Guinn’s neutron-activation Analysis

      In 1977, Dr. Vincent P. Guinn, professor of chemistry at University of California, Irvine, and one of the foremost practitioners of NAA in the world, was contacted by the House Select Committee on Assassinations to reanalyze the remaining bullet fragments by NAA. In the 13 years that had elapsed since the FBI’s analysis, modern Ge(Li) gamma-ray detectors had become available. These new semiconductor detectors eliminated most of the potential interferences that had plagued the earlier Na(I) scintillation counters. Reanalyzing the fragments offered the prospect of more precise and accurate results.
Guinn was an ideal choice for other reasons. First was his general education in analytical chemistry. He held A.B. and M.S. degrees in chemistry from USC (1939 and 1941, respectively), and a Ph.D. in physical chemistry from Harvard (1949). He had studied radioisotopes at the Oak Ridge Institute of Nuclear Studies, Tennessee, in 1952. He was a fellow of the American Nuclear Society, the American Academy of Forensic Scientists, and a member of the American Chemical Society.
Second, he was a specialist in NAA. He had published many scientific articles on the subject, had been an advisor to the Atomic Energy Commission, and had made a training film on NAA. He came upon the NAA scene right after pioneers such as Dr. Auseklis K. Perkons and Dr. Robert E. Jervis. Guinn's first areas of study involved paint chips and residues from firearms. By April 1965, he was conducting seminars on the criminalistic use of NAA, and had demonstrated that inorganic materials from a crime scene could be analyzed and used more precisely than organic materials such as soil. In 1964 Guinn had been involved in the first criminal case in which NAA was admitted as evidence in a courtroom (Jurgen Thorwald, Crime and Science: The New Frontier in Criminology, Harcourt, Brace & World, Inc., New York, 1967, pp. 434435, 450). 
Third, he and his associates had been studying the trace-element compositions of all sorts of bullets, specifically for forensic purposes, for a decade. He had published on the subject, and was the American expert on the application of NAA to bullets. He had even analyzed Mannlicher-Carcano bullets from each of the four production runs at Western Cartridge Co., East Alton, Illinois, that had been supplied to him by University of Kansas forensic pathologist Dr. John Nichols.
[1] These studies showed him that the composition of MC bullets differed much more from bullet to bullet than other brands did. This fact led Guinn to think that he might be able to determine the number of bullets that hit the two men from the composition of the fragments alone. He might also be able to state which fragments came from which bullets. Had any weapon other than a Mannlicher-Carcano rifle been used in the assassination, the bullets and fragments would not be differentiable.
      Guinn's colleague Dr. Nichols had become most actively involved in the fight to get copies of the FBI's NAA data and to reanalyze the fragments. Also involved was Robert P. Smith, who was a colleague of Dr. Cyril Wecht and who had cowritten an article with him on Wecht's examination of the autopsy materials. Smith had also told Texas Congressman Henry Gonzales, the first chairman of the HSCA, that he should try to get Guinn to reanalyze the samples.
In April 1975, Dr. Nichols succeeded in getting a copy of the FBI’s 1964 raw data from their NAA of the fragments. Nichols flew the 70 pages out to Guinn, whole immediately checked their calculations. He initially agreed with Hoover’s conclusions that the data were inconclusive, and he and Nichols called publicly for the fragments to be reanalyzed with NAA and the new, much-improved Ge(Li) semiconductor detectors that had come into use after the FBI had analyzed the fragments. When the HSCA called in early summer 1977 to ask whether he could really do a better job than the FBI had done a decade earlier, he replied affirmatively, and the job was his.

The samples that Guinn analyzed
Guinn reanalyzed the samples in mid-September 1977 during three working days at the TRIGA Mark I nuclear reactor on the campus of the University of California, Irvine. Table 14 shows the complete list of samples that Guinn analyzed, together with the original weights of the fragments (as received by the FBI in 1963) and the weights of the portions of each that Guinn analyzed. Note that only two of the three tiny fragments from the rear floor of the limousine were still available, and that their weights were 10–20 milligrams lower than originally (because of portions removed for earlier analyses). Also note than in some cases, Guinn analyzed very large portions of the remaining fragments, even the entire fragments, and then returned them to the National Archives virtually unaltered.

Table 14. Bullets and fragments analyzed by V. P Guinn with NAA.
Specimen Description Original weight (mg) Total weight analyzed by NAA (mg)

Group I

CE 399 (Q1) Bullet from stretcher 10,277 10.7
CE 567 (Q2) Bullet fragment from front seat cushion 2,890 50.5
CE 843 (Q4) Larger metal fragment from the President’s head 107 41.9
CE 843 (Q5) Smaller metal fragment from the President’s head 9.7 5.4
CE 842 (Q9) Metal fragments from arm of Governor John Connally 32 16.4, 1.3
CE 840 (Q14) Metal fragments from rear floorboard carpet 58; 45 33.4; 33.8
Group II
CE 573 (Q188) Alleged Walker bullet   16.3
CE 141 (Q8) Unfired WCC/MC cartridge from Oswald’s rifle   24.3; 6.3

Guinn’s procedures and their compromises
Guinn used an analytical procedure that he and associates had developed previously. First, the samples were given a very short irradiation and count, to measure the 24-second
110Ag, the 93-second 124m1Sb, and the 5.1-minute 66Cu radionuclides.[2] A sample was irradiated for 40 seconds, allowed to decay for 40 seconds, and counted on a Ge(Li) semiconductor gamma-ray detector for 40 seconds. The photopeaks used to calculate the results were 658 keV for silver, 498 keV for antimony, and 1039 keV for copper. Because this procedure went so fast, all samples and associated standards (except CE 141) were analyzed twice during a single day’s work.
The next day, all the samples and standards were irradiated for one hour, allowed to decay for about one hour, and counted for five minutes each. This procedure measured longer-lived isotopes of antimony and copper (2.8-day
122Sb and 12.8-hour 64Cu) with better precision. Photopeaks used in these calculations were 564 keV for antimony and 511 keV for copper. Because of time constraints, the samples could only be analyzed once each by this longer procedure.
Guinn’s final results for silver were averages from the two runs with the short procedure. His final results for antimony and copper came from the long procedure because it gave better data than the short procedure. This meant that his data for antimony and copper represented only single determinations.
Preparing the samples and analyzing them required only three days in the laboratory. Immediately afterward, the samples were returned under armed guard to the National Archives in Washington.
It is very important to understand that Guinn’s analytical procedures involved multiple compromises. One major compromise was the speed of irradiating, cooling, and counting. In particular, the long irradiation, cooling, and counting could have yielded better results if it had been spread over several days. The short procedure was also rushed. Another major compromise was the lack of replicate analyses. As we will see below, the FBI’s replicates ultimately made their data far more valuable than Guinn’s, in spite of their systematic errors.

Guinn’s results
Guinn’s results for silver and antimony are shown in Table 15.
[3] Two features of these data should be understood clearly: only single fragments are reported; and the uncertainties refer to only the counting procedure, not to other aspects of the measurement process or to variations within a specimen. Full uncertainties of measurement will be roughly twice as high, and variations within a specimen are several times higher than counting uncertainties. The sole exception is Q14, where both fragments listed in Table 14 were analyzed for Table 15. Thus the uncertainty for Q14 represents heterogeneities in the sample, as well as counting uncertainties. Q5, the smaller fragment from the President’s head, with a mass of only 5.4 milligrams, was analyzed but not reported. The smaller piece of Q9, fragments from the Governor’s wrist, had a mass of only 1.3 milligrams and was also analyzed but not reported. Thus for all intents and purposes, Guinn’s results are for single samples.

Table 15. Guinn’s NAA results for silver and antimony in bullets and fragments.
Specimen Description Silver, ppm Antimony, ppm
CE 399 (Q1) Bullet from stretcher 8.8±0.5 833±9
CE 842 (Q9) Largest metal fragment from Connally's arm 9.8±0.5 797±7
CE 567 (Q2) Bullet fragment from front seat cushion 8.1±0.6 602±4
CE 843 (Q4) Larger metal fragment from the President’s head 7.9±0.3 621±4
CE 840 (Q14) Metal fragments from rear floorboard carpet 8.2±0.4 642±6

      Guinn’s results are very similar to those of the FBI 13 years earlier. Data for silver are almost identical, and will not be discussed further because they do not differ much from fragment to fragment. Guinn’s data for antimony agree well with the FBI’s Run 4, as seen in Figure 9. No error bars are shown here because they are discussed in detail in the section on heterogeneity of antimony. Thus, Guinn and the FBI produced nearly identical results for the fragments, even though different pieces were analyzed.

Figure 9. Guinn’s results for antimony vs. those of the FBI’s run 4.

Guinn’s conclusions
Guinn reached three key conclusions from his results, as reported in his Analytical Chemistry paper: (1) to a “high probability,” the fragments all came from Mannlicher-Carcano bullets; (2) there was positive evidence for two and only two bullets; and (3) the results grouped in a way that supported the single-bullet theory. The next sections discuss these results separately.

The fragments are all from WCC/MC bullets
The full statement of this result is that antimony in all the bullet fragments is “in the unusual (though not necessarily unique) concentration ranges of WCC/MC bullet-lead samples.”
[4] This made it highly probable that they actually were Mannlicher-Carcanos. It is highly unlikely that any of the fragments came from any other kind of ammunition.
The justification for these conclusions comes for Guinn’s analyses of many different kinds of ammunition. As early as July 1971, Guinn had reported the concentrations of antimony and various other trace elements in 36 kinds of ammunition.
[5] Although Mannlicher-Carcano ammunition was not tested here, three lots of 0.38-caliber bullets from Western Cartridge Company were. The original table of data is reproduced here as Table 16 below.

Table 16. Concentrations of elements in 36 lots of bullets. (Table I from "Comparison of bullet lead specimens by nondestructive neutron activation analysis." H. R. Lukens, B.S. and V. P. Guinn, Ph. D. Journal of Forensic Sciences, Vol. 16, No. 3, pp. 301–308, July 1971.)


Concentration of elements in lead

Caliber Make Sb, ppm Ag, ppm Other, ppm
0.45 Krasne (reload) 10,000   5600 Sn; 15 Al
0.45 Hensley & Gibbs 28,000   10 Al
0.45 Military <70   11 Al
0.38 Hensley & Gibbs 44,000   3400 Sn; 75 Al
0.38 United States Cartridge Co. <30 87  
0.38 United States Cartridge Co. 15,500 0.67  
0.38 Union Metallic (UMC) 820 2.2 22,000 Sn; 2.2 As


Remington-UMC 130 2.6 16,000 Sn; 27 As
0.38 Remington-UMC 17,600 4.1  
0.38 Remington-UMC 20,700 8.8  
0.38 Peters 19,400 7.6  
0.38 Peters 29,900 0.77  
0.38 Remington-Peters 9200 3.6  
0.38 Remington-Peters 8700 3.1  
0.38 Remington 8500a    
0.38 Western 25,900a    
0.38 Western 30,000 2.6  
0.38 Western 440 11.5  
0.38 Winchester 760 16.0 167 As
0.38 Winchester 30,900 4.9  
0.38 Winchester 13,800 0.56  
0.38 Winchester 14,700 0.56  
0.38 Winchester 15,500 3.9  
0.38 Winchester 12,300 1.8 31 As
0.38 Federal 18,900    
0.30 Remington 12,000   12 Al
0.30 Sierra 31,000   5.3 Al; 2100 Sn
0.30 Nosler 22,000   6.5 Al
0.22 Hornady 35,000   5.4 Al
0.22 Lapua 12,200b   1.1 Al; 10,400 Sn; <5 As
0.22 Sears 12,600c   1.3 Al; 1,800 Sn; 285 Asd
0.22 Imperial 9900c   <4 As
0.22 Peters 8700c   345 As
0.22 Remington 8500c    
0.22 Western 5400   78 As
0.22 Crossman <200 39  

aAverage of 10 bullets from a single box.
Average of 50 bullets from a single box.
Average of 20 bullets from a single box.
Average of 5 bullets from a single box.

      Already by 1971, Guinn had focused on antimony as the principal indicator element in bullet lead. The reason antimony was so useful was that it was added to some leads up to 4% or so as a hardening agent. This made its concentration range from 10 or 20 ppm un virgin lead to 40,000 ppm in hardened lead, a much larger range than for elements that were not added to the lad. Guinn listed the concentrations of antimony in each of the 36 lots of bullets. He also listed the concentrations of 1–3 other elements if he was able to measure them. The 36 concentrations of antimony are shown in Figure 10.[6] A logarithmic scale had to be used for antimony because its concentrations ranged over nearly four orders of magnitude.

Figure 10. Concentrations of antimony in 36 lots of bullets.

      The range of antimony in Mannlicher-Carcano bullets reported by Guinn, 20–1200 ppm, is also shown, for comparison. It corresponds almost exactly to a group of seven bullets that is below the other 29 by nearly an order of magnitude. (The cause of the two distinct groups is probably just hardened lead versus unhardened lead in the bullets.)
      By contrast, Guinn found that silver varied much less in concentration in the suite of bullets—two orders of magnitude vs. the four for antimony. The reason is probably just that silver is not deliberately added to bullets. Guinn also found that the concentration of silver in Mannlicher-Carcano bullets falls in the middle of the range of concentrations for other types of bullets (Figure 11). The reason for this is also presumably that silver is not added to lead the way antimony is.

Figure 11. Concentrations of silver in 36 lots of bullets.

      The data from Figures 10 and 11 can be used to find the probability of getting false positives, that is, the chance that piece of lead that appears to be from a Mannlicher-Carcano bullet is actually from another kind of bullet. We can calculate the probabilities for fragments in general and for those from the assassination. Consider first the general case where a bullet or a fragment contains antimony in the WCC/MC range. The chance that it is not a WCC/MC bullet is shown in Figure 10 as 7/36, which is the fraction of Guinn's 36 test bullets that had the concentration of a Mannlicher-Carcano but were something else. Thus by using only antimony, there is a 7/36 (19%) chance of falsely identifying the bullet as Mannlicher-Carcano.
A 19% chance of a false positive is pretty high. The probability can be lowered by also using silver, the most useful element that Guinn found next to antimony. With silver alone, 4 of 19 bullets (21%) of other types had concentrations in the range of Mannlicher-Carcanos (Figure 11). So silver alone gives 20% false positives.
    But silver with antimony is better. The simplest approach, incorrect here, would be to just multiply the two probabilities of false positives and get the probability of having simultaneous false positives, (7/36)(4/19) = 4.1%. But since only 19 of the 36 test bullets contained measurable silver, we should limit the comparison to those 19 bullets in which both antimony and silver were measured. Although we could do this from the table, a clearer way is to do it graphically, by plotting one element versus the other. Figure 12 shows such a scatterplot for Sb vs. Ag in MC bullets, other types, and the JFK fragments. The zones of concentrations of antimony and silver for the MC bullets are shown by the dashed horizontal and vertical lines, respectively. The rectangular area between the two sets of lines represents the place where false positives are found. Taking the multiplicative approach for false positives used above, the figure for WCC/MC bullets would be (5/19)(4/19) = 5.5%. This is equivalent to predicting that 5.5% of the 19 other types of bullets, or one bullet, would fall within the MC ranges and give a false positive. In actuality, two of the 19 bullets fall within the MC zone, for a probability of 2/19 = 10.5%. Given the small number of bullets being considered here, the difference between one and two is not significant. We may then conclude that the probability of an apparent MC bullet really being something else is 5%10%.

Figure 12. Scatterplot of Sb and Ag in WCC/MC bullets compared with other bullets and fragments from the assassination.

      The fragments from the assassination can be treated in the same way (Figure 13). Here, however, the ranges of concentration are much smaller. The predicted occurrence of false positives would be (2/19)(2/19) = 1.1%, or 0.2 bullets. The actual occurrence is is no bullets (0%), which is not surprising in view of the 0.2 bullets predicted above. But this is not the right calculation, which must be the fraction of non-MC bullets falling within the area of the fragments divided by the fraction of MC bullets falling within the same area. [The earlier calculation for MC bullets as a whole was a special case of this more general one, where the denominator was 100% (of the MC bullets falling within the MC area).] Since the fraction of MC bullets falling within the range of the assassination fragments is (1/12)(7/12) = 4.9%, the true percentage of false positives becomes [(2/19)(2/19)]/[(1/12)(7/12)], or 22.8%. This latter figure should be regarded as a very rough one because of the small number of cases that went into calculating it. To see just how rough it is, one need only note that no bullets of either kind fell within the narrow zone of assassination fragments. It is probably better then to revert to the earlier calculation for MC bullets as a whole, with a 5%–10% probability of being some other kind of bullet. This assessment agrees with Guinn's characterization of “highly unlikely.” It is also agrees with his statement in Analytical Chemistry that his results “have demonstrated that, to a high degree of probability, all of the bullet-lead evidence specimens are of WCC/MC 6.5-mm brand…”

Figure 13. Scatterplot of Sb and Ag in fragments from the assassination compared with WCC/MC bullets and other bullets.

There are two distinct groups of fragments that make physical sense
This result expresses the central issue raised by the neutron-activation analysis. Two groups of fragments that are statistically distinct would lend considerable support to the single-bullet theory, to a single head shot from the rear, and consequently to a simple, straightforward view of the assassination. One big group of fragments would weaken or destroy these ideas.
For this point Guinn reported (in Analytical Chemistry) that “the specimens show clearcut evidence for the presence of two, and only two, WCC/MC bullets—one of a composition of 815 ppm Sb and 9.3 ppm Ag, the other of a composition of 622 ppm Sb and 8.1 ppm Ag.”
Was Guinn right? His results can be seen in Table 15 above. Given his uncertainties, the two groups do indeed appear to be distinct. (Much more about those uncertainties below.) For people who prefer to see results graphically, the same data (for Sb only) are shown in Figure 14. They lead to the same conclusion that the groups are distinct.

Figure 14. Guinn’s results for antimony in the fragments.

    The other extremely important aspect of Guinn's results was that the membership of the two groups make physical sense. The group with 815 ppm Sb was composed of the stretcher bullet and the fragment from Connally's wrist. This group obviously represented the body shot, and probably the SBT as well. The group at 622 ppm Sb contained the larger fragment from the front seat, the smaller fragment from JFK's brain, and the small fragments from the rear carpet. These fragments obviously represent the head shot. As we will see below, there is less than 1 chance in 50 that these fragments grouped this way randomly—some organizing force was clearly at work.

The groups agree with the FBI’s groups
Another very important feature of Guinn’s results is that his two groups agree closely with the groups revealed by the FBI’s data—the samples fall into the same groups, the groups are about as tight, and the groups are separated by about the same amount. This is shown in Figure 15. Note how closely Guinn’s results overlap one of the FBI’s runs (#3). This extreme similarity validates both sets of analyses.

Figure 15. Guinn’s results for antimony vs. the FBI’s four runs.

The groups support two or more bullets
Recall from above that Guinn described his results as constituting clear evidence for two and only two bullets. He is, of course, correct, in the sense that two groups most obviously mean two bullets of different compositions. But we must be clear that the results are also consistent with more than two bullets, provided that the fragments from the extra bullets just happened to match the results from the first two bullets. (The probability for this per fragment is estimated below at 3%.) The problem is that this interpretation is complex as well as unlikely—it requires that the extra bullets look like the first two bullets. Guinn’s data provide no direct, positive evidence for additional bullets; they just allow for that rather contorted, remote possibility. The spirit of the Principle of Parsimony does not allow us to consider this complex hypothesis as long as simpler ones remain, however, so we must reject it.

Implications for the assassination
Guinn’s results carry at least two important implications for the JFK assassination. First, they provide the simple (and intuitively satisfying) picture that two bullets from one rifle accounted for all the fragments recovered and tested. The fact that it was Oswald’s rifle agrees with all the other validated physical evidence that is available.
Second, Guinn’s results provide strong support for the single-bullet theory by showing that the small fragment from Connally’s arm (Q9) is indistinguishable from the stretcher bullet (Q1). Considering how closely the concentrations of antimony are to each other, the very strong onus falls upon those who claim that Q9 did not come from Q1. Recall that this result was the first direct piece of physical evidence to support the SBT. Although it doesn’t prove the SBT, it is strongly consistent with it. (By strong consistent, I mean consistent under circumstances where it would be unlikely to be consistent. Specifically, I mean that since the tiny fragment Q9 could in principle have had an antimony anywhere between 0 and 1200 ppm, the fact that at 797 ppm it fell so close to the 833 ppm of Q1 must be viewed as strongly significant. It’s rather like a difficult prediction in physics—it doesn’t prove the question at hand but it means more than an easy prediction that comes true.)
Thus Guinn’s NAA strongly supported the Warren Commission’s view that Lee Harvey Oswald was responsible for all the wounds to both Kennedy and Connally.

Challenges to Guinn's results
    Guinn's results were challenged repeatedly by the critical community. All the charges have now been answered satisfactorily. This sections deals with only the most important of them.

    Guinn had previously worked for the Warren Commission. I have never understood this challenge other than representing guilt by association, where even the taint of the association can't be proven. Here is the passage from the HSCA hearings (Volume I, p. 556) where Mr. Fithian raises the subject with Guinn near the end of his testimony.

    Q. Dr. Guinn, this is not meant to be an embarrassing question, but I think I must ask it. Mr. Chairman, a recent article in the New Times magazine stated that you have worked for the Warren Commission and, therefore, your conclusions for this committee would be implicitly biased.
    Did you ever work for the Warren Commission or work for the FBI in connection with the analysis of these evidence samples?
    A. Neither one. I think Mr. Wolf called my attention to the existence of this article, which I haven't seen, but I never did anything for the Warren Commission, and although I know people in the FBI, I have never done any work for them.
    Q. But it is correct, isn't it, that the Warren Commission had the FBI perform neutron activation analysis measurements on the bullet-lead specimens in 1964?
    A. Yes, sir.
    Q. How did you find this out?

Guinn then proceeded to describe how Dr. John Nichols had received the FBI's raw data by a FOIA request and how Guinn had recalculated their results.
    The New Times article in question probably was derived from an article in the New York World Telegram & Sun of 28 August 1964 (just about a month before the Warren Report was released) that charged that Guinn had applied NAA to paraffin casts from Oswald's hand and cheek when he was working as tnechnical director of the activation analysis program at Gulf General Atomic back in 1963 and 1964. The source of that allegation had apparently been a UPI press release of 27 August 1964 that reported on a talk Guinn had given at the International Conference on Recent Advances in Activation Analysis in Glasgow. The press release and the article in the World Telegram & Sun included these statements:

    Dr. Vincent P. Guinn, 46, head of the activation analysis program of the general atomic division of General Dynamics Corp., has been working on the [paraffin cast] problem with the Federal Bureau of Investigation.
    "I cannot say what we found out about Oswald because it is secret until the publication of the Warren report…"

  The story was apparently false. On 18 September 1964, Special Agent John Gallagher of the FBI (who had analyzed the fragments by NAA for the FBI) contacted Dr. Spofford G. English of the AEC about the errors, and Dr. English contacted Guinn. Guinn then wrote a strongly worded letter on 25 September to the World Telegram & Sun. It included statements like "It is the worst job of reporting I have ever seen…Your version was shot full of atrocious misstatements. Worse yet, the writer had the gall to make up his own statements, then put them into alleged direct quotations attributed to me." Nevertheless, the story lived on, and appeared in Mark Lane's Rush to Judgment and elsewhere. Wallace Milam offered this lukewarm acceptance of Guinn's version in Milam's article of August 1994 and Starks's web site:  

It appears that Guinn may well have been misquoted, though it is difficult to imagine a wire service and/or newspaper having the audacity to manufacture such sensational quotes and statements. At the very least, when Guinn denied under oath before HSCA in 1978 that he had ever worked with the FBI or on the Kennedy case, he was being consistent with the position he took in his letter of September 25, 1964, and was not contradicting his record in the case, but was implicitly contradicting the account published in the World-Telegram & Sun.

  Further information on this topic can be found in Appendix G of Milam's article on Guinn (posted on Starks's web site at ) and in a 1994 memo by Alan Rogers, "In defense of Guinn's integrity," posted to alt.conspiracy.jfk on 18 February 1994 and reposted to alt.assassination.jfk on 4 March 2001 by Joe Durnavich.

    Wholesale planting of fragments with different weights and identities? Guinn's testimony led to avoidable confusion over the identity of the "particles" he analyzed and whether they could have been planted. Through an unfortunate choice of words, Guinn muddied a simple subject. When reading the passage below, keep in mind that "little particles" or "little pieces" does not mean tiny original fragments, but small fractions (aliquots) removed from bigger particles for analysis. The fact that Guinn no longer had the original "little pieces" cut off and analyzed by the FBI thus meant very little, for he could (and did) take another one from that fragments that were large enough. The confusing passage begins on page 561 of HSCA Volume I. Mr. Fithian is still questioning Guinn.

    Q. Now, then; did you test exactly the same particles that the FBI tested in 1964?
    A. Well, it turns out I did not, for reasons I don't know, because as they [the FBI at Oak Ridge National Laboratory] did the analysis, they did not destroy the samples either.
    Q. So?
    A. The particular little pieces that they analyzed, I could just as well have analyzed over again, but the pieces that were brought out from the Archives—which reportedly, according to Mr. Gear, were the only bullet-lead fragments from this case still present in the Archives—did not include any of the specific little pieces that the FBI had analyzed.
    Presumably those are in existence somewhere, I am sure nobody threw them out, but where they are, I have no idea.
    Q. And the 1964 equipment wouldn't have consumed them, either?
    A. No.
    Q. What was the state of the knowledge at that time in terms of storing radioactive materials? Would there have been any prospect that someone not adequately informed, such as perhaps the FBI at that time or law enforcement people, would have been a little leery about keeping radioactive materials in their files?
    A. I wouldn't think so. I am sure by that time they knew enough about the safety aspects that, considering the very small amount of radioactivity in them, they would have rightly considered them to be perfectly harmless. Also, the little activity in them soon died out.

In other words, Guinn was saying that the FBI analysts had taken little pieces from the larger fragments, analyzed them, and either filed them away or discarded them rather than returning them to the National Archives. The samples that Guinn received thus all weighed different from the originals, and the actual samples analyzed were gone. Except for Q4,5, Q9, and Q14, which he analyzed in their entirety, he then analyzed new aliquots from the originals.
    Conspiracists have seized on Guinn's remarks and proclaimed that Guinn was analyzed bogus samples that had planted. But they didn't think the subject through before they proclaimed. It is very easy to show that Guinn's samples had to have been real.
    First, there are good reasons for the weights of the larger fragments to have differed from the FBI's weights by the time Guinn got them—the FBI had removed aliquots for analysis rather than irradiating the entire fragment. This was the responsible procedure to follow. Second, there was good reason for the "little particles"—the aliquots for analysis—to be gone; Oak Ridge National Laboratory simply disposed of them as radioactive trash. This is common practice in analytical laboratories, and although it was a bad decision because they were such important pieces of evidence, it is easy to see how it could have happened in the routine course of things. The problem was likely the number of cooks' hands stirring the broth, so to speak. First there was John Gallagher of the FBI, who actually analyzed the samples under detail to ORNL. As soon as he finished counting their radioactivity, he probably had to hurry back to the FBI's central laboratory in Washington, where is as chief spectroscopist had many other responsibilities. Then there were Juel and Emery, the two employees of ORNL who oversaw Gallagher's analyses but who were one step removed from formal responsibility. They probably put the samples into some kind of storage without giving the matter any special thought. At some later time, somebody had to decide whether to keep the samples, somebody who might well have been yet another person. So it is easy to imagine that the final decision to discard the samples could have been taken by someone at least two steps removed from the original process, and who therefore had little sense of their importance. This is the way of large government facilities. It is also easy to imagine Gallagher himself giving permission to trash them, hardly imagining that there would ever be a call for reanalysis. The fact that they were ultimately discarded implies nothing about any sort of conspiracy.
    In fact, it would make no sense to switch fragments between the FBI's NAA and Guinn's, for that would make Guinn's results disagree with the FBI's and the tampering become obvious. Anyone who jumps from Guinn's different weights to switched fragments hasn't thought through this critical point. The only way to justify switching fragments between the two NAA analyses would be if the conspirators would be sure that the FBI's results would be meaningless, and neither they nor anyone else could know that at the time.
    The fragments also could not have been switched (or planted) before the OES on the night of the assassination. First, there simply wasn't enough time: Q1 and Q9 were retrieved that afternoon, Q4,5 at autopsy late that evening, and the other fragments from the limousine as it was searched late that night after it had been flown to Washington. Second, the immediate chains of custody were too tight: most of the fragment had gone directly into the possession of the FBI. And third, only a very few people were privy to the list of fragments and their properties. The only reasonable time to have switched fragments would have been between the OES and the FBI's NAA, a period of about six months.
    But any kind of planting/switching would have required an all-powerful conspiratorial force, one that could dominate at a minimum the FBI, the National Archives, the Warren Commission, and probably the Executive Branch. That Force would also have to have ensured that all employees in all these agencies would keep the secret, not just of the planting, but also of the larger conspiracy that lay behind the assassination and required the planting. As we know all too well, Washington simply does not work that way.
    But the strongest reason why no fragments were planted is that their altered story would then be at odds with the rest of the physical evidence, which tells a remarkably consistent and redundant story. The planted fragments would then have stood out like a sore thumb. That would require all the rest of the physical evidence to have been falsified as well. For example, if the false fragments had been meant to conceal a shot from another rifle, the three cartridge cases on the sixth floor would also have to be faked, as would the long brown package found by the sniper's nest. The conspiracists would also have had to conceal all physical traces of that second weapon. If the planted fragments were intended to conceal a shot from the right front, the Zapruder film would have to have been faked, too, for it shows the head snapping forward, the bulk of the material from the head flying forward, and a rear of the head that was intact after it exploded. The autopsy report stating that the left part of the brain was undamaged would have to be faked, too, as would the photos and X-rays of the head. Collectively, these things could not have happened.
    When all these difficulties and impossibilities of faking the entire suite of physical evidence are considered, only one conclusion is possible: all the fragments are genuine.

    A typical sequence of changes in weights from taking repeated aliquots from a large sample is shown for Q4 in Figure 16 below.

Figure 16. The aliquots taken from Q4 for the three sets of analyses.

    The full set of decreasing weights of the fragments is shown in Table 17 below. Note how in all cases the weights for a given analysis are compatible with the weights available (the original weights less the weights used for the earlier analysis or analyses. This consistency from one analysis to the next is strong evidence that the fragments were not tampered with along the way.

Table 17. Masses (mg) of fragments received and analyzed.
Identity Origin Original mass Mass used for OES Mass remaining for FBI's NAA Mass used in FBI's NAA Mass remaining for Guinn's NAA Mass used in Guinn's NAA
CE 399 (Q1) Stretcher 10,277* 10 10,267* 17.73 10,249 10.7
CE 567 (Q2) Front seat 2890* 10 2880 68.87 2811 50.5
CE 569 (Q3) Rt. fr. door 1361* 10 1351 NA 1351 NA
CE 843 (Q4) JFK brain 107 10 97 28.82 68 41.9
CE 843 (Q5) JFK brain 9.7 NA** 9.7 3.22 6.5 5.4
CE 842 (Q9) JBC wrist 32 10 22 5.33 17 16.4, 1.3
CE 840 (Q14) Limo rear fl. 58; 45; 45 10; NA; NA 48; 45; 45 12.4; 10.65; 9.70 36; 34; 35 33.4; 33.8; ***
CE 841 (Q15) ??? None listed          

NA = Not analyzed
† Entire remaining fragment analyzed
* Includes copper jacketing
** Most likely not analyzed because mass less than needed
*** Fragment missing

    Readers who feel more comfortable with graphs than with tables can view the same data in Figure 17 below. The results are the same as above—the weights of all the particles decrease smoothly and monotonically.

Figure 17. Progressive decreases in weights of fragments with additional analyses.

    Heterogeneity of antimony. The biggest and longest-lasting criticism of Guinn's analysis centered on the fact that his tracer element, antimony, varied enough in his "background" WCC/MC bullets to cause the two tight groups of 2 + 3 fragments to overlap. This criticism has lasted for nearly a decade. It has only now been put to bed. The section below illustrates how Guinn instinctively dealt with it. The full section that follows ("Wallace Milam and the Heterogeneity of Sb") illustrates some of the ways in which Guinn was attacked. The rest of this whole document resolves the controversy by showing that the basic idea ultimately proved to be irrelevant.

Hints about the heterogeneity of Sb
Recall that WCC/MC bullet lead has the unusual property of differing in Sb from one bullet to the next, even within a box of bullets derived from the same production run. Recall also that these differences in concentration stem from the fact that the Western Cartridge Company made its lead by mixing virgin lead and recycled bullet lead that had been hardened with up to 20,000 ppm antimony, or 1000 times greater than in virgin lead. A natural consequence of this mixing process was that antimony would have to vary somewhat within a bullet as well as from bullet to bullet because the scale of heterogeneities in the vat of molten lead would not necessarily match the size of a bullet. In other words, the major differences in antimony from bullet to bullet must translate into some degree of difference within bullets. Guinn was almost caught in a catch-22 situation with respect to heterogeneities—the very heterogeneities he needed to distinguish bullets could also hurt him by making fragments from the same bullet appear as different as those from different bullets. In order for Guinn's scheme to work, the heterogeneity of WCC/MC bullets had to be great enough to to tell the bullet apart but not great enough to create heterogeneities within bullets that would wreck his scheme. Guinn was walking a fine line indeed.
Guinn found the expected (in hindsight) intrabullet heterogeneities and noted them in his testimony to the HSCA. His various comments showed that he was aware of the fine line he was walking—he had to admit them but downplay their significance. As a result, he used general phrases like bullets being “clearly distinguishable,”
wrist fragments differing “considerably” from fragments in the other group, the stretcher bullet matching the wrist fragments “so closely” that “CE 399 did cause the injuries to Governor Connally’s wrist, and that the overall results being “consistent with” the single-bullet theory. Guinn's wording does not necessarily mean that he was misrepresenting anything, however. He sensed (correctly, as it turns out) that the fragments were carrying a very strong message about the number of bullets that hit the men, but that this message was easily muddied by the heterogeneities within the bullets. Guinn was conflicted, and was searching for the most appropriate way to express confidence in the presence of conflicting data.
    It is instructive to examine Guinn's comments on the results versus the intrabullet heterogeneity. In his 1979 article in Analytical Chemistry, he wrote that his earlier studies of WCC/MC bullets had revealed that "Although individual bullets were fairly homogeneous in their antimony and silver contents, they exhibited a great heterogeneity from bullet to bullet—even within the same production run and even within an individual box of 20 cartridges." [Emphasis added.] Also, "the Sb variability within an individual bullet, particularly, is usually several times larger than the measurement precision on an individual sample." As for conclusions, he wrote "All of the Dallas samples are in the unusual (though not necessarily unique) concentration ranges of WCC/MC bullet lead; and the specimens show clear-cut evidence for the presence of two, and only two, WCC/MC bullets…" and "The nondestructive instrumental neutron activation analysis results have demonstrated that, to a high degree of probability, all of the bullet-lead evidence specimens are of WCC/MC 6.5-mm brand, there there is evidence for the presence of portions of two—and only two—such bullets, and that the Connally stretcher virtually intact bullet indeed caused the fracture wound of Governor Connally's wrist—a previously hotly disputed part of the Warren Commission's theory." [Emphasis added.]
    In Guinn's report to the HSCA, 8 September 1978, he wrote the following passages:

    Guinn's testimony to the HSCA (Volume I, pp. 489–567) contains passages such as these:

    After the HSCA finished questioning Dr. Guinn, they allowed him five minutes to make any statement he wished, as was their custom with witnesses. Guinn used part of his time to further clarify the issue of heterogeneities within and between WCC/MC bullets. Here is the relevant portion of his statement.

    "And then, if you are tying to match this piece and this piece—which really both came from the same bullet but you don't know it and you are trying to prove it—you have to take into account another factor [besides analytical uncertainty]—how homogeneous is that bullet? Is every piece that you take from a bullet the same? And the answer is: No; they are not. The individual bullets are fairly homogeneous, but there are significant variations within them.
    We have a great deal of background data, specifically on WCC Mannlicher-/Carcano bullet lead, that isn't in the report, but we use that as the backup which shows that the variation within a bullet is significant.
    So when you start to compare numbers, it turns out, for example, on the antimony numbers, roughly speaking, if you take the plus or minus that is shown and multiply it by about 6, that will take care of all of those variabilities within the sample, as well as the small measurement uncertainties.
    The variation within that individual bullet is then taken into account, and then you find out that two samples indeed match one another as closely as could be expected. For example, the CE-399 sample gave a measured value of 833 parts per million antimony, whereas the CE-842 sample showed 797. Well, any grammar school  boy will tell you 797 is hot the same as 833. But when you consider that the 833 is plus or minus about 50 and the 797 is plus or minus about 50, then you see that you can't distinguish one from the other. They are indistinguishable, but, by the same token, the other samples which are only about roughly 620 plus or minus a smaller amount, in that case about 20 or 30—they very clearly not only match one another, but they also widely differ from this 800 figure.
    But some of that is explained in the text of the report. You can't just take the numbers from the table and blindly go ahead; you have to read the fine print as well to see that everything is properly taken into account.
    In any event, though, I think the results have come out in a fairly clean-cut fashion. We didn't predict any particular way they would come out; they just fell out this way. And, as I say it led me to reexamine the FBI data more carefully than I had done earlier. I frankly was very surprised to see that even their data, somewhat fuzzier, et cetera, still feel right into the same picture."

    So you see the care that Dr. Guinn had to take to describe his results in a way that emphasized their strengths as he saw them and downplayed their weaknesses, which he probably considered more apparent than real. In the end, he has been proven absolutely correct, but not without a great deal of soul-searching that was caused by Wallace Milam's overstated and ultimately unjustified alarm-sounding combined with Guinn's regrettable word-choosing. That episode is discussed in the next section.

[1]H.R. Lukens and V.P. Guinn, J. Forensic Sci., 16, 301 (1971); H. R. Lukens, H. L. Schlesinger, V.P. Guinn, and R. P. Hackleman, “Forensic Neutron activation Analysis of Bullet-Lead Specimens,” U.S. AEC Report GA-10141, 48 pages, 1970.

[2]V. P. Guinn and M. A. Purcell, “A very rapid instrumental neutron activation analysis method for the forensic comparison of bullet-lead specimens,” Journal of Radioanalytical Chemistry, 39, pages 85–91 (1977)

[3]As given in the Analytical Chemistry paper listed in an earlier reference

[4] From the Analytical Chemistry paper.

[5]H. R. Lukens and V. P. Guinn, “Comparison of bullet lead specimens by nondestructive neutron activation analysis,” Journal of Forensic Sciences, 16, pp. 301–308 (1971).

[6]In the few cases where antimony was below Guinn’s detection limit, half the limit is plotted instead of the concentration.

Ahead to Milam and Heterogeneity of Sb
Back to FBI Tries NAA
Back to NAA and the JFK Assassination
Back to NAA