928 A.2d 165 (2006)
395 N.J. Super. 175
STATE of New Jersey, Plaintiff
v.
Bryan SHARP, Defendant.
Superior Court of New Jersey, Law Division, Camden County.
Decided December 5, 2006.
A. Victoria Shilton, Assistant County Prosecutor, for the State (Camden County Prosecutor's Office).
Leonard S. Baker, for Defendant Bryan Sharp (Mitnick, Josselson, DePersia, Baker & Schweiger, attorneys).
SCHUCK, J.S.C.
Following a fire of December 30, 2003, that destroyed his residence in Barrington, New Jersey, the defendant in this criminal matter, Bryan Sharp, was charged with separate counts of second degree aggravated arson under N.J.S.A. 2C:17-1(a)(1) and N.J.S.A. 2C:17-1(a)(3). He now presents the court with the following pretrial motions, the first two of which present issues of first impression not previously addressed by any reported New Jersey decision:
(1) The defendant seeks to preclude Deputy Fire Marshal Medlar from giving his expert opinion that the fire in question was started by arson. Defendant contends that the process-of-elimination technique for determining the cause and origin of a fire is not "reasonably reliable" under N.J.R.E. 703 and must be excluded as a net opinion.
(2) The defendant seeks to preclude Deputy Fire Marshal C. Jay Houck from giving his expert opinion that the alert to certain locations at the fire scene by Blondie, an accelerant-detection canine, is proof that an accelerant was present near the origin of the fire. The laboratory reports in this case came back negative for accelerants. Therefore, the defendant moves for exclusion, claiming such an opinion would be based on a novel scientific theory that a canine's sense of smell is more accurate than laboratory instruments. The defendant likewise contends that any reference at all to the accelerant-detection canine should be excluded as unduly prejudicial under N.J.R.E. 403.
(3) The defendant also seeks to bar that aspect of Deputy Fire Marshal Medlar's opinion indicating that an accelerant "possibly" was used at the fire scene. Again, the defendant contends that this is an impermissible net opinion.
Following an evidentiary hearing, and for reasons set forth more fully herein, the court answers these questions as follows:
(1) The opinion of Deputy Fire Marshal Medlar, developed through a process-of-elimination technique, that the fire was caused by arson, is based upon a generally accepted and reasonably reliable scientific methodology, correctly applied to the facts before the court, and is therefore not an inadmissible net opinion.
(2) In the absence of laboratory confirmation, Deputy Fire Marshal C. Jay Houck's opinion that an accelerant-detection canine alerting to a particular location is probative of the presence of an accelerant, is not based upon a generally accepted scientific theory and, accordingly, is inadmissible. Further, evidence of the canine alerting in combination with other evidence of the fire investigation to show the reasonableness of taking samples from certain locations, even with a limiting instruction, is not admissible, and the risk of undue prejudice of such evidence would substantially outweigh its probative value under N.J.R.E. 403.
(3) Deputy Fire Marshal Medlar's opinion that an accelerant "possibly" was used to set the fire is an inadmissible net opinion.
BACKGROUND
The December 30, 2003 fire that consumed defendant's home originated near the Christmas tree in defendant's living room, igniting the adjoining walls and rapidly spreading throughout the residence. The defendant, his wife, and their two adult sons all escaped unharmed.
Robert M. Medlar, a Camden County Deputy Fire Marshal, responded to the scene and conducted an investigation into the cause of the fire. Deputy Medlar concluded that the fire resulted from arson, relying in part on the findings of Deputy Fire Marshal C. Jay Houck. During the initial investigation, Deputy Houck's accelerant-detection canine, Blondie, "alerted," indicating the presence of flammable liquids near the Christmas tree. In Deputy Houck's expert opinion, Blondie's reaction indicates to a reasonable degree of scientific certainty that an accelerant was used to start the fire. Notably, lab reports of samples taken from the area where Blondie alerted came back negative for an accelerant. See Laboratory Report of Jill Ann Dolan, Bureau of Alcohol, Tobacco, Firearms & Explosives, May 28, 2004.
I. Is State Expert Medlar's Opinion that the Fire Was Caused by Arson an Inadmissible "Net Opinion"?
N.J.R.E. 702 allows an expert to testify in the form of an opinion. The testimony must: (1) concern a subject matter beyond the ken of the average juror; (2) be relative to a field of study that is state of the art; and (3) come from an expert with sufficient expertise. State v. Fortin, 178 N.J. 540, 597, 843 A.2d 974 (2004). N.J.R.E. 703 requires that the opinion be based on a relevant methodology and that the methodology be correctly applied to the facts before the court. New Jersey courts have interpreted N.J.R.E. 703 as requiring the expert's method of reaching his opinion to be one which produces "reasonably reliable" results. State v. Kelly, 97 N.J. 178, 210, 478 A.2d 364 (1984).
In New Jersey, a method of reaching an opinion is reasonably reliable if it is generally accepted in the scientific community. See Frye v. United States, 293 F. 1013 (D.C.Cir.1923). General acceptance, which is the Frye standard, requires the strict application of the scientific method and an "extraordinarily high level of proof based on prolonged, controlled, consistent and validated experience." Rubanick v. Witco Chemical Corp., 125 N.J. 421, 436, 593 A.2d 733 (1991). However, it is not necessary to demonstrate the "total or absolute infallibility of the technique" in question. Romano v. Kimmelman, 96 N.J. 66, 80, 474 A.2d 1 (1984).
To establish that proffered scientific evidence is reasonably reliable, the proponent may offer evidence of the following types: "(1) the testimony of reliable or knowledgeable experts; (2) authoritative scientific literature; [and] (3) persuasive judicial decisions which acknowledge such general acceptance of expert testimony." State v. Moore, 188 N.J. 182, 206, 902 A.2d 1212 (2006) (quoting Windmere Inc. v. Int'l Ins. Co., 105 N.J. 373, 379, 522 A.2d 405 (1987)). What is more, when the State seeks to advance scientific evidence in criminal cases, given the important liberty interests of the defendant that are at stake, the reliability of that evidence is subject to heightened scrutiny. See Windmere Inc., supra, 105 N.J. at 378, 522 A.2d 405 (requiring reliability be "clearly established" in the criminal context); see also State v. Moore, supra, 188 N.J. at 206-07, 902 A.2d 1212, State v. Cavallo, 88 N.J. 508, 516-26, 443 A.2d 1020 (1982) (applying the Frye test in a criminal context).
The arson investigation involved in this matter is a subject beyond the ken of the average juror and is at a state of the art which the court recognizes. Also, Deputy Medlar's credentials are not disputed. Thus, the strictures of N.J.R.E. 702 are met. The final question is whether Deputy Medlar's opinion satisfies N.J.R.E. 703, requiring that the method be "reasonably reliable" and "clearly established."
In the present case, Deputy Medlar's opinion on the cause of the fire is based on the process of elimination. Those courts reaching this issue have sanctioned the process-of-elimination approach for determining the cause of a fire. Breidor v. Sears, Roebuck and Co., 722 F.2d 1134, 1138 (3d Cir.1983); see also Gichner v. Antonio Troiano Tile & Marble Co., 410 F.2d 238 (D.C.Cir.1969) (admitting the opinion of a fire investigator on the "probable cause" of a fire, reached by eliminating all other potential causes). In contrast, courts have refused to sanction expert opinions that a fire resulted from one of two causes or "possibly" resulted from a specified cause. Lanza v. Poretti, 537 F.Supp. 777, 785 (E.D.Pa.1982). These opinions are merely speculative, or in legal jargon, they are "net opinions" which do not meet the reliability requirements of N.J.R.E. 703. In this case, Deputy Medlar's testimony does reach an ultimate conclusion after ruling out all other possibilities. Accordingly, Deputy Medlar can testify to his opinion on the cause of the fire. Such an opinion on causation, reached through the process of elimination, is not a net opinion.
II. Is State Expert Houck's Opinion that the Dog's Alert Indicates the Presence of an Accelerant an Opinion Reached under a Generally Accepted Scientific Theory and Therefore Admissible?
To determine whether an expert's opinion is based on a generally accepted scientific theory, courts examine expert testimony, scholarly literature on the subject, and judicial recognition of the theory on which the expert relies. State v. Moore, supra, 188 N.J. at 206-07, 902 A.2d 1212. While the reliability of trailing dogs has been widely recognized and generally accepted by the courts, with thirty-two states approving of admission of trailing by bloodhounds, courts have been much less receptive to accelerant-detection canine "alerts" not confirmed by laboratory analysis. See State v. Buller, 517 N.W.2d 711, 712-14 (Iowa 1994) (holding bloodhound trailing reliable); compare, e.g., People v. Acri, 277 Ill.App.3d. 1030, 1034, 214 Ill.Dec. 761, 662 N.E.2d 115 (1996) (holding uncorroborated alerts are not generally accepted). What follows is a review of the expert testimony, authoritative literature, and judicial opinions that this court has considered in evaluating the reliability of this evidence.
1) Expert Testimony
The experts in this case disagree as to the significance of an accelerant-detection canine alert, but agree that the standard texts on which they rely require laboratory confirmation. In this case, the laboratory samples collected based on the dog alerting to particular locations at the fire scene all came back negative for the presence of an accelerant. Deputies Medlar and Houck agreed that the texts upon which they rely call for laboratory confirmation, but instructed the court that between the industry manuals and the dog they would both "trust the dog."
More than blind trust is necessary, however, to reach the reliability level required of scientific expert testimony. Deputy Houck testified that no spreadsheets exist tracking how many alerts of Blondie have been confirmed by the laboratory, thus no record exists by which to gauge the reliability of Blondie's nose. This is compounded by the fact that substantial portions of the contents of defendant's home would qualify as pyrolysis products, which, as discussed in the next section, can cause a canine to give a false alert. This court cannot rely on the record that the experts have made in this case for admission of accelerant detection canine testimony under the heightened scrutiny as to reliability required in criminal trials by Windmere, 105 N.J. at 378, 522 A.2d 405 and Cavallo, 88 N.J. at 516-26, 443 A.2d 1020.
2) Scientific and Legal Writings in the Field
There is substantial scientific agreement that unconfirmed canine alerts are not reliable in the absence of laboratory confirmation, largely due to the fact that canines cannot meaningfully discriminate between background pyrolysis products and accelerants. This is the position of defendant, while the State takes the position that fire investigators can rely on canine alerts in the absence of laboratory confirmation if the totality of the other circumstances supports their conclusion.
Significantly, at the hearing in this matter, Deputy Medlar agreed that he relies on the National Fire Protection Association (NFPA) 921 guide and John D. DeHaan, Kirk's Fire Investigation (6th Ed. 2006) treatise. The NFPA 921 guide reads in pertinent part:
[1] 16.5.4.7.1 In order for the presence or absence of an ignitable liquid to be scientifically confirmed in a sample, that sample should be analyzed in a laboratory. . . . Any canine alert not confirmed by laboratory analysis should not be considered validated.
[2] 16.5.4.7.2 Research has shown that canines have responded or have been alerted to pyrolysis products that are not produced by an ignitable liquid and have not always when an ignitable liquid accelerant was known to be present.
[3] 16.5.4.7.3 Specifically, the ability to distinguish between ignitable liquids and background materials is even more important than sensitivity for detection of any ignitable liquids or residues. Unlike explosive- or drug-detecting dogs, these canines are trained to detect substances that are common to our everyday environment. . . . [M]erely detecting [traceable] quantities [of these substances] is of limited evidential value.
[4] 16.5.4.7.5 The proper objective of the use of canine/handler teams is to assist with the selection of samples that have a higher probability of laboratory confirmation.
[5] 16.5.4.7.6 Canine ignitable liquid detection should be used in conjunction with, and not in place of, the other fire investigation and analysis methods described in this guide.
[(emphasis added).]
Kirk's Fire Investigation treatise reads in pertinent part:
"The canine cannot discriminate between the residues of carpet glue used to repair a section of carpet from the residues of isoparaffinic solvent used as an accelerant. Despite the advice of the IAAI Forensic Science Committee and the NFPA Technical Committee on Fire Investigations (NFPA 921) a number of courts have admitted evidence of canine alerts as indications of the presence of ignitable liquid accelerants in the absence of laboratory identifications. Studies have shown that even the best-trained canine teams cannot discriminate all possible pyrolysis products and background contaminants from accelerant and will alert to a percentage of targets that really do not contain ignitable liquid accelerants. Without a specific verifiable identification of just what is present [by a lab], the investigator cannot decide the significance of such positive alerts."
Id. at 543. (emphasis added).
Further, this court has reviewed some of the scholarly literature published in the Journal of Forensic Science, which is in accord with the NFPA 921 guide and Kirk's Fire Investigation. See, Katz and Midkiff, Unconfirmed Canine Accelerant Detection: A Reliability Issue In Court, 43 J. FORENSIC SCI. (Mar. 1998) (noting the disagreement of courts and commentators on the sensitivity of canine detection of accelerants as compared with laboratory analysis); Tindall and Lothridge, An Evaluation of 42 Accelerant Detection Canine Teams, 40 J. FORENSIC SCI. (July 1995) (concluding on the basis of a Finellas County Forensic Laboratory Study that the accuracy and reliability of accelerant detection canines varies and identifying the significant possibility of false negatives from pyrolysis products); Broadus, Brock, Dabdoub, Griffith, Kurz, Schultz and Sparks, Effect of Background Interference on Accelerant Detection by Canines, 41 J. FORENSIC SCI. (September 1996) (concluding from a study of 34 accelerant detection canines that false alerts from burnt carpeting material, aerosol cans and other pyrloysis products require laboratory confirmation to ensure reliability).
3) Judicial Opinions
Most courts have held that uncorroborated canine alerts are novel scientific evidence, not generally accepted in the scientific community of arson investigators. See State v. Schultz, 58 P.3d 879 (Utah App.2002) (uncorroborated alert is novel scientific evidence); People v. Acri, 277 Ill.App.3d 1030, 1034, 214 Ill.Dec. 761, 662 N.E.2d 115 (1996) (uncorroborated alerts are not generally accepted); Carr v. State, 267 Ga. 701, 704, 482 S.E.2d 314 (1997) (reliability of uncorroborated canine alerts is questionable); Farm Bureau Mut. Ins. Co. of Ark., Inc. v. Foote, 341 Ark. 105, 117, 14 S.W.3d 512 (2000) (testimony that accelerant detection canine is more reliable than laboratory equipment is without scientific validity); but see Fones v. State, 765 So.2d 849, 850 (Fla.App.2000) (permitting testimony of accelerant detection despite absence of lab results); Reisch v. State, 1993 WL 227264, *2, 1993 Del.Lexis 229 at *4-6 (1993) (affirming admitting accelerant detection testimony in absence of lab tests paired with testimony that the dog was better than 50% reliable); State v. Buller, 517 N.W.2d 711, 712-14 (Iowa 1994) (requiring accuracy foundation be laid and making determination of admissibility subject to N.J.R.E. 403 balancing).
In this matter, evidence uncorroborated by laboratory testing of the accelerant-detection canine's alert to certain locations at the fire scene should be barred under the Frye "general acceptance" standard. The scientific theory at issue that a dog's nose is more accurate than laboratory equipment is simply not supported by experts on fire causation, by scientific literature on the subject, or by judicial opinions. Such a tenuous scientific foundation must be subjected to intense scrutiny in a criminal trial where the liberty interests of the accused are at stake. See Windmere, 105 N.J. at 378, 522 A.2d 405 and Cavallo, 88 N.J. at 516-26, 443 A.2d 1020.
The State contends, alternatively, that, at a minimum, it should be permitted to present evidence to the jury of the complete fire investigation in this matter, including evidence of the dog alerting at certain particular locations at the fire scene. This would then demonstrate that fire officials were not acting randomly or otherwise unreasonably in selecting certain samples to send to the lab for analysis as to the presence of possible accelerants. Further, the State contends that their fire causation expert, Deputy Medlar, should be permitted to opine that based upon the combination of the alerting dog and the entirety of the other information gathered as a result of his investigation, the fire was, indeed, the result of arson. The court could then give a limiting instruction to the jury to the effect that they should not consider the fact that the canine alerted to particular spots at the fire scene to conclude that an accelerant was used or that the fire was arson.
In this regard, the court considers the applicability of N.J.R.E. 403. Under this rule, the court may exclude relevant evidence "if its probative value is substantially outweighed by the risk of undue prejudice, confusion of the issues, or misleading the jury." Here, given the lack of reliability of unconfirmed canine alerts under the Frye standard discussed above, the probative value of the canine alert is minimal. In fact, evidence of a dog trained to detect fire accelerants alerting in a particular area carries a substantial risk of confusing the issues and misleading the jury. The prejudicial effect of such testimony is great. A jury could infer that an accelerant was used in that spot, disregarding the lack of laboratory confirmation, and wrongly conclude that this particular evidence is indicative of arson and by extension defendant's guilt. Such testimony tends to go to the heart of the issue in this particular trial whether the fire was started accidentally or purposely. Thus, the prejudicial effect substantially outweighs the probative value of this testimony, and the court is not satisfied that a limiting instruction would be sufficient to cure the problem.
Accordingly, since an uncorroborated alert by an accelerant-detection canine cannot form the basis of an opinion on fire causation, Deputy Houck's entire testimony is irrelevant and shall be excluded. Deputy Medlar's testimony shall be limited so as not to include any mention of the accelerant-detection canine. Neither of the attorneys shall elicit testimony about an accelerant-detection canine for any purpose throughout the course of the trial. The court notes, however, that by stipulation, the jury shall be advised that Deputy Medlar did not select randomly the samples to be submitted for laboratory analysis, but followed a sampling methodology recognized in his field.
III. Is State Expert Medlar's Opinion that an Accelerant "Possibly" Was Used to Set the Fire an Inadmissible "Net Opinion"?
The defense claims that the portion of Deputy Medlar's opinion stating the fire "possibly" was aided by an accelerant is an impermissible "net opinion" which runs afoul of N.J.R.E. 703. The State contends that the totality of the circumstances (burn patterns, interviews, speed with which the fire spread, etc.) supports a conclusion that an accelerant possibly was used.
Notably absent from Deputy Medlar's testimony were any ascertainable methodology or scientific process by which he reached this opinion. Also lacking was any indication of the probability of accelerant use, or even a commitment that an accelerant was most likely used. The ring of the testimony was that of speculation rather than science.
An expert must state the "why and wherefore" of his opinion to satisfy N.J.R.E. 703. Rosenberg v. Tavorath, 352 N.J.Super. 385, 401, 800 A.2d 216 (App. Div.2002). When opining on the cause of an event, an expert must state his opinion in terms of probabilities. See Scully v. Fitzgerald, 179 N.J. 114, 128, 843 A.2d 1110 (2004); see also State v. Harvey 121 N.J. 407, 581 A.2d 483 (1990). Here, the expert must opine that the purported cause was "most likely" the cause or was the "probable cause" of the fire. See Scully, supra, 179 N.J. at 128, 843 A.2d 1110. Recently, the Supreme Court of New Jersey excluded a fire chief's opinion that it was his "best guess" that a discarded cigarette started the fire in question. Ibid. It is well settled that opinions which amount to guesswork, because they fail to give probabilities, are inadmissible under N.J.R.E. 703. See Pelose v. Green, 222 N.J.Super. 545, 549-51, 537 A.2d 745 (App. Div.), certif. den. 111 N.J. 610, 546 A.2d 530 (1988) (guess or conjecture impermissible); Sheerr v. Evesham Tp., 184 N.J.Super. 11, 445 A.2d 46 (Law Div.1982) (educated guess insufficient); Princeton Research Lands v. Upper Freehold Tp., 4 N.J.Tax 402, 407 (1982) ("gut feelings" are not evidentiary support).
Here, Deputy Medlar testified that his opinion that an accelerant aided the fire in question was based on the fact that the canine alerted on scene. Given the exclusion of the canine's alert elsewhere in this opinion, Deputy Medlar's opinion as to the presence of an accelerant is based entirely on circumstantial evidence not normally relied upon by experts in his field and is not stated in terms of probabilities as required. Thus, that portion of his opinion at to the "possible" use of an accelerant must be excluded as an impermissible net opinion.
CONCLUSION
Expert Houck's opinion that an accelerant was used to start the fire in this case is inadmissible, because the canine alert on which he relies is unreliable in the absence of laboratory confirmation. Further, any reference to the accelerant-detection canine "alerting" to particular locations at the fire scene is barred under N.J.R.E. 403, in that the risk of undue prejudice substantially outweighs its probative value. Expert Medlar's opinion that an accelerant "possibly" was used to start the fire is an inadmissible net opinion because he fails to commit to this conclusion and rests his opinion on no scientific foundation. However, expert Medlar's opinion that the fire was caused by arson (as opposed to an accidental cause) is admissible, as he utilizes a process-of-elimination technique that is generally accepted in the field of fire investigation.