SCIENCE IN THE LAW FORENSIC SCIENCE ISSUES

By

David L. Faigman Uni t·C'rsity of' Cnlifnmia

Hasting,; Colll'!JI' of the Law

David H. Kaye Arizona Stale Ull it:ersit y

College of'Laic

Michael J. Saks Arizona Stole Uniuersity

Colle ge of'La u:

Joseph Sanders Uniuersity of Houston

Law Cell/e r

AMERICAN CASEBOOK SERIES ®

~ .. WEST GROUP

A THOMSON COMPANY

CHAPTER 7

FIRES , ARSONS AND EXPLOSIONS

Table of Sections

A. LEGAL ISSUES Sec. 7- 1.0 The Legal Response to Expert Testimony on Fires and Explosions

7-1. l Introduction . 7-1.2 Before Kumho Tire: Ase Fire Experts Subject to Daubert Scrutiny? 7-1. 3 Other Aspects of Admissibility and Exclusion of Fire Experts.

7- 1.3.1 The Qualifications of the Expert. 7-1.3.2 Novelty. 7-1.3.3 Basis in Experience versus in Empirically Tested Knowl-

edge. 7-1.4 Accelerant-Detecting Can ines. 7-1.5 On Whom Can Fi.re Experts Rely for the Data on Which They Base

Their Conclusions? 7- 1.6 Who Bear s the Burden of Proof in a Daubert Hear ing? 7-1. 7 l\'atur e of Scrutin y Following Ku mho Tir e. 7-1. 8 State Cases.

7-2 .0 The Scientific Basis of Eiq:ien Tes timony on Fi.res. Arsons , and Explosions 'i-2 .1 lnt roductorv Discussio11 of the Science.

7-2. 1.1 F ield Investigations . l l l Test Burns. [2] Accelerant Detecting Canines. [31 Sniffer s.

7-2. 1.2 Laborato n · Analvsis. I 11 Classificati; n of Petroleum Products. [2] Ident ification (Individu alizat ion ) of Petroleum Prod ­

ucts. 7- 2.1. 3 Sources .

I l] Authoritative Pub licatio ns. [21 Periodical Literature .

7- 2.2 Areas of Scientific Agreement and Disagreement. 7-2.2 .1 FieJd Investigations.

Il l Th e Behav ior of Fire. [2] Acciden tal Fires. [3] Electrical Activity. [4] Cause and Effect . [5] Black Holes. [6] "Melt~d" Steel. [7] Crazed Glass. [8] Concrete Spalling. [9] Colors of Smoke and Fire. flO] Compute r Modeling . [ 11] Fatal Fires. (12] Explos ions . (13] Smoke Detector s. [14] Stolen Autos Recovered Burned. [15] Presumption of Accidental Cause. [16] The " Negative Corpus". I 17] Certainty of Opinions. l 18] Conflicting Opinions.

338

Sec. 7-1.1 FIRES, ARSONS AND EXPLOSIONS 339

_Sec.. 7-2.2.2 Laboratory Analysis.

[1] Standard Methods of Sample Preparation. {2] Classification of Ignitable Liquids.

. [3] Detection of Explosives. 7-2.3 Future Directions.

Appendix-Glossary.

Westlaw Electronic Research

See Westlaw Electronic Research Guide preceding the Summary of Contents.

A. LEGAL ISSUES

§ 7-1.0 THE LEGAL RESPONSE TO EXPERT TESTIMONY ON FIRES AND EXPLOSIONS

§ 7-1.1 Introduction Fire and explosion investigation consists of a highly varied mixture of

methods, techniques, and principles. Consequently, there is not just one actual or potential body of relevant scientific research on which experts may depend, but many, some of which are more sound and others which are less sound. 1 Fire investigators can be found who rely on such tools as electronic sniffers, accelerant detecting canines, and gas chromatography; examination of electrical arc beads, metallurgical examination, burn patterns, crazed glass, concrete spalling; and consideration of reports of the color of the smoke and the fire, blood cltemistry, and other indications from human remains found at the fire scene. Some of these clues are derived from sound science. Others are nothing more than a set of more or less shared beliefs that may or may not be true. 2 An opinion is then reached by these clues being processed through each investigator's personal experience, beliefs and assumptions-in addition to or instead of any well tested model for analyzing fire evidence.3

Such profusion presents ~urts with a dilemma. In considering admissibil­ity, should members of a field of expertise be required to elucidate each of the components on which they rely, and to establish the validity of each compo­nent? Or should the courts make a general, global, judgment about the field, and trust the expert, under questioning by counsel, to spell out the details? Historically, the courts generally followed the latter strategy, and experts, once permitted to testify, were given wide latitude to offer opinions based on whatever the expert thought reasonable.! The task-at-hand analysis called for

§ 7-1.0 1. This state of affairs cont1·asts with most

of the kinds of expertise with which the chap­ters of this book are concerned. But it is not unique. For example, the testimony of medical examiners or accident reconstructionists de­pends on the witness bringing together a po­tentially wide range of principles or assump­tions on which an opinion is based.

2. As the field begins to test its beliefs, some are confirmed and others are found to be false. See infra § 2.1.l[lJ.

3. "As circumstantial proof of the incendi­ary origin of a fire, arson investigators rely most heavily upon a rather amorphous group of so-called burn or arson indicators." ANDRE A. MoENSSENS ET AL., Scil:NTmc EVIDENCE IN CIVIL AND CRlMINAL CASES 416 (4th ed. 1995). .

4. Thus, it is not possible to list the pivotal or leading case by which each jurisdiction per­mitted each component of the expertise. And after Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2H 469 (1993) (hereafter, Daubert), and Kumho

..

340 FIRES, ARSONS AND EXPLOSIONS Ch. 7

by Daubert and elaborated upon by Kumho Tire would appear to have brought those carefree days to an end.5 Task-at-hand analysis requires a court to identify the precise knowledge and skills invoked by an expert witness and to focus its gatekeeping responsibilities on those particular expert claims. The approach elaborated by Kumho Tire follows from the logic of Daubert. Daubert requires a finding of validity of the basis for an expert witness's opinion. That gatekeeping requirement would be defeated if multiple techniques were al­lowed to hide behind a global claim of expertise. In other words, Daubert appears to require that an expertise be unpacked, so that the court can permit those methods and principles it is persuaded are valid, and only those, to be offered to the factfinder. Kumho Tire makes that requirement explicit.

Expert testimony on the causes of fires went through a period of initial judicial resistance as being an inappropriate subject matter for expert opin­ions.6 Gradually, the courts began to reverse themselves, allowing fire investi­gators and others to offer expert testimony on causes of fires.7 The courts came to focus more on the training and experience of the proffered expert rather than on the validity of the proffered expertise, their assumption being that valid knowledge existed and the only issue was whether the witness possessed it in sufficient quantity. Of course, the line that divides enough training and experience 8 from not enough 9 may not be a very bright one. But deciding on an expert witness's "qualifications" is easy compared to evaluat­ing the validity lor lack of validity) of the knowledge held by the field represented by the witness. 10

As this chapter explains, some of the scientific predicates long relied upon by investigators and admitted by courts were later found by research on those

Tire Co. v. Carmichael. 526 U.S. 137, 119 S.Ct. 1167, 143 L.Ed.2d 238 ( 1999) (hereafter, Kum­ho Tire) the case law admitting fire and explo­sion experts under a notion of broad, global expertise has been rendered irrelevant to fu­ture fire admission decisions, at least in Dau­bert/Kumho jurisdictions.

5. See ScIE:-.cE 1:s THE LAw, STA.'-DARDS, STATIS­

TICS A.'-D REsE..utCH Issn:s, Chapter 1, § 3.4. 6. State v. Watson, 65 Me. 74 (1876) (testi­

mony of a fireman on the spread of fire and aspects of fire behavior excluded as within the scope of common experience); Neal v. Missouri Pac. Ry. Co., 98 Neb. 460, 153 N.W. 492 (Neb. 1915) (fireman's opinion as to where a fire started excluded as invading the province of the jury); People v. Grutz, 212 N.Y. 72, 105 N.E. 843 (N.Y. 1914) (an assistant fire marshal should not have been permitted to express his opinion about the origin of a fire because the physical facts could be readily understood by the jury when properly described); Sawyer v. State, 100 Fla. 1603, 132 So. 188 (Fla. 1931) (witness in arson case, in this instance the chief of a fire department, may not as a general rule give an opinion whether fire was of incen­diary origin because this is "a question for the jury to determine, and upon which they can usually form their own opinion without any need of expert advice"); Beneks v. State, 208 Ind. 317, 196 N.E. 73 llnd.1935) (expert wit­ness cannot give his opinion as to origin of fire

because jury can draw conc:lusions from ob­servable facts that can be testified to by ex­pert).

7. An example of testimony by an "other": Brown v. Eakins, 220 Or. 122, 348 P.2d 1116 (Or.1960) (an electrician who regularly investi­gated fires was qualified as an expert on the causes of fires, despite lack of any formal fire investigation training).

8. Billings v. State, 503 S.W.2d 5i (Mo.App. 1973) (fireman with 4.5 years experience and investigation of two dozen fires found to be qualified); State v. Wilbur, 115 R.I. 7, 339 A.2d 730 (R.I. 1975) (seven years experience as fire­fighter, several months arson investigation training, and three years as a fire inspector found qualified).

9. Sperow v. Carter, 8 Pa. D. & C.2d 635 (1957) (a part time fire chief who had been fighting fires for 20 years found not qualified); State v. Barnett, 480 A.2d 791 (Me.1984) (a fire chief with 18 years experience fighting fires, extensive formal training in fire fighting, and fire investigation experience as a consul­tant t.o the Air Force and several large corpora­tions found not qualified).

10. For example, consider an astrologer with a successful practice for 25 years, exten­sive education and training, and numerous cer­tifications and awards.

Sec. '7-1.2 FIRES. ARSONS AND EXPLOSIONS 341

investigative methods to be less valid than the experts or the courts thought them to be. 11 In some instances, principles of fire investigation that lacked a sound scientific basis led to convictions for arson and homicide by araon that later were vacated. 12

Consider the courts' responses to some specific elements of fire investiga­tion lore. Concrete spalling has been allowed as conclusive evidence that a fire was started by incendiary means, typically with little or no question raised about the validity of the asserted relationship between spalling and arson. 13

Crazed glass has been relied upon as "indicating a fast spreading fire."" Expert testimony about burn patterns has been admitted as providing evi­dence that a fire was of incendiary origin. 15 These and other beliefs of fire investigators have since been called into doubt by empirical research. M Simi­larly, some courts recently have been admitting dog "alerts" as evidence of arson without requiring hard evidence of the accuracy of the canines, and in particular without considering empirical evidence of the level of false positive errors in the method-data which have given great pause to many arson evidence professionals. i;

These examples might make clear why courts could profitably unpack this broad area of forensic expertise into its component sub-areas and separately consider the validity of each. Judicial skepticism is expected to return to some of these areas.

§ 7-1.2 Before Kumlao Tire: Are Fire kperts Subject to Daubert Scrutiny?

In deciding whether asserted fire and arson expertise had to be tested under Daubert, some courts made a distinction between whether the expert was offered as. or claimed to be, a " scientific" expert, in which case Daubert was applied , or whether the expert was sailing under the flag of " technical or other" expertise, in which case the expert was t.ested by a lesser standard.

The most interesting of these cases is Michi1an Millers Mutual Insuronce Corp. v. Benfield.11 The Millers bad sought a declaratory judgment against its

11. S. infra § 2.1.1(1] . U. See, e.g., Staie .... Knapp, No. CR78779

(Superior Court of Arizona, Maricopa County, Feb. 11, 1987); State v. Girdler , No. 9809 (Su­perior Court of Arisona. Maricopa County, Jan. S. 1991); ,_ alao, Girdler .... Dall, 869 F .Bupp. 1279 (D.Am.1994 ).

13. S., e.g., Reed .., . Allataie Ina. Co., 376 So.2d 1300 (La.App. 2d Cir .1979); s.curity Ina. Co. of Hartford v. Duclda, Inc.. 148 F .2d 273 (5th Cir.1981); State .... Danakin. 122 N.H. 817, 451 A.2d 396 (N.H.1982); LeForp v. Nation­wide Mu1. Fire Ina. Co., 82 Omo App.3d 692, 612 N.E .2d 1318 (Ohio App.1911); American Mfrs. Mut. Ina. Co. v. General Motors Corp., 582 So.2d 934 (I.Upp. 2d Cir.1991) (unlike tbe other eaaa in thia note, the apert drew a neptive inference: becauae thenl wu no spall­ing, the fire at iuue wu thoupl not to have beat anon ). &t ,. infro I 2.2.1(8).

14. McReynoldl .... Chmvkee IDS. Co., 815 S.W.2d 208 (Tenn.App.1991) . Blu see infra I 2.2.1(7) .

16. People V. Calvin Thomas, 65 Cal.2d 698, 56 Cal.Bptr. 306,423 P.2d 233 (Cal.1967); People V. Swain, 200 Cal.App.2d a«, 19 Cal . Rptr . 403 (1962); Commonwealth v. W'111188ki, 214 Pa.Super. 397, 2.57 A.2d 62' (Pa.Super. 19691; State v. Reil P . KeU.,, 901 S.W..211 193 (Mo.App. W.D.1995); Stai. .... Sw_amon. 19915 WL 2388158 CMinn.App.1996); Staie v. Bouchil­lion. 1996 WL 7154-44 (Tenn.Crim.App.1995 ); Staie .... Bernier, 1995 WL 70337 (CAan.Su­per .19915); State v. Haggood, 36 Coon.App. 7153, 653 A.2d 216 (Conn.App.1996); Adami•- Ten­~ Farmers Mui . IDS. Co., 898 S.W..211 216 (Tenn.App.1994). But ,ee infra ff 2.1.l(l) and 2.2.1(1] .

11. See infra I 2.1 .1(1).

17. Ser infra§ 1.4.

18. The opinion of the United Staam Dis­trict Court for the Middle Diltric:t of Florida, No. 93-1283-CIV-T-l 7 A, is unreported.

342 FIRES, ARSONS AND EXPLOSIONS Ch. 7

insured, Benfield, precluding payment of fire insurance benefits on several grounds, notably that the fire had been intentionally set. The district judge excluded the testimony of the insurance company's expert, finding that his proffered testimony did not meet Daubert's reliability criteria, adding that the testimony did not meet the requirements of Frye19 either.

Michigan Millers appealed to the 11th Circuit,2° arguing that its expert should have been admitted-not because the expertise was scientific, but because it was not scientific, therefore should not have been subject to Daubert, and should instead be admissible on a lesser standard as "experi­ence-based" expertise.21 Appellants cited several cases where testimony had been admitted over Daubert-based objections.

The International Association of Arson Investigators (IAAI) submitted an amicus brief urging the admission of the expert testimony. The brief argued that a fire and arson expert who is qualified by conventional criteria and who is not presenting any novel scientific evidence should not have to pass Daubert scrutiny. That argument was based on the clearly mistaken (but surprisingly common) belief that Daubert was narrowly focused on novel scientific tech­niques and methodologies. Since fire investigation is not a novel scientific technique, the amicus argued, Daubert should not apply.22 The IAAI brief also argued that fire and arson investigation is neither a science nor strictly based on science, but its asserted validity rests instead on training and experience, and for that reason as well Daubert was inapplicable-a view later rejected unanimously in Kumho Tire. 'l3

In addition, a member21 of the IAAI, upon seeing the organization's amicus brief, wrote and submitted his own amicus brief, in which he argued that although the field once was unscientific, it has been making important strides in recent years, and that judicial toleration of unscientific arson investigation certainly would not inspire the field to continue to develop itself as an empirically grounded science.25

The Court of Appeals applied an analysis following its decision in the case of Carmichael u. Samyang Tire, lnc. 26 It concluded that because the fire investigator in question had held himself out as an expert in fire science, Daubert criteria did indeed apply to the issue of the admissibility of his testimony, and the Circuit Court upheld the District Court's exclusion of the testimony. The expert had performed no tests, taken no samples, and could not adequately explain how he had reached his conclusion. The court cited

19. Frye v. United States, 293 F. 1013 (D.C.Cir.1923).

20. 140 F.3d 915 (11th Cir.1998). 21. "Millers argues that Carmichael [v. Sa­

myang Tire, Inc., 131 F.3d 1433 (11th Cir. 1997)] made clear that the Daubert criteria apply only to scientific testimony, and the tes­timony of their expert was not based on scien­tific principles but rather was based on his years of experience, and on his skill and experi­ence-based observations." Id. at 920.

22. Cf. Daubert, 509 U.S. at 593 n. 11. 23. The Supreme Court's view implies that

these questions could and should be put to those who claim a valid expertise on the basis of training and experience: What knowledge

was imparted to you in your training? How can we know if it was valid? What knowledge did you acquire through your experience? How can we know if it was valid? (After all, if training and experience without further inquiry estab­lish expertise, then astrology would be admissi­ble.)

24. John Lentini, author of the scientific status portion of this chapter.

25. Both the IAAI brief and Lentini's brief were placed on the IAAI homepage, at fire­investigators.orgiwwwboard.

26. 131 F.3d 1433 (11th Cir.1997), later to be known as Kumho Tire u. Carmichael.

Sec. 7-1.2 FIRES, ARSONS AND EXPLOSIONS 343 /

General Electric v. Joiner for the propositi~n that courts are not required to admit opinions based on nothing more than the ipse dixit of the expert.27 On the other hand, it held that the "experience-based" testimony of a different expert, the local fire investigator employed by Sarasota County, could be admitted for the jury's consideration. Because the fire in question was an obvious arson fire, however, the Court set aside the trial court's directed verdict and remanded the case for a new trial.

The case ultimately was settled prior to a second trial, but the repercus­sions of the case continued to reverberate through the fire investigation community. AJ3 a result of the 11th Circuit's reliance on the witness's self­characterization as either a scientific witness or an experience-based witness, some insurance company attorneys began counseling fire investigators to identify themselves as "experience-based" experts, in an effort to avoid scrutiny under Daubert.

Still trying to avoid judicial scrutiny of its asserted expertise, the IAAI again submitted an amicus brief when Carmichael v. Samyang Tire (of which Benfield had been progeny), now known as Kumho Tire Co., Ltd. v. Carmicha­el28 was heard by the United States Supreme Court. Joining an amicus brief filed by the International AJ3sociation of Chiefs of Police, Mothers Against Drunk Driving, the National District Attorneys AJ3sociation, and numerous other organizations, their brief in Kumho Tire argued that the trial judge's gatekeeping responsibility should not serve as an obstacle to testimony based on "technical" or "specialized" knowledge or it would threaten much of the expert evidence that law enforcement organizations offer.29

The Supreme Court in Kumho Tire, of course, unanimously rejected such arguments. Had the decision gone the other way, and "technical and other specialized" experts were exempted from scrutiny under Daubert, their expe­rience, intuition and ipse dixit would have been transformed from shortcom­ings into the very basis of their expertise. As a result of Kumho Tire, Daubert­based objections to fire investigation testimony not based on good science have increased. 30 Few of the rulings on Daubert challenges to fire investigators have reached the appellate level, but decisions affecting admissibility are quickly and widely circulated among fire investigators through trade publica­tions and electronic bulletin boards.

In still another repercussion of the Benfield decision, the National Fire Protection AJ3sociation Technical Committee on Fire Investigations, the com­mittee responsible for the preparation and maintenance of NFPA 921, Guide for Fire and Explosion Investigations, received several public proposals to eliminate from the Guide any reference to science or the scientific method. (This despite the fact that Kumho Tire had rendered that means of evading scrutiny unavailing.) The proponents of this change wanted to substitute the

27. General Elec. Co. v. Joiner, 522 U.S. 136, 118 S.Ct. 512, 139 L.Ed.2d 508 !1997).

28. 526 U.S. 137, 119 S.Ct. 1167, 143 L.Ed.2d 238 (1999).

29. Brief Amici Curiae of Americans for Effective Law Enforcement, Inc.; Criminal Jus­tice Legal Foundation; Grand Lodge of Frater­nal Order of Police; International Association of Arson Investigators; International Associa-

tion of Chiefs of Police; Mothers Against Drunk Driving; National Association of Police Organizations, Inc.; National District Attor­neys Association; and National Sheriffs; Associ­ation; and Police Law Institute, submitted to the United States Supreme Court in Kumho Tire Company, Ltd. v. Carmichael (October 19, 1998).

30. See infra § 1. 7.

344 FIRES, ARSONS AND EXPLOSIONS Ch. 7

words "systematic approach" for "scientific method." The Technical Commit­tee rejected these proposals, while attempting to deal with the misperception that a scientific fire investigation requires a complete reconstruction of the fire. The 2001 edition of NFPA 921 makes reference to "cognitive" testing as well as experimental testing as a means of testing a hypothesis within the structure of the scientific method. Cognitive testing, as used in this context, means mentally comparing all of the collected data with the proposed hypoth­esis, and rejecting the hypothesis if it cannot account for all of the data.

The Benfield court was not alone in the question it pondered, as fire investigators sought shelter from Daubert. Before Kumho Tire unanimously resolved the confusion about whether Daubert's gatekeeping requirement applied only to the "scientific" prong of Rule 702, or whether it applied to all kinds of expert opinion evidence, a number of courts struggled with the question in the context of fire causation. Typically, these cases relied on an expert's "experience" to substitute for systematic empirical or theoretical knowledge on the facts in dispute, and allowed the asserted expert to testify. 31

After Kumho Tire and after the recent revision to Fed. R. Evid. 702 it is doubtful that the cases cited in the margin could be regarded as following the requirements of Federal evidence law.

§ 7-1.3 Other Aspects of Admissibility and Exclusion of Fire Ex­perts

A number of other aspects of the problem of what is required for proffered fire expert testimony to be admitted, some of them hinted at in the Benfield discussion, supra, are considered next.

§ 7-1.3.1 The Qualifications of the Expert The Supreme Court's admissibility cases make clear that the admissibility

of claimed expertise is distinct from the qualifications of the expert. Qualifica­tions of fire and arson experts vary considerably, and the courts have done little to distinguish among the various kinds of proffered experts, ranging from fire department employees with minimal science training to bachelors level engineers to chemists with doctorates. Suitable qualifications are neces­sary but not sufficient for the admission of the proffered expert testimony.32

31. Talkington v. Atria Reclamelucifers Fa­brieken BV (Cricket BVl, 152 F.3d 254 (4th Cir.1998) (holding Daubert inapplicable to ex­pert testimony about the cause of fire in a house, because the defendant acknowledged the testimony was not science and was not based on scientific principles or data, but was based instead on the expert's training and ex­perience); Polizzi Meats, Inc. v. Aetna Life & Casualty Co., 931 F.Supp. 328 (D.N.J.1996) (holding that the lack of any scientific or other systematic empirical basis for their testimony on the origin and cause of fire at insured's building did not preclude testimony of insur­er's expert witnesses); Patterson v. Conopco, Inc., 999 F.Supp. 1417 (D.Kan.1997) (holding it acceptable for an expert to "rely on his experi­ence," though he had never actually conducted any tests on whether human hair could sustain

a fire and, presumably, could point to no data on the question); Fireman's Fund Insurance Co. v. Xerox Corp., 30 F. Supp. 2d 823 (M.D.Pa.1998) (holding Daubert inapplicable and admitting an expert's opinion on how a fire started at a copy machine, concluding that the expert was "not relying on any particular methodology or technique," and that he had "reached his . . . conclusions by drawing upon general electrical engineering principles and his twenty-five years of experience investigat­ing electrical accidents."). For further discus­sion of experience-based expertise in fire cases, see infra § 7-1.3.3.

32. Cf. Allstate Ins. v. Maytag, 1999 WL 203349 (N.D.111.1999), discussed in more detail infra, a case which illustrates reliance on cre­dentials and no scrutiny of the basis of the proffered testimony.

Sec. 7-1.3.1 FIRES, ARSONS AND EXPLOSIONS 345

Kumho Tire went further, emphasizing the need to evaluate the precise "task at hand" -the factual issue which expertise is being offered to resolve. The implications of Kumho Tire suggests that the proffered expert must be qualified on precisely the expert issue at bar, rather than some general or global expertise in fire investigation.

Though fairly superficial scrutiny usually is involved in passing on qualifications, it may be worth noting that one can only be qualified as an expert in a subject on which an expertise exists. In other words, one cannot be an expert on something in which there is no expertise. Thus, where a technique (e.g., dog sniffing for accelerants) is being challenged, qualifications and admissibility of the expertise will be entangled.

One example of rather stringent scrutiny of qualifications, interwoven with scrutiny of doubtful foundations for the experts' proffered opinions, is Weisgram v. Marley Co.,33 where the Court of Appeals for the Eighth Circuit found the district court's wholesale admission of various types of fire causa­tion experts to constitute reversible error. The Court of Appeals judged all three experts to be "unqualified" and their opinion testimony "speculative." A city fire captain who investigated the fire in the home was held not qualified to give expert testimony as to whether or not the baseboard heater in the home malfunctioned, or how the heater might have ignited other objects, and his opinions were found to be without foundation. A fire investi­gator's testimony that the baseboard heater was defective and caused the fire was held to be unsupported by sufficient foundation because no studies had been conducted to test the investigator's theory: the Court found that instead it was based on pure speculation. Finally, the Court held that the expert testimony of a metallurgist that thermostat contacts on the baseboard heater were defectively designed was not supported by sufficient foundation because the metallurgist had little knowledge about this particular model heater or this type of heater. Moreover, upon excluding the testimony of these wit­nesses, the Court of Appeals directed that judgment be entered against the proponent of this expert evidence, which decision was appealed to the United States Supreme Court.

Specifically, the plaintiffs complained that it was "unfair" to direct a verdict against them without giving them a chance to procure admissible expert testimony. The Supreme Court rejected this argument: "Since Daubert, . . . parties relying on expert evidence have had notice of the exacting standards of reliability such evidence must meet. It is implausible to suggest, post-Daubert, that parties will initially present less than their best expert evidence in the expectation of a second chance should their first try fail. We therefore find unconvincing [the plaintiff's] fears that allowing courts of appeals to direct the entry of judgment for defendants will punish plaintiffs who could have shored up their cases by other means had they known their expert testimony would be found inadmissible. "34 On the specific procedural question presented, the Court held "that the authority of courts of appeals to direct the entry of judgment as a matter of law extends to cases in which, on

33. 169 F.3d 514 l8th Cir.1999). 34. 528 U.S. 440, 456, 120 S.Ct. 1011, 145 L.Ed.2d 958 !2000).

FIRES, ARSONS AND EXPLOSIONS Ch. 7

excision of testimony erroneously admitted, there remains insufficient evi­dence to support a jury's verdict.'' 35

§ 7-1.3.2 Novelty

Occasionally a court mistakenly treats Daubert as applying only to "novel" types of expertise, and in effect grandfathers in a non-novel yet questionable expertise by refusing to scrutinize it. The district court in Polizzi Meats, Inc. v. Aetna Life & Casualty Co.36 did so with remarkable vehemence considering that the court was misreading Daubert. The defendant insurer refused to pay the plaintiff's claims resulting from a fire that had destroyed its place of business. Aetna asserted that the plaintiffs had set the fire intentionally. In a partial summary judgment motion, the plaintiffs argued that Aetna's experts had not produced "scientific proof' of the cause of the fire, and therefore they should be barred from testifying. The district court held that this "astounding contention is based on a seriously flawed reading of . . . Daubert. . . . [ which] addressed the standards to be applied by a trial judge when faced with a proffer of expert scientific testimony based upon a novel theory or methodology. Nothing in Daubert suggests that trial judges should exclude otherwise relevant testimony of police and fire investigators on the issues of the origins and causes of fires."3; Apparently the district court overlooked the gloss on novelty given in Daubert, namely: "[W]e do not read the requirements of Rule 702 to apply specially or exclusively to unconven­tional evidence. Of course, well-established propositions are less likely to be challenged than those that are novel, and they are more handily defended. "38

An even more confused opinion is Jugle v. Volkswagen of America, lnc. 39

This case involved a young man who was burned to death in a Volkswagen Jetta. The defendant sought in limine to exclude the plaintiff's experts' testimony, but that motion was denied. One expert offered the opinion that the catalytic converter had caused a wax used on the floor pan to ignite, which in turn had caused plastic fuel lines to melt and burn. A second expert offered the opinion that the converter had ignited the fuel lines directly, even though this theory was contradicted by the first expert's data. The Jugle court held Daubert inapplicable to the expertise or to the opinions at issue in the case. "Because the opinions of Dr. Jacobson and Mr. Cole are not based on novel scientific techniques, the Court need not test their opinions against Daubert's four factors."40 The opinion went on to say that "[t]he Court must, however, still assess the reliability and fit of the proposed experts' opinions,"41 and for this proposition it quoted the language in Daubert which explained that Rule 702's gatekeeping requirements, explicated in Daubert, were not limited to novel or unconventional scientific evidence. Finally, the court found the methodology of the experts reliable, though it provided virtually no descrip­tion of it in the opinion!1

35. Id. at 457. 40. Id. at 580. 36. 931 F.Supp. 328 (D.N.J.1996). 41. Id. at 580. 37. Id. at 336. 42. Id. at 581. 38. Daubert n. 11, at 593. 39. 975 F.Supp. 576 (D.Vt.1997).

Sec. 7-1.3.3 FIRES, ARSONS AND EXPLOSIONS 347

§ 7-1.3.3 Basis in Experience versus in Empirically Tested Knowledge

It is clear, after Kumho Tire, that no expert witness of any kind may pass through the Daubert gate unless and until the court is properly satisfied that the testimony is based on valid principles. What is not yet clear is what criteria courts may or must use in making this assessment, especially since they will vary to a greater or lesser extent with the specie of expert evidence being offered.

In the area of fire and arson, as in other areas, courts have been presented with the dilemma of deciding whether expertise must be based on adequate empirical testing or whether a looser accumulation of individual "experience" is sufficient. One such case is Patterson v. Conopco, Inc. 43 This was a wrongful death case involving a woman who died from burns and smoke inhalation in a bathroom fire. The decedent's plaintiffs claim was that hair spray made by the defendant had caused this fatal accident. The trial court denied challenges to the expert testimony of a chemist and a fire investigator. The chemist's report concluded that the hair spray contained flammable materials that are easily ignited and will propagate a fire. The fire investiga­tor's testimony purported to explain how the accident could have occurred. The defendant argued that the chemist's conclusion depended on the assump­tion that human hair by itself could not sustain combustion, a "fact" based only on the chemist's "experience," not any systematic testing. The court held, nevertheless, that it was acceptable for an expert to "rely on his experience" and to take that more casually acquired knowledge into account in forming an opinion.u This decision may or may not have survived Kumho Tire's emphasis on the need of a court to evaluate the specific "task at hand."

Because Allstate Insurance Co. v. Maytag Corp . .is was decided after Kum­ho Tire, the magistrate judge had no doubt that Daubert applied to fire experts, yet he appears to have been content to rely on the credentials and experience of the experts, rather than scrutinizing with particularity the basis of their asserted knowledge, the content of the knowledge, and determining whether it was sufficiently sound that the opinions flowing from it would be dependable. The judge also did not find the objection that one expert's theory of causation was unsupported by any testing to raise any barriers to the witness offering his opinion on causation. On the other hand, the judge did draw some lines. Because the defendant's expert was a mechanical engineer with a great deal of experience with the cooktops which were the disputed source of a house fire, and not an expert in fire causation, the judge limited his testimony to the nature of the cooktop and why in his opinion it could not have been the source of the fire. In admitting the testimony of both experts, the magistrate judge ruled that they were "based on deductions from various known technical facts which appear to have at least a theoretical basis. "46 One might have expected the judge to follow the example of Justice Breyer in

43. Patterson v. Conopco. Inc., 999 F.Supp. 1417 (D.Kan.1997).

44. "It is true, as Conopco asserts, that Armstrong's conclusion was based in part on his stated opinion that human hair by itself (i.e. without an outside fuel source) will not sustain combustion and that this view in tum

was founded in large part on Armstrong's per­sonal experience. That fact alone, however, does not make it an illegitimate basis for con­sideration." Id. at 1420.

45. 1999 WL 203349 (N.D.Ill.1999).

46. Id. at 5.

348 FIRES, ARSONS AND EXPLOSIONS Ch. 7

Kumho Tire and state with some particularity what the underlying knowledge was, and why it was or was not sound.

This is an issue that undoubtedly will continue to be confronted. Courts would be well advised to unpack the claimed "experience" in order to discover what was learned from it and whether that something supports a valid and reliable expert opinion. As stated in the commentary to the recently revised Fed. R. Evid. 702, "the witness must explain how that experience leads to the conclusion reached, why that experience is a sufficient basis for the opinion, and how that experience is reliably applied to the facts. "4;

§ 7-1.4 Accelerant-Detecting Canines

Without requiring evidence of the ability of dogs in general to detect accelerants, or data concerning the accuracy of the particular do'§ooin question, some courts have allowed the dogs' handlers to testify concerning the pres­ence of accelerants at fire scenes based on the "alerts" of their dogs. This has occurred despite the newness of the technique and the applicability of Daubert in those jurisdictions. 48 Such research as has been conducted suggests canines used in this capacity are prone to making false positive errors. 49 Accordingly, the professional association of arson investigators has cautioned against reliance on canines for the detection of accelerants.50

Though most courts admitting canine accelerant detection evidence have done so without awareness of the research and professional association prohibition on the evidence, the U.S. Court of Appeals for the Second Circuit has gone to considerable lengths to twist the data and the cautions into unrecognizable form so as to uphold a district court's admission of dog sniff evidence. In United States v. Marji,51 the Court of Appeals evaluated the district court's admission of dog alert evidence and found no error.

47. Advisory Committee Notes, Amend­ments to Fed.R.Evid. 702 (effective December 1, 2000!.

48. Admission of evidence of dog alert evi­dence was upheld in: Reisch v. State, 628 A.2d 84 (Del.Supr.1993); State v. Buller, 517 N.W.2d 711 (Iowa 1994). Evidence from canine arson investigation was admitted in other cases, but the admissibility was not challenged. See e.g., Auto-Owners Insurance Company v. Ogden, 667 So.2d 743 (Ala.1995) (table) (aff'd; rehear­ing denied; all opinions withdrawn); State v. Bernier, 1995 WL 70337 (Conn.Super.1995); In re W.T.B., i71 So.2d 807 (La.App.2000).

49. The Illinois State Police Bureau of Fo­rensic Sciences conducted an experiment in which they placed known quantities of known substances in containers, and tested the sensi­tivity (how small a sample can be detected) and selectivity (distinguishing accelerants from other pyrolyzed substances) of both dogs and gas chromatographs or mass spectrometers. They found that while one dog did quite well, other dogs "were indicating on the pyrolyzed carpeting and foam padding samples, as well as on pine wood .... " They conclude that because dogs are "not very selective," "a positive alert must always be corroborated by the laborato-

ry." George Dabdoub et al., Accelerant Detec­tion Canines and the Laboratory, 1995 PRO­CEEDISGS OF THE ~!ERIC..\.-.: :\cADE.'dY OF FoREXSIC ScIENCES 19. See§ 2.1.1[2].

50. The International Association of Arson Investigators adopted an official position which stated that until there is sufficient research to confirm that canines actually can discriminate between real accelerants and the wide array of other ignitable compounds. "(a]ny alert or in­dication not confirmed by laboratory analysis must be considered a false positive . . . for the purpoees of origin and cause determination." They concluded: "If the forensic laboratory examination of the sample is negative for the presence of identif1&ble ignitable liquids, any positive indication by the canine of that sample must be deemed as not relet•ant." IAAI Foren­sic Science Committee Position on the Use of Accelerant Detection Canines, (Sept.1994). See infra § 2.1.1(2].

51. 158 F.3d 60, 63 (2d Cir.1998) ("Al­though the defendant cites some studies and a proposed amendment to the National Fire Pro­tection Association's Guide for Fire and Explo­sion Investigations to the effect that dog-sniff evidence is not always reliable, all that these

Sec. 7-1.5 FIRES, ARSONS AND EXPLOSIONS 349

The Court of Appeals seems to have presumed the soundness, and therefore the admissibility, of dog sniff evidence. The Court misconstrues the nature of error rate data when it states that dog alerts are "not always reliable," and it implies that the dogs usually are reliable. The Court essen­tially begs the empirical question to be decided. Reliability of a technique cannot be measured on an instan~by-instance basis but only in the aggre­gate, and the aggregate measure inevitably will show some level of accuracy (hits and correct rejections) and some level of inaccuracy (misses and false alarms). By the standard used by this panel, few if any techniques could ever be found so unreliable that they would have to be excluded. When the Circuit Court states that "all that these sources suggest is that special weight should not be assigned to dog-sniff evidence in the absence of any corroborating evidence," it misrepresents what those sources say and what they mean. The fire and arson field regards dog sniff evidence as merely a preliminary screening test that gives leads to an investigator but which ought not be offered as evidence unless and until it has been confirmed by laboratory testing. That is a far cry from cautioning against giving "special weight" "to dog-sniff evidence in the absence of any corroborating evidence." The Court transforms the field's advice that the evidence should be given no weight into a conclusion that it should be given no more than normal weight.

Finally, the Circuit Court concluded that "even if we were to assume arguendo that the district judge's decision to allow this expert testimony was erroneous" the error was harmless. The Circuit Court would have accom­plished the same result, while placing itself on ground more consistent with the empirical evidence on dog alerts, to have concluded that, although the district court erred by admitting the dog-sniff evidence without sufficient basis for believing in its dependability, the error was harmless in light of other evidence in the case on the question of the presence of accelerants.

In other cases, courts found the use of dogs to establish the presence of accelerants to be unreliable and therefore inadmissible.32

§ 7-1.5 On Whom Can Fire Experts Rely for the Data on Which They Base Their Conclusions?

The case-specific information on which fire and arson experts base their conclusions sometimes includes the observations or inferences of others. To what extent can a fire investigator rely on the statements and conclusions of others? In Westfield Insurance Co. v. Harris, 33 a fire marshal relied for most of his information upon the investigator hired by one of the parties, thereby malcing his own conclusions the product of someone else's investigation and

sources suggest is that special weight should not be assigned to dog-sniff evidence in the absence of any corroborating evidence. We con­clude that the trial judge did not abuse his broad discretion under Daubert in admitting the testimony. We note further that, even if we were to assume arguendo that the district judge's decision to allow this expert testimony was erroneous, there was substantial addition­al evidence offered at trial demonstrating that an accelerant was used by the defendant to start the fire.")

52. People v. Acri, 277 lli.App.3d 1030, 214 Ill.Dec. 761, 662 N.E.2d 115 (lli.App.1996); Carr v. State, 267 Ga. 701, 482 S.E.2d 314 (Ga.1997); Farm Bureau Mutual Insurance Company of Arkansas, Inc. v. Foote, 341 Ark. 105, 14 S.W.3d 512 (Ark.2000) <adopting Dau­bert as the test for admissibility of expert evi­dence in Arkansas, and fmding that dog sniff evidence for the detection of accelerants failed the test).

53. 134 F.3d 608 (4th Cir.1998).

350 FIRES, ARSONS AND EXPLOSIONS Ch. 7

less than independent. In light of this, the district court had the fire marshal's evidence stricken from the record. The Court of Appeals vacated and remand­ed, holding that "it is within the fabric of the State Fire Marshal's official duties to receive and rely on insurance company information about fires within his jurisdiction."

In Minersville Safe Deposit Bank & Trust Co. v. BIC Corp.,"' the plaintiffs claimed that a fire had been caused by a BIC lighter. Their experts, a fire marshal and an independent fire investigator, based their opinions about causation on the statements of a three year-old boy who said he had started the fire with a lighter. The district court in this case held that reliance on the child's statement was inadmissible hearsay, excluded the experts' testimony, and granted the defendant's motion for summary judgment'.

§ 7-1.6 Who Bears the Burden of Proof in a Daubert Hearing?

At one level the answer to the question in the heading should be obvious: the proponent of evidence always bears the burden of persuading the court that the conditions for its admission are met. But the question has neverthe­less led to confusion in Daubert hearings on a variety of expert areas, and one particularly confusing instance arose in the context of a proffer of expert evidence on fire causation.

Maryland Casualty Co. v. Therm-0-Disc, Inc. 55 involved a fire that began in a clothes dryer, allegedly due to a malfunctioning part. The district court initially placed the burden on the opponent of the expert witness to show that the testimony could not meet the requirements of Daubert. Here is the Fourth Circuit's recitation of what happened:

Neither party disputes that, at the beginning of the Daubert hearing, the district court told Jim Rothschild, counsel for Therm-0-Disc, "[y]ou have the burden of proor' with regard to Rodems's testimony. However, counsel immediately corrected the district court on this point, and from then forward it appears that the district court understood both the demands of Daubert and its own role as "gatekeeper," and conducted the hearing accordingly. Immediately after the objection, the district court withdrew its call for Mr. Rothschild to come forward and show that Rodems's testimony was not admissible, and called Rodems himself to the stand to explain the basis for that testimony. This Rodems did under both direct and cross-examination. After several hours of testimony, the dis­trict court determined that, although it had some reservations about the proffered basis for Rodems 's opinion, "the defendant has failed to estab­lish ... [that] Mr. Rodems relied upon a scientific principle that was not valid. ,,;;s

Although the Court of Appeals goes to considerable (and confusing) lengths to try to establish both that there is no burden of persuasion in a 104(a) hearing and that the district court did not err because it placed the· burden where it was supposed to be, the final sentence in the quotation above-offered in support of the correctness of the district court's hearing process-should leave

54. 176 F.R.D. 502 (E.D.Pa.1997). 56. Id. at 784. 55. 137 F.3d 780 (4th Cir.1998).

Sec. 7-1.7 FIRES, ARSONS AND EXPLOSIONS 351

readers wondering if, at the end of the day, the district judge did understand who bore the burden of persuasion, if there is a burden of persuasion.

The confusion in Daubert hearings is perhaps understandable, because the first voice heard is that of the opponent of proffered expert testimony. This has given some lawyers and judges the impression that the opponent has the burden of convincing the court that the witness does not meet Daubert's requirements. This impression may be all the more compelling when the expertise being challenged is a type that has become familiar to the courts. But the correct procedure is the opposite of that.

The opponent of expert evidence need only make a showing sufficient to convince a trial court that the objection to the evidence is not frivolous; this triggers a Daubert hearing under Rule 104{a). In a Rule 104(a) hearing on the question of admissibility of expert evidence, "These matters should be estab­lished by a preponderance of proof. "57 This statement of a standard of proof implies that the hearing is analogous to a civil bench trial. The proponent has the initial burden of production and the ultimate burden of persuading the court that the proffered expert evidence satisfies Rule 702, as interpreted by Daubert and Kumho Tire. The trial judge serves as the factfinder.38 And the judge must believe the expert's opinion by a preponderance of the evidence.39

Thus, if both parties sat mute, the court would have to rule against the party with the burden of persuasion, namely, the proponent of the evidence. If at the end of the hearing, the evidence on the evidence were in equipoise, again the court would hav~ to rule against the proponent of the evidence.

§ 7-1. 7 Nature of Scrutiny Following Kumho Tire Because Kumho Tire closed off the major route by which fire and arson

experts had been seeking to evade scrutiny under Daubert, one might expect courts to be subjecting experts of many kinds, including fire and arson experts, to more rigorous inspections following Kumho. Though limited in number, and revealing a ratio of about 50:50 (close scrutiny to relaxed scrutiny), it does seem that the courts are indeed beginning to step up the level of rigor employed in their evaluations of fire and arson expertise under Daubert and Kumho Tire. It might be misleading to point out that every one of the cases of more vigorous scrutiny and exclusion occurred in the context of a civil action, because almost all of the cases in which the expertise was challenged were civil cases.

In Weisgram v. Marley,6Al discussed in greater detail above, the Eighth Circuit Court of Appeals found plain error in the admission of three expert witnesses who had been permitted to testify by the District Court. A fire department officer who investigated the fire in the home was held not qualified to give expert testimony as to whether or .not a baseboard heater in the home malfunctioned, or how the heater might have ignited other objects.

57. Daubert 509 U.S. at 593 n. 20. 58. Note that Rule 104(a) also provides:

"In making its determination [the court] is not bound by the rules of evidence except those with respect to privileges."

59. This still does not answer the question of how much error is too much error to be admissible. For example, would expert predic-

tions that are shown by all of the available evidence on the evidence to be correct only 51% of the time be sufficiently "reliable" to be admissible? Would expert predictions that are wrong more often than they are right be ad­missible?

60. 169 F.3d 514 (8th Cir.1999).

352 FIRES, ARSONS AND EXPLOSIONS Ch. 7

A fire investigator's testimony that the baseboard heater was defective and caused the fire was held to be unsupported by sufficient foundation because no studies had been conducted to test the investigator's theory. Finally, the Court held that the expert testimony of a metallurgist that thermostat contacts on the baseboard heater were defectively designed was not supported by sufficient foundation because the metallurgist had little knowledge about this type of heater or this particular model heater. The rulings of the Court of Appeals were affirmed by the Supreme Court. 61

In Pride v. BIC Corp.62 the plaintiff's three experts were excluded by the District Court (following the recommendation of a Magistrate Judge), and the exclusion was upheld by the Sixth Circuit Court of Appeals. The plaintiff's husband had burned to death in a fire which the plaintiff argued had been caused by the defendant's defective cigarette lighter. In a thoughtful and detailed opinion, the Court of Appeals upheld the exclusion of a mechanical engineer, an analytical chemist, and a local fire inspector, all of whose methods it found inadequate to support their opinions.

Werner v. Pittway Corp. 63 was a suit alleging defective smoke detectors. The District Court excluded the plaintiff's expert witness, finding that the expert offered nothing but a "bare conclusion" regarding the type of detector the plaintiffs' used, and offered no explanation whatsoever regarding the reasoning process that would have permitted him to reach this conclusion.

In Donnelly v. Ford Motor Co. s-1 the plaintiff claimed that his Ford had a defective ignition switch, which caused the automobile fire in which he was injured. The expert's report asserted his experience and his conclusion to "a reasonable degree of engineering certainty" that "any fire in f one of the defendant's vehicles), in which arson has been eliminated as the cause, that has its origin under the driver side dash and in the area of the steering column can be directly linked to the vehicle ignition switch and system. "00

The report lacked any explanation of the expert's reasoning. A supplemental report contained an explanation insofar as the expert rebutted the defendant's experts, but the court held that "is not a substitute . . . for setting forth the reasoning or methodology by which he formed the opinions in his own reports . . . . "66

Finally, in Comer v. American Electric Power, 67 the plaintiff claimed a power surge caused a fire that seriously damaged his house. The defendant power company's expert agreed that arcing in a panel distribution box led to the fire, but the two sides's experts disagreed about the cause of the arcing condition. Awaiting guidance from the Supreme Court in Kumho Tire, the district court delayed ruling on the motion until after the jury returned a verdict for the plaintiff. In granting a motion for judgment as a matter of law the court held the plaintiff's expert evidence inadmissible. The court held that the plain tiff's expert's testimony was based only on personal knowledge and experience, lacked a more sound basis, and therefore his opinions were nothing more than unsupported speculation, inadmissible under Daubert,

61. 528 U.S. 440, 120 S.Ct. 1011, 145 65. Id. at 49. L.Ed.2d 958 (2000).

62. 218 F.3d 566 (6th Cir.2000J. 66. Id. at 50.

63. 90 F.Supp.2d 1018 (W.D.Wis.2000). 67. 63 F.Supp.2d 927 (N.D.Ind.1999).

64. 80 F.Supp. 2d 45 (E.D.N.Y.1999J.

Sec. 7-1.7 FIRES, ARSONS AND EXPWSIONS 353

Joiner and Kumho Tire. In commenting on the "lack [ofJ factual, technical, or scientific support," 68 the court expressed particular disapproval of the expert's flexibility in reaching opinions:

Indeed, one marvels at the breath-taking ease with which [the expert] offers his 'expert' opinions, surpassed only by his apparent ability to change them based on nothing more than the mere suggestion of counsel. With such an approach, unshackled as it is to any sort of factual, scientific or technical analysis, [the expert's] testimony easily accommo­dates whatever theory or time interval is needed by his client. 69

On the other hand, a number of other cases admitted expert witnesses with minimal scrutiny of the methodology and reasoning (if any) underlying the proffered testimony to determine if it rested on a valid, and therefore admissible, foundation. These include the following cases.

Call v. State Industries 70 involved a claim that the defendant's water heater caused the fire that destroyed the plaintiffs' home. The trial resulted in a jury verdict for the plaintiffs. On appeal, the defendant challenged, inter alia, the admissibility of the plaintiff's expert witnesses. The Tenth Circuit Court of Appeals offered only a cursory review of the challenged expert testimony, never explaining how it, or the lower court, reached the conclusion that the testimony was reliable, or what factors it used to reach that conclusion. This would seem to be the very abuse of discretion (a failure to explain what factors were used, why they are appropriate, and why they lead a court to its conclusion on admissibility) that Justice Scalia warned against in his concurrence in Kumho Tire.

Allstate v. Maytag,71 discussed in more detail above, relied entirely on the qualifications of the experts and made no assessment of the basis of the proffered knowledge to opine whether a cooktop was the source of a house fire.

Abu-Hashish v. Scottsdale Insurance Co. 72 allowed the defendant's experts to testify from their observations of burn patters that the fire at issue was not accidental, with no deeper scrutiny of how they could infer causation from what they observed. 73

Cooper v. Toshiba Home Tech. Corp. 74 admitted a plaintiff's expert who would opine that the fire was caused by a defective kerosene heater made by the defendant. The court admitted the expert on the basis that he was qualified, having "written or presented on issues of fire and arson over three hundred times .... [and] he holds a patent for an anti-flareup device for kerosene heaters." 75 As to the expert's failure to test his hypothesis that a flareup had caused the fire in this case, the court held that such testing was not required, but only that "[t]he expert's conclusions simply must not be 'subjective belief or unsupported speculation.' '' 76

68. Id. at 937. 69. Id. at 935. 70. 221 F.3d 1351 (10th Cir.2000) (Table). 71. 1999 WL 203349 (N.D.111.1999). 72. 88 F.Supp.2d 906 (N.D.lll.2000). 73. Fire and arson experts have subscribed

to numerous beliefs about the relation of fea-

tures of a burned structure's remains and the cause of the fire, recently determined to be unreliable indicators. See infra§ 2.1.1[1].

74. 76 F. Supp. 2d 1269 (M.D.Ala.1999).

75. Id. at 1278.

76. Id.

354 FIRES, ARSONS AND EXPLOSIONS Cb. 7

In United States v. Gardner? a criminal case, the court engaged in no apparent scrutiny of the government's expert, but merely asserted by ipse dixit that the government's proffered expert met the standard for admission.

§ 7-1.8 State Cases

Three fire cases from the state courts are noteworthy.

In Mensink v. American Grain,18 the Iowa Supreme Court confronted the same question the U.S. Supreme Court did two years later, in Kumho Tire, namely, whether (Iowa's version of) Daubert applies to all experts or only to "scientific" experts. The plaintiff had been seriously injured by a dust explosion in a grain elevator that was struck by lightning. The plaintiffs argued, inter alia, that the elevator could have and should have installed lightning protection devices, and offered a retired professor of electrical engineering to testify to that opinion. The jury found for the plaintiff. The defense appealed, and the case was reversed and remanded on grounds other than the expert witness issue.

As to the expert issue, the defendants argued unsuccessfully that the plaintiffs' expert should have been excluded under Daubert because he had no experience with grain elevators, though he had extensive experience in lightning protection. The court also rejected the argument that the basis of the expert's opinion failed Daubert's reliability criteria and was inadmissible for that reason. Daubert was held to be inapplicable because, the Court concluded, it is limited to "evidence of a complex nature," which the Court took to mean scientific evidence as opposed to "technical or other specialized knowledge." In this case, the expert had considered factors like the type of building construction and topography of the surrounding area, which the court believed were readily understood by lay jurors. This narrow reading of Daubert limits it to "complex" or "scientific" expert evidence, and to the explicit "Daubert factors:" rather than taking the case to stand for the more general proposition that all expert testimony must be found to be valid before it can be admitted. Kumho Tire, of course, took this latter tack, holding that all expert evidence, regardless of how it is labeled, must be found to be valid before it can be admitted. Whether Iowa or other state courts will follow the Supreme Court's broader reading in Kumho Tire remains to be seen.

At trial in the Texas case of Doyle Wilson Homebuilder, Inc. v. Pickens,19

the plaintiffs won damages for the value of their 18-month old home which had burned to the ground while they were away. Their theory of the fire was that defective or improperly installed wiring caused the fire, and their claim was brought against the general contractor. Though it reversed on other grounds, the Court of Appeals rejected the defendant's challenge to the admissibility of the plaintiffs' electrical engineer (no challenge having been made against the plaintiffs' other expert, a fire investigator). The case is interesting for the remarkably thoughtful and detailed care the court of appeals gave to reviewing the details of the experts' testimony. Yet despite that detailed care, the court nevertheless accepted without question the soundness of whatever theory or empirical data or experience underlay the

77. 211 F.3d 1049 (7th Cir.2000). 79. 996 S.W.2d 387 (Tex.App.-Austin 1999). 78. 564 N.W.2d 376 (Iowa 1997).

Sec. 7-2.1 FIRES, ARSONS AND EXPLOSIONS 355

opinions offered. This case highlights the difficulty counsel have in seeing that some experts may (or may not) be offering poorly grounded opinion (where in this case counsel for the defense challenged one but not the other expert). And the difficulty judges sometimes have in looking underneath the opinions and procedures of experts to try to discern the validity of the knowledge upon which it purports to stand.

Finally, In re W.T.B.80 involved a Louisiana juvenile who was adjudicated delinquent for setting fire to a high school building. On appeal he argued that the trial court should not have accepted expert testimony that the fire in question was started with an accelerant. An engineer employed by the school's insurer concluded the fire was caused by an electrical problem. But the Louisiana Court of Appeals affirmed. Citing Daubert, the court noted that the fire investigator was well qualified and had 23 years training and experience. His opinion was "based upon the pattern of damage in [the area where he determined the fire had started] and the reaction of the dogs used in the investigation that were trained to sniff the presence of accelerants." 81 Since some of the field's beliefs about burn patterns have been found to be fallacious82 and dog sniff evidence produces what the field regards as an unacceptably high risk of false positive errors, 83 this is a good example of why a court ought not to be relying on vague assurances from the expert witness but should require hard evidence on the soundness of the proffered testimony before admitting it. As to the contradictory report by the insurer's engineer, that was properly treated as competing evidence which the finder of fact has the duty to weigh andjudge. 84

B. SCIENTIFIC STATUS

by

John J. Lentini*

§ 7-2.0 THE SCIENTIFIC BASIS OF EXPERT TESTTh10NY ON FIRES, ARSONS, AND EXPLOSIONS

§ 7-2.1 Introductory Discussion of the Science

The scientific study of fires, arsons, and explosions is unique among the forensic sciences for two reasons. First, the fire or explosion tends to destroy the very physical evidence which can be used to establish the cause, so in the

80. 771 So.2d 807 (La.App.2000). 81. Id. at 813. 82. See infra § 2.1.1[1]. 83. See supra§ 1.4 and infra§ 2.1.1[2]. 84. When a trier of fact "is confronted with

a decision of which expert opinion to credit, a determination of the weight of evidence is a question of fact which rests solely with the trier offact." Id.

* John J. Lentini is a fire investigator and chemist who manages the fire investigation division of Applied Technical Services of Mar­ietta, Georgia. He is a fellow of the American

Academy of Forensic Sciences and the Ameri­can Board of Criminalistics, holds certificates from the National Association of Fire Investi­gators and the International Association of Arson Investigators. He chairs the ASTM Committee Responsible for developing forensic science standards, and is a principal member of the National Fire Protection Association's Technical Committee on Fire Investigations. Mr. Lentini has investigated more than 1500 fires, analyzed more than 20,000 samples of fire debris, and testified on more than 200 occasions.

356 FIRES, ARSONS AND EXPLOSIONS Ch. 7

case of arson, it is first necessary to prove that a crime has been committed. Second, the vast majority of practitioners of this "scientific" endeavor are not scientists and have little, if any, scientific training or education. While there are other forensic disciplines where technical skills learned on the job may provide adequate training (e.g., fingerprints, firearms identification, and handwriting comparison), it is difficult to argue that reasonable conclusions about fires can be drawn by individuals who have a limited understanding of the chemistry and physics of fire pattern development. Yet, the vast majority of practitioners do not possess a bachelors degree. With the exception of the chemists, who spend most of their time in the laboratory and most of their efforts on detecting minute quantities of flammable, combustible, or explosive material, the people who investigate fires and explosions got their experience one fire at a time, as fire fighters, and later as fire investigators. As such, the body of knowledge used by investigators is divided into two parts: the scientific literature and the anecdotal reports of field investigators.

Because of the extensive destruction of physical evidence, those who investigate fires in the field, known as "cause and origin investigators," rely heavily on eyewitness testimony, and in the absence of that, a frequent occurrence, rely even more heavily on their previous experiences in analyzing small bits of evidence. Fire investigation has been likened to putting together a jigsaw puzzle, where the pieces are all scattered, but additionally, the pieces are frequently missing, and those that are present are frequently unrecogniza­ble.

A fire investigator puts this puzzle together and reaches conclusions by comparing observations with expectations. The expectations have been devel­oped from training and experience, but that training and experience may not necessarily have a solid scientific foundation. For this reason, it is imperative that before an investigator's opinion is taken seriously, the efforts taken to "calibrate" the investigator's expectations should be scrutinized. Most impor­tantly, the presumptions which the investigator carries into each fire scene should be determined, as these presumptions will have a significant impact on the expert's credibility.

Observations that one investigator will use to show incontrovertible evidence of an incendiary origin might be found by another investigator to be an unimportant indicator of a secondary event that occurred long after the origin. There are major areas of disagreement on the ability of investigators to "read" burn patterns, particularly in fires that have burned for extended periods of time. There is also disagreement about an investigator's ability to interpret the condition of wires as evidence of electrical arcing, which might have caused the fire or may be a result of the fire. There are numerous other chicken-and-egg problems that arise in fires, due to the destructive nature of the event.

A major consequence of the destruction of the physical evidence is that very few criminal arson cases are brought. Most of the litigation surrounding fires occurs in the civil arena. Insurance oompanies are much more likely to deny a payment based on the belief that their insured committed arson than is a prosecutor to bring an arson case against that same individual with the same evidence. This is at least partly due to the lower standard of proof in civil cases. Likewise, in cases where a product or service defect is alleged to

Sec. 7-2.1.1 FIRES, ARSONS AND EXPLOSIONS 357

have caused a fire, the impetus to settle based on overwhelming evidence is frequently absent, because the evidence is seldom overwhelming, at least as compared to other cases. In the case of arson, the first task of the prosecutor is to prove that a crime has been committed-a task that in most other cases is much more easily accomplished, or hardly even necessary.

There is reasonably good agreement among forensic scientists regarding the proper testing of physical evidence in the laboratory. Consensus standards exist for most routine tests of fire debris. Standardization of field practices, however, is still controversial. One impetus for the standardization of the fire investigation field is the realization by fire investigators (and, indeed, most forensic scientists) that standards may be the key to admissibility. Another impetus for standardi~ation springs from efforts at certification of both laboratory and field investigators. Because examinations are required to grant certification, a standard body of knowledge from which to develop such examinations also is required. Field investigator certification began in the mid-eighties, after a five-year development period. Laboratory investigator certification development began in the mid-seventies, but certification did not become universally available until 1993. Field investigators may obtain certifi­cation from either the International Association of Arson Investigators (IAAI) or the National Association of Fire Investigators (NAFI). Laboratory analysts may obtain certification from the American Board of Criminalistics (ABC).

§ 7-2.1.1 Field Investigations Just as the type of evidence examined and the type of people examining

the evidence differ from the field to the laboratory, the approach to the scientific analysis of fire behavior is radically different between the field and the laboratory.

[1] Test Burns

During the 1970s and 1980s, the Center for Fire Research at the National Bureau of Standards, now known as the National Institute of Standards and Technology (NIST), conducted hundreds of excellent test burns, and charac­terized the behavior of fire up to the point of flashover. Flashover is a transitional phase in compartment fires in which temperatures rise to a level sufficient to cause ignition of all exposed combustible items in the compart­ment. Most structure fires will eventually achieve flashover, unless there is intervention by fire fighters or unless there is an unusual occurrence which allows the release of the fire gases, thus preventing the heat build-up.1

In a typical flashover scenario, an item of burning fuel, typically a piece of furniture, releases heat and smoke into the room, but in its early stages, the fire is unaffected by the room itself. This is known as the ''free-burning'' stage, and the behavior of the fire at this stage is relatively simple and easily explained (heat rises). When the fire begins to interact with its enclosure, its behavior becomes much more complex. As the fire progresses, a layer of hot gases begins to form at the ceiling, and gradually banks down, becoming

§ 7-2.0 GATIONS 17 (PuB. No. 921) (1995) [hereafter, 1. NATIONAL FIRE PRoncnoN AssocIATION, NFPA 921].

STANDARD GtiiDE FOR FIRE AND ExPLOSioN INVEST!-

358 FIRES, ARSONS AND EXPWSIONS Ch. 7

thicker and more charged with energy. Once the gas layer reaches a tempera­ture in the neighborhood of 1100 to 1300F, the radiant heat coming from the gas layer is sufficient to ignite common combustibles. 2

Unfortunately, the work of the Center for Fire Research was aimed at characterizing the behavior of materials up to the point of flashover, for purposes of improving the safety of structures and contents. Of the hundreds of fires conducted, none were examined to look at the aftermath, so there are almost no data from scientifically conducted test burns which give the field investigator any clues about what type of "burn patterns" remain behind after flashover has been achieved.

Other test burns take place on a regular basis, and are usually conducted at weekend seminars sponsored by local chapters of the International Associa­tion of Arson Investigators (IAAI). The reproducibility, and therefore, the validity of these tests varies widely from test to test, depending on the dedication of the test organizers. Many of these "burn exercises" are conduct­ed merely to familiarize new investigators with what a flammable liquid pour pattern looks like, and to provide extinguishment exercises for fire crews.3

The vast majority of burn exercises conducted over the years have been performed with these limited goals in mind. This approach has resulted in many trainees getting a one-sided view of fire investigation which has unfor­tunately been passed on to each successive generation of investigators.

As a result of criticism of this practice, fire investigation groups are now beginning to try to simulate accidental fires, and to collect more data from the fires they set. A properly instrumented test burn may have as many as two hundred thermocouples and several radiometers collecting data. A typical test burn conducted by professional fire investigators has fewer than ten thermo­couples and no radiometers. The behavior of the fire is usually recorded on video tape. ·

Two types of test burns have been attempted. The vast majority of test burns are set up to test one or more hypotheses about the general behavior of fire. If the test burn is narrowly focused in terms of the questions it seeks to answer, it is possible for useful information to be derived. Frequently, however, because structures that are available to burn are a rare resource, multiple burns are scheduled for the same structure, but the validity of subsequent tests is questionable.

The second type of test burn aims to reconstruct a particular fire, even though it is generally accepted that no two fires are alike, and an exact reconstruction is impossible. A simple change, such as leaving an interior door open when it should be closed, can drastically affect the behavior of a fire. About the best that can be hoped for is to reasonably reproduce a fire in a single compartment. This requires an exact match of interior finish and furnishings, something which is difficult to ascertain after a severe fire. Because of the time and enormous expense ($10,000-$100,000) involved in full scale test burns, they are usually conducted only when there have been multiple deaths or when the damages are in the millions of dollars.

2. Id. 3. NATIONAL FIRE PROTECTIO!'." AssocIATION,

STANDARD ON LIVE FlRE TRAINING EVOLUTIONS IN STRUCTURES (PUB. No. 1403) (1992).

· ec . 7- 2.1.1 359

The U.S. Fire Administration released a new report on the study of fire patterns in July of 1997.' A Fire Pattern Research Committee conducted ten full-scale fire tests, four at NIST headquarters in Gaithersburg, Maryland, two in residences in Florence, Alabama, and four in residences in Santa Ana. California. All of the test fires were instrumented and recorded, and the results of the tests are presented in a 210-page report. Many of the concepts, investigative systems, dynamics of pattern production, and pattern analysis concepts put forward in NFPA 921 were confirmed by the program's testing. Several of the "old fire investigators' tales" and fire investigation misconcep­tions that are repudiated in NFPA 921 were also shown to be u.nsu.bstan.tiated by the program testing.

The "old investigators' tales" whose repudiation was confirmed by this testing included:

• Wide V's versus narrow V's (which were erroneously thought to reflect the "speed of a fire")

• Crazing of window glass (which was erroneously thought to indicate rapid heating-it actually indicates rapid cooling-a much less signifi­cant phenomenon )

• Char blisters and speed of fire (large, shiny blisters were thought to indicate a rapid fire, while small flat blisters were thought to indicate a slower fire)

• Window sooting/staining (formerly thought to signify the type of fuel that had burned )

• Color of smoke and flame (also thought to be indicative of the type of fuel that was burning)

Throughout the ten test burns, it became apparent that a major factor in fire pattern development, namely ventilation, was the least understood . The study concluded that much more research needs to be directed at studying the effects of ventilation on the development of fire patterns.

(21 Accelerant Detecting Canines

In the early 1980s, the Bureau of Alcohol, Tobacco & Firearms and the Connecticut State Police pioneered the use of accelerant detecting canines. This practice has spread as its efficacy has become more apparent. 6 One of the central problems in fire investigation, particularly in arson investigation, is the location of suitable samples for submission to a laboratory so that the presence of ignitable liquids can be confirmed. Because most of these liquids ~ve an odor, it is surprising that it took as long as it did until the concept of accelerant detecting canines was explored. The canines have the ability to improve the efficiency of a fire investigation, by going straight to the location

4. hDD.u. EMDGENCY MANAGBJi02ff AoZNCr , U. S. Fire Adminiatratioll. USFA Fm BURN PATTERN ~PaoolWI ,08 THE STUDY OF FIJU: PArn11N8 (FA 178) (Jul.1997).

I. Meliua F. Smith, EvidentiaJy 1-­Surroundinr Accelerants Detected by Canines (Auguet 9, 1998) (Presentation to the Annual Meetiq of the American Bar Auociation ); Mi-

chael Kurz et al., Eualuation of Canino for Accelerant Detection at Fire Scena, 89 J . Fo. RENSIC Sa . 1528 (1994); Georp Dabdoub et. al., AccelerantDdmio,a ConinaandtlteLabortst&­ry, PIIOCUDINOII 0, TBS ANNuA1. Mamo OF TD AMlaicAN /t£ArJDlv OF FoDN'lllC 8cmfCZs 19 (1996).

360 h. 7

of the accelerant, and decreasing the likelihood of the submission of negatives t.o the laborat.ory, thus saving an enormous amount of resources . Like many scientific advances , however, the law has gotten ahead of it, and there are now individuals testifying as t.o the presence of ignitable liquids at a fire scene based on "alerts " from their canines , even though the laborat.ory has failed to confirm the indication . Given that there have been few scientific studies and even fewer published research papers on the subject of canine proficiency , this is a disturbing trend.' Most scientists in the fire investigation field hold that unless there is a positive laborat.ory analysis t.o confirm a canine alert, the alert is not useful in determining the cause of the fire, and is, therefore, irrelevant, both to the fire investigat.or and t.o the trier of fact.1

This widely held view has been codified in NFPA 921, discussed below. Because of concern over some misguided court decisions allowing the testimo­ny of dog handlers regarding unconfirmed alerts , the NFP A passed a "Tenta­tive Interim Amendment " (TIA) to its 1995 edition of the Guide for Fire and Explosion Investigations . Fearing that if they waited t.o voice their concerns in the 1998 edition of the document , the rapidly developing law on the subject would be too hard t.o change. Consequently , the Technical Committee on Fire Investigations declared (and the NFPA Standards Council agreed) that a "judicial emergency " existed. The TIA stated that the only legitimate uses for a canine were the selection of samples that had a higher probability of testing positive , and the establishment of probable cause for a warrant to search further . The TIA echoed the concerns expressed by the IAAI Forensic Science Committee , and was carried forward into the 1998 and 2001 editions of NFPA 921.8 ·

As a result of the publication of the TIA, judges began to follow the guidance of the fire investigation communit y, and exclude evidence of uncon ­firmed alerts . In a murder case in Georgia a conviction was overturned because the trial judge allowed testimony about unconfumed alerts into evidence, • and an appellate court in Illinois held that the trial judge appropri­ately excluded similar evidence ."

[3 ] Sniffers

Dogs were preceded into fire scenes by electronic sniffers , devices which first were developed in order to detect combustible gases in mines and in utility installations. These devices, while useful in eliminating negative sam­ples , are widely believed to be prone t.o providing false positive alerts. A positive alert by an electronic sniffer is generally not accepted as an indication of the presence of ignitable liquids.

Commercially available electronic sniffers generally incorporate detection devices similar to those used on gas chromatographs . The simplest machines use thermal conductivity detectors . Flame ionization, photoionization and

8. State v. Buller , 517 N.W.2d 711 aowa 1994 ).

7. IAAI Foremic Science Committee a. al. , Position on At:cd,ran.t Ddection Canina (adopted September, 1994), 45 Pm • All90N INVEff!OAroR 22 (1994 ).

8. See NFPA 921, at ff 12-5.9 and 14-5.8.4.

t. Carr v. State, 267 Ga. 701. '82 S.E..2d SU (Ga. 1997 ).

10. People v. Acri, 277 ffl.App.3d 1030, 214 W.Dec. 761, 662 N.E.2d 115 (Dl.App.1996).

Sec. 7-2.1.2 FIRES, ARSONS AND EXPWSIONS 361

solid state models have been used. The more complex detectors are sometimes unable to withstand the fire scene environment. Solid state units with comparison modules, sold by Pragmatics, are probably the most popular unit in use today.

§ 7-2.1.2 Laboratory Analysis

[ 1] · Classification of Petroleum Products

Most of the people with science degrees who are interested in the study of fires conduct their investigations in the laboratory, where they examine samples brought to them by field investigators. The laboratory analysis of fire debris is one area of forensic science where there is a near consensus on methodology and termino_logy. Because of this consensus, there has been a considerable amount of research published on the characterization of ignitable liquid residues recovered from fire debris. Much of the credit for this general consensus goes to the International Association of Arson Investigators Foren­sic Science Committee and ASTM Committee E 30 on Forensic Sciences, which publishes Standard Test Methods for the separation and identification of ignitable liquid residues. 11

The gas chromatograph (GC) has been the primary instrument used in the identification of petroleum based hydrocarbon liquids, which are the most commonly used accelerants. An identification not based at least in part on GC is probably invalid. Gas chromatographic techniques that are significantly at variance with the ASTM standards have a lower likelihood of being valid.

The identification is based on pattern recognition and pattern matching. The "pattern" arises because petroleum distillates are complex mixtures of up to three hundred compounds. There is, for example, no such entity as a "gasoline molecule." Gasoline, and most petroleum distillates, are resolved by the chromatographic column into separate compounds, usually in ascending order by size (molecular weight). The laboratory analyst comes to recognize the patterns produced by particular classes of petroleum products, and the "match" is made by overlaying the chromatogram from the sample extract onto the chromatogram of a known standard.

There is some room for judgment in this pattern matching, and it is for this reason that the use of mass spectrometry, coupled with gas chromatogra­phy, has gained much wider acceptance in recent years. While gas chromatog­raphy produces a pattern of peaks, the mass spectrometer is capable of identifying the compounds which produce the peaks. With gas chromatogra­phy, it may be possible to confuse patterns produced by background materials with patterns of petroleum based accelerants. This will be less likely to happen when gas chromatography/mass spectrometry (GC/MS) is used, since there is much less guesswork about the identity of the compounds causing the peaks on the chart. Caution is still required, however, since a piece of carpeting (or any combustible) breaking down in the process of combustion

11. AMERICA.'- Soc!ETY FOR TEsnNo AND MATE­RIALS, STANDARD 1'EsT METHoo FOR IGt.TIABLE LIQ­UID REsmuES IN ElrrRACTS FROM SAMPLES OF FmE DEBRIS BY GAS CHROMATOGRAPHY CPu:e. No. E 1387) (2000); AIIEmcAN 8ocJETY FOR 1'Em.'lG AND MATB-

RIALS, STANDARD TEST METHOD FOR IGNITABLE LIQ•

UID RF.slDUES IN ExTaACTS FROM 8AMPLES OF FmE DEBRIS BY GAS CHROMATOGRAPHYtMAss SPECTROME­

TRY (Pu:e. No. E 1618) (2000).

362 FIRES, ARSONS AND EXPLOSIONS Ch. 7

may produce compounds that also are found in petroleum distillates. Thus, ASTM E 1387 and E 1618 advise analysts that merely detecting benzene, toluene, and xylenes, or higher molecular weight aromatics is not sufficient for identifying gasoline. The relative concentrations of all of the compounds of interest must be such that a recognizable pattern is produced.

In recent years, proficiency tests, manufactured by Collaborative Testing Services (CTS) and sponsored by the American Society of Crime laboratory Directors (ASCLD), have revealed that the error rate for laboratories using only gas chromatography is significantly higher (50 to 100% higher) than laboratories using GC/MS.12 These same CTS studies show a decreasing reliance on GC alone (at least among CTS subscribers-arguably the better laboratories in the field), and may result in future changes in ASTM stan­dards.

[2] Identification (Individualization) of Petroleum Products

Considerable work has been done regarding the individualization of ignitable liquids in order to tie a suspect to a source. Matching of gasolines has been demonstrated, but as a fire progresses and the gasoline becomes more evaporated, the identification becomes more difficult. The most useful components for individualizing a sample of gasoline are, unfortunately, the components which evaporate first in a fire. The individualization of other petroleum distillates has not been extensively studied, due mainly to the difficulty of the task. Unless there is some unusual compound dissolved in the petroleum distillate, identification is generally regarded as difficult, if not impossible.

Recent studies by environmental chemists, attempting to measure the age or source of petroleum discharges, have identified several classes of higher molecular weight compounds, more likely to survive a fire, which can be used for individualization. Considerable additional work is required before this technology will be applicable to fire debris analysis. 13

§ 7-2.1.3 Sources

[1] Authoritative Publications

The industry standard for fire investigation is known as NFPA 921, STANDARD GUIDE FOR FIRE AND EXPLOSION INVESTIGATIONS.14 The National Fire Protection Association is a nonprofit organization which promulgates all types of codes related to fires, including building codes, equipment specifications, guidelines for certification of individuals, and a guide for fire investigation. NFPA 921 is produced and maintained using the NFPA consensus process, which has been approved by the American National Standards Institute (ANSI). The Technical Committee on Fire Investigations, which drafted

12. COLLABORATIVE 'rEsTING SERVICES, ~­BI.ES ANAJ.'YSIS REPORT No. 9716 (1998); CoLLABO­RATIVE TEsnNG SERVICES, F'LAMMABLES A.~AI.YSJS

REPoRT No. 9816 (1999); COLLABORATIVE TESre.G

SmMCES. F'LAM?4ABLBS A.~AJ.YSIS REPORT No. 99-536 (2000).

13. S.A. Stout & A.D. Uhler, Chemical "Fingerprinting" of Highly Weathered Petrole­um Products, PllocEEDINGS OF THE AMERICA?( AcMJEWi OF FORENSIC Sc:IENCES (Feb.2000).

14. Supra note 1.

Sec. 7- 2.1.3 363

NFPA 921, consisted of twenty-nine individuals, with membership strictly regulated by NFPA guidelines, including specified numbers of public officials, academics, insurance industry representatives and private experts. The gener­al public has the opportunity to comment on all proposed language in the standard., and on all proposed changes. In general, however, only NFPA members are made aware of pending standards. The committee then votes on whether to accept or reject the comments from the individual submitters, some of whom may be memben of the committee. The first edition of NFPA 921 was published in 1992, with no dissenting votes from committee mem­ben.

When this chapter was first published in 1997, the industry standard for fire investigation was the 1995 edition of NFPA 921, GUIDE FOR Fnu: AND EXPLOSION INvEsTJGATIONS. As a result of the receipt of more than 150 proposals for changes, the 1998 edition, which became effective in February of 1998, contained many substantial changes and clarifications.

One of the more interesting changes was the removal of the word "misconception" from the titles of many-sections. A significant portion of the fire investigation community was offended by the use of the word "misconcep­tion" in the first two editions of the GUIDE. Proponents of the change argued that while the misconceptions might exist in some fire investigaton' minds, the first two editions had cleared up many of those misconceptions. In most cases, the text of the chapter section was left intact, but the title was changed. For example, the sections entitled "Misconceptions about Char" and "Miscon­ceptions about Spalling" had their titles changed to "Interpretation of Char" and "Interpretation of $palling." The cautions regarding the potential for misinterpretation of these two artifacts remained in the text.

Another significant change was the removal of the section on certainty of opinions. u A consensus within the Committee developed regarding removal of these various levels of certainty. The Committee felt that they had been mistakenly equated by the legal community with burdens of proof-an easy mistake to make since some of the terms and their definitions plainly adopted terminology that borrowed from and tracked legal burden of proof concepts. Whatever the intention of the fire and arson community may have been, they provide an example of how to invite (and get) confusion. An attempt had been made to clear up the confusion between the 1992 and 1995 editions, but it wu not successful. Elimination of the terminology altogether was judged to be the best available course. Now there is no standardization of certainty-of-opinion language among fire investigators.

In the 1998 edition, the chapter on electricity and fire and the interpreta­tion of electrical artifacts was significantly expanded and improved. A chapter was added on building fuel gas systems and a large number of references were added to the 8%J)lanatory material in the appendix.

Between 1998 and 2001, the NFPA received 183 proposals for changing NFPA 921, and over 500 comments on the Technical Committee's handling of those proposals.

11. ~ in the 1997 edition of Uus EvmENa:: THB I.Aw AND Scmia or ExPDT TE:m-cbapter , John J. Lentini, Fira, ATIOft8 and llc»ff (David L. Paigman. Duid H. Kaye, Mi­E.,q,losioM: Fldd Invatwationa: Cmointy of chael J. Saks & Joaeph Sanden eds., 1997). Opi.raio,I,, I ~2.2.1(14], ill MODBII!,' Scmmnc

364 FIRES, ARSONS AND EXPWSIONS Ch. 7

The current edition of NFPA 921, released in February, 2001, includes a rewritten chapter on vehicle fires, and new chapters on fire deaths, human behavior in fires, analytical tools (including computer assisted fire modeling), wildfire investigations and building systems. Additionally, the GmoE will be organized into three sections outlining what a fire investigator should know, how a fire investigator should conduct a routine investigation, and special topics in fire investigation.

Other significant additions to the 2001 edition include a discussion of spoliation of evidence, and a definition of those necessary activities that should not be considered spoliation, as well as a discussion of the process of elimination, an attempt to come to grips with the so-called "negative corpus" determination. 16

The Technical Committee and the NFPA as a whole continued to endorse the scientific method as the appropriate way to investigate fires, turning back proposals to eliminate the word "science" in favor of a so-called "systematic" approach.

In late 1997, the Justice Department began working on national guide­lines for fire and arson scene investigation. These guidelines were modeled after NATIONAL GcmELIXES FOR DEATH l:NVESTIGATION, a research report published by the National Institute of Justice (NIJ) in December of 1997.17 The Techni­cal Working Group assembled by NIJ at first recommended that the Justice Department simply purchase 15,000 copies of NFPA 921 and mail them to the nation's law enforcement agencies and fire departments. This suggestion was not adopted, and work on the national guidelines continued for three more years, culminating in the publication of a finished pamphlet in June 2000.18

These national guidelines recommend a general procedure for the handling of fire and arson scenes, and specifically direct responsible officials to find a fire investigator capable of ·conducting a scientific scene inspection according to the recommendations of NFPA 921. The text urges adherence to other nationally accepted standards, such as those published by ASTM.

A similar guide was published at the same time dealing with the responsi­bilities to explosion or bombing scenes. 19

Most fire investigators will, on cross examination, concede that NFPA 921 represents the industry standard for the conduct of fire investigations al­though it is "only a guide." Perhaps the most important concept embodied in NFPA 921 is the recognition that fire investigation must be based upon the scientific method.20 This may seem obvious, but until recently, fire investiga­tors based their conclusions upon their "technical knowledge" gained through education, training, and experience. The existence of NFPA 921 makes it more difficult for the investigator to rely solely upon anecdotal experience.

18. See infra., § 2.2.1[16].

17. NATIONAL MEmOOLEGAL REvmw PANEL. NA­

TIONAL GUJDELINES FOR DEATH INVESTIGATION (Na­tional Criminal Justice Reference Service CNCJ 167568) (1997)).

18. TEcHN!CAL WoJUID;G GROUP ox FIREtARsoN

ScENE INVESTIGATION. USDOJ, Fm A.''D ARsoN SCEm: E'VmENCE: A Gt.'IDE FOR Pt.'BLIC 8An:TY PER-

soNNEL (June 2000), at http://www.ojp.us­doj.gov/nij/scidocs2000.htm.

19, TECHNICAL WORKING GROUP FOR BoMBING

SCENE INvEBTIGATION, USDOJ, A Gmm: FOR EXPLO· SION AND BoMBING SCENE INVESTIGATION (June 2000), at http://www.ojp.usdoj.gov/nij/sci­docs2000.htm.

20. NFPA 921, supra. not.e l, at 9.

Sec. 7-2.2.1 FIRES, ARSONS AND EXPLOSIONS 365

As far as other authoritative sources or learned texts in the field, there is only one text that any group of fire investigators will be likely to agree upon as being authoritative, and that is KIRK'S FIRE INVESTIGATION, by John D. DeHaan. 21 This excellent basic text, which is understandable to nonscientists, explains most of the aspects of fire investigation that a typical investigator is likely to encounter. The first edition of this book was authored by Paul Kirk in 1969,22 and it was the standard reference for over a decade. To appreciate the changes and improvements in the understanding of fire dynamics and fire investigation that have occurred since 1969, a review and comparison of the successive editions of this book is useful. This popular reference text on fire investigation, came out in its 4th Edition in 1997. There were several additions to the text, as DeHaan attempted to bring it more in line with NFPA921.

[2] Periodical Literature

There are several periodicals to which fire investigators may subscribe, and the reliability of the · information in these periodical varies widely. FIRE TECHNOLOGY, a peer-reviewed journal, generally deals with highly technical aspects of fire behavior, such as computer modeling and the behavior of large liquid pool fires. Very few fire investigation articles have been published in FIRE TECHNOLOGY.

The most widely read fire investigation publication is THE FIRE & ARSON INVESTIGATOR, the official publication of the International Association of Arson Investigators. Publication in this journal may reflect a thorough peer review (for the more technical articles) or simply editorial review (for news reports or "oped" pieces). Peer review began only in 1996. The IAAI, until then, elected instead to publish more of a newsletter than a scientific journal, in the belief that everyone is entitled to their own opinions and that the exclusion of articles, even highly technical scientific articles, because of the objections of peer reviewers was equivalent to "censorship."Fortunately, that view has changed.

FIRE FINDINGS, an independent publication, contains an interesting mix of fire-related news and reports of tests conducted by the magazine staff. Peer review is spotty, but much useful data can be found in this journal, and it is likely that review will increase as the magazine develops more depth in its editorial staff.

§ 7-2.2 Areas of Scientific Agreement and Disagreement

§ 7-2.2.1 Field Investigations

There are few fields where the ability of experts to disagree after viewing the same evidence is more of a problem than in fire investigation. Because so much of the physical evidence is destroyed by a fire, it is a rare fire that can be examined by more than one expert yet have only one conclusion reached about it. Generally, the more severe the fire, the less likely two individuals are to agree as to its cause. Certainly, the more thorough the investigation and

21. JOHN D. DEHAAN, KIRK'S Fm: bm:snGA, TION (4th ed.1997).

22. PAUL KIRK, Fm: INVESTIGATION (1969).

366 FIRES, ARSONS AND EXPLOSIONS Ch. 7

the more information that is actually collected, the more likely these individu­als are to be able to agree as to the cause of a fire.

[1] The Behavior of Fire

With respect to the behavior of fire, all investigators will agree that there is a fire triangle consisting of heat, fuel, and oxygen. Some investigators will expand the triangle into three dimensions, and describe a fire tetrahedron, the fourth point of which is a sustained chemical reaction.23 Fire investigators will all agree that heat rises, and that the principal means of fire spread are conduction, convection, and radiation.

When they actually begin to describe how a particular fire spread, however, many fire investigators ignore everything but convection, that is, the phenomenon which causes warm air to rise. This phenomenon also causes a fire to spread out in a "V" shaped pattern until it reaches an obstruction. Thus, seeking the bottom of the "V" shaped burn pattern should lead one to the origin.2~

Unfortunately, this indication of a fire's origin is useful only when the fire is extinguished prior to achieving total room involvement. Once a fire has progressed beyond a certain point, items that were located near the top of the room catch fire and fall down, causing secondary ignitions, and additional "V" shaped burn patterns. 25 These may be falsely interpreted as evidence of a second point of origin. Since everyone agrees that an accidental fire can begin in only one place, multiple points of origin are generally considered to be an indicator of arson.

There is considerable disagreement in the fire investigation community about the ability of fire investigators to determine multiple origins when the fires have burned together. Some fire investigators insist that they have the ability to determine multiple points of origin even if a room has flashed over. They do this by looking at holes in the floor, and because holes in the floor represent "low burns," these are equated with multiple origins. NFPA 921 advises the investigator to be wary of numerous conditions which could result in "apparent" multiple origins.26

Low burns are also taken as an indication of the presence of flammable liquids. Since heat rises and fire burns up, the presence of a low burn is sometimes taken as an indication that there was "something" on the floor which held the fire down. While ignitable liquids will accomplish this task, radiation does an equally fine job of burning floors. If total room involvement has been achieved, there is no reason for a floor not to be burned. 27 The failure to take into account the effects of radiation may be a result of a general lack of understanding of this common and important means of heat transfer. 28

When the floor is burned, there is a tendency on the part of some fire investigators to believe that, unless it has burned in a perfectly uniform manner, radiation can be ruled out as the cause of the low burning. This is

23. NFPA 921, supra note 1, at 10. 27. Id. at 33.

28. l>EHAAN, supra note 21, at 28. 24. Id. at 34. 25. Id. at 33. 26. Id. at 129.

Sec. 7-2.2.1 FIRES, ARSONS AND EXPLOSIONS 367

based on the perception of flashover as being a uniform phenomenon. 29

Actually, the uniformity of the flashover event breaks down within the first few seconds. Additionally, most synthetic floor coverings have a tendency to tear open during a fire, thus leaving parts of the floor exposed and other parts of the floor covered. These alternating exposed and covered areas can result in the production of patterns, particularly on wooden surfaces, which look remarkably like patterns produced by flammable liquids. 30 This is a fact which is not yet accepted by many fire investigators, but the realization of their misperceptions has led to several well publicized reversals of convictions.31

The 2001 edition of NFPA 921 contains several new photographs of what one would expect a flammable liquid pour pattern to look like, but which were actually created by radiation alone.32

The series of test burns conducted under the auspices of the USF A, discussed above, has generated much interesting data, but there has not been unanimous agreement on the correct interpretation of the data. Several comments critical of the series of tests were received by the NFPA Technical Committee on Fire Investigations when it proposed citing the final test report in NFPA 921.33 Many of these criticisms cited the "incomplete" nature of the data, because funds ran out before all of the proposed tests could be complet­ed. These arguments really did not focus on the quality or the interpretation of the data. Other, more legitimate, criticisms of the report focused on its conclusions about the cause of certain types of burn patterns, believed to have been caused by unique ventilation parameters, and on the suggestion of the report's authors that _much meaningful data could be found in the depth of the "calcination" of gypsum wallboard.

[2] Accidental Fires

The vast majority of fire scene investigators receive their training as "arson investigators." There are many fires, however, that are accidental in nature and which result in civil litigation. The trend toward subrogation in the insurance industry picked up considerable strength in the 1980s and shows no signs of abating. Thus, arson investigators are now called upon to determine accidental causes, with an eye toward pinning the blame on a manufacturer, or a provider of a service, or anyone other than the named insured. Competent investigators have the good sense to call in the appropri­ate engineering discipline once they have determined that a particular device is located at the origin. Most engineers are unable to determine the origin of a fire, but most fire investigators are unable to independently determine the cause of failure of an appliance or system.

29. Barker Davie, Flashover, 11 NAT'L FIRE & Al!soN REP. 1 (1993).

30. John D. DeHaan et. al., The Lime Street Fire, 43 FIRE & ARsoK INVESTIGATOR, Sep. 1992, at 41; NFPA 921, supra note 1, at 38.

31. State v. Knapp, No. CR 78779 (Superi­or Ct. of Arizona, Maricopa County, Feb. 11, 1987); State v. Girdler, No. 9809 (Superior Ct. of Arizona, Maricopa County, Jan. 3, 1991l.

32. NFPA 921 (2001), supra note 1, at § 4-17. 7.2.

33. Technical Committee Documentation, NFPA, Report on Proposals for the Fall, 2000, Meeting, at www.nfpa.org.; Technical Commit­tee Documentation, NFP A, Report on Com­ments for the Fall, 2000, Meeting, at www. nfpa.org.

368 FIRES, ARSONS AND EXPLOSIONS Ch. 7

[3] Electrical Activity

Fire investigators agree that the electrical system can be used as a fire detector, in that the first point on an energized electrical circuit which is compromised by a fire is likely to be the first and only point on that circuit where arcing occurs.:M There is some disagreement, however, about the characterization of arcing. There are many fires where one investigator will point to a piece of melted copper and identify it as arcing, while another investigator will look at the same piece and state that the melting of copper was caused by local fire temperatures in excess of the melting point of copper (1981 degrees F). The determination as to whether a bead of melted copper was caused by electrical or thermal activity can usually be resolved through an examination by a practiced electrical engineer. 35

Some study has been conducted on the examination of arc beads, in order to determine whether they were created in an atmosphere full of smoke or in a smoke-free atmosphere. The theory is that an arc bead created in a smoke­free atmosphere was created toward the beginning of the fire, and may, in fact, have been the cause of the fire.36 The ability to determine the elemental content of the atmosphere at the time an arc bead was created, however, has not been repeatedly demonstrated, and the significance of the atmospheric chemistry is a subject of debate.37 Arcing occurs in almost all fires, and almost all arcing events are the result of a fire, rather than the cause of it. The search for the primary arc, however, has resulted in many disagreements among fire investigators and electrical engineers. The fact is that electrical arcing is not responsible for a large number of fires, but an arc can be shown to be a competent ignition source, although the typical arc lasts less than a second. Thus, electrical arcs are often mistakenly blamed for causing fires.38

Other electrical sources are frequently cited, correctly or incorrectly, as the cause of a fire. Heat producing appliances (clothes dryers, portable heaters, stoves, and ovens) are the most frequent causes of fires started with electrical energy.39 Ballasts from fluorescent lights are probably the most frequently falsely accused electrical devices. When we consider the millions of these devices in use, it is not hard to imagine that there will be a ballast found within ten feet of the origin of almost any commercial fire.

Electronic equipment such as computers, stereo systems, and televisions often are blamed for fires. Television sets manufactured in the early 1970s were responsible for a very high number of fires. By 1993, the Consumer Products Safety Commission estimated that incidence was down to 0.4% of all residential fires, 2100 incidents causing 30 deaths. 40

Each proposed electrical fire cause deserves careful evaluation. Some "indicators" of electrical causation, like some indicators of arson, have been studied and shown to be less valid than previously thought. Oversized fuses or

34. Richard Underwood & John J. Lentini, Appliance Fires: Determining Responsibility, 7 NAT'L FIRE & ARsoN REPORT 1 (1989).

35. Bernard Beland, Examination of Elec­trical Conductors Following a Fire, 16 Foo: TECH. 252 (1980).

86. Robert Anderson, Surface Analysis of Electrical Arc Residues in Fire Investigation, 34 J. FORENSIC Sci. 633 (1989).

37. Bernard Beland, Examination of Arc Beads, 44 FIRE & AllsoN INvEsTIGATOR, Jun.1994, at 20.

88. DEHAAN, supra note 21, at 263.

89. U.S. CONSUMER PRODUCTS SAFETY COMMIS­

SION, 1986 NATIONAL Foo: Loss ESTIMATES (Oct. 1988).

40. DEHAAN. supra note 21, at 261..

See. 7-2.2.1 FIRES, ARSONS AND EXPLOSIONS 369

breakers, unless very much larger than required, are generally incapable of supplying sufficient current to overheat a circuit. The condition of insulation on a cable may yield some information about overcurrent, particularly if the insulation has melted loose from the conductor. A comparison must be made with a similar unheated wire, however, to determine the original tightness of the insulation. The lack of loose insulation does not rule out overcurrent. 41

The most frequently encountered problem in the examination of electrical evidence is one of cause and effect. Did the wire short and start the fire, or did the fire burn the insulation and cause the wire to short?

[ 4] Cause and Effect

The same type of chicken-and-egg, or cause and effect, argument applies to other systems found in buildings, as well. The gas system, for instance, is frequently compromised by a fire, resulting in leaks observed after the fire. It is the goal of the fire investigator to determine whether the leak existed before the fire. This is often a more difficult question than the science is capable of handling, particularly if the leak occurs in a combustible line. Metallurgists can be of some assistance in determining the reason for a fracture, and can sometimes tell whether the metal broke while it was hot or cold.

The compromise of electrical and fuel systems by fire, and the confusion that it creates, is even more evident in vehicle fires. As a general rule, fire investigators will agree that, while fires can start in the engine compartment and move to the passenger compartment, the reverse is seldom true. Some of the early fire investigation literature pertaining to vehicles suggested that almost all vehicle fires were intentionally set, as all of the fires contained certain "indicators~' of excessive heat.u These texts are now generally regard­ed as incorrect, as it has been shown that regardless of ignition method, the temperature achieved by a vehicle fire will approach 2000 degrees F, resulting in buckling and warping of body panels, melting and flowing of window glass, and a loss of seat spring temper. Thus, the intensity and duration of a vehicle fire cannot be interpreted as indicating or not indicating the presence of accelerants. 43

In a structure fire, an investigator who can narrow the origin down to a three by three foot square is considered a hero. In a vehicle fire, a three by three foot square is the starting point, and unless the exact cause can be determined, the fire investigator will be looked upon as a failure. Thus, frequently investigators will seize upon a burned fuel line, or an arced electrical wire as the cause of a fire, when the evidence argues equally that the observed phenomenon is actually an effect.

[5] Black Holes

Perhaps no type of fire is more difficult to investigate than the "black hole," a structure fire in which everything is reduced to ashes. Despite the

41. Id. at 216. 42. NATIONAL AUTOMOBILE THEFT BUREAU,

MANuAL FOR INVESTIGATION. OF VEHICLE F'mEs (1986).

43. DEHMN. supra note 21, at 179.

370 FIRES, ARSONS AND EXPLOSIONS Ch. 7

difficulties of these investigations, fire investigators have been known to claim the ability to detect multiple origins in completely consumed structures, or to state, based on "indicators," that a fire burned hotter than normal or faster than normal.

The studies done by the Center for Fire Research have tended to put to rest the diagnosis of "faster than normal." If a piece of upholstered furniture is ignited, it can bring a room to total involvement within five minutes. Time to flashover as low as ninety seconds has been reported. 44 Once flashover occurs in a particular room, extension into nearby rooms can be exceedingly rapid, involving entire houses in as little as fifteen minutes.

The fire which burns "hotter than normal" has, in the past, been identified by examination of the metals and other noncombustible materials within a structure, in order to get a handle on the temperature which the fire achieved. In 1969, Kirk advised noting melted metals because:

The investigator may use this fact to his advantage in many instances, because of the differences in effective temperatures between simple wood fires and those in which extraneous fuel, such as accelerant, is present. 45

Contrast this advice with DeHaan, Kirk's successor, in 1991:

While such melted metals cannot and should not be used as proof that the fire was incendiary, the fire investigator should note their presence, extent and distribution. Such information can be of help in establishing differences between normally fueled and ventilated accidental fires and those produced by enhanced draft conditions or unusual fuel loads from accelerants in incendiary fires.46

Thus, the modern text recognizes what blacksmiths and metallurgists have known for centuries: that increased ventilation can lead to increased temperatures. Despite this knowledge, fire investigators have frequently re­lied on the presence of melted copper to indicate a hotter than normal (accelerated) fire. This was particularly true of fires where the melting was found at floor level.

Melted steel was considered to be even more indicative of a "hotter than normal" fire. Steel has a melting temperature of 2100-2700 degrees F, depending on its elemental content. Multiple areas in a structure which exhibit melted steel, then, have been considered as indications of the use of flammable liquids to accelerate a fire.

Actually, it has been demonstrated that the flame temperature above a pool of flammable liquid is no greater than the flame temperature of a well­ventilated wood fire.47 The purpose of an accelerant is to make the fire burn faster, by involving more materials sooner than they would be otherwise involved. These fires do not necessarily bum at higher temperatures. The rate of heat release is higher in an accelerated fire, as the BTUs are released over a shorter period of time. The temperature of the fire, however, and its ability to melt items such as steel and copper, is actually no different from that of an unaccelerated fire.

44. NFPA 921, supra note 1. at 18. 45. KIRK, supra note 22, at 145. 46. DEHAAN, supra note 21, at 173.

47. Richard Henderson & George Lightsey, Theoretical Combustion Temperature, 3 NAT'L

FIRE & .AllsoN REPORT 7 (1985).

Sec. 7-2.2.1 FIRES, ARSONS AND EXPLOSIONS 371

After the catastrophic Oakland fire of 1991, Lentini, Smith, and Henderson conducted a study to determine the validity of the "indicators of arson." They studied copper. steel, and glass in fifty of the houses that had been burned to completion. Most of the houses exhibited multiple "indica­tors'' of arson, even though they are known to have burned in an accidental fire.48

[6] "Melted" Steel

Metallurgical laboratory analysis conducted as a follow-up to the Oakland study revealed that it is not possible to determine by visual inspection alone whether a piece of steel, particularly a small mass piece of steel such as a bed spring, has melted or merely oxidized. This distinction can only be made by microscopic examination of a polished cross section of the metal. Thus, steel that had been characterized as "melting," at temperatures of up to 2700 degrees F, may have actually been exposed to temperatures as low as 1300 degrees F for long periods of time, and gave an appearance that was wrongly interpreted as melting.49

[7] Crazed Glass

Glass is another material that changes as a result of exposure to the heat of a fire. Many texts have referred to the crazing of glass as an indication of rapid heating, and one widely circulated handbook went to far as to state that crazed glass was an indicator of nearby accelerants. 50 Crazed glass was also used as an important indicator in the trial of Ray Girdler, whose conviction was later overturned based on new scientific evidence.51 Experiments conduct­ed after the Oakland fire study revealed that no amount of rapid heating would cause crazing, but that rapid cooling, caused by the application of a water spray, would cause crazing in all cases, whether the glass was heated rapidly or slowly. 52

[8] Concrete Spalling

Spalling is the explosive chipping up of concrete, caused by the applica­tion of heat. This phenomenon has been the subject of more rhetoric, and probably less research, than most of the other issues in fire investigation. It is one of the most misunderstood and improperly used evidentiary features in the field, 53 and was the basis of an unfortunate case in Alabama which resulted in a major punitive damage award against the insurance company which presented spalling as evidence of incendiary origin. In that case, the fire had reduced a two story house to a pile of rubble about a foot deep on top of the concrete slab basement floor. The fire investigator (the second one hired

48. John J. Lentini et al. Baseline Charoc­teristics of Residential Structures Which Htwe Burned to Completion: The Oakland Experi­ence, 28 FIRE: TECHNOLOGY 195 (1992).

49. Id. 50. JoHN BARRACATO, BUllNING. A GUIDE ro

Fm hm:mGATION 4 (AETNA Casualty and Surety Company) (1986).

51. State v. Girdler, No. 9809 (Superior Ct. of Arizona, Maricopa County, Jan. 3, 1991).

52. John J. Lentini, Behavior of Glass at Elevated Temperatures, 37 J. FORENSIC Sci. 1358 (1992).

53. NFPA 921, supra note 1, at 27.

372 FIRES, ARSONS AND EXPLOSIONS Ch. 7

by the insurance company) cleared off a narrow area about ten feet in length and discovered that the floor was spalled. He then declared that a "trail of spalling" existed, and was incontrovertible proof of an arson fire. Despite the fact that the defendant's investigator found that the entire slab was spalled, this "trail" evidence was presented, resulting in the court rendering and the Supreme Court upholding the following characterization: "The presentation of his [the investigator's] testimony borders on the perpetration of a fraud upon the Court."N

There is an "old school" which holds that concrete spalling is an indication of the presence of flammable or combustible liquids, as well as a cadre of scientists (none of whom have published in a peer reviewed journal) who hold that it is impossible for a flammable liquid to cause spalling.iili In the middle are the vast majority of fire investigators, who believe that spalling is just another facet of the "burn pattern," which may or may not indicate the presence of a flammable or combustible liquid, depending on the situation. Most fire investigators have seen containers of ignitable liquids which have spilled their contents onto a concrete floor, and in the exact place where the flammable liquid was located, spalling has occurred. The extrapolation of this anecdotal experience to all fires is, of course, an error, as is the contention that flammable liquids cannot cause spalling under any circumstances.

[9] Colors of Smoke and Fire

Other indicators of unusual fire behavior that have fallen by the wayside include the color of smoke and the color of the flame. When these indicators first were promulgated by the teachers of fire investigation, they were considerably more valid than they are today. Wood and cellulose products tend to have a gray to white smoke, and burn with a yellow flame. Petroleum based products, such as most common ignitable liquids, burn with a sooty orange flame and produce large quantities of black smoke. In the past, it was thus possible to reach conclusions about what was burning, particularly in the early stages of a fire. In the modern structure, however, a large portion of the interior finish and furnishings consists of petroleum based products in the form of plastic films and fibers. A burning couch is just as likely to produce thick black smoke as is a burning pool of flammable liquid.

Once again, it is useful to contrast Kirk in 1969 with DeHaan in 1991. According to Kirk, "The presence of much black smoke, especially in the early stages of a building fire, is highly indicative of the presence and burning of a highly carbonaceous material, typical of many fire accelerants. "56 DeHaan, on the other hand, advises, "The combustion of [such] polymers contributes largely to the formation of greasy or sticky dense soot found at many fire scenes, and is responsible for the dense black smoke more frequently noted during the early stages of structure fires." 57 Not only is smoke color an unreliable discriminant between normal and abnormal fuels, it has lately been

54. United Services Auto. Ass'n v. Wade, 544 So.2d 906 (Ala.1989J.

55. Dennis Canfield, Causes of Spalling Concrete at Elevated Temperatures, 34 FmE & ARsoN INVESTIGATOR, Jun.1984, at 22.

56. Kmx, supra note 22, at 61.

57. DEHAAN. supra note 21, at 80.

Sec. 7-2.2.1 FIRES, ARSONS AND EXPLOSIONS 373

found that even ordinary wood fires can produce black smoke in the low oxygen conditions which occur following flasbover.58

(10] Computer Modeling

As a result of the work conducted at the Center for Fire Research, several computer programs have been developed to predict the spread of a fire, given certain assumptions. 59 If certain facts are known about the configuration of the compartments and fuel packages in a building, and the spread of a fire, these can be plugged into the equation, and the initial conditions derived. Since it is the initial conditions that are of most interest in determining the origin and cause of a fire and its spread from that point, this is the best use to which these computer programs can be applied.

There are currently two types of computer modeling programs for fires: zone models and field models. A zone model divides each compartment into an upper zone and a lower zone, and predicts the conditions in each zone as a function of time. Zone models are useful for situations where a rough approximation will do, and have been used to closely predict, for instance, when flashover will occur, given a specific fire on a specific fuel package. Zone models can be run by a proficient modeler in a few hours on a personal computer. A typical zone model assumes that every part of the zone is uniform with respect to temperature and smoke concentration. Consequently, while the model may be able to predict when any sprinkler head might activate, it will be less reliable in predicting the activation of a particular sprinkler head.

Field models (also known as computational fluid dynamics or CFD models) are much more complicated. They divide each compartment into thousands or tens· of thousands of small volumes, and calculate the fire's progress through each volume. This makes the field models much more precise, but they are much more difficult to work with compared to zone models. A multi-compartment field model may require the use of a mainframe computer for one to four weeks in order to perform all of the calculations.

The information required for both field and zone models is the same. A good description of the required inputs, as well as the limitations of computer models has been added to the 2001 edition of NFPA 921.

As in other areas of fire investigation, two experts provided with the same program can plug in different assumptions and reach different conclusions about the spread of the fire. This is because of the large number of variables which affect the fire's behavior. Although computer modeling has been touted as a method for testing an investigator's hypothesis, the vast majority of computer models that are likely to be presented to a jury will demonstrate, but not prove, an expert's opinion.

Because of their ability to graphically present the growth of a fire, computer models are becoming commonplace in fire litigation.

58. NFPA 921, supra note 1, at 21.

. . ~ .... -.

59. HAROLD NELSON. FPETOOL USERS GUIDE (National Institute of Standards and Technolo­gy Pub. No. 4439) {1990) .

374 FIRES, ARSONS AND EXPLOSIONS Ch. 7

(11) Fatal Fires

Fires that involve fatalities are more likely to become the subject of civil or criminal litigation than fires that cause only property damage. The method­ology of investigating a fatal fire is exactly the same as the methodology involved in investigating a property fire, except that there is one important piece of evidence provided in the fatal fire: the body. In those cases where the victim dies at the scene, the body can provide invaluable information as to the condition of the atmosphere at the time of death.

Carboxyhemoglobin (COHb) and blood alcohol readings are imperative for a proper understanding of what occurred. Low carbon monoxide content in a victim's blood suggests that they were rapidly overcome by heat, and died from burn injuries, rather than smoke inhalation, the most common cause of fire death. Highe;r carbon monoxide concentrations (around 50%), on the other hand, suggest exposure to a gradual build-up of smoke. Still higher levels of CO suggest brief exposure to very high concentrations of toxic smoke. Such exposures are typical of victims found away from the origin of a fire. Those intimate with the originating fire are unlikely to be still breathing by the time the fire produces high concentrations of CO. Fire extending from the room of origin undergoing flashover can rapidly spread deadly concentrations of CO throughout the building. 60

Low carbon monoxide (CO) concentrations have been used to indicate to a fire investigator that he was looking at an arson, rather than an accidental fire. The problem with this indication is that, like fire damage itself, carbon monoxide poisoning is a result of both the intensity of the exposure (carbon monoxide concentration) and the duration of the exposure. Exposure to a high concentration for a short period of time may result in the same carboxyhemo­globin level as exposure to a low concentration for a long period of time. For instance, exposure to a. concentration of 0.05% CO (500 parts per million) for two to three hours will result in a COHb level of 30o/c. The same result is achieved by exposure to a concentration of 1% CO (10,000 parts per million) for one to five minutes. 61 Of course, a COHb concentration of zero is an indication that the victim was not breathing, and suggests that death may have preceded the fire.

An excellent review of carbon monoxide data compilations has been published by Gordon Nelson.62

The effects of incineration can also lead to mischaracterization of the events leading up to the victim's death. Muscle contraction caused by expo­sure to heat results in a pugilistic pose, which has led investigators to see the victim as fighting off an assailant. 63 Other artifacts of incineration include neck contusions, which have been interpreted as evidence of strangulation, and skull fractures, caused by the expansion of cranial contents, which have been misinterpreted as evidence of bludgeoning. 64 The knowledge and experi-

60. DEHAAN, supra note 21, at 380. 61. Id. at 311. 62. G.L. Nelson, Carbon Monoxide and Fire

Toxicity: A Review and Analysis of Recent

63. DEHAAN, supra note 21, at 308.

64. State v. Girdler, No. 9809 (Superior Ct. of Arizona, Maricopa County, Jan. 3, 1991).

Work, 34 Fm: TEcliNOLOGY 39 . ..:(:::.19:.:9::::8.:.:,). __________________ _

Sec. 7-2.2.1 FIRES, ARSONS AND EXPLOSIONS 375

ence of the medical examiner with burn victims should be carefully scruti­nized before allowing these sorts of conclusions into evidence.

(12] Explosions

Procedures for investigating an explosion are similar to those used in fire investigations. A more detailed examination of the surrounding area is gener­ally required, particularly in the case of chemical explosions. 66

Historically, explosions have been difficult to define, because there are several types of explosions, some of which are difficult to distinguish from rapid combustion. For this discussion, let us describe an explosion as an event having the following four characteristics: high pressure gas, confinement or restriction of the pressure, rapid production or release of the pressure, and change or damage to the confining or restricting structure or vessel.

Two major types of explosions may occur: mechanical explosions, such as steam boiler explosions, and chemical explosions, which encompass combus­tion explosions and the detonation of high explosives.

In a mechanical explosion, no chemical or combustion reaction is neces­sary, although mechanical explosions caused by boiling liquid and expanding vapor (BLEVE) frequently happen as a result of heating a sealed container of liquid in a fire. If the liquid is flammable, a chemical explosion may follow the mechanical explosion.

Chemical explosions may be caused by the sudden ignition of dusts, gas/air mixtures, or vapor/air mixtures. These are known as combustion explosions. An explosion in a cloud of smoke from a pre-existing fire is known as a backdraft. Most of the explosions described so far are accidental in nature. Explosions fueled by chemicals whose primary function is to explode are more likely intentional.

All explosions, whether mechanical or chemical, are grouped into two categories: low order and high order. Low order explosions are characterized by a widespread "seat" or no "seat," and by the movement of large objects for short distances. High order explosions are characterized by a well defined "seat," where the energy of the explosion creates a shattering effect, and typically a crater. High order explosions tend to project small objects for long distances.

Determination of the origin or epicenter of an explosion is carried out by searching the perimeter of the scene, locating and documenting projected debris, and developing force vector diagrams. This task may be complicated by secondary explosions, which appear to have more than one "origin." Once the origin is observed, conclusions can be drawn about the type of fuel involved and, if necessary, samples selected for laboratory analysis.

While the types of materials involved in commercial or industrial explo­sions are too numerous to cover in this chapter, the potential fuels for residential explosions are very limited. Unless the explosion is a backdraft, easily recognized by the smoke staining on the projected objects, the potential

66. NFPA 921, au.pro note 1, chapter 13.

376 FIRES, ARSONS AND EXPLOSIONS Ch. 7

sources _of fuel are limited to natural and LP gas, and flammable liquid vapors.

NFPA 921 contains an excellent discussion of the techniques of explosion investigation, and a recent National Institute of Justice guide dealing with the responsibilities of responders to explosion or bombing scenes contains much useful information. 66

(13] Smoke Detectors

According to statistics compiled by the National Fire Protection Associa­tion, residential fire deaths in the United States have dropped from a high of 6,015 in 1978 to 3,360 in 1997.6; The National Smoke Detector Project-a joint project among the Consumer Product Safety Commission, the Congressional Fire Services Institute, the U.S. Fire Administration, and the National Fire Protection Association-issued a major report in October 1994 on the use of home smoke detectors, and characterized the home smoke detector as the fire safety success story of the decade. According to the 1994 report, smoke detectors cut the risk of dying in a home fire by roughly 40%. In the ten years ending in 1995, the death rate from fires in homes with a smoke detector present was 45'k lower than the death rate from fires in homes with no smoke detector present. 6$

Of course, once technology comes into being that can save lives, certain failures of the technology become occasions for tort litigation. 69 The National Smoke Detector Project study found that nearly . all of the smoke detectors that failed to operate did so because their batteries were either dead or disconnected. Some research, however, has indicated that for certain types of smoldering fires, the most common type of detector, the ionization detector, does not respond as quickly to the large particles generated by smoldering fires as a different type of detector, the photoelectric detector. 70 The general consensus of the scientific community involved in smoke detector research, and the vast majority of the literature,71 however, supports the proposition that the differences in response time are not significant with respect to smoldering fires, and the ionization detector's faster response to the more immediately dangerous flaming fire makes it the detector of choice. In recent litigation, smoke detector manufacturers have been sued for failing to incorpo­rate a photoelectric detector into their smoke alarms, and the plaintiffs have had some success. i2

66. See supra note 19. 67. 1997 Fire Loss in the US, 92 NFPA

JOURNAL, Sept./Oct.1998, at 72. 68. CoNSUMER PRooucr SAFETY Co}OIJSSION,

SMOKE DETECTOR 0PERABILITI' St'RVEY: REroRT ON FINDINGS (1994).

69. See Mark Grady, Why Are People Negli­gent? Technology, Nondurable Precautions, and the Medical Malpractice Explosion, 82 Nw. U. L. REv. 293 (1988J.

70. R.G. BILL. THE RESPONSE OF SMOKE DETEC,

TORS TO SMOLDERING-STARTED F'mEs IN A HOTEL

OccUPANCY (Factory Mutual Research, Nor­wood, MA) (1988).

71. R. Bukowski & N. Jason leds.) b,T'L FIRE DETF.CTJON BIBLIOGRAPHY 1975-1990 (NIS­TIR 4661, Building and Fire research Labora­tory, Gaithersburg, MD) (1991).

72. See, e.g., Gordon v. BRK Brands, Inc., No. 992-0771 <Circuit Ct. of the City of St. Louis, July, 1999) (settled after a $50 million verdict), or Mercer v. Pittway Corp., 616 N.W.2d 602 (Iowa 2000) ($16.9 million trial verdict for compensatory and punitive dam­ages, reversed in part and remanded for new trial).

Sec. 7-2.2.1 FIRES, ARSONS AND EXPLOSIONS 377

Most smoke detectors use a small amount of radioactive material to ionize particles of air in the detection chamber. This causes a current to be conducted between two electrodes. The presence of small particles of smoke in the ionization chamber interferes with this passage of current and triggers an alarm. In photoelectric detection chambers, there is a light emitting device and a light detecting device. The light emitting device is aimed away from the detection device, but the presence of smoke particles causes light to be reflected to the detection device, which sets off the alarm. Photoelectric detectors are not as sensitive to particles smaller than one micron (character­istic of flaming fires) as are ionization detectors. Ionization detectors are not as sensitive to particles larger than one micron (characteristic of smoldering fires) as are photoelectric detectors. All fires produce a wide range of particle sizes, and both types of detectors have been evaluated and found to provide adequate warning.73 It is possible to build a smoke alarm that utilizes both types of detectors, and the argument has been advanced that alarms that incorporate only ionization detectors are therefore dangerous and defective. Unfortunately, when the smoke detector manufacturers put combination units on the store shelves, they stayed there. Consumers seem to be motivated largely by cost in their selection of smoke alarms. Litigation surrounding smoke detector design is likely to continue, but as of this writing, there have been few appellate decisions on the subject.

[14] Stolen Autos Recovered Burned

This common scenario requires almost no investigation to determine that the fire was intentionally set. The chances that a vehicle happened to catch fire accidentally after it was stolen is almost not worth considering .. The question in cases such as this is not whether the car was set on fire, but who did it. If an insurance company can prove that it was their insured who set the fire, or arranged the theft and fire, the company can avoid payment. Historically, this has been difficult to prove.

A new technique, bearing some resemblance to traditional toolmark analysis, purports to be able to determine the "last key used" to move a vehicle. This technique has no support in the relevant scientific community of firearms and toolmark examiners, but has nonetheless proven popular with insurers, and has been admitted over Daubert objections in several jurisdic­tions. Challenges are rare because the stakes are usually too low to support the involvement of adverse experts to refute the claim of the "forensic locksmith."

The proponents of this technique submitted a proposal to the NFP A to include it as a tool for vehicle fire investigations, but the Technical Committee rejected the proposal because there was no scientific evidence supporting the validity of the technique.14

[15) Presumption of Accidental Cause

Because the individual making the arson case frequently lacks scientific training, the presumptions which that individual carries into a fire scene

73. Ionization Versus Photoelectric: Choos- 74. See Report on Comments, supra note ing the Right Smoke Detector, 30 BUD..DING Om- 33. CIAL & ConE ADMIN., Nov./Dec.1996, at 17.

378 FIRES, ARSONS AND EXPLOSIONS Ch. 7

should be closely scrutinized. Just as the assumptions which are plugged into a computer model can affect the outcome of the analysis, so will the assump­tions which a fire investigator carries with him into a fire scene affect the outcome of his investigation.

Because of the large amount of evidence destroyed in a fire, it is possible to "prove" almost any fire scene to be the result of arson, if one is bent on doing so. This idea is conveyed by DeHaan: "If an investigator decides that a fire is arson before collecting any data, then only data supporting that premise are likely to be recognized and collected. " 75 DeHaan, of course, was inspired by Holmes (Sherlock., not Oliver Wendell), who stated, "It is a capital mistake to theorize before one has data. Insensibly, one begins to twist facts to suit theories, instead of theories to suit facts." 76

Many fire investigators will state that they carry no presumptions into a fire scene with them, and rely on an objective evaluation of the evidence to reach their conclusions. NFPA 921 urges upon investigators the scientific method of hypothesis development and hypothesis testing. The question is: Should there be a hypothesis before all of the evidence has been observed? It could be argued that the proper presumption to carry into a fire scene is a presumption of accidental cause, i.e., all fires are presumed accidental until proven otherwise. Such a presumption protects the presumption of innocence accorded to individuals. Many states have codified this presumption of acci­dental cause into the standard jury charge for arson, but whether codified in a particular jurisdiction or not, the fire investigator who fails to apply the presumption of accidental cause to all fires will eventually make an erroneous declaration of arson.

The error will result from a misinterpretation of circumstantial evidence. In nearly every fire case, it is circumstantial evidence which allows the cause of the fire to be deduced. Likewise, in nearly every arson case, the corpus delicti is proven by circumstantial evidence, and the jury is read the standard circumstantial evidence charge. Mr. Holmes described the perils of circum­stantial evidence in The Bascombe Valley Mystery:

Circumstantial evidence is a very tricky thing. It may seem to point very straight to one thing, but if you shift your own point of view a little, you may find it pointing in an equally uncompromising manner to something entirely different/ 7

In many instances, if there is one survivor of a fire, particularly a fatal fire, and the fire is determined to have been the result of arson, then only one conclusion can be reached-the survivor did it. This is because, in the investigator's "opinion," the survivor's description of events, which typically depict an accidental fire, is "impossible," and therefore, the survivor is lying. This is exactly what happened to Ray Girdler. Judge James Sult, who presided over the first trial and sentenced Girdler to life in prison, wrote in his opinion remanding the case for a new trial:

75. DEHAAN. supra note 21, at 4. 76. Arthur Conan Doyle, A Scandal in

Bohemia, in THE A.,-xoTATED SHEBLOCK HOLMES (William S. Baring-Gould ed., 1967!.

77. Arthur Conan Doyle, The Boscombe Valley Mystery, in THE ANNOTATED 8m:m.ocK HOLMES (William s. Baring-Gould ed., 1967).

Sec. 7-2.2.1 FIRES, ARSONS AND EXPLOSIONS 379

The newly-discovered evidence would probably change the verdict upon a retrial of this case. Several considerations support this finding: ...

[A]t the trial of the case, the State claimed, based on then understood fire investigation evidence, that Mr. Girdler's account of the fire was impossi­ble and, therefore, false. The new evidence shows that Mr. Girdler's observations of the fire are consistent with a flashover fire of innocent origin.7s

If the state's investigator had the proper scientific approach to fire investigation, or even admitted the possibility that an explanation other than burning flammable liquids (none were detected in laboratory analysis) existed, the erroneous conviction might have been avoided.

(16] The "Negative Corpus"

Since the advent of scientifically-based fire investigation, one of the thorniest issues for fire investigators has been the determination of fire cause when the evidence has either burned up or been taken from the scene by the firesetter. "Negative corpus," short for negative corpus delicti is fire investi­gator shorthand for the determination that a fire was incendiary based on the lack of evidence of an accidental cause. Such determinations have generally been held in low regard by the proponents of scientific fire investigation, but that has not prevented their introduction into evidence. The case of Michigan Millers Mutual Insurance Corp. v. Benfield19 was a "negative corpus" determi­nation. When fire investigators testify that a fire was intentionally set, "the elimination of all potential accidental causes" is frequently added to other evidence of incendiary activity.

The NFPA Technical Committee on Fire Investigations struggled with the concept of "negative corpus" for several years. Despite the lack of a demonstrable ignition source, many fires can be stated to have been set based on the absence of any other possibilities. The Committee's challenge was to limit the abuse of the negative corpus determination, and to put legitimate determinations of incendiary activity into the context of the scientific method. The result of the Committee's work, published in the 2001 edition of NFPA 921 is as follows:

Process of Elimination. Any determination of fire cause should be based on evidence rather than on the absence of evidence; however, when the origin of a fire is clearly defined, it is occasionally possible to make a credible determination regarding the cause of the fire, even when there is no physical evidence of that cause available. This may be accomplished through the credible elimination of all other potential causes, provided that the remaining cause is consistent with all known facts. For example, an investigator may properly conclude that the ignition source came from an open flame even if the device producing the open flame is not found at the scene. This conclusion may be properly reached as long as the analysis producing the conclusion follows the Scientific Method as discussed in Chapter Two.

78. State v. Girdler, No. 9809 (Superior Ct. of Arizona, Maricopa County, Jan. 3, 1991).

79. 140 F.3d 915 (11th Cir.).

380 FIRP.8, ARSONS AND EXPLOSIONS ····-.. ..--:- -:- Cb. 7

"Elimination," which actually involves the testing and rejection of alter­nate hypotheses, becomes more difficult as the degree of destruction in the compartment of origin increases, and is not possible in many cases. Any time an investigator proposes the elimination of a particular system or appliance as the ignition source, the investigator should be able to explain how the appearance or condition of that system or appliance would be different than what is observed, if that system or appliance were the cause of the fire.

There are times when such differences do not exist, for example, when a heat producing device ignites combustil?les that are placed too close to it, the device itself may appear no different than if something else were the ignition source.

The "elimination of all accidental causes" to reach a conclusion that a fire was incendiary is a finding that can rarely be scientifically justified using only physical data; however, the "elimination of all causes other than the application of an open flame" is a finding that may be justified in limited circumstances, where the area of origin is clearly defined and all other potential beat sources at the origin can be examined · and credibly eliminated. It is recognized that in cases where a fire is ignited by the application of an open flame, there may be no evidence of the ignition source remaining. Other evidence, such as that listed in Section 19-3, which may not be related to combustion, may allow for a determi­nation that a fire was incendiary . In a determination of an accidental cause, the same precautions regarding "elimination" of other causes should be carefully considered.

Note that nowhere in the above quotation does the term "negative corpus" appear.

The above language represents a compromise between the presumption of accidental cause, and the knowledge that in many cases, particularly where the ignition source is an open flame, incendiary fires may leave behind little physical evidence of their cause.80

[ 17) Certainty of Opinions

Few legal issues other than the cause of fires rely so heavily on the opinion of the investigator. Even in the case of explosions, which may be equally destructive or more destructive than fires, the fact that an explosion occurred drastically limits the number of potential causes.

Fire investigators have struggled with the question of certainty for years, raising such questions as whether an investigator's "comfort level" with his opinion should be stronger in a criminal case than in a civil case. On cross examination, many investigators will admit that they are not infallible, yet nevertheless go on to assert that there is no other possible explanation for their observations than what they have offered.

The uncertainty about certainty has generated much discussion in the fire investigation community, as illustrated by the discussion in the previous

80 . ATIO!\'AL FIRE PR OTECTION AssOW,T ION, GAT!Ol\'S ( P UB. No. 921) (2001). ST ANDARD GtnDE FOR FIRE AKO EXPI..OSIO!\" b "VESTI-

Sec. 7-2.2.2 FIRES, ARSONS AND EXPLOSIONS 381

edition of this chapter. Because it seemed impossible to separate a codification of "comfort level" from legal burdens of proof, the Technical Committee on Fire Investigations voted in 1998 to remove the discussion about levels of certainty from the document.

[18] Conflicting Opinions

There is a curious notion in the fire investigation community that every fire investigator is entitled to his own opinion about the cause of a fire. There is even a tacit recognition of the possibility of investigators reaching different conclusions after making the same observations of the same fire scene in the International Association of Arson Investigators CooE OF ETHICS, which in­cludes the rule, "I will remember always that I am a truth seeker, not a case maker.'' 81

Unfortunately, due to the lack of scientific training in the discipline, many investigators do not understand the concept of a "professional" opinion. Certainly, very few investigators will grant their physicians the same right to make a misdiagnosis based on their observation of a set of symptoms. If two doctors disagree on a diagnosis, the doctors regard it as their duty to cooperate and attempt to reach the correct conclusion. They would be uncom­fortable knowing that one of them was wrong if they did not do so. Such cooperation in the search for the truth, particularly when arson is alleged, is so far a relatively rare occurrence in fire investigations. Fire investigators with differing views most often leave it up to the trier of fact to decide who is right, even though the legal fact finder is likely to be less knowledgeable about the substance of the expert testimony than either investigator.

§ 7-2.2.2 Laboratory Analysis Unlike the field investigation of fires, there are considerably more areas

of agreement and fewer areas of disagreement in the laboratory analysis of fire debris, and since the early 9Os, a near consensus has developed in the scientific community regarding the proper techniques to be applied to samples of fire debris in which it is suspected that ignitable liquid residues are contained. Two chemists, looking at the same data from a fire debris sample, are more likely to agree on its interpretation than are two field investigators looking at the same fire scene, but disagreements still occur, and these are usually due to one of the chemists failing to meet industry standards.

[1] Standard Methods of Sample Preparation

The industry standard for the laboratory analysis of fire debris is embod­ied in ASTM E 1387, STANDARD TEST METHOD FOR IGNITABLE LIQUID RESIDUES IN EXTRACTS FROM SAMPLES OF FIRE DEBRIS BY GAS CHROMATOGRAPHY.82 A second standard, E 1618, STANDARD TEST METHOD FOR IDENTIFICATION OF IGNITABLE LIQUID RESIDUES IN EXTRACTS FROM SAMPLES OF FIRE DEBRIS BY GAS CHROMATOGRAPHY/MASS 8PECTROMETR-v83 also has been adopted by ASTM. It is agreed almost unani­mously in the forensic science community that gas chromatography is an

81. Th'TERXATIONAL AssocIATJON OF ARsoN IN. 83. Id. VESTIGATOBS, lAAI CoDE OF ETHICS (1949).

82. Supra note 11.

382 FIRES, ARSONS AND EXPLOSIONS Ch. 7

essential requirement for the identification of common petroleum-based prod­ucts. Gas chromatography/mass spectrometry and gas chromatography/infra­red spectrophotometry, known as "hybrid" techniques, provide more informa­tion, but are basically more sophisticated versions of gas chromatography. Gas chromatography has been the accepted method of analyzing petroleum prod­ucts since the 60s, but there have been considerable improvements in the field. These improvements and variations on the technique of gas chromatog­raphy are reported in peer reviewed journals such as the JOURNAL OF FORENSIC

ScIENCES, the Jon.VAL OF THE FORENSIC ScIENCE SocIETY, ANALYTICAL CHEMISTRY,

and others.

There have also been numerous improvements in sample preparation techniques over the years. These improvements are also likely to be docu­mented in the literature, and most commonly used sample preparation ~­niques are described in ASTM standards.

Headspace analysis (ASTM E 1388) is the simplest of the sample prepara­tion techniques. This method is rapid, but not highly reproducible, and not highly sensitive to the heavier hydrocarbons such as those found in diesel fuel. The sample is warmed and a syringe is used to withdraw a small volume of the air above the sample, known as the headspace. This headspace is then injected directly into the gas chromatograph.~

Steam distillation (ASTM E 1385) is a classical technique which relies on the immiscibility of oil and water. A visible oily liquid can be separated from the sample, and then diluted or injected directly into the gas chromatograph. This technique is time consuming, and is not sensitive to very low concentra­tions of ignitable liquids, which are often all that remains in fire debris samples. When applied to a sufficiently concentrated sample, the visible liquid that the technique produces, however, can make a very convincing exhibit. 85

When the jury can actually see the recovered liquid, and perhaps smell it and see it burn, they will not likely feel the need to understand the intricacies of gas chromatography.

Solvent extraction (ASTM E 1386) is another classical technique which is highly sensitive, but which has the disadvantage of dissolving materials other than the ignitable liquid residues of interest. It is also dangerous, and destructive of evidence. This is a technique best applied to very small samples and to the problem of determining what was inside a now empty container. 86

Headspace concentration techniques (ASTM E 1412 and E 1413) employ an adsorbent to trap volatile materials present in the headspace above a warmed sample. These adsorption/elution techniques are highly sensitive, highly reproducible, and the passive headspace concentration technique is both simple to use and essentially nondestructive of evidence. The sample can be analyzed repeatedly, by different laboratories if necessary. Passive heads-

84. AMERICAN 8oc:IETY FOR TESTING AND MATE­RIALS., STANDARD PRArncE FOR SAMPLING oF HEADS­PACE V APOBS FROM Fm DEBRIS SAMPLES (Pua No. E 1388) (1995).

85. AMERICAN Socrrn FOR TESTING A.'\i> MATE­RIALS, STANDARD PRACTICE roR SEPARATIO:S A.,i> CON­

CENTRATION OF IGNITABLE UQUID REslDUES FROM

FIRE DEBBIS SAMPLES BY STEAM Dlsnu.ATION (PuB No. E 1385) (1995).

86. .AMERICAN 8ocmTY FOR Tf.sTJNG AND MATE­RIALS, STANDARD PRACnCE FOR SEPARATION AND CoN­

CENTRATION OF IGNITABLE LIQUID REsiDUES FROM FIRE DEBBIS SAMPLES BY SoLVENT ExTRACTION (PuB. No. E 1386) (1995). -------------

Sec. 7-2.2.2 FIRES, ARSONS AND EXPLOSIONS 383

pace concentration is rapidly becoming the "method of choice" m modern forensic science laboratories. 87

All of the above sample preparation techniques are scientifically valid. Sample size, ignitable liquid concentration, and the analyst's experience and preference will determine which method of separation is selected. Regardless of separation technique, the analytical methods recognized as valid are limited to those involving gas chromatography.

[2] Classification of Ignitable Liquids

Until recently, it was generally believed that ignitable liquids could be placed into one of five major classes88 or a sixth "miscellaneous" class. Innovations by the petroleum industry, however, made the classification system used by forensic scientists become cumbersome and confusing. Conse­quently, the ASTM Committee on Forensic Sciences has changed the struc­ture of its classification system to utilize descriptive class names, and to dispense with class numbers. Distinctions within any one of these classes are very difficult, if not impossible.89 Once an ignitable liquid has been exposed to a fire, its character changes to the extent that its source is very difficult to identify. Some work has indicated that source identification is possible if a sample is less than 30% evaporated (i.e., at least 70% of the original weight remains). There are times, however, when ignitable liquids are mixed, produc­ing a unique pattern which can conceivably be identified with a source.

The exposure of a petroleum distillate to a fire results in its evaporation, with the lower boiling point compounds being preferentially evaporated over the higher boiling point compounds. This results in an increase in the average molecular weight of the mixture. It is also generally recognized that it is not possible to distinguish whether a sample has been exposed to a fire or to room temperature evaporation. A sample of petroleum distillate that has burned to 50% of its original volume or weight will give a gas chromatographic pattern which is indistinguishable from a sample that has evaporated to that point.

[3] Detection of Explosives

Because of the relative rarity of bombing incidents, the cadre of scientists regularly dealing with the detection and identification of explosive residues is very small. Most private laboratories have only primitive explosive detection capabilities, and most state and local government laboratories are not much better equipped. Techniques for explosive detection and identification appear in the literature, but few laboratories are capable of repeating the published analyses. Techniques used by explosives chemists are as varied as the explo­sives themselves. The following are techniques used in the federal laboratories

87. AMERICAN SOCIETY FOR TESTING AND MATE­

RIALS,. STANDARD PRACTICE FOR SEPARATION AND

CONCENTRATION OF foJl.1TABLE LIQUID RESIDUES FROM

FIRE DEBRIS SAMPLES BY PASSIVE HEADSPACE CON­

CENTRATION (PUB No. E 1412) (1994); AMERICAN

SOCIETY FOR TESTING AND MATERIALS., STANDARD

PRACTICE FOR SEPARATION AND CONCE?\"TllATION OF

IGNITABLE LIQt,1D RESIDUES FROM Fuu: DEBRIS SAM-

PLES BY DYNAMIC HEADSPACE CONCENTRATION (PUB. No. E 1413) (1995).

88. Class l: light petroleum distillates, Class 2: gasoline, Class 3: medium petroleum distillates, Class 4: kerosene, Class 5: heavy petroleum distillates.

89. ASTM E 1387, supra note 11.

384 FIRES, ARSONS AND EXPLOSIONS Ch. 7

on a routine basis: thin layer chromatography, gas chromatography, gas chromatography/mass spectrometry with chemical ionization, infrared spec­trophotometry, high performance liquid chromatography, energy dispersive x­ray analysis, x-ray diffraction, and capillary electrophoresis, one of the newer techniques.

As in the analysis of petroleum distillates in fire debris, the critical first step in the analysis of explosive residue is the separation of the residue from the debris. The salts which are the products of the explosive reaction are removed from the debris by a cold water extraction, while the unreacted or partially reacted residue of the explosive itself is removed using an organic solvent. These concentrated extracts are then analyzed by one, or usually several, of the above techniques.

For explosions caused by fuels other than chemicals designed to explode, gas chromatography is the usual method of analysis. Gasoline, the most common fuel for explosive vapor/air mixtures, is detected as described previ­ously. More sophisticated gas chromatography is required to detect ethane, which is found in natural gas, but not in sewer gas. Odorization of natural and LP gases is frequently an issue in explosion cases. Quantitation of the odorant level may be accomplished by gas chromatography or through an odor panel. Reagent tubes can be used to detect the ethyl mercaptan or thiopane used to odorize fuel gases.

§ 7-2.3 Future Directions The laboratory analysis of fire debris is about as "settled" as any forensic

science is ever likely to be. The gas chromatograph/mass spectrometer can provide almost total characterization of complex mixtures to allow for un­equivocal identification of the petroleum products that are likely to be used as accelerants. The techniques of sample preparation have reached the practical limit of what is desirable to detect. More sensitive levels of detection increase the risk of identifying ignitable liquid residues that are part of the normal background. The simplicity of the techniques available to achieve current levels of detection provides little impetus to improve the techniques. The impetus in the field is generally to improve the quality of work done by laboratories that have yet to adopt techniques that are generally recognized as valid. As the ASTM methods first adopted in 1990 and 1991 come into more widespread use, laboratories that fail to follow these minimum standards can expect to see their results challenged more frequently and more vigorously.

With the lack of a frontier, more laboratory scientists are stepping out into the field, and applying their scientific skills to the understanding of the behavior of fire. The National Fire Protection Association (NFPA) and the National Institute of Standards and Technology (NIST) are both looking at ways to repeat the experiments of the 70s and 80s, but this time, the researchers will look at the aftermath, rather than just the fire itself. Numerous test burns should be recorded in the next few years, and the information which comes out of them should greatly improve the quality of field fire investigation work.

As more canines are brought into the field of accelerant detection, a body of knowledge, including peer reviewed research, is likely to come into being. The use of accelerant detecting canines may free up large amounts af fire

Sec. 7-2.3 FIRES, ARSONS AND EXPWSIONS 385

investigators' time, allowing overworked state and local officials to concen­trate on those fire scenes most likely to result in prosecutable arson cases.

Computer modeling is likely to assume a much larger role in the future, particularly as data come in from more test burns, and can be used to test a model's predictions.

Certification of field investigators by the International Association of Arson Investigators or by the National Association of Fire Investigators is becoming more common. Neither certification program guarantees the compe­tence of the witness or the correctness of his findings, but the programs do serve a useful purpose in encouraging the fire investigation community to identify some areas of agreement and to study areas of disagreement.

Certification of laboratory analysts through the American Board of Crimi­nalistics began only in 1994, so it will be some time before there is a large core of certified fire debris chemists. As more scientists leave the laboratory to do field research in the area of fire behavior, the understanding of fire behavior is likely to improve, and the quality of fire investigations is likely to benefit from the application of a scientist's natural skepticism to the outdated or unsupported beliefs held by many field investigators. While there are still far too many cases of incorrect fire analyses, the profession is moving incrementally toward a more accurate calibration of expectations.

Appendix

Glossary

Accelerant. An agent, often an ignitable liquid, used to initiate or speed the spread of fire.

Arc. A luminous electric discharge across a gap. If the arc generates sufficient energy, an arc bead may be formed. An arc bead is a round globule of re­solidified metal at the point on an electrical conductor where the arc occurred.

Capillary Electrophoresis. An analytical separation technique which utilizes electric charge to separate and analyze sub-milligram quantities of chemical substances. Capillary Electrophoresis is useful in many types of analytical chemistry, including the detection of explosives and gunshot residues.

Compartment Fire. Any fire which occurs inside an enclosure. Once a fire has progressed beyond the initial free-burning stage, it interacts with the floors, walls, and ceilings of the enclosure and behaves differently from a free­burning fire.

Flashover. A transition phase in the development of a compartment fire in which surfaces exposed to thermal radiation reach ignition temperature more or less simultaneously and fire spreads rapidly throughout the space.

Gas Chromatography (GC). An analytical method for separating and identify­ing mixtures of compounds. A compound's solubility in a stationary phase versus its solubility in a mobile phase allows separation of similar compounds due to subtle differences in physical or chemical properties. Most gas chroma­tography performed on ignitable liquid residues relies on differences in boiling points to effect the separation. GC is the fundamental first step in the analysis of any ignitable liquid residue. The output of the GC is known as a chromatogram. Infrared Spectrophotometry (JR). An analytical method which measures the absorbance of radiation having a wavelength slightly longer than the wavel­ength of visible light. This method is used to characterize the functional groups present in a sample, and is frequently applied to polymers and drugs. The utility of IR is limited in ignitable liquid residue analysis because most ignitable liquids are mixtures, and infrared spectrophotometry requires pure or nearly pure compounds in order to yield meaningful data. The output of the IR spectrophotometer is known as an absorbance spectrum.

Mass Spectrometry (MS). An analytical method that begins with the breaking up of the compounds of interest by the application of chemical or electrical energy, followed by a measurement of the size and number of ions produced in the ionization step. Like other spectral techniques, mass spectrometry requires pure compounds in order to yield meaningful data. The purification for most mass spectral analysis is accomplished via gas chromatography. Typically, the MS is attached to the output side of a gas chromatograph (GC/MS) column. The output of the mass spectrometer is known as a mass spectrum.

386

Ch. 7 GLOSSARY 387

Odorization. The addition of small concentrations of substances to a fuel gas

in order to make it detectable by smell. The two common fuel gases, natural

gas and LP gas, have no odor. Odorants such as ethyl mercaptan or thiophane

must be added to fuel gases in order to make them detectable at a concentra­

tion not over one-fifth of the lower limit of flammability.

Radiometer. A collection of thermocouples encased in a solid conductive metal

jacket (e.g., copper) which is cooled by water. By measuring the voltage

difference between the thermocouples exposed to the fire and the thermocou­

ples exposed to the water, and taking into account the surface area of the

case, the radiative flux in watts per square centimeter (or kilowatts per

square meter) can be measured directly.

Thermocouple. A device consisting of two dissimilar metal wires which convert

heat energy into electrical energy. A voltage measuring device is attached to

the wires and the temperature at the junction of the wires can be calculated.

This is usually accompiished electronically, and the thermocouple readout,

known as a pyrometer, reads directly in degrees For degrees C.

Thin Layer Chromatography (TLC). A chemical analytical procedure which

separates compounds by their solubility in a solvent and the tenacity by which

these compounds adsorb (adhere) to a thin sheet of silica gel spread out on a

glass plate. Once separated, the spots of analyte can be further characterized

by exposure to a developing agent, which causes the spots to change color. As

in all chromatographic analyses, a comparison is made between a known

substance and an unknown substance. TLC may be used for the separation of

drugs and explosives, and also for the characterization of dyes in automotive

gasoline.