The Harmony of the Physical Evidence

	Kenneth A. Rahn
	The Warren Report and Its Legacy
	A conference sponsored by the AARC and the Wecht Institute
	17–19 September 2004
	Washington, D.C.

The major types of physical evidence

	Rifle
	Bullets and fragments
	NAA
	Cartridge cases
	Fingerprints
	Brown paper bag
	Wounds
	Clothing
	Zapruder film
	Autopsy X-rays and photographs
	Revolver, shells, and jacket from Tippit shooting
	Dictabelt

The physical evidence makes a “physical framework,” a strong web that might better be called an impregnable wall. The NAA knits it all together.

The NAA is truly the “Queen of the physical evidence.”

Bullets and Fragments Received by the FBI

Locations of Fragments

The Two NAA Analyses

	By FBI in May 1964 (for WC)
		Contained multiple systematic errors.
		Kept secret by FBI.
	By Vincent P. Guinn of UC Irvine in 1977 (for HSCA)
		Highly publicized.
		Showed two clear groups of fragments.
		Agreed with FBI’s analysis.

The goal:

	Try to determine how many bullets hit the men by seeing whether the little fragments (too small to have ballistic engravings) could be associated chemically with the larger, engraved fragments.

Result: Two clear groups, each with one big fragment and one or two little ones. Positive evidence for two and only two bullets, both fired from Oswald’s rifle.

	One of the strongest results from any of the physical evidence.

Slide 10

The Two Groups Are Distinct.

Tests of Distinctness

	Just look at individual samples.
	Just look at means and standard deviations.
	Test difference of means.
	General Linear Model analysis.

1. The samples.

Slide 14

2. Means and standard deviations of the groups

Expanded view—means differ by 8 σ.

3. Test significance of means.

Find underlying distribution of Sb in WCC/MC bullets

Try Gaussian (Normal) First

	This is standard statistical practice.

Not a normal distribution—too skewed

Try Lognormal Distribution

	Standard statistical practice—least difference from normal

Lognormal distribution works fine.

Fragments from assassination fall with the others.

Testing the means

	Use lognormal distribution.
	Means and standard deviations of groups are 6.71±0.04 and 6.43±0.03.
	Two-sided p < 0.0028. (< 1/360 chance that the means are the same)

Same Answer With Normal Distribution

	Group 1: 815 ± 25 ppm (3.1%)
	Group 2: 623 ± 22 ppm (3.5%)
	Two-sided p < 0.0028. (< 1/360 chance that the means are the same)

4. General Linear Model analysis

"Assumes no underlying distribution."

	Assumes no underlying distribution.
	“Group” variable has F-statistic that corresponds to probability of 0.001 to 0.0001.
	That means probability of only 1/103 to 1/104 that the groups arose by chance.
	Agrees with the three previous tests.

Significance of Distinct Groups

"All fragments from Oswald’s rifle"

	All fragments from Oswald’s rifle.
	Oswald’s rifle was fired that day.

Slide 30

"No fragments or cartridge cases..."

	No fragments or cartridge cases were planted.
	Must have been a forward snap. (Bullet from rear must snap head forward.)

For anyone who doubts the forward head snap…

"Predicts proper speed of forward..."

	Predicts proper speed of forward snap (with simple physics).
	Big rearward lurch not from frontal hit. (Both bullets hit from rear; lurch has properties of something other than bullet.)

"Renders locations of entrance and..."

	Renders locations of entrance and exit wounds to JFK’s head moot.
		Bullet came from Oswald’s rifle in TSBD (ballistic engraving on front-seat fragment).
		Bullet passed through JFK’s head (NAA match to head fragment).
		Bullet came to rest on front seat.
		So don’t need to know where it entered and where it exited.

"Renders location of JFK’s back..."

	Renders location of JFK’s back wound moot.
		Bullet came from Oswald’s rifle in TSBD (engravings on CE 399).
		Bullet had to pass through Kennedy’s body (DBH).
		Bullet hit Connally’s arm and left fragment (NAA match to CE 399).
		So don’t need to know details of passage through JFK’s body.

"Renders offset of holes in..."

	Renders offset of holes in JFK’s clothing moot. (Same reasoning as above.)
	Invalidates all conspiracy theories with other shooters or planted bullets. (To be demonstrated shortly.)

"Leads to best shooting scenario"

	Leads to best shooting scenario.
		First shot early (Z150–160 or so). Missed and hit street or grass. Rushed shot as car passed under tree.
		Second shot around Z222–224. Passed through both men and recovered as CE 399 (stretcher bullet).
		Third shot Z312–313. Passed through right rear of JFK’s head. Exited as 3 large fragments, 2 of which were recovered from front seat. The 3rd (large piece of lead core) flew over windshield and down to Tague.

"Brings Oswald much closer to..."

	Brings Oswald much closer to the crime:
		His rifle did it all.
		His prints on rifle and boxes.
		His clipboard.
		His paper bag.
		His backyard photo with rifle and pistol.
		His flight from TSBD.
		His killing of Tippit in cold blood.
		His attempt to kill again in Texas Theater.

Two Bullets From Oswald’s Rifle Did It All.

The Improbability Of Conspiracy Theories

One Random Match

Two Random Matches

Five Random Matches

In Other Words, All Popular Conspiracy Theories Eliminated Mathematically

Recent Objections by Stu Wexler

George A. Miller, Harvard University, “The magical number seven, plus or minus two: some limits on our capacity for processing information,” in The Psychological Review, Vol. 63, No. 2, March 1956.

“My problem is that I have been persecuted by an integer. For seven years this number has followed me around, has intruded in my most private data, and has assaulted me from the pages of our most public journals. This number assumes a variety of disguises, being sometimes a little larger and sometimes a little smaller than usual, but never changing so much as to be unrecognizable. The persistence with which this number plagues me is far more than a random accident. There is, to quote a famous senator, a design behind it, some pattern governing its appearances. Either there really is something unusual about the number or else I am suffering from delusions of persecution.”

Examples of his harsh statements on the NAA and our procedures

"11/30/2002,"

	11/30/2002, aaj: “Larry and Ken both need to actually acquaint themselves with the literature on the subject before they respond [to Stu in the newsgroups]. They made fundamentally wrong assumptions before they did their work. They flouted convention. They failed to go to actual forensic chemists to find out how REAL forensic chemists couch their conclusions and why.”
	12/01/2003, aaj: “It’s difficult to argue with people who assume they are right and remain willfully ignorant of up-to-date published material on protocols, convention and statistics.”

"11/28/2003,"

	11/28/2003, aaj: “What Ken cannot deny is that his entire case rests upon an edifice of junk science. Without any primary sources, without any background in the statistical arguments being brought against the entire science as a whole, Larry and Ken have put forth a contrived scenario where Oswald had the unfortunate luck of choosing the only known brand of bullets “exempted” form the National Academy of Sciences report.”
	7/06/2004, aaj: “Ken knows full well that his own analysis flies completely in the face of the last 10 years of best-practice in the area of Comparative Bullet Lead Analysis.”

Stu’s core objections to the JFK NAA and why they are wrong

They are based on the 2004 report
Weighing Bullet Lead Evidence
by the Committee on Scientific  Assessment of Bullet Lead Elemental Composition Comparison, National Research Council of the National Academies

Brief Summary of The Report

	Subject: Compositional Analysis of Bullet Lead (CABL), used to examine potential links between bullets or fragments from crime scenes and boxes of bullets found in the possession of a suspect.
	Findings: The links are often overstated. Procedures and interpretations need to be tightened. CABL still useful.

Stu’s objections center on:

	The viability of CABL
	The scenarios for using it
	The JFK bullet leads
	The resulting statistics

The Viability of CABL

	Stu: CABL is a failed technique.
	NRC Report: “In many cases, CABL is a reasonably accurate way of determining whether two bullets could have come from the same compositionally indistinguishable volume of lead.”

Stu: “Vincent P. Guinn’s legacy is essentially going down the toilet.” aaj, 11/21/2003

NAS Committee Member (February 2004): “This [report] should not affect Vince’s legacy.”

Stu: “…at it’s more recent incarnation, the requirement is at *least* 7 elements.” (8/25/2004, aaj)

NAS Report, p. 20: “Although little power to detect matches would be lost if Ag or Bi were dropped from the analytical procedure, using ICP-OES, no time or effort would be saved by measuring five rather than seven elements.”

Slide 56

The Scenarios

	Stu, 11/30/2003, aaj: “There is absolutely no difference [between conventional CABL and its application to the JFK case].”
	Reality: CABL considers broad question—matching bullets or fragments from crime scene to boxes of bullets found somewhere else. JFK NAA is narrow—bullets and fragments from same crime scene. The JFK situation is much tighter, and much stronger conclusions can rightly be drawn from it.

The JFK Equivalent To CABL

	Collect one or two fragments from the crime scene.
	Find Oswald’s box of bullets.
	Analyze them both and compare the compositions.

The Bullet Leads

	Stu: MC bullets are not unique.
	Reality: They are fundamentally different from the bullets considered by the NRC report.

“CABL assumes that a ‘source’ of bullet lead is homogeneous.”

	NRC report, p. 9

The NRC report gives tables to back up this claim of homogeneity.

Within-Bullet Variations Are
Very Small

Within-bullet variabilities are only a few percent

Slide 64

Variations within larger masses are also very small.

"Wires,"

	Wires, Slugs, and Bullets (p. 83): “The extrusion process used to produce the wire from a billet is thought to negate the inhomogeneity due to segregation during solidification… Koons and Grant have sampled wires produced from billets from a pour and found that concentrations remained constant (that is, within analytical precision) over several billets.”

"Wires,"

	Wires, Slugs, and Bullets (p. 83, cont.): “It is reasonable to assume that cutting the wire to produce the slugs and pressing the slugs to form the final bullets produce no substantial segregation of elements in the lead.”

"Pigs,"

	Pigs, Ingots, and Billets (p. 83): “The homogeneity of ingots, pigs, and other large blocks of smelted lead is not an issue…”
	Melt (p. 82): “It is reasonable to assume that a given batch of molten lead exhibits sufficient mixing (such as convective stirring because of the heating process) for compositional homogeneity to develop quickly in the melt, assuming that there are no additions to the molten vat during pouring.”

Report gives data for these larger masses

Bullet-to bullet variabilities are still   only a few percent

Larger masses (melts or lots) become distinguishable, however.

Lot-to-lot variabilities can be greater

But WCC/MC bullets are qualitatively different.

	They vary hugely, both within a bullet and between bullets.

Slide 74

Slide 75

Contrast between CABL and MC lead

	CABL lead is homogeneous up to a melt (a few tons). Differences are first found between melts.
	MC lead is heterogeneous within a bullet, but melts are indistinguishable.
	Huge qualitative difference.
	CABL lead becomes heterogeneous at a ton or more, whereas MC lead is heterogeneous at 0.1 ounce.
	The two types of lead are entirely different.

Stu ignores this huge difference.

	12/27/2003, aaj: “We have absolutely no reason to think that these bullets are fundamentally different from any other bullet.”
	12/27/2003, aaj: “You have no primary source material whatsoever to support the notion ‘exceptionalism’ claim.”

He mocks the idea that MC bullets are different.

	12/27/2003, aaj: “He [Larry Sturdivan] is certain, without ANY primary information, that the vats of lead were made in a way that is completely different from any other known bullet. He completely speculates here, without information from Western.”
	11/30/2003, aaj: “Ken and Larry are claiming that there are bullets ARE UNIQUELYMADE and therefor impervious to the claims against CBLA. They have “special properties” and are not ‘normal.’ … Do you honestly buy that nonsense?”
	12/27/2003, aaj: “Larry and Ken want you to buy into the notion—unsupported at all by any primary information—that the 650–850ppm bullets were some sort of abherrant ‘profile’ coming from a poorly mixed vat.”

"11/30/2003,"

	11/30/2003, aaj: “Don’t you think the argument that Ken and Larry are positing—that Oswald chose the only known bullets immune from the charges against Comparative Bullet Lead Analysis as a whole—is incredibly contrived and should be treated with great skepticism?”
	11/26/2003, aaj: “Oswald was really unlucky in choosing the only ammunition known to man that CBLA can apply to. They know this because Larry Sturdivan, without any primary documentation or confirmation from a metallurgist, has speculated about a hypothetical vat and reverse engineered to get bullets that fit his conclusion. Very solid science there.”

Stu on randomness of bullets:

	11/28/2003, aaj: “…bullet boxes are essentially non-random samples, mixed haphazardly, packaged conviently, without any true a priori knowledge about the number of vats/compositional groups.”
	11/29/2003, aaj: “Bullet boxes are not random samples.”
	8/25/2004, aaj: “Every single thing Ken or Guinn knows about MC leads comes from a small sample that we have every reason to believe is biased, or could be highly skewed.”

The Reality Of Randomness

	Sb follows a smooth, overlapping  lognormal distribution in all four MC lots. That shows that all four lots were sampled randomly.

Slide 82

This extremely important result:

	Justifies our calculations on false matches.
	Keeps the two groups whole.
	Shows that only two bullets hit.
	Harmonizes the other physical evidence.
	Makes sense of the mechanics of the assassination.

Furthermore: The lognormal distribution of Sb in the MC bullets is the same distribution as the default case used by the NRC panel and others:

"p."

	p. 39: “In our analysis, the data are log-transformed.”
	p. 40: “Obviously, the assumption of log-normality is not fully supportable for this element [Cd].”
	p. 41: “The 1,837-bullet set was used primarily to validate the assumption of lognormality in the bullet means…”
	p. 41: “The data on composition of each of the seven elements generally, but not uniformly, appear to have a roughly lognormal distribution.”
	p. 46: “Carriquiry et al. (2002) utilize the assumption of mixtures of lognormal distributions in their analysis of the 800-bullet data set.”

Summary of Stu’s claims:

	All wrong.

The reason for Stu’s harsh campaign:

	His scenario for the assassination is driven by weak evidence and threatened by the NAA.

Stu’s scenario in part

	JFK was hit by three MC bullets from two Mannlicher-Carcanos. They came from two shooters in the rear.
	The first hit was to the body at Z-190, but not from TSBD or Oswald. Evidence includes JFK’s motions in the Z-film, other photos, and reactions of witnesses. The bullet was not CE 399.

"Second hit was to JBC’s..."

	Second hit was to JBC’s back, around Z-224 by another MC bullet, probably from Oswald. It may have been CE 399. But CE 399 did not hit Connally’s wrist.
	Connally’s wrist was damaged by a fragment found in the car (front seat?), possibly from the head shot.
	The third hit was to JFK’s head, at Z-313, from an MC to the rear. Not clear if it was Oswald’s.

"Second shooter from the rear..."

	Second shooter from the rear was required because there was no SBT and the interval between 190 and 224 was too short for a single shooter.

What Stu needs

	A second shooter in the rear.
	BUT HE HAS NO PHYSICAL EVIDENCE FOR THAT SHOOTER.
	That failing, he needs a second rifle in the rear.
	BUT HE HAS NO PHYSICAL EVIDENCE FOR THAT RIFLE.
	That failing, he needs a third MC bullet.
	BUT HE HAS NO THIRD BULLET, BECAUSE THE NAA ONLY ADMITS TWO.

So Stu invokes an accidental match from an undiscovered third bullet from an undiscovered second MC rifle fired by an undiscovered second shooter.

This theory has weak evidence (witness reactions, etc.) driving strong evidence (NAA).

All Stu’s claims are wrong.

	The two scenarios are identical, but they are qualitatively different.
	(In fact, the NAA works far better in the JFK case than in the general CABL case.)
	MC bullets are the same as the others, but real data reveal that they are hugely different.
	They are packaged and distributed the same as other bullets, but that cannot be documented and is almost certainly wrong.
	That 14 samples are too few to represent the population, but that is belied by their actual distribution.

"There is no reason to..."

	There is no reason to invoke the lognormal distribution, but that is the default distribution used in the NRC report and by the Iowa State group.
	There is no primary documentation for the unusual properties of the WCC/MC vats, but that is belied by actual data on within-bullet heterogeneities.
	Guinn’s reputation is “down the toilet,” but the NRC panel disagrees.
	CABL is failed science, but the NRC panel disagrees.
	Larry’s statistics are flawed, but they follow simply and straightforwardly from the clear lognormal distribution.

Final point—Our alleged inability to pass peer-review

	11/25/2003, aaj: “Is it any wonder these guys [Ken and Larry] can’t pass peer-review?!?!”
	12/28/2003, aaj: “I’VE GOT A PRETTY DARN GOOD IDEA THAT HE IS BEING REJECTED AND WHY.”
	7/06/2004, aaj: “This is why Ken couldn’t get published in a peer-reviewed journal if he started one himself.”

Ken, 7/07/2004, aaj: “You have stated bluntly in the last sentence that I cannot get published in a peer-reviewed journal, and you have stated that several times before. So if Larry and I do indeed get our papers published in peer-reviewed journals, will you publicly retract these previous statements and admit that you were wrong?”

"Stu replies,"

Stu replies, 7/08/2004, aaj: “If you got your actual work published and not some massively edited work that didn’t include the 1-element argument or the 2% false positive argument, I would happily retract. I want to see a piece that says you can definitely determine the number of bullets fired from a rifle based on fragments recovered at the crime scene using 1 solitary element and to a 98% certainty. Is that going to happen Ken?”

Our Two Peer-Reviewed Articles

	K. A. Rahn and L. M. Sturdivan “Neutron activation and the JFK assassination, Part 1. Data and interpretation.” Journal of Radioanalytical and Nuclear Chemistry 262 (1): 205–213, 2004
	L. M. Sturdivan  and K. A. Rahn “Neutron activation and the JFK assassination, Part 2. Extended benefits.” Journal of Radioanalytical and Nuclear Chemistry 262 (1): 215–222, 2004

They contain:

	The two analyses.
	The two groups.
	The uniqueness of Sb.
	The distinctiveness of WCC/MC lead.
	Guinn’s conclusions.
	The problem of heterogeneity and its resolution.
	Chance matches and how they eliminate conspiracy theories.
	Elimination of planting.
	Locations of head and back entry rendered irrelevant.
	Best shooting scenario.
	NAA as Rosetta Stone for the assassination.

I don’t know what Stu will say today, but…

If he says that CABL is dead:

	Remind him that’s not what the NRC committee said—they still support it.

If he says that Guinn’s reputation has gone down the toilet:

	Remind him that’s not what the NRC committee member said: “Guinn’s legacy is still intact.”

If he says that you can’t get a valid distribution from 14 bullets:

	Show him the distribution with the four lots completely overlapping.
	Remind him that the NRC, the FBI, and others use this lognormal distribution as their default.

If he says that MC bullets are no different from the others:

	Remind him that actual data from Guinn and the NRC don’t say that. WCC/MC lead is 10 to 100 times more heterogeneous than the other leads.

If he says that you can’t eliminate an accidental match from a third bullet:

	Agree with him, but add that the chance of that match no greater than 2%.
	Then ask him whether he wants to base his theory on an idea that is at least 98% certain of being wrong.

If he says that Tom Pinkston’s analysis of 10 bullets from a single  MC lot destroys Larry’s distribution:

	Remind him that no valid distribution can be formed by ignoring three out of four lots, as he and Tom do.
	Explain to him how random sampling can easily produce “clumps.”

If he says that we can never get published:

	Show him our two articles in JRNC that contain the full story.

If he says that Oswald was unlucky enough to use super-special ammunition:

	Congratulate him for getting one right.

Stu has failed on his great quest. The NAA remains untouched

Thank you!

President John F. Kennedy