(concurring in part, dissenting in part):
I believe the military judge erred in admitting the DNA evidence concerning the PCR Amp-FLP analysis in the Apo-B region on microstains found in appellant’s rental car and on an axe located in appellant’s work area. However, I also believe the error was harmless and am convinced beyond a reasonable doubt that appellant is guilty of Ms. Yolanda Pengson’s murder.
A relatively new procedure for analyzing minute quantities of DNA developed the questionable evidence. This evidence was then used to tie fibers from appellant’s rental car and a microstain on the axe to Ms. Pengson’s charred remains. Appellant argues the evidence is unreliable and should not have been admitted.
Scientific evidence must be both relevant and reliable to be admissible. Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469, 480 (1993). Reliability involves a determination that the application of the principle produces consistent results; while validity involves a determination that the principle supports what it purports to show. Id., 509 U.S. at -, fn. 9, 113 S.Ct. at 2795, fn. 9, 125 L.Ed.2d at 481, fn. 9.
The trial judge has a “gatekeeping” responsibility in determining whether the methodology is scientifically valid and whether the methodology can be applied to this case. Id, 509 U.S. at -, 113 S.Ct. at 2798-99,125 L.Ed.2d at 482. The trial judge may consider certain factors in making this determination. These include:
1. Can the theory or technique be tested and has it been tested?
2. Has the theory or technique been subject to peer review?
3. What is the general acceptance of the theory or technique among the relevant scientific community?
4. What is the known or potential rate of error?
Id, 509 U.S. at---, 113 S.Ct. at 2798-99, 125 L.Ed.2d at 482-3. We will ad*638dress each of these considerations based on the facts of this case.
Testing of the Theory or Technique
Testing involves the development of a scientific hypothesis and testing that hypothesis to determine if the results can be falsified. Id., 509 U.S. at-, 113 S.Ct. at 2796, 125 L.Ed.2d at 483. In other words, does the testing produce reliable, reproducible results? See United States v. Bonds, 12 F.3d 540, 558 (6th Cir.1993).
Dr. Pflug testified that he has conducted DNA testing using the Amp-FLP procedure for over three and one half years. During ' the last two years, he has been looking at the Apo-B region of DNA examining over 4000 samples from about 120 cases. He indicated that there are several other laboratories in Germany, Austria, and Switzerland that use this technique.
However, it is also significant that no other criminal laboratories use the Amp-FLP analysis in the Apo-B region. In fact, both the FBI and the Royal Canadian Mounted Police have specifically rejected Amp-FLP testing in the Apo-B region. Additionally, Dr. Eisenberg’s testimony indicated that numerous experts in the area of DNA testing throughout the world agreed that Dr. Pflug’s methodology of analyzing the Apo-B region yields inconclusive results.
There is no question that the scientific community is testing PCR Amp-FLP analysis with inconsistent findings. However, the military judge should have focused his analysis on the Amp-FLP process as it applies to the Apo-B region. In this narrow sub-area, Dr. Pflug and his associates appear to stand alone in their support of forensic use of ApoB analysis. I believe that Daubert contemplates something more than initial testing and development of a procedure, but less than general acceptance before it is admissible in court.
Peer Review and Publication
PCR analysis was the subject of over 1400 papers published between 1985 and 1990. At least 50 of these papers dealt with the Amp-FLP procedure. It is unclear how many papers dealt with testing in the Apo-B region but four papers written in that time frame were introduced at trial.
Daubert indicates that submission of a theory to the scientific community increases the scrutiny of the hypothesis and the possibility that flaws in the process will be discovered. Daubert, 509 U.S. at-, 113 S.Ct. at 2797, 125 L.Ed.2d at 483. Again, I believe the focus of this peer scrutiny, as anticipated by Daubert, should be on the specific analysis in the Apo-B region and not general PCR or Amp-FLP analysis. Despite the publication of only a few papers on analysis in the ApoB region, there appears to be considerable controversy about the procedure as well as outright rejection by laboratories who reviewed it.
General Acceptance
Daubert indicates that the trial judge may consider the general acceptance of the theory or principle in determining the admissibility of scientific evidence. However, this should not be the sole consideration. Id, 509 U.S. at-, 113 S.Ct. at 2796-97,125 L.Ed.2d at 483. A known technique that has attracted only minimal support within the scientific community may be viewed with skepticism. Id, 509 U.S. at-, 113 S.Ct. at 2797, 125 L.Ed.2d at 483. However, neither newness nor lack of certainty renders the evidence inadmissible per se. United States v. Stifel, 433 F.2d 431, 438 (6th Cir.1970), cert, denied 401 U.S. 994, 91 S.Ct. 1232, 28 L.Ed.2d 531 (1971).
General acceptance involves considerations of both the theory and the methodology. Bonds, 12 F.3d at 562. Thus, in this case, it is apparent that the theory of PCR analysis is generally accepted. Testimony indicated that PCR analysis is possibly the most widely recognized technique currently used in research science. It further appears from Dr. Holland’s testimony that the methodology of Amp-FLP analysis is somewhat accepted in the scientific community but has not been admitted in court. But, based on the evidence, the methodology of Amp-FLP analysis in the Apo-B region does not meet general acceptance standards. As previously indicated, not only does no other criminal labora*639tory use this form of analysis, but several have also specifically rejected it. The fact that it is used in one private laboratory in the United States and several, but not all, German speaking laboratories does not overcome the evidence of its rejection by numerous others.
Known or Potential Error Rate
The record of trial contains no evidence of the error rate of this procedure. Dr. Pflug briefly discusses the fact that his laboratory is subject to a quality inspection by the German GENDAP group. He also indicated that there is proficiency testing within this GENDAP group. However, he does not describe the results of this quality testing or give any indication as to an error rate. Likewise, those experts who questioned the procedure did not discuss error rate.
Daubert also suggests that the trial judge should consider the known or potential rate of error of the procedure. Daubert, 509 U.S. at-, 113 S.Ct. at 2797, 125 L.Ed.2d at 483. When a procedure has been generally accepted, “it is implicit that the rate of error is acceptable to the scientific community as well.” Bonds, 12 F.3d at 560. On the other hand, a procedure that has not been generally accepted may have an unacceptable error rate.
To me, the error rate directly equates to the reliability of the procedure — does the analysis produce consistent results? I have been spoiled by the abundance of information available in the urinalysis area to support the reliability of that procedure. Retests, blind samples, and verification tests are routinely used to authenticate consistent results. This degree of evidence is not necessary in every case, but I find the lack of any comparable evidence in this case significant.
I am not willing to accept the unspecified quality control testing done by GENDAP as establishing an acceptable error rate without more information. Evidence of the error rate or detailed descriptions of the methods of quality control testing and its results may be available and, if so, could overcome many of the weaknesses in the analysis of the ApoB region. However, without this evidence it is virtually impossible to evaluate the reliability of the procedure in view of the other evidence questioning it.
Admissibility of the Evidence
In addition to the Daubert factors, there are several other areas that I considered prior to finding the evidence inadmissible. In general, these factors go to the responsibility of the trial judge and the overall reliability of the analysis in the Apo-B region.
First, Daubert clearly gives the trial judge gatekeeping responsibilities in determining the admissibility of the expert testimony on scientific evidence. This implies more than a mere acquiescence in the position of the proponent of the evidence and more than the use of laboratory procedures. Here, the military judge did not meet his gatekeeping responsibilities. His instructions to the members, in effect, concedes that point when he states: “Second, with respect to the PCR (Amp FLP) DNA test results in this case, there is a dispute as to its validity or value. Third, my ruling in permitting you to evaluate the evidence of the DNA test results does not mean that I have determined that such evidence has any particular value in the case.” To me, if the trial judge determines the evidence is admissible under Daubert, this implies a determination that the evidence is valid, has value, and will help the members.
Second, Daubert implies that every step in the expert’s analysis should be based on scientific knowledge which requires a step-by-step evaluation of the Daubert factors. In re Paoli R.R. Yard PCB Litigation, 35 F.3d. 717, 745 (3rd Cir.1994). “This is true whether the step completely changes a reliable methodology or merely misapplies that methodology.” Id. I do not believe, based on the totality of the evidence presented, that the Amp-FLP procedure focused on the Apo-B region meets the reliability standard of Daubert. In contrast, Judge Gamboa believes that the focus of the admissibility analysis should be on the general acceptance of the Amp-FLP procedure and not the locus of the testing in the Apo-B region. I disagree because if the testing in the Apo-B region does not produce reliable, consistent results, then the evidence should not be admissible.
*640Third, PCR Amp-FLP analysis in the Apo-B region is not reliable, in part, due to its subjective interpretation of the results. Although PCR analysis has been admitted in criminal trials in the United States, the particular form of testing used was the DQ-Alpha and not Amp-FLP. See Oregon v. Lyons, 124 Or.App. 598, 863 P.2d 1303 (1993); Spencer v. Commonwealth of Virginia, 240 Va. 78, 393 S.E.2d 609 (1990). A significant distinction of the DQ-Alpha procedure is that the results do not rely on the subjective interpretation of the expert. Lyons, 863 P.2d at 1309.
The results of the Amp-FLP analysis in the Apo-B region are solely interpreted through visual observation. Here, despite the fact that technicians performed the procedures, there is no evidence to suggest that anyone other than Dr. Pflug reviewed these samples or would have interpreted the results in the same manner. Dr. Eisenberg stressed that this lack of computer verification was a component of others’ rejection of the process. It is also a possible reason why Amp-FLP analysis is not used for forensic purposes and has not been admitted in any court.
I find this subjective interpretation of the results to be especially significant in a developing procedure. For example, when Dr. Pflug initially examined the samples, he identified a new allele designation (marker) that was not present in any of the prior 4000 samples he had examined in developing his comparison ladder. He called this marker a “55” because it looked like “55” was the appropriate distance above his known “53” marker. He subsequently identified other “55” markers and a “57” marker when he tested samples from Philippine nationals. He testified that other markers may exist particularly among non-Caucasian populations. Dr. Eisenberg testified that while Dr. Pflug refers to these designations as 33, 35, 37, and so forth, other scientists refer to them as 34,36, 38, and so on. This subjectivity in identifying the allele designators, the lack of cohesiveness of scientists in the field examining the Apo-B region, and the lack of a mechanical verification of the designators impact the reliability of the analysis.
Fourth, in evaluating scientific evidence there is sometimes a very fine line between methodology and conclusions. Daubert says admissibility should focus on methodology and not conclusions. Daubert, 509 U.S. at -, 113 S.Ct. at 2797, 125 L.Ed.2d at 484. However, the trial judge should not exclude evidence when there is a flaw in the expert’s investigative process that renders the conclusions incorrect unless the flaw is such that the expert’s conclusions lack good grounds to support them. In re Paoli R.R. Yard PCB Litigation, 35 F.3d at 746.
Here, the witnesses often spoke in terms that normally apply to conclusions. For example, the term “inconclusive” was frequently used. Additionally, the experts indicated their personal preferences for human controls, use of identical gels to test the substances, and other internal procedures for conducting the analysis. However, the testimony also emphasized certain technical weaknesses in the methodology and the lack of general acceptance of analysis in the ApoB region. I then considered these weaknesses coupled with the fact that all results are visually determined by one individual without any type of computer or mechanical verification. The totality of this information persuades me that the use of the term “inconclusive” encompassed more than just conclusions, but goes to the actual methodology of Apo-B analysis.
Thus, after evaluating the Daubert criteria as well as examining the evidence in this case, I conclude that the military judge abused his discretion in admitting this evidence. My opinion is limited to the facts of this case. I am not saying that the procedure and methodology of Amp-FLP analysis in the Apo-B region would not be admissible given additional evidence answering somé of my concerns.
Impact of Inadmissible Evidence
Having decided the evidence was inadmissible, I must now determine the impact of that evidence. The general rule is that reviewing courts must determine whether “there is a reasonable probability that the evidence complained of might have contributed to the conviction.” Chapman v. Califor*641nia, 386 U.S. 18, 23, 87 S.Ct. 824, 827-28, 17 L. Ed.2d 705 (1967), quoting Fahy v. Connecticut, 375 U.S. 85, 86-87, 84 S.Ct. 229, 230-31, 11 L.Ed.2d 171; United States v. Morgan, 40 M. J. 405, 412 (C.M.A.1994). In this case, I believe the admission of this evidence did not contribute to the conviction or materially prejudice the substantial rights of the accused. The government’s case against appellant, although circumstantial, was strong and conclusive. Article 59(a), UCMJ, 10 U.S.C. § 859(a); see also United States v. Giambra, 38 M.J. 240, 242 (C.M.A.1993).
First, I note that this was an extremely well litigated trial. Every weakness of the Amp-FLP analysis in the Apo-B region was thoroughly detailed by trial defense counsel before the members. This testimony was such that I believe the members gave this evidence little or no weight in their deliberations.
Next, I believe that there is more than sufficient evidence to prove appellant’s guilt beyond a reasonable doubt independent of this evidence. Appellant possessed the motive and opportunity to kill Ms. Pengson. Motive — he wanted to break off their relationship, considered Ms. Pengson a second class citizen, and was concerned that his wife knew he had a girlfriend. Opportunity — Ms. Pengson told others she was going to meet appellant on Saturday night and appellant received a phone call from an oriental sounding woman at work late Saturday night. Additionally, the German police officer told appellant about Ms. Pengson’s death and indicated that a witness said Ms. Pengson left her house to meet appellant on Saturday night. Appellant responded that he met Ms. Pengson on Saturday. While he did not state when he met her, a reasonable implication based on the prior discussion is that he met Ms. Pengson on Saturday night. If he met her on Saturday night, this would make him the last known person to see her alive. The autopsy report, while inconclusive as to the time and manner of death, indicated that Ms. Pengson’s stomach contents revealed food similar to what she had eaten earlier Saturday evening. The coroner speculated that she died within 4-6 hours of her last meal. Additionally, appellant’s comings and goings from Thursday through Monday were unusual, frequently contradictory, and his known whereabouts were such that he could have murdered Ms. Pengson and disposed of her body.
It is difficult to determine whether appellant had the means to murder Ms. Pengson as the exact manner of death is unknown. However, there is evidence that the head and hands were severed with an ax or machete. An ax was found in appellant’s work area. This ax was distinctive because it was not dusty or in its protective case as were the others. Additionally, a screening test using a Heglar stick revealed the presence of blood on the ax.
Other circumstantial evidence can also be interpreted as pointing towards appellant’s guilt. On Friday, he rented a car to make a trip to Hannover, yet he was scheduled to work at his part time job that weekend. He told some individuals that his car was not working properly, but loaned it to a friend with car problems. The friend had no mechanical problems with the car. The rental car was clean when he picked it up, but was returned with a dark, dry, crusty spot in the trunk.
Additionally, the evidence indicated that hairs found in the ladies rest room in the building where appellant worked matched hairs identified as being Ms. Pengson’s. These hairs were distinctive both because of a bend and a cut edge.
Several other incidents over the weekend are also questionable. On Sunday evening, about 2300 hours, appellant called his military supervisor and asked for Monday off because he was working Sunday night. However, appellant ended his Sunday shift at about 1800 hours. Around 0330 hours on Monday morning, appellant drove the rental car into his work area claiming he had to work on his car. His personal vehicle was not in this area. He left about 45 minutes later.
On Monday morning about 0700 hours, a witness identified a car similar to appellant’s rental car leaving the area where the corpse was found. However, another witness said the rental car was not similar to the observed *642car. Appellant returned the rental car about 0950 and subsequently picked up the keys for his personal car from his friend. The odometer reading of the rental car reflected 554 kilometers. This reading may be consistent with travel to the location of the corpse which was about 30 minutes from base. It is inconsistent with appellant’s initial stated use of seeing a friend in Hannover which was a round trip of 704 kilometers from base.
Finally, while confined in a German prison, appellant told a German police officer that he could get the death penalty if tried in an American court. He asked the officer what would happen to him if he were tried in a German court for the offense. The officer told him he could not receive the death penalty. Appellant then replied that he would be better off in a German court. Again later, he questioned what a prisoner did in a German jail. During this same period appellant indicated that when he thought of Ms. Pengson, “his head was empty and he only sees black.” Appellant never denied having anything to do with Ms. Pengson’s murder.
In conclusion, I find that the record contains insufficient evidence of reliability to support the military judge’s decision to admit the evidence of Amp-FLP testing in the Apo-B region. However, I find no reasonable probability that this DNA evidence contributed to the conviction because of the inherent weaknesses in the analysis and the overall strength of the government’s case.