replication by other scientists. Otherwise, defendants and their attorneys who lack confidence in ODH’s certificates of approval will subject every per se case to the intense scrutiny this court has undertaken in this case.
The decision to approve RepCo. Batch No. 93002 on the basis of unwritten and conflicting approval criteria, lack of written laboratory protocol, standards, and review is indeed questionable. However, the decision to approve RepCo. Batch No. 93002 was also based on a summary data sheet prepared by a laboratory analyst. This summary sheet purports to show that the ODH testing resulted in the solution testing within 0.8% of the manufacturer’s target value, well within any arguable scientific limit.
Doubt was cast upon the 1993 approval, though, by a confusing ODH memo issued in January 1996. The 1993 approval was then questioned by the Director of Health, who in response to inquiries from the media and general public, requested an audit of the 1993-1996 approvals of calibration solution. (Testimony of Dr. Peter Somani, pp. 77-78.) If the subsequent 1996 data had replicated the 1993 data the court could say with assurance that the decision in 1993 was based on substantial and valid scientific evidence. That, however, is not the case. The data calculated in 1996 was vastly different from that in 1993 even though the data was extracted from the same chromatograms. The ability to replicate the same results from the original raw data is the hallmark of scientific reliability. These conflicting results strongly suggest a lack of scientific reliability. It is particularly disturbing to note that when Dr. Sutheimer and Ferguson measured the 1993 chromatograms and constructed a five-point calibration curve using the same scientific methods believed to be used by Porter, not a single calculation was confirmed. Batch No. 93002, which Porter certified as nearly perfect, failed every criterion and had to be rejected. This, of course, calls into questions the validity of the 1993 test.
The following chart illustrates the discrepancies in the December 6 assay calculations:
*41 *Numbers in parenthesis denote the sample number Mean Concentration (2699) (2700) (2701) (2702) Percent RSD (2699) (2700) (2701) (2702) Calibration Verif. % Recovery #1 #2 03 04 05 Method Validation Samples % Recovery #1 02 03 04 05 Grand Mean % of Target Value Difference
1993 Using a Jive paint calibration curve 21 1.2248 1.2208 1.2217 1.2269 1.36 0.48 0.34 1.02 100.4% 100.1% 98.7% 99.0% m.4%, 102.4% 99.0% 98.9% 99.4% 101.1% 1.2236 100.5% .5%
1996 Using a Jive point calibration curvefrom 9300222 1.307254 1.327859 1.320783 1.323016 2.17 0.60 1.05 1.81 176.8% 108.0% 106.8% 106.8% Not listed 105.0% 109.6% 107.7% 109.2% 111.8% 1.319728 108.4% 8.4% ,v,i,
1996 Using a four point calibration curve from Batch 93390. 1.183145 1.203072 1.196229 1.198389 2.32% 0.64% 1.12% 1.93% 98.5% 97.6% 96.5% 99.9% n/a 93.4% 24 97.8% 95.9% 97.4% 99.9% 1.195209 98.2% 2.8%
1996 Using a four point calibration curve from Guth Batch 93390 23 ant dropping QC1. 1.18314Í 1.203072 1.19622Í 0.64% 1.12% 1.93% 98.5% 97.6% 96.5% 99.9% N/A droppet 97.9% 96.0% 97.4% 99.9% 1.19520$ 98.2% 2.8%
*42Either the information in 1993 was in error or the information in 1996 was in error. The 1993 data was calculated by the Ohio Department of Health. As noted, the ODH lab that performed the analysis did not have any written protocol, and had conflicting standards and a lack of rigorous scientific review. In addition, James Ferguson, the most experienced expert in the field of gas chromatography, testified that when the solutions were first sent to the organic laboratories of ODH, the lab had a difficult time analyzing these solutions. “They probably worked on methods six to nine months before they assayed their first batch successfully.” During the 1996 audit, Dr. Craig Sutheimer and James Ferguson also testified that the chromatograms for RepCo. Batch No. 93002 were examples of “very poor chromatography.” Dr. Peter Somani testified that he was willing to rely also on the 1996 data audit. He stated that he told Dr. Sutheimer and Ferguson to “give me the full picture, give me the good news, bad news, I want to know if these solutions that have been approved were approved appropriately or not.” (Testimony of Dr. Peter Somani, pg. 32, lines 1-5.) He also testified that he was aware of the ramifications of Dr. Sutheimer’s or Ferguson’s discovering a solution that should not have been approved. He stated that “[t]he ramification is that if the solutions were approved and they should not have been approved, then the use of those solutions in any Breathalyzers will invalidate the results.” (Testimony of Dr. Peter Somani, pg. 36, line 1-4.)
Based on the facts before the court, the 1993 approval of RepCo. Batch No. 93002 was based on false or erroneous information contained on the summary data sheet. Our inquiry, however, does not stop there. The court must also determine whether there was any external evidence that the approval decision could have been based upon in 1993. The prosecution maintains that results from testing of RepCo. Batch No. 93002 by Northwest Toxicology should provide an independent analysis upon which to base an approval. The court disagrees. The approval process by ODH has been described as a quality assurance measure. As part of that quality assurance measure, the ODH would analyze four samples from four bottles in a batch. This quality assurance measure which apparently *43failed in 1993 cannot be replaced by a less stringent quality assurance measure conducted by Northwest Toxicology. The president of RepCo. testified that he sent only two bottles of solution to Northwest Toxicology. Northwest would then conduct analysis on five samples from each bottle. Northwest did not perform an independent analysis to determine the concentration. To do that, the experts agree, a sample of the square root of the number of bottles plus one would need to be analyzed. For example, for a batch of 2,000 bottles, that would be somewhere between 44-45 bottles that would need to be tested. (Testimony of James Ferguson, pg. 97, lines 14-17.) Further, Northwest Toxicology was hired by the manufacturer. If Northwest Toxicology had performed an independent determination rather than a check of the concentration of Batch No. 93002 or if their actual chromatograms had been introduced into evidence, the court might have been persuaded that the Northwest Toxicology results would justify an approval decision. However, such evidence was not before the court.
The prosecution also asserts that Blaine Keckley, as an employee of the Ohio State Highway Patrol, had the occasion to assay various calibration batch solutions and, based on his experiences, never found a problem with solutions provided by manufacturers, including RepCo. Again, the purpose of ODH testing was one of quality assurance not of reputation. To base criminal convictions on a lesser standard would be a great injustice and would make the presumption of innocence a meaningless phrase.
Additionally, the prosecution argues that both Keckley and ODH performed stability analysis on RepCo. Batch No. 93002 more than one and a half years after it was manufactured and found that it contained the amount of alcohol it was supposed to contain. While conceivably these studies may suggest reliability, the fact that they were not available in 1993 is problematic. The ODH stability study was not even performed until April 3, 1996. Nor was the stability study done with the same sample bottles. To base an approval in 1993 on analysis performed years later is impossible. To adopt that line of reasoning, the court would have to ignore the multitudes of calibration checks performed in the interim, and the selective retention and exclusion of data which apparently were done in order to ensure approval of the calibration solution.
Finally, the prosecution and the Ohio Department of Health have argued that the 1996 data audit supports the decision in 1993 to approve RepCo. Batch No. 93002. The court agrees that the data was extracted and calculated by experts in the field, but it does not agree that the audit in any way supports an approval. The defense has shown by a preponderance of the evidence that the data did not even meet the criteria set forth by Dr. Sutheimer and Ferguson. The December 3, 1993 assay did not meet criteria for acceptable chromatography. The December 6, 1996 assay did not meet the 1996 acceptance criteria when calculated with *44a five-point calibration curve from RepCo. Batch No. 93002. The internal standard references were missing. Some of the running standards were mixed. The December 6, 1993 assay data did not meet the correlation coefficient requirement of .999 when fitted to a five-point calibration curve from Guth Batch No. 93390. The December 6,1993 assay did not meet the acceptance criterion of “Extreme QC Recovery” within ± 5%.” It had an extreme QC1 recovery of 93.7% which is simply unacceptable. Interestingly, the extreme QC recovery is not listed even on the summary form submitted to Dr. Somani. Instead, the mean QC recovery is listed under the “Extreme QC Recovery” column. Moreover, the 1996 data results contained many calculation errors and misplaced values. When confronted with data errors, Dr. Sutheimer, Ferguson and Steve Wagner all concurred with the recalculations provided by the defense.
What is evident is that after using an external five-point calibration curve from RepCo. Batch No. 93002 and obtaining a grand mean 8.4% off from the manufacturer’s target value, the batch should have been analyzed with an internal method. The internal standard, however, was missing. So without the ability to go back and reassay the batch in 1993 for a third time, attempts were made to salvage the data and approve the batch by Dr. Sutheimer and Ferguson.
The data audit performed at the request of the Director of Health showed by a preponderance of the evidence that the data sheet prepared in 1993 by ODH was in error. Therefore the decision to approve RepCo. Batch No. 93002 was based on false or erroneous information. It is an abuse of discretion to rely on this data or the certificate of approval prepared in 1993. To approve the batch solely on the results of a less stringent quality control check provided by the manufacturer would likewise be an abuse of discretion. Further, it is impossible to substitute new stability analyses performed in 1996 by the Ohio Department of Health and the Ohio State Highway Patrol as a basis for an approval decision that took place in 1993. The results obtained in 1996 using the 1993 chromatograms persuade this court that the batch should not have been approved in the first instance. Since there was not substantial evidence to support ODH approval of RepCo. Batch No. 93002 in 1993, this court is unable to permit this evidence to be used against the defendant in the trial of this case.
Accordingly, defendant’s motion to suppress the results of his BAC test is hereby sustained.
Motion granted.
. For example, although Leonard Porter testified that he would look to see if the test results were within ± 5% of the target value (testimony of Leonard Porter, pg. 23), ODH protocol written in 1994 states that the findings must be within ± 2.5% of the target value. See Defendant’s Exhibit DDL
. See £h. 1.
. Breath Alcohol Content.
. "Gas chromatography” is a separation technique involving passage of a gaseous moving phase through a column containing a fixed absorbent phase; it is used principally as a quantitative analytical technique for volatile compounds. McGraw-Hill Dictionary of Scientific and Technical Terms, Fifth Edition.
. The summary sheet does not state the number of assays. This information was derived from the 1996 data audit.
. A ‘‘calibration curve” is a plot of calibration data, giving the correct value for each indicated reference.
. An "external standard” is literally an injection of a solution without determining the area or without evaluating the area of the internal standard. An "internal standard” is a marker or a material that is put in along with the target analyte, in this case alcohol. For an internal standard, the ratio of ethanol to the internal standard will stay the same. (Dr. Sutheimer, pp. 32-33.)
. An acceptable grand mean should have been ± 5% by Porter’s testimony or ± 2.5% based on other testimony by ODH employees. This grand mean was 8.4% of the target value.
. Guth is another manufacturer of calibration solution.
. The standard for purposes of the 1996 data audit was a correlation coefficient of .999.
. An "outlier” is defined as a data point or other item more than ± 3 standard deviations. (Bellotto, pg. 115.)
. There were a total of ten reference calibration standards and only four were used for the final results,
. It is also important to note that RepCo. Batch No. 93002 was one of four batches which were the topic of a meeting with laboratory personnel from ODH. Because of problems either with registration or quality control, Dr. Sutheimer and Ferguson had to discuss the batches and go through the lab personnel notebooks in detail. (Ferguson, pg. 100.)
. See Defendant’s Exhibit B-25.
. See Defendant’s Exhibit I, pp. X9-X15. "Data Unacceptable” (X9), "Mean of Calibration Verification Recovery 129.2%" (XI5).
. See Defendant’s Exhibit I, pp. X4-X8.
. Calculations in the form of a "Summary Sheet” were not presented into evidence. However, see Defendant Exhibit M, and Defendant’s Exhibit W2. Testimony of Dr. Robert J. Bellotto, Jr., pp. 10-11. Testimony of James Ferguson.
. See Defendant’s Exhibit I, pp. X1-X3. Also see corrected version, State’s Exhibit 7 and Defendant’s Exhibit HH1.
. See Defendant’s Exhibit I, pp. X1-X3. Also see corrected version, State’s Exhibit 7 and Defendant’s Exhibit HH1.
. For a calibration to be acceptable, it has to have a degree of linearity, and that degree of linearity over all the samples, the deviation of each sample around this mean, is called the "least squares method.” The correlation coefficient must be, by the 1996 ODH rules, .999 or better. This is also reported as R2(.999*.999 = .998) (Testimony of James Ferguson, pp. 90-91.) *37The acceptance criteria used by Dr. Sutheimer and James Ferguson for the 1996 data audit indicated an acceptance criterion of R2 greater or equal to 0.998. (Defendant’s Exhibit H, cross-reference Defendant’s Exhibit B27.) The correlation coefficient then must be at least .999. (Testimony of Dr. James Belotto, pg. 13.) See, also, Defendant’s Exhibit M.
. Data extracted by Andrew Smith, ODH lab employee. See Defendant’s Exhibit B-25.
. Data extracted by James Ferguson, reviewed by Dr. Craig Sutheimer. See Defendant’s Exhibit I — pp. X4-X6.
. Data extracted by James Ferguson, reviewed by Dr. Craig Sutheimer. See Defendant’s Exhibit I — pp. X1-X3. See, also, corrected documents State’s Exhibit 7, and Defendant’s Exhibit HH1.
. This would have caused the batch to be rejected unless there was a way to drop the data point. (See testimony of Dr. Craig Sutheimer, pp. 62-63.)