The FBI Laboratory: An Investigation into Laboratory Practices and Alleged Misconduct in Explosives-Related and Other Cases (April,1997) Office of the Inspector General, Department of Justice

SECTION C: WORLD TRADE CENTER BOMBING

On February 26, 1993, an explosion occurred at the World Trade Center, in New York City, New York, resulting in six deaths, numerous injuries, and substantial property damage. An investigation was undertaken by the FBI, as lead agency, with the assistance of other agencies including the ATF and the New York City Police Department. Several defendants were indicted, tried, and convicted in a case dealing primarily with the Trade Center bombing--United States v. Salameh, which was tried from September 1993 to March 1994 in the Southern District of New York. A broader case, which included evidence of the Trade Center bombing (United States v. Omar Ahmed Ali Abdel Rahman a/k/a Sheik Omar ), was tried in 1995 in the Southern District of New York, resulting in the conviction of the defendants.

Prior to the Salameh trial, Whitehurst complained about several matters, all of which were resolved to his satisfaction prior to trial. On January 8, 1996, Whitehurst submitted to the OIG an 80-page critique of the Salameh testimony of SSA David Williams, an examiner in the Explosives Unit. Whitehurst covered a multitude of topics and concluded that Williams misrepresented the truth, testified outside his area of expertise, and presented testimony biased in favor of guilt.

To investigate Whitehurst's allegations, we interviewed Whitehurst, Williams, EU Chief J. Thomas Thurman, Special Agent Steven Burmeister (an examiner who worked on the case), former MAU Chief James Corby, CTU Chief Roger Martz, other examiners and employees at the FBI Laboratory, a chemist at the Eglin Air Force Base, persons who allegedly discussed the case with Williams prior to the Salameh trial, other FBI and ATF personnel (some of whom worked at the scene of the blast), and other persons associated with the case. The interviews of Whitehurst, Williams, Thurman, and Martz were sworn and transcribed. Additionally, we considered relevant trial transcripts, pertinent FBI documents, and applicable literature in the field of explosives.

As explained below, we conclude that in the Salameh trial Williams gave inaccurate and incomplete testimony and testified to invalid opinions that appear tailored to the most incriminating result. Regarding most of Whitehurst's many other allegations, we either find them meritless or conclude that any error was insignificant. We first address the allegations relating to Williams' Salameh testimony (Section II), then the pre-trial issues (Section III), followed by our conclusion (Section IV).

II. Testimony of SSA David Williams in the Salameh Trial

David Williams testified at length on direct examination in the Salameh case regarding several areas, including the following: his manufacture of urea nitrate pursuant to formulas found in manuals seized in the case; his calculation of the amount of urea nitrate that could have been produced based on certain chemical purchases; and the possible explosives used at the bombing and their weight, based on the damage at the scene. On cross-examination Williams elaborated on some of these subjects and opined specifically that the main explosive used in the bombing was urea nitrate. The principal allegations relate to these areas of Williams' testimony. We address first the FBI's manufacture of urea nitrate (Section A), then Williams' opinions on defendants' capacity to make urea nitrate and on the explosive used in the bombing (Section B), then Williams' testimony regarding an attempt to modify Whitehurst's dictation (Section C), and finally the other allegations concerning Williams' testimony (Section D).

A. FBI's Manufacture of Urea Nitrate

Whitehurst asserts that Williams falsely testified that Williams manufactured urea nitrate pursuant to formulas in certain blue manuals that were seized in the case and were linked to the defendants. Whitehurst maintains that Williams in fact did not manufacture any urea nitrate and that the explosive was made by other Laboratory personnel who did not use the formulas in the manuals. First we will summarize Williams' testimony; then we will present the facts found in our investigation and our analysis of the issues.

Williams testified that he had experience in manufacturing or putting together urea nitrate. He further testified that in manufacturing the urea nitrate I actually used two formulas that were removed from one of the blue manuals. (The blue manuals were manuals in Arabic and English for home-made bomb-making.) Williams further testified that the formula recommends that you mix the urea to the nitric acid in a one-to-one range;. . .[i]t suggests that you mix by amount 60 parts of urea to 63 parts nitric acid. He further testified, When I made a large quantity of urea nitrate in the large plastic tubs, it was very heavy. On both direct and cross examination, Williams used both the first person, singular pronoun I and the first person, plural pronoun we to describe who made the urea nitrate.

On cross-examination he testified:

When asked whether he concocted a bomb with some of the urea seized in the searches, Williams responded: I did. In the early tests in the summer, I used some of the urea from Mallory [the location of one of the searches] and made small one-pound bombs of urea nitrate and detonated it.

Williams further testified to the production of urea nitrate at the Eglin Air Force Base in Florida in August 1993. When asked why he used an outdoor laboratory there, he stated, I didn't want to have any of the fumes bother myself or any of my workers. Williams testified that we started with smaller batches of 20 pounds of urea and 20 pounds of nitric acid. On cross-examination, Williams listed the persons who worked on the project with him including Whitehurst, Steven Burmeister, agents from the Jacksonville office of the FBI, technicians in the Explosives Unit, and personnel from the Air Force Base. He then testified:

Williams further testified:

Williams testified that he used two formulas from the blue books to make the urea nitrate. The first (G.Ex. 2781, p.172) was in Arabic and English. The second formula (G.Ex. 2783T, p.2) was entirely in Arabic.

Personnel in the FBI Laboratory made several batches of urea nitrate prior to the Salameh trial. Several small batches were made in the spring and summer of 1993, and approximately 1200 pounds were made at Eglin Air Force Base in August 1993.

The first two batches were made in test tubes by Chemist James Molnar on March 8 and 9, 1993. He followed the procedures set forth in Davis, The Chemistry of Powder & Explosives 372-73 (1943) ( Davis book ). For the second batch, he calculated a synthesis yield of 97%. He wrote up his findings.

The next batch was made by Chemist Mary Tungol. She also followed the procedures set forth in the Davis book. She also prepared a formula for the synthesis of urea nitrate in a four or five gallon quantity. In summary, she calculated the amount of water (2 gallons), urea (20 pounds), and nitric acid (8.7 liters) needed to produce a theoretical (100%) yield of 42.5 pounds of urea nitrate. Tungol made smaller batches (5 to 10 pounds of urea nitrate) using a percentage of the quantities in the formula. These batches were taken to the FBI range at Quantico, Virginia, and detonated.

Another batch was prepared by Whitehurst and Burmeister at Quantico pursuant to the Tungol formula. It would not detonate because it had not been properly dried.

In August 1993, Williams, Whitehurst, Burmeister, and other FBI personnel manufactured approximately 1200 pounds of urea nitrate at the Eglin Air Force Base in Florida. Williams and Whitehurst jointly decided to undertake this project, and both helped set it up, including the acquisition of the necessary personnel, equipment, and materials. The mixing occurred outdoors. Reagent grade (99% pure) or technical grade (about 97%) urea and reagent grade (70.4%) or technical grade (67%) nitric acid were used, as well as distilled water. Whitehurst and Burmeister did the mixing in plastic trash cans surrounded by ice water to cool the solution. Although the evidence is conflicting, the recipe they followed was apparently based on the one developed by Tungol. First, the urea was weighed and dissolved in the distilled water. Then the nitric acid was put in. Several batches were mixed at the same time. Whitehurst and Burmeister wore protective clothing during the mixing. After a precipitate (the urea nitrate) formed, the liquid was filtered through a funnel. The urea nitrate was then put on drying trays, which were put in drying ovens provided by Eglin. The urea nitrate was allowed to dry overnight. Personnel from Eglin then weighed and bagged the urea nitrate. It took about three or four days to produce the 1200 pounds of urea nitrate.

We conclude that the basic point of Williams' testimony--that Williams personally manufactured urea nitrate pursuant to formulas found in the blue manuals--was inaccurate in two respects. First, no one in the FBI used the formulas from the blue manuals to manufacture urea nitrate. Second, Williams' role in the manufacture of the urea nitrate by the FBI was much more limited than his testimony described. We reach these conclusions for the following reasons.

In his testimony Williams indicated that he personally took the formulas from the blue books, followed them, and was able to produce the explosive urea nitrate. Neither Williams nor anyone else in the FBI actually did this. The first (test tube) batch, by Molnar, was made pursuant to the information in the Davis book. From then on, the Tungol formula (also based on the Davis book) was used. All of the formulas (Molnar's, Tungol's, and the Arabic) used the same essential ingredients (urea and nitric acid). The weights and concentrations in the FBI's formulas, however, were different from the weights and concentrations in both of the Arabic formulas referred to in Williams' testimony.

The first formula from the blue books (G.Ex. 2781) sets out the chemical equation for the reaction and states that urea and diluted nitric acid (34%) should be mixed. The formula does not prescribe dissolving the urea in water before adding the diluted nitric acid (34%). Williams testified that the numbers 60 and 63 on the exhibit meant that the formula suggests that you mix by amount 60 parts of urea to 63 parts nitric acid. The numbers 60 and 63 are the molecular weights of urea and nitric acid and were noted underneath the chemical equation. A 60 to 63 ratio by weight is theoretically the correct ratio for the reaction, but only if both substances are in the same concentration. Here, the manual prescribes that the nitric acid be diluted to 34%, which would require a ratio of 60 parts urea to about 189 parts nitric acid (63 divided by .34) if the urea was 100% pure, or some other ratio if the urea was less pure.

The second formula (G.Ex.2783T) is closer to, but is not exactly, what the FBI followed. This formula indicates that 200 grams of urea should be dissolved in water, and then 200 grams of diluted nitric acid should be put in. No mention is made of reagent or technical grade products or distilled water. The manual's translated discussion of how to dilute nitric acid is difficult to understand. In a 1997 interview Burmeister told us he construes the discussion to mean that the nitric acid should be diluted to 35% purity. Thus, the weights and concentrations of G.Ex.2783 differed from those in the formulas used by the FBI, and, as explained in note 41, infra, it is unclear whether G.Ex. 2783 could effectively produce urea nitrate. In any event, prior to the Salameh trial no one in the FBI attempted to decipher the dilution procedure and actually dilute nitric acid pursuant to it; nor did anyone in the FBI otherwise attempt to make urea nitrate pursuant to this formula.

In his OIG interview Williams did not say that he or anyone else in the FBI actually manufactured urea nitrate by literally following the formulas in the manuals. Despite Williams' trial testimony that the early (pre-Eglin) batches were made using the formulas in the manuals, he testified in the OIG interview that he did not know what formulas were used in the only two pre-Eglin batches he was aware of.

As for Eglin, Williams testified at the interview as follows: He did not see the written formula Whitehurst and the other members of the team were following and did not know whether it was the formula from the manuals. However, based on Whitehurst's verbal instruction to the team, Williams thought that the formula from the manuals appeared to be the formula that we were also using. Williams further testified that on the first day of mixing at Eglin he received a fax of one of the translated Arabic formulas; he showed it to Whitehurst and the Eglin chemists and asked how it compared to what they were doing; and they said it was the same.

Whitehurst, Burmeister, and the Eglin chemist at the scene of the mixing (Paul Bolduc) told the OIG that they could not recall telling Williams that a formula in the fax was the same as the formula the FBI was using at Eglin. Two bomb technicians present at Eglin, however, recall the conversation. FBI Comments at 10.

After Williams' OIG interview, we obtained a copy of the fax Williams received at Eglin. The fax includes two Arabic formulas and their translations. The first formula (First Fax Formula) is one of the two formulas Williams testified in Salameh he used to make urea nitrate and became G.Ex. 2781. The second formula in the fax (Second Fax Formula) is different from the two Arabic formulas Williams testified he relied on to make the urea nitrate.

As noted above, the First Fax Formula (G.Ex. 2781) differs from the formula used at Eglin in that the fax formula fails to prescribe that the urea should be dissolved in water prior to the addition of the nitric acid and further states that the nitric acid itself should be diluted to a 34% concentration. As indicated above, at Eglin the urea was first dissolved in distilled water, and then reagent grade (70.4%) or technical grade (67%) nitric acid was added. The fax formula, moreover, is essentially a chemical equation with molecular weights. It does not include a specific amount of 34% nitric acid to be added to a specific amount of urea.

The Second Fax Formula is quite different from the Eglin formula. The Second Fax Formula uses human or animal urine as an ingredient. The formula sets forth a procedure for evaporating and filtering the urine; then 90% nitric acid is added to the urine filtrate at a ratio by volume of one part acid to three parts urine.

Thus, the fax formulas were different from the formula the FBI used to manufacture urea nitrate at Eglin, and no one in the FBI at Eglin attempted to manufacture urea nitrate from the fax formulas.

Because he was not a chemist, Williams lacked the expertise to determine on his own whether a fax formula was the same as the formula Whitehurst was following. When Williams testified at the OIG interview that the formulas seemed to be the same because both used a 60 to 63 ratio by weight of urea to nitric acid, it is clear he did not understand that the ratio of the weights must take into account the concentrations of the ingredients. Because the concentrations of the ingredients at Eglin were different from the concentrations in the Arabic formula, the ratios of weights would have to be different as well. Additionally, the formulas were different with respect to the form of the urea (solid versus water solution) and the absence in the Arabic formula of specific amounts (in pounds or liters) for the ingredients.

As we have noted, the Eglin and fax formulas utilized the same basic ingredients but were different as to the weights, concentrations, and the form of the urea. Nevertheless, Williams and the bomb technicians maintain that Williams was told by a chemist that the Eglin formula and the Arabic formula were the same. In his trial testimony Williams should have made the source of his information clear. Instead of testifying that I made the urea nitrate at Eglin and in the pre-Eglin batches pursuant to the Arabic formula, he should have said that he had no personal knowledge of what formulas were used, that comparing chemical formulas is a matter beyond his expertise, but that, when Williams asked, a chemist told him that the Arabic and Eglin formulas were the same.

We conclude that Williams' trial testimony that the formulas from the manuals were the source from which the FBI manufactured urea nitrate was incorrect. The source of the formulas used by the FBI was the Davis book. Moreover, Williams told us that he did not know or did not have a clue as to what formulas were used before Eglin and that he had no idea as to the source of the Eglin formula. Williams' testimony concerning the use of the Arabic formulas was seriously flawed.

Williams also gave inaccurate testimony about his role in the FBI's manufacture of urea nitrate. Regarding the batches before Eglin, he had no role other than attempting to dry some of the product and was not even aware of all the batches. Thus, his testimony that I made the early batches of urea nitrate was apparently false. Williams responded at the OIG interview:

Williams acknowledged that [p]erhaps they were a bad choice of words. We are troubled by the choice of words. Williams' testimony that I performed some Laboratory procedure implied that he was in a position to know something about that procedure--when in fact he was not. Thus, instead of saying I made the pre-Eglin batches of urea nitrate pursuant to the Arabic formulas, Williams should have testified to the truth--that he was not involved in those batches and did not know what formulas were used.

As for Eglin, Williams' testimony on direct that I made approximately 1300 pounds of urea nitrate in Florida, and his testimony on cross that he supervis[ed] the mixing process, was inaccurate.

According to Williams' OIG interview, the decision to manufacture the large quantity of urea nitrate in Florida, and the planning for the project, were jointly undertaken by him and Whitehurst, but Whitehurst decided how to make the explosive and what formula to use. Special Agent Burmeister stated in his OIG interview:

In his OIG interview, Paul Bolduc, the Eglin Air Force Base chemist present for the mixing operation, characterized Williams' role in the mixing process as that of a gofer.

We find that Williams' role in the mixing operation was to provide manual assistance under the direction of Whitehurst and Burmeister.

Accordingly, we conclude that Williams' trial testimony on direct examination that I made the urea nitrate at Eglin, and his testimony on cross-examination that he supervis[ed] the mixing process, was incorrect. The reference in his trial testimony to the other FBI personnel at Eglin as my workers could be interpreted to manifest an intent to downplay the role of the others and to aggrandize his own. Williams' exaggeration of his role erroneously suggested that Williams was an expert in the manufacture of urea nitrate, that he was in a position to know how the FBI made its urea nitrate, and that therefore he could say authoritatively that it was manufactured pursuant to the formulas in the blue books. Williams' flawed testimony about the manufacture of urea nitrate was the first of numerous errors he committed in the Salameh trial.

An important part of Williams' Salameh testimony consisted of his opinions concerning (1) the capability of the defendants to manufacture urea nitrate and (2) the main explosive ( main charge ) used in the World Trade Center bombing. We conclude that Williams' testimony about these subjects was deeply flawed.

As noted above, urea nitrate is made by combining urea with nitric acid. Regarding the defendants' capacity to make urea nitrate, Williams subtracted the amounts of urea and nitric acid recovered in the searches from the amounts the defendants ordered from chemical companies. From the amounts of urea and nitric acid missing, he calculated that the defendants could have produced approximately 1200 pounds of urea nitrate.

Williams then rendered opinions concerning the main explosive used in the World Trade Center bombing. On direct examination, based on the damage at the scene, he opined that the main charge consisted of about 1200 pounds of a category of explosives that included urea nitrate. On cross-examination, he went further and rendered a specific opinion that the bulk of the main charge was urea nitrate.

Taken together, the opinions concerning the defendants' capacity to make urea nitrate, and the likelihood that urea nitrate was used in the bombing, were incriminating in view of the uniqueness of the criminal use of urea nitrate. Williams testified that his research revealed only one prior use of urea nitrate as an improvised explosive charge--in a pipe bomb in 1988. If such an unusual explosive was indeed used at the World Trade Center, the defendants' link to a bomb factory and storage facility capable of making the precise amount of urea nitrate allegedly used at the Trade Center would substantially contribute to the proof of guilt.

Williams' opinions were important for another reason. Normally, the way a crime laboratory determines the main charge of an exploded bomb is by finding unconsumed particles or distinctive byproducts of the explosive among the residue. The search for such particles is made by a forensic chemist. In the FBI at the time of the World Trade Center case, the chemists specializing in the examination of explosives residue were Whitehurst and Burmeister, who were assigned to the MAU. One problem for the prosecution in the World Trade Center case was that the MAU chemists did not find any residue identifying the explosive. Thus, the normal way of scientifically determining the main charge was unavailable. Williams' purported identification of the explosive filled that void.

Williams calculated the amount of urea nitrate the defendants could have produced from the amounts of urea and nitric acid that were missing--i.e., from the amounts ordered minus the amounts recovered in searches of premises associated with the defendants. To make such a calculation, the area of chemistry known as stoichiometry must be applied. Stoichiometry concerns molecular weight relationships in chemical reactions. In this instance, the chemical reaction was: one molecule of urea plus one molecule of nitric acid produces one molecule of urea nitrate. As previously noted, each of these molecules has a different mass or weight. The molecular weight of urea is 60; that of nitric acid is 63; and that of urea nitrate is 123. Thus theoretically (100% yield), 60 grams of urea plus 63 grams of nitric acid produces 123 grams of urea nitrate. For every 60 grams of urea, 63 grams of nitric acid is required. (Similarly, for every 60 pounds of urea, 63 pounds of nitric acid is needed.)

Determining the potential amount of urea nitrate that could have been produced requires a determination, first, of the limiting reagent because it is the chemical that will run out first. For example, with only 63 grams of nitric acid, one could only produce 123 grams of urea nitrate even with an unlimited amount of urea. In this example, the nitric acid would be the limiting reagent.

Once the limiting reagent is determined, the potential amount of urea nitrate can be determined with a simple calculation: If urea was the limiting reagent, for every 60 grams (60 pounds) of urea that was missing, the perpetrators potentially could have produced 123 grams (123 pounds) of urea nitrate. If nitric acid was the limiting reagent, for every 63 grams (63 pounds) of nitric acid that was missing, the perpetrators potentially could have produced 123 grams (123 pounds) of urea nitrate.

One additional factor must be taken into consideration: the purity of the components. The calculations above assumed that the components were 100% pure. If, for example, the urea was only 50% pure, you would need twice as many grams (or pounds) of urea as indicated above: 120 grams (or 120 pounds) would be needed for every 63 grams (63 pounds) of 100% pure nitric acid. Similarly, if both components were less than 100% pure, appropriate adjustments would have to be made.

On March 7 or 8, 1993, Williams provided a list of the missing components to a forensic chemist in the CTU (Thomas Jourdan) and asked him to calculate the potential amount of urea nitrate that could have been produced. Jourdan made the calculations and reported back to Williams, Agent Richard Hahn, and possibly EU Chief J. Christopher Ronay. It appeared to Jourdan that they did not understand his explanation of how nitric acid was the limiting reagent, so Jourdan prepared a memorandum explaining his calculations and gave it to Ronay and Williams and probably to Hahn.

Based on the figures Jourdan had, he determined that the nitric acid was the limiting reagent, and determined that the upper limit was the production of 1821 pounds of urea nitrate. Jourdan used a 97% yield instead of 100% because a staff member (this was James Molnar, see p.85, supra) had achieved such a yield in the Laboratory. Jourdan also noted that [r]ecovered empty bottles of HNO3 [nitric acid] indicated usage of about equal portions of 70.4% (reagent grade) nitric acid and 67% (technical grade) nitric acid. He defined limiting reagent as stoichiometrically you run out of it first, and stated that ordinarily, urea is the limiting reagent to make sure the urea nitrate is not adulterated with unreacted urea, which would inhibit the explosive's effectiveness.

At the time Williams testified at the Salameh trial, his figures regarding the missing components were different (presumably updated) from the ones given to Jourdan. At the time of the trial it was determined that 1200 pounds of urea and 1694 pounds of nitric acid were missing. See G.Ex. 862. Using these figures and Jourdan's basic methodology, a proper stoichiometric calculation would be as follows: Jourdan assumed, as we will do here, that the concentration of the urea was 100% and the average concentration of the nitric acid was 68.7%. A quantity of 1694 pounds of 68.7% nitric acid is the equivalent of 1164 (1694 x .687) pounds of 100% nitric acid. Since, as noted above, 63 pounds of nitric acid is needed for every 60 pounds of urea, 1164 pounds of 100% nitric acid is inadequate to achieve a complete reaction of 1200 pounds of 100% urea. Accordingly, the nitric acid was the limiting reagent.

For every 63 pounds of completely reacted nitric acid, 123 pounds of urea nitrate is theoretically (100% yield) produced. Therefore, with a 100% yield, 1164 pounds of nitric acid would produce 2273 pounds of urea nitrate. A 97% yield, as obtained by Molnar, would produce 2205 pounds of urea nitrate.

In his testimony in the Salameh trial, Williams was asked to calculate how much urea nitrate could be produced from the missing urea and nitric acid. Williams first addressed the concept of a limiting reagent:

Next, Williams addressed the issue of yield. He testified that in a laboratory type environment the [b]est case scenario would be in the neighborhood of 90 percent. He then testified:

Williams then testified:

Later in his testimony Williams referred to the amount of urea nitrate that could have been made as about 1,200 pounds.

We have reached several conclusions regarding Williams' testimony.

First, Williams lacked the requisite scientific knowledge to testify competently in this area. When Jourdan initially discussed the calculation of potential urea nitrate, Williams appeared to Jourdan not to understand the concept of a limiting reagent. His testimony makes clear that he never learned the concept. Urea is not always the limiting reagent and was apparently not the limiting reagent here. Moreover, in his memorandum Jourdan explicitly defines limiting reagent as stoichiometrically you run out of it first and finds nitric acid to be the limiting reagent based on the information he was given. Accordingly, Williams' testimony was inconsistent with the Jourdan memorandum.

Moreover, assuming that urea was the limiting reagent in this case, Williams' numbers do not add up. Because, as earlier noted, 60 pounds of fully reacted urea will produce 123 pounds of urea nitrate, 1200 pounds of urea will produce a theoretical (100% yield) of 2460 pounds of urea nitrate. A 90% yield would produce 2214 pounds (not 2100 pounds), and a 60% to 70% yield would produce 1476 to 1722 pounds (not 1200 to 1800 pounds). The errors in Williams' calculations conveniently produced a range that included the exact amount of urea nitrate--1200 pounds--that he later testified was used in the Trade Center bombing.

Second, Williams' discussion of laboratory yield was problematic. Williams testified that in a laboratory type environment the [b]est case scenario would be a yield in the neighborhood of 90 percent. In his OIG interview Williams said he got the 90% figure from Whitehurst or Burmeister, although they do not confirm this. Assuming they said it, we nevertheless question Williams' choice of words, which implied that his testimony about laboratory yield was based on his own expertise. A laboratory yield for a chemical reaction is obviously outside Williams' area of expertise. He told us in his OIG interview that he had no way of knowing, independent of the chemists, the accuracy of the 90% number, but believed he could rely on the opinion of other experts in his testimony. An expert may rely on opinions of other experts if this is the normal practice in the field. See Fed. R. Evid. 703. Accordingly, Williams would have been fully justified, in rendering his own opinions, in relying on the chemist's statement about yield. For example, he could have testified, My opinion is based in part on the statement of Chemist W, who told me 90% is the best yield. But if he had so testified (with an attribution for the yield statement), the court would have known on whose expertise the 90% number rested. But that is not what Williams did. He did not attribute the 90% number to anyone else, but rather continued to give the impression that he was speaking from his own expertise, which was misleading.

The failure to attribute the 90% figure was particularly inappropriate here because at this point in Williams' testimony he was apparently testifying about the manufacture of urea nitrate based on his personal experience in making it. Because the 90% figure was not based on that experience, Williams should have revealed the source of the yield number.

Third, Williams' trial testimony about non-laboratory yield was unscientific and speculative, was based on improper grounds, and appears tailored to correspond with his estimate of the amount of explosive used in the bombing. Williams testified that in reality, in a non-laboratory environment, I would expect that and, as a matter of fact, you would get somewhere around a 60- to 70-percent yield.

When asked in his OIG interview the basis for this testimony, he explained that it was based on three factors. The first factor was the yield at Eglin. He said the yield there was 1158 pounds of urea nitrate from 1600 pounds, or 1500 pounds, give or take, of ingredients (urea and nitric acid). A yield of 1158 pounds from 1600 pounds would be 72%; a yield from 1500 pounds would be 77%. Williams described the Eglin operation as a pseudo-laboratory environment.

The second factor was Williams' observations during the searches of the defendants' alleged bomb factory and storage facility. During these searches he observed evidence of a lot of spillage of urea nitrate, which was more than at Eglin.

When asked whether the evidence of spillage suggested a yield much lower than 60-70%, Williams identified the third factor he considered to determine non-laboratory yield :

We are deeply troubled by Williams' rationale. The first factor used--the yield at Eglin--is problematic. To use Williams' words, Eglin was a pseudo-laboratory environment, in which chemists did the mixing. It is impossible to say whether the typical non-laboratory environment --if there is one--would be better or worse than Eglin. Assuming it would be worse because of an absence of chemists, one could only speculate about how much worse. Further, improvised (i.e., homemade ) explosives are sometimes produced by chemists; so an assumption that non-chemists made the explosive would be invalid.

The second factor was also inappropriate. Williams' trial testimony about a non-laboratory yield was offered as an expert opinion based on his experience making urea nitrate. He was asked what the yield typically would be in a non-laboratory setting. By basing that opinion on residues found at the defendants' storage facility and bomb factory, Williams really offered an opinion on the yield he thought the defendants would have had, but masked it in the guise of a general opinion. Moreover, it is pure speculation to say what the defendants' yield would have been from the discovery of some urea nitrate crystals evidencing spillage.

The third factor, however, is the most problematic. There is a degree of ambiguity as to what exactly Williams meant. In essence, he said he based his testimony about non-laboratory yield in part on the amount of chemicals missing (amounts purchased minus amounts recovered at the storage facility). Our interpretation of the passage is this: Williams apparently assumed the Trade Center bomb was made from the chemicals missing from defendants' storage facility. He estimated, as he later testified, that the main charge at the Trade Center weighed 1200 pounds. He then divided 1200 by the weight of the applicable amount of missing urea and nitric acid to give him an estimate of defendants' yield. He then considered defendants' yield to help him determine non-laboratory yield generally.

Based on the amount of urea and nitric acid missing from the defendants' facility, they had the capacity to produce urea nitrate in an amount in excess of 2000 pounds if the yield was high (over 90%) and in an amount less than 1200 pounds if the yield was low (below 50%). Williams testified at trial that the amount of the explosive used in the Trade Center bombing was about 1200 pounds. If the defendants' yield was substantially below 90% but not below 50%, a good match could be obtained between the amount the defendants could have produced and the amount supposedly used in the bombing. By setting the non-laboratory yield at 60 to 70 percent, Williams obtained a good match.

The purpose of a criminal trial, of course, is to determine guilt. The issue of guilt is the ultimate question to which all others are directed. In contrast, Williams began with a presumption of guilt as a foundation on which to build inferences. (As we shall see below, this is not the only time in the Salameh trial that Williams so utilized a presumption of guilt.) The agent simply assumed that the perpetrators produced a 1200 pound bomb at the Trade Center using the urea and nitric acid missing from the defendants' facility, and that yield (the amount used at the bombing divided by the amount missing) informed his testimony about non-laboratory yield, which was presented to the jury as a general number applicable to all non-laboratory environments.

It appears Williams may have worked backwards --that is, he may have first determined the result he wanted (here, that the defendants could have produced 1200 pounds of urea nitrate, the amount he estimated was used in the bombing) and then tailored his testimony about yield to reach that result. We are deeply troubled by this possibility.

We conclude that a competent expert cannot give a narrow range for the yield in a non-laboratory environment. A commercial production facility or a meticulous chemist in a garage can potentially achieve a yield as high as that produced in a laboratory. On the other extreme, careless persons without knowledge or skill may be unable to produce the explosive at all (0% yield) or may achieve only a very low yield. Accordingly, we find that Williams' testimony about non-laboratory yield was invalid and beyond his area of expertise.

Fourth, had Williams or another witness performed the stoichiometric calculation correctly, the result--a 100% yield of about 2273 pounds of urea nitrate with a real possibility of a much lower figure in a non-laboratory setting--would have been perfectly acceptable to the prosecution's theory of the case. Williams seemed to have pushed the envelope to get to 1200 pounds--his estimate of the weight of the explosive used in the bombing. Such exacting symmetry was unnecessary.

In sum, we conclude that Williams' testimony about the potential production of urea nitrate was outside his area of expertise and deeply flawed, and his excesses were unnecessary to an effective presentation of the prosecution's case.

Having established the defendants' capacity to manufacture 1200 pounds of urea nitrate, Williams went on to render an opinion in the Salameh trial that the main explosive charge in the Trade Center bombing was 1200 pounds of urea nitrate. This testimony was also seriously flawed.

An important part of Williams' opinion concerning the explosive used at the Trade Center was his determination of the velocity of detonation (VOD) of that explosive based on his assessment of the damage at the scene. Attachment C: A Primer on Explosives and Velocity of Detonation, infra, defines VOD and is a necessary foundation for the discussion that follows. The significance of the VOD determination was that it provided a basis for Williams' opinion concerning the type of explosive used in the bombing.

Williams testified at the Salameh trial to the VOD of urea nitrate:

When asked at his OIG interview the basis for these figures, Williams stated that they were a rough estimate from information I had obtained from different sources. The information was allegedly received orally from persons Williams regarded as knowledgeable sources within the field of explosives. These sources told him, [I]t's approximate. These fellows had not worked with it. And wherever they got the information from, this is what I had received from them. Williams told us there was very little literature on the subject. He continued:

After the OIG interview we obtained Williams' notes for the World Trade Center case. There is nothing in the notes indicating that the VOD of urea nitrate is 14,000-15,500 feet per second.

The notes, however, do contain two copies of page U103 of the Encyclopedia of Explosives and Related Items (U.S. Armament Research and Development Command 1983) ( Encyclopedia ), a standard text in the field. Page U103 contains the following:

(Abbreviations in original.) A VOD of 3400-4700 meters per second converts to about 11,155 to 15,420 feet per second. In the OIG interview, Williams stated that he reviewed the Encyclopedia regarding the VOD of urea nitrate before he testified in Salameh.

Also among the case notes is a notation of 12-15,500 FPS, without further elaboration, on a sheet from Williams' notepad. In a letter in August 1996 Williams commented on this notation:

In his OIG interview and correspondence, Williams named only three knowledgeable sources within the field of explosives who he allegedly consulted prior to his testimony--Tom Dowling and Fred Smith of the Institute of Makers of Explosives and Paul Cooper of Sandia National Laboratories. In his OIG interview Dowling stated that he did not recall talking to Williams or talking to anyone from the FBI about the VOD of urea nitrate after the Trade Center blast, but said he was reasonably sure he talked to FBI employees on the telephone about other aspects of urea nitrate. Dowling said that if he had been asked about the VOD of urea nitrate, he would have consulted his reference material and given the caller the information he had. Dowling had only one reference book that contained the VOD of urea nitrate--the Encyclopedia. Smith stated in his OIG interview that he did not recall that anyone ever asked him about the VOD of urea nitrate, that he would not have known the VOD, and that to answer the inquiry he would have consulted the Encyclopedia. Cooper stated in his OIG interview that he was pretty sure no one from the FBI called him to ask about the VOD of urea nitrate and that if someone had called he would have had to perform research or calculations to determine the VOD.

In addition to the Encyclopedia, our own literature search found only one text setting forth the VOD of urea nitrate (Urbanski, Chemistry and Technology of Explosives 469-70 (1965)), and it contained the same VOD as the Encyclopedia--3400 to 4700 meters per second.

Williams testified at his OIG interview that after the Salameh trial (and before the Rahman trial) [w]e detonated the explosives [the urea nitrate] we made at Eglin and measured the VOD to be 12,100 feet per second. Williams characterized this measured VOD as substantially less than 14,000.

Williams' Salameh testimony about the VOD of urea nitrate was, at best, incomplete and, at worst, knowingly incorrect. The Encyclopedia, a standard text in the field of explosives, indicated that urea nitrate has a VOD of about 11,155 to 15,420 feet per second. Although the applicable page of this text was in Williams' notes and although prior to his testimony he had consulted it, he nevertheless testified, without qualification, that the VOD is 14,000-15,500 feet per second.

Williams claimed in his OIG interview that he based his testimony about the 14,000-15,500 feet per second VOD of urea nitrate on oral statements from persons outside the FBI. The interviews of Dowling, Smith, and Cooper, and the absence of supporting documentation in the case notes, leave us with grave doubts about the veracity of this claim. In any event, these oral opinions allegedly came from persons who had not worked with urea nitrate, and Williams did not know the basis of their opinions. Assuming Williams received such opinions, we conclude that it was inappropriate for him to blindly rely on them and ignore the Encyclopedia. At a minimum, Williams should have told the court he was relying on outside opinions, and he should have supplemented those opinions in court with the information from the Encyclopedia.

Finally, in his August 1996 letter, Williams came up with a completely new reason for his testimony about the VOD of urea nitrate:

This new explanation for Williams' trial testimony is not helpful to Williams' position. First, we do not find it credible. It is inconsistent with both his trial and OIG testimony, and we think that if this were the real reason for his trial testimony he would have mentioned it at the OIG interview. At the OIG interview Williams mentioned the Encyclopedia but limited his remarks to: I know I definitely looked at the Encyclopedia of Explosives, and I don't recall specifically what it had said at that point. The August 1996 explanation came after we confronted Williams with page U103 from the Encyclopedia, and the new explanation appears contrived to accommodate that text. Second, Williams' trial testimony did not purport to be an estimate of the VOD of the urea nitrate made either at Eglin or by the perpetrators. Rather, it was put forth as the general range for the VOD of urea nitrate. The 14,000 feet per second figure was explicitly limited at the trial to smaller quantities, which would be inapplicable to both Eglin and the perpetrators. Third, density was not the only variable mentioned in the Encyclopedia; the confinement also varied (paper versus steel tube) and may have had as significant an impact on VOD as the density. Thus, Williams' new explanation is based on a misconstruction of the Encyclopedia. Fourth, Williams' statement in the new explanation that he thought the urea nitrate used in the bombing had a high density is speculative. If, as seems unlikely, the new explanation is the true explanation, Williams should have given the same information in court as he did in his letter--namely, that the VOD for urea nitrate is about 11,155-15,420 feet per second, but that he thought the VOD of the main explosive was at the high end of that range for certain specific reasons. The new explanation reflects adversely on Williams' credibility and competence.

We conclude that the 14,000-15,500 VOD range for urea nitrate that Williams gave at the Salameh trial was clearly too narrow, and appears tailored to correspond to the estimates in his report (14,000 feet per second) and in his testimony (14,000-15,500 feet per second) of the VOD of the main explosive used at the Trade Center. In his trial testimony about the VOD of urea nitrate, Williams failed in his responsibility to provide the court with complete and accurate information.

Having told the jury that the VOD of urea nitrate was about 14,000 to 15,500 feet per second, Williams went on to testify as follows to the VOD of the main explosive at the Trade Center:

We conclude that Williams' VOD opinion lacked a sufficient scientific and empirical foundation.

At the outset we note that Williams has been inconsistent as to his estimate of the VOD of the main charge at the World Trade Center. In his report dated July 1, 1993, he stated that the explosive main charge was a high explosive having a velocity of detonation (VOD) of approximately 14,000 feet per second. In his Salameh testimony in February 1994, he gave a VOD of somewhere between 14,000 and about 15,500 feet per second, with a little bit of give on each side of that. Later, in the Rahman trial in April 1995, Williams testified:

In his OIG interviews in February and March 1996 he also stated that his VOD estimate included a 2000 feet per second tolerance on either side of the 14,000-feet-per-second estimate--i.e., a range of 12,000 to 16,000 feet per second. Finally, in a letter to the OIG in August 1996, Williams stated: The other reason that I testified as to the VOD damage in the Trade Center, is that from the damage I witnessed, it appeared to me that the improvised explosive device was faster tha[n] 11,000 and slower than 16,000.

Thus, Williams has given four estimates of the VOD for the main charge: approximately 14,000 feet per second (his report), 14,000 to about 15,500 feet per second with a little give on each side of that (Salameh trial), around 14,000 feet per second with a bracket on both sides of a couple thousand feet (Rahman trial, OIG interviews), and between 11,000 and 16,000 feet per second (letter to the OIG).

We observe that Williams' adjustment from 14,000 (report) to 14,000-15,500 feet per second (Salameh trial) coincided with his Salameh testimony that the VOD of urea nitrate was 14,000-15,500 feet per second. His change from 14,000-15,500 (Salameh trial) to 12,000-16,000 feet per second (Rahman trial) occurred after Williams discovered that the VOD of the urea nitrate made at Eglin was 12,100 feet per second. His change to 11,000-16,000 feet per second (August 1996 letter) occurred after we pointed out to him that the Encyclopedia gave the VOD range of urea nitrate as about 11,155 to 15,420 feet per second. The circumstances of the four estimates imply that Williams changed his VOD opinion for the main charge in order to maintain a match with the VOD of urea nitrate.

We conclude that Williams' inconsistencies severely undercut the credibility of his VOD opinion for the main charge.

Williams testified in the Salameh trial that he considered several observations to determine the VOD of the Trade Center bomb:

The problem with this testimony is that Williams never explains how the observations compute to 14,000-15,500 feet per second. That he observed evidence of heaving as opposed to brisance --i.e., the damaged materials appeared to have been pushed rather than shattered --only necessarily excludes military explosives such as C4 with VODs in excess of 18,000 feet per second. Nowhere in his testimony does Williams explain how he narrowed the broad heaving range of high explosives (about 3000 to 18,000 feet per second) to 14,000-15,500 feet per second.

In his OIG testimony Williams elaborated further on his rationale. He stated he considered the damage to the component parts of the suspect vehicle and other witness materials around there, the concrete, the steel, the vehicles, the people. He stated that because he found pitting and cratering within four feet, and evidence of heaving and no tearing within eight and a half feet, of the seat of the explosion, [t]hat put me into an area of somewhere between 12,000 and 16,000 feet per second . . . I didn't find any pitting or cratering eight feet away; but four feet away, I did. He continued:

Williams stated he considered a host of other things, including [t]he bodies, the burning. He elaborated:

That's my question, sir.

The thrust of Williams' OIG testimony is that he considered numerous factors bearing on VOD, which he then filtered through his experience to produce his VOD estimate. We find Williams' application of his methodology flawed, because it is essentially an unscientific, unverifiable process of intuition. This is apparent from some of the language Williams used to describe his method of determining the VOD and weight of the explosive: I can roughly get a feel that it was a very hot explosive or not (emphasis added); [w]hat caused me to guess a velocity of detonation (emphasis added); [t]hese things produced an impression on me (emphasis added).

The application of the methodology is one of rough[] . . . feel[ings], guess[es], and impression[s]. There was a complete absence of empirical data to support any of the inferences Williams made from the various factors he identified. For example, Williams emphasized that the pitting and cratering within a radius of 4 feet from the seat of the explosion, when combined with only heaving without pitting and cratering within 8.5 feet, showed a velocity of detonation of 12,000-16,000 feet per second. But neither Williams nor the FBI has data to support that thesis. Moreover, in the Oklahoma City case (see Part Three, Section G, infra) Williams found pitting and cratering 12 feet from the seat but nevertheless estimated the VOD to be 13,000 feet per second in that case, effectively undercutting the primary basis he claimed for his VOD opinion in the World Trade Center case.

The same could be said for the conclusions he drew from observing certain victims' bodies--e.g., the way in which an arm was severed, an eye injury. Williams and the FBI have no data or other basis for concluding that the nature of those injuries meant the VOD was 14,000-15,500 feet per second.

Williams' attempt to justify a specific VOD estimate in the Oklahoma City case is similarly unpersuasive and supports our view of the inappropriateness of attempting to fix a narrow VOD range from an assessment of the blast damage. In his Oklahoma City report, Williams estimated the VOD of the main charge to be 13,000 feet per second. Williams explained in his OIG interview that he reached his VOD opinion by considering the explosive damage at the crime scene in light of his experience. He cited approximately fifteen different factors that contributed to his opinion--such as, the damage to the vehicle containing the bomb, the size of the crater, the lip of the crater, evidence of heaving, the damage to the concrete, the size of the vehicle fragments, pitting and cratering, the movement of parked cars, and the damage to parking signs. As in the World Trade Center case, however, the difficulty arose when Williams attempted to explain how he got from the observed damage to the specific VOD. For example, he contended that the size of the fragments contributed to his opinion. But neither Williams nor the FBI can cite any empirical studies linking specific sized fragments to specific VODs. Williams stated in his Oklahoma City interview that he had no documentation or experimentation to support his premises regarding the various factors and that he relied solely on his memory of explosive experiences spanning 10 to 15 years. We conclude that this is an inadequate basis for rendering a specific VOD opinion from observations of blast damage.

Agent Thurman, the current EU Unit Chief, stated in his OIG interview that normally an EU examiner will only determine from the damage whether the explosive was high or low, heaving or brisant. With the exception of differentiating between a high explosive and a low explosive, the arbitrary, we do not, as a rule, go in the reports and state that it's X' number of feet per second. Indeed, Thurman, who has been in the EU for about 14 years, has never himself opined a specific VOD from a damage assessment, but has limited himself to opinions about high versus low, brisant versus heaving, explosives. In fact, Williams is the only examiner Thurman is aware of who has attempted to find a specific VOD from a damage assessment, and attempting to make such findings is not part of the EU training. Williams also believes he is the only EU examiner to have rendered a specific VOD opinion from the explosive damage. Furthermore, as noted above, Williams is unaware of any literature stating that an explosives expert properly may render such a VOD opinion. We also are unaware of any such literature. It thus appears that Williams may be unique, both within the FBI and within the community of explosives experts generally, in his willingness to render such specific VOD opinions.

We have no doubt that an experienced explosives examiner may properly draw certain inferences from observations at a crime scene. For example, an experienced expert will be able to discern the difference between the damage left by a high versus a low explosive, and can differentiate the damage caused by a heaving high explosive (like most commercial products) versus a brisant (like most military explosives) high explosive. Similarly, an observation of pitting and cratering will tell an experienced expert that the explosive used was a high explosive with a VOD typically in excess of about 10,000 feet per second. All of this involves the use of experience to recognize certain distinctive characteristics of explosive damage.

Going further, however, and attempting to infer from the damage a specific VOD is a process that appears to have no precedent either in the literature or at the FBI. We believe it is unprecedented and unjustifiable because the differences in damage caused by explosives with different specific VODs are insufficiently distinctive to allow an experienced expert to say that certain damage will only result from an explosive with a particular VOD.

No database exists at the FBI that correlates specific VODs with particular damage or with the many other variables identified by Williams. We conclude that Williams in fact has no objective basis for estimating a specific VOD from an inspection of the crime scene.

Accordingly, we conclude that Williams' specific VOD opinion of 14,000-15,500 feet per second for the main charge at the World Trade Center lacked an adequate scientific and empirical basis.

Having testified that the VOD of the Trade Center explosive was 14,000 to 15,500 feet per second, Williams went on to testify about the type of explosives that fit that range. We will first summarize that testimony and then analyze it.

On direct examination, Williams testified as follows:

Williams further testified that he was able to rule out quite a bit of the slurries, water gels and emulsions because of the failure to find microballoons or tipper ties among the debris at the scene. Williams testified that microballoons are tiny glass balloons that are included in some emulsions to add air space, and tipper ties are the wire ends of water gels. He testified that he would have expected to find microballoons if the explosive at the Trade Center had been an emulsion and find tipper ties if it had been a water gel. He then concluded by identifying urea nitrate as within the category of a fertilizer-based explosive that would have that velocity of detonation consistent with the damage that [he] saw.

On cross-examination, one of the defense counsel (Mr. Campriello) attempted to recapitulate Williams' earlier testimony but misstated it, leading to the following:

First, Williams testified on direct examination that because of that type of damage . . . the type of explosives that fit in that bracket are very limited. Assuming the VOD testified to at the Salameh trial--14,000-15,500 feet per second--there are many different explosives that fit in that bracket. Using the VOD testified to in the Rahman trial--12,000-16,000 feet per second--there are even more that qualify. Williams testified in his OIG interview that a lot of different explosives meet the 14,000-15,500 feet per second VOD range. For example, the 1980 Dupont Blasters' Handbook ( Dupont ) lists six prill products, four water gels, and two dynamites with a VOD within the 14,000-15,500 feet per second range, and more within the 12,000-16,000 feet per second range. The 1968 Canadian Industries Limited Blasters' Handbook lists three products with velocities in the 14,000-15,500 feet per second range. The 1995 Dyno Nobel Inc. Explosives Engineers Guide ( Dyno ) lists twenty-seven products with velocities in the 14,000-15,500 range.

Williams' testimony about the very limited type of explosives that fit in the 14,000-15,500 feet per second bracket was literally correct, because the many commercial products within that range fall into certain categories or types--namely, dynamites, water gels, emulsions, and fertilizer (e.g., ANFO) products. We are concerned, however, that the court may not have understood that within each type there are numerous commercial products meeting the 14,000-15,500 feet per second range.

Second, Williams testified that the VOD of water gels and emulsions are just a tad on either side of that parameter of detonation [14,000-15,500 feet per second]. This testimony was incorrect. There are several commercially available water gels and emulsions with VODs within the 14,000-15,500 feet per second bracket. See Dupont at 71; Dyno at 1-2.

Third, Williams testified at trial that he could rule out some of the explosives that met the range--namely, the emulsions and the water gels because of a failure to find microballoons and tipper ties in the debris. Williams contradicted this testimony at his OIG interview.

As for the microballoons, if used they may have been made of resin and likely consumed in the blast. More fundamentally, however, any microballoons used would have constituted only about five percent of the total explosive mixture. No residue of the main explosive was recovered at the Trade Center. If residue of the component constituting ninety-five percent of the charge was not recovered, it should be no surprise that remains of the five percent component were not found. Williams conceded at his OIG interview that the failure to find the microballoons meant only that it's possible that they were not there. Williams added, I couldn't eliminate them, because we didn't find anything.

Similarly, the failure to find tipper ties did not rule out water gels. Williams testified at his OIG interview as follows:

In his OIG interview, Williams told us: Because I did not find any evidence of any of the other commercial explosives does not necessarily mean that they were not used. Accordingly, we conclude that Williams should not have testified at trial that he could rule out the slurries, water gels, and emulsions.

Fourth, in his OIG interview Williams stated that, based on his assessment of the damage at the scene, he really could not make any type of identification of the explosive used at the Trade Center:

(Emphasis added). Williams' acknowledgment at the OIG interview that, based on the crime scene, the main explosive could have been anything differs significantly from the opinions he rendered at the Salameh trial. At the trial Williams testified that his observations at the scene enabled him to help the court determine the explosive that may have been used in the blast. Now he has admitted that there's no way I could have called any kind of explosive. In light of Williams' OIG testimony, we are deeply troubled that his testimony on direct examination may have misled the court.

In sum, we conclude that Williams' direct examination was inaccurate and misleading, and suggested too strongly that a fertilizer-based explosive like ammonium nitrate or urea nitrate was used in the Trade Center bomb.

Even more troubling than Williams' direct examination was a part of his cross-examination in which he rendered an incriminating opinion based on speculation beyond his scientific expertise. On direct, Williams identified a category of explosives that fit the VOD and damage that he observed at the post-blast scene. This category included but was not limited to urea nitrate. At his OIG interview (as discussed above), Williams was emphatic that he could not identify a specific explosive based on his observations at the crime scene.

Nevertheless, Williams testified on cross-examination that the bulk of the explosive was, in my opinion, urea nitrate. See also on the same page of cross-examination: I believe urea nitrate was the bulk of the constituent in that bomb with other explosive materials. At his interview we asked Williams how he could render such an opinion, and he answered: the reason I was able to do that in testimony was because I had the benefit of the search sites, the storage sites, the bomb factory and, of course, viewing the evidence from the crime scene. Williams continued:

Williams' use of the auxiliary searches to render an opinion that the bulk of the main charge was urea nitrate was improper for two independent reasons.

First, Williams improperly based his expert opinion that urea nitrate was the main charge on the fact that urea nitrate and other materials had been associated with the defendants. This error is analogous to the one Rudolph made in Psinakis when he relied on the fact that stripped detonating cord had been found outside the defendant's house as a basis for his identification of PETN on a knife. See Part Three, Section A, supra. By basing his opinion on the collateral evidence associated with the defendants, Williams improperly engaged in speculation beyond his scientific expertise.

Williams portrayed himself as a scientist and rendered opinions as an explosives expert. As such, he should have limited himself to conclusions that logically followed from the underlying data and the scientific analyses performed. Here, Williams' scientific analysis of the cause of the explosion rested on an examination of the damage at the post-blast scene. He should not have based his opinions, in whole or in part, on evidence that was collateral to his scientific examinations, even if that evidence was somehow connected to the defendants. For Williams to identify the main charge as urea nitrate based on evidence that the defendants had or could make that compound is comparable to a firearms expert identifying the caliber of a spent bullet based on the mere fact that a suspect had a handgun of a particular caliber.

Earlier in the cross-examination Williams rejected defense counsel's suggestion that Williams was trying to infer that the items seized at the locations associated with the defendants must have been the items that were used in the World Trade Center (emphasis added). Williams testified then that he was only saying that the items seized could have been used in the Trade Center explosion. This was a valid scientific assessment of the defendants' capability and an appropriate rejection of the suggestion that the cause of the explosion could be determined scientifically from the evidence associated with the defendants. Williams should have maintained this approach throughout his cross-examination.

Evidence associated with the defendants is logically relevant to the blast's cause only under the following chain of reasoning:

This chain of reasoning is objectionable because it is not scientific and because it uses a presumption or inference of guilt (point two) as a building block in the analysis. The question of the defendants' guilt is the ultimate issue. It should not be presumed as a foundation for further analysis. By basing his urea nitrate opinion on the collateral evidence, Williams implicitly accepted as a premise the prosecution's theory of guilt. This was improper.

Moreover, even assuming defendants committed the bombing and had the capacity to make a urea nitrate bomb, that did not necessarily mean urea nitrate was used at the Trade Center: the defendants, for example, may have disposed of the urea nitrate elsewhere and used another explosive in the bomb, or they may have converted the urea nitrate to nitro urea and used that explosive. Williams' opinion based on the collateral evidence was thus not only unscientific but also speculative, and it therefore fell well below the minimum standards required of competent forensic scientists.

Finally, because Williams failed to reveal that his urea nitrate opinion was based not on his independent scientific examination but on speculation from the mere fact that defendants could have made urea nitrate, the court was unable to put the opinion in its proper perspective, and a danger arose that the opinion would be given undue weight in support of the prosecution's case.

Second, the context of the questioning that led to Williams' identification of urea nitrate appears limited to an opinion based only on Williams' assessment of the damage at the crime scene. On direct examination Williams' opinion regarding the type of explosive used was explicitly [b]ased on the damage and [his] estimated velocity of detonation. It is obvious that the applicable cross-examination was an attempt to get Williams to repeat what he said on direct examination, which defense counsel misunderstood. See, e.g.: Correct me if I'm wrong. If I understood you correctly, you indicated . . . . Moreover, defense counsel, in the applicable cross-examination, explicitly asked about the possible bomb that could have caused this kind of damage. . . . [W]as whatever caused it [the damage] just this one possibility or were there other possible bombs as well . . . ? The court's questions about ANFO, moreover, make clear that the court believed the applicable examination related to Williams' assessment of the damage at the scene. Further, Williams' ready affirmative answer to the court's question Could it be ANFO? suggests Williams understood that the inquiry related to the damage at the scene.

It must be remembered that establishing that the explosive used at the World Trade Center was urea nitrate was extremely damaging to the defendants' case. Evidence linked the defendants to a bomb factory and storage facility containing evidence of urea nitrate or the ingredients for urea nitrate, an explosive rarely used in a criminal device. Williams' testimony on cross-examination, therefore, that the bulk of the explosive was, in my opinion, urea nitrate was very incriminating.

In this context, it was unprofessional and misleading for Williams, without explanation, to base such an incriminating opinion on a factor (the auxiliary searches) so different from the factors previously relied on (VOD and damage at the scene).

In sum, when Mr. Campriello asked Williams, Could it have been another kind of bomb or no? , the question, reasonably interpreted, meant: Could it have been another kind of bomb or no, based on your expert analysis of the damage at the crime scene? In any event, even if the questioning was inept, Williams had an obligation to restrict his opinions to his scientific analysis and to refrain from speculating about what the main charge must have been based on the defendants' capacity to manufacture a particular explosive. Williams' answer to Campriello's question should have been compatible with the answer he gave us: [The main explosive] could have been anything. We conclude that by answering instead, [T]he bulk of the explosive was, in my opinion, urea nitrate, Williams failed in his responsibility to provide the court with an objective, unbiased expert opinion.

Williams testified at the Salameh trial as follows concerning the weight of the explosive used in the Trade Center bomb:

In his OIG interview he explained further:

Some of the same considerations that apply to Williams' testimony about VOD apply here. First, his analysis is intuitive, unscientific, and imprecise: you're able to kind of estimate how much explosive (emphasis added); Williams testified on cross-examination that he was speculating about the weight of the explosive; [t]hese things produced an impression on me. Second, the weight estimate was dependent on the VOD estimate ( If you vary one, of course, you have to vary the other ), and as discussed above the VOD estimate was itself speculative.

Third, EU examiners normally do not estimate the quantity of explosives because the placement and confinement of the explosive has such a significant effect on the amount of damage. As EU Chief Thurman told us:

On the other hand, Williams' estimate of the quantity of explosives was quite broad: 1000-1500 pounds, with 1200 pounds as a ballpark figure. The thrust of his trial testimony about quantity was that it was a rough estimate: you're able to kind of estimate how much explosive. Viewing agent Williams' estimate of weight in that light, we conclude that it was within his expertise to render such an opinion.

Whitehurst alleges that Williams gave inaccurate testimony regarding an attempt by Williams to modify a report (dictation) written by Whitehurst. The evidence supports Whitehurst's claim.

On June 15, 1993, Whitehurst submitted dictation to Williams for inclusion in the official reports of the case. The dictation included the following language:

(Italics added.)

After receiving Whitehurst's dictation, Williams asked James Corby, Whitehurst's Unit Chief, whether the sections of the dictation that are italicized above could be removed. According to Corby, Williams wanted those things deleted. Corby refused to alter the dictation. A meeting was held with James Kearney, the chief of the SAS, Alan Robillard, the Assistant SAS Chief, Corby, and Williams. Kearney and Robillard decided to leave the dictation substantially unchanged, and Williams agreed to this decision.

Regarding the passages Williams wanted taken out, Williams told us at the OIG interview:

At the Salameh trial, Williams testified as follows:

A. Not the phraseology, the format.

Williams went on to testify about making some innocuous changes in the format of a report other than Whitehurst's June 15, 1993, dictation quoted above.

Although defense counsel's questions lack precision, we think a fair construction of them implicated Williams' attempt to modify Whitehurst's June 15, 1993, dictation. The sections Williams wanted deleted from that dictation provided innocent explanations for the residue results as alternatives to a more incriminating explanation--e.g., urine and fertilizer as alternatives to urea nitrate. Accordingly, when counsel asked Williams, And when we talk about format, the specific part of the format that you didn't like is when those opinions gave alternate reasons for finding some residue. Am I correct? , Williams erred when he answered, That's not correct. Similarly, when counsel asked, Well, when they said that, say like for urea nitrate, in those reports when it said, urea nitrate could have came from sewage, you were dissatisfied with those kinds of conclusions; weren't you? , Williams again erred when he answered, No, I was not. We conclude that Williams' answers to these questions were, at a minimum, misleading.

In his January 8, 1996, letter to the OIG, Whitehurst made numerous other allegations concerning Williams' testimony in Salameh.

1. In his testimony Williams attempted to distinguish high from low explosives by saying that the velocity of high explosives is above, and the velocity of low explosives below, 3000 feet per second. This is technically incorrect (see Attachment C, infra), but a common error, which was harmless here.

2. Whitehurst criticizes Williams' general testimony about dynamite. We find Williams' testimony substantially accurate and within his area of expertise. Any technical errors (e.g., what is or is not carbonaceous ) were harmless and insignificant.

3. Williams was technically incorrect when he testified urea nitrate which is urea and nitric acid, or nitro urea, urea with sulfuric acid. Urea nitrate does not consist of urea and nitric acid; urea and nitric acid when mixed form a new substance, urea nitrate. Nitrourea is made by mixing urea nitrate with sulfuric acid. Although these errors are inconsequential, it may have been preferable for a chemist to testify to these matters.

4. Williams' attempts to explain how nitroglycerin will precipitate from a methanol solution and how nitroglycerine decomposes were poor. A knowledgeable chemist could have provided better explanations. Nevertheless, Williams was asked the questions, and he no doubt did his best to answer them accurately. Williams should have told the prosecutor ahead of time that these matters would be best left to another witness.

5. Williams was asked what the components of urea nitrate are, and he said, urea and nitric acid. We think the answer was a fair response to the question. Urea and nitric acid are the ingredients, which when mixed form a new substance, urea nitrate. One definition of component is ingredient. Webster's Ninth New Collegiate Dictionary 270 (1990).

6. Whitehurst claims that Williams testified falsely that he (Williams) researched the use of urea nitrate in the United States. This claim is apparently based on the fact that Whitehurst did research on the subject. That Whitehurst did some research does not mean Williams did not. Williams insists that he did some research. Accordingly, we conclude that Whitehurst's claim is unfounded.

7. Whitehurst criticizes Williams' testimony about the possible explosive uses of certain materials. Generally, we have no problem with Williams' testimony on this subject, and believe it was within his area of expertise. Williams can be second-guessed on certain matters (e.g., the discussion of phenol ), but any errors were harmless and insignificant.

8. Whitehurst's claim that Williams cannot consider the results of a chemist's analysis in rendering Williams' own opinion is frivolous.

9. Whitehurst criticizes Williams' description of nitrocellulose. We think that Williams' description was accurate for one form or type of nitrocellulose, but was not a good generic description.

10. Despite Whitehurst's criticism, we find that Williams' testimony about the use of smokeless powder and lead azide as initiators is substantially correct.

11. Despite Whitehurst's criticism, we are not concerned with Williams' testimony that when he arrives at a blast scene he look[s] for structural damage to see what repairs have to be done. Obviously, an EU examiner will not himself direct the repairs, which will be handled by appropriate experts.

12. Contrary to Whitehurst's claim, it is within an explosives examiner's expertise to identify explosive damage on metal.

13. Whitehurst complains that Williams testified outside his area of expertise when he discussed the matching of two pieces of tape. Williams has only been qualified in the FBI Laboratory in the areas of explosives and toolmarks. In the testimony challenged by Whitehurst, however, all Williams did was describe the measurements and observations he made, which was merely a factual description. This testimony was given without objection. We think it was permissible for Williams to answer the questions asked.

14. Whitehurst criticizes Williams' testimony about blast damage to portions of a truck. Although Williams is not a metallurgist, we think it was within his area of expertise to testify that he observed blast damage to the truck.

15. Whitehurst criticizes Williams' testimony about freezing and frozen nitroglycerine. We, however, find no contradiction in saying that the process of freezing nitroglycerine is dangerous, but that frozen nitroglycerine is stable.

16. We disagree with Whitehurst's assertion that because some of the pieces of debris were the size of toothpicks the main charge at the Trade Center could not have been a heaving explosive.

17. Finally, Whitehurst complains that some of Williams' testimony did not meet the test of Daubert v. Merrell Dow, 113 S. Ct. 2786 (1993), because Williams did not use the scientific method, which involves the testing of hypotheses. Although evidentiary questions are beyond the scope of this Report, we note that the discussion of expert testimony in Daubert was limited to scientific . . . knowledge and not technical, or other specialized knowledge. 113 S. Ct. at 2795 & n.8. Much of Williams' testimony could be viewed as based on technical or other specialized knowledge within the meaning of Daubert.

Several controversies occurred, and were resolved to Whitehurst's satisfaction, before the trials in the World Trade Center case.

Immediately after the Trade Center bombing, the chemists in the FBI Laboratory specializing in explosives residue analysis (MAU chemists Whitehurst and Burmeister), went to New York City to conduct examinations at the blast scene. That left no chemists specializing in explosives residue analysis at the laboratory in Washington. When specimens were sent back to the laboratory for examination, the examinations were conducted by chemists in the CTU, Unit Chief Roger Martz and Lynn Lasswell.

Specimen Q23 was a tire fragment recovered from the crime scene. Lasswell analyzed it with solid probe mass spectrometry and concluded that urea nitrate was detected on the specimen. Martz as unit chief approved Lasswell's conclusion, which was incorporated in an official report and distributed April 12, 1993. This conclusion would have been extremely helpful to the prosecution because it would have tended to establish that urea nitrate was used in the Trade Center bomb.

Whitehurst and Burmeister disagreed with Lasswell's conclusion on the ground that the instrumental results only really showed the presence of urea and nitric acid, which could have originated from substances other than urea nitrate--e.g., urine, fertilizer, car exhausts, or ice melter. Whitehurst's and Burmeister's objections, however, were overruled.

Whitehurst and Burmeister then prepared a blind test for Martz by submitting to him specimens they claimed were from the Trade Center evidence. In reality, Whitehurst and Burmeister prepared one sample from Whitehurst's urine and another by mixing ammonium nitrate fertilizer and urea. According to Burmeister, the results were close enough that you wouldn't be able to tell the difference from running a sample of urea nitrate. (Martz insists he never rendered an opinion that these samples were urea nitrate, but said only that the instrument detected urea and nitric acid.) With the blind test results, Whitehurst and Burmeister went to Assistant Section Chief Robillard, who scolded them for making the blind test.

Eventually, Corby directed Whitehurst to make a review of Lasswell's results and to write a new dictation. Whitehurst made the review and wrote the dictation. Whitehurst's dictation was incorporated into a new official report amending the April 12, 1993, report. The new report is dated July 1, 1993. At the Salameh trial, Burmeister testified in accordance with Whitehurst's dictation. Martz told the OIG in 1996 that he no longer agrees with Lasswell's original dictation because the results could have been produced by urea and nitrates rather than urea nitrate.

Ultimately, the FBI Laboratory correctly resolved the controversy concerning Q23, although the resolution procedure ( blind tests, etc.) was flawed. Moreover, the chemist who examined Q23 should have been trained in the explosives residue protocol.

The Barringer Ion Mobility Spectrometer (IMS) tests for the presence of particular molecules. When a sample is introduced, a graph is produced with peaks. Certain substances have distinctive graphs or peaks. If a distinctive peak is produced, an inference can be drawn that a particular substance is present. The manufacturer programs the memory of the instrument to identify common explosives such as nitroglycerine. The user of the instrument can also program the memory to identify certain peaks.

Lasswell introduced a urea nitrate sample in the IMS and produced a particular peak. He then programmed the memory of the instrument to indicate the presence of urea nitrate whenever that peak reappeared. When specimen Q65 was submitted to the IMS, a graph was produced, and the machine automatically identified one of the peaks as urea nitrate.

When Whitehurst reviewed Lasswell's instrumental results to prepare the dictation that went into the July 1, 1993, official report, he examined the IMS graph for Q65. Whitehurst took the position that the peak was not for urea nitrate specifically, but was just a nitrate peak that would be produced by certain nitrates, including but not limited to urea nitrate. Based on this, Whitehurst took issue with Lasswell's decision to program the memory of the IMS to identify the particular peak as urea nitrate. He wrote the OIG (in one of his first submissions to us) as follows:

Whitehurst stated in a letter to the OIG that the analytical output was purposely altered to read <urea nitrate' in order to deceive the innocent reader of the computer printout. This claim is grossly overstated and without merit.

Both Lasswell and Martz insist that the IMS was used only as a screening mechanism to determine whether urea nitrate was possibly in the specimen. Lasswell asserted that when he identified the presence of urea nitrate in Q65 in his original dictation, he relied on instruments other than the IMS.

Whitehurst acknowledged in his OIG interview that the IMS could properly be used as a screening device for urea nitrate. Moreover, in his own dictation for Q65, Whitehurst stated as follows:

(Emphasis added). When Whitehurst was asked at his OIG interview whether he was saying that Lasswell intentionally tried to create false information, Whitehurst stated, No, no.

We conclude that the implication in Whitehurst's assertion--that Lasswell engaged in something like scientific fraud, [which was] unethical, wrong and very damning to the defense position in this matter --is unfounded. Although labeling the peak on the IMS graph as a urea nitrate peak was potentially misleading (because the peak could be caused by other nitrates), the IMS could properly be used as a screening device for urea nitrate. Accordingly, we find that Lasswell engaged in no misconduct in his work with the IMS.

At one point in the Trade Center investigation the government was preparing affidavits for search warrants and wanted to use an examination by Whitehurst that found nitroglycerine on a specimen. Although Whitehurst found nitroglycerine, he refused to make a positive identification because of the possibility of contamination by a bomb technician. Instead, he was only prepared to say that the results were consistent with the presence of nitroglycerine on the specimen. Williams argued strongly for Whitehurst to make a definite assessment. Whitehurst considered this argument to constitute undue pressure to get me to change the wording in my report.

Although we do not know the exact words Williams used, we find no impropriety in Williams discussing the matter with Whitehurst to determine whether a more definite conclusion could be reached. Ultimately, the report was not changed.

Additionally, Williams changed the format of one of Whitehurst's dictations when Williams issued one of the official reports. With a series of specimens, Whitehurst set forth each instrument he used to examine each specimen. Williams made a list of all the instruments and said one or more was used with each specimen, and then just set forth the results with respect to each specimen. Williams also replaced the language None of these explosives were detected on the specimens with Analysis was conducted with negative results. Whitehurst protested the changes, and a new report was issued containing his dictation verbatim.

We consider the changes in format innocuous. One of the reasons Williams gave for the changes, however, is troubling. In referring to Whitehurst's habit of always setting forth, at length, the technical examinations made, Williams stated: [I]f I've got to retype this there's always the possibility of a typographical error and it's a pain in my neck to do it everytime.

A principal examiner (PE) is supposed to include verbatim in the official report the dictation of an auxiliary examiner (AE) unless the AE and the AE's Unit Chief agree to the change. In the Trade Center case Williams was the PE and Whitehurst an AE. The verbatim-inclusion rule is fundamental and should not be broken at any time. The burden of retyping a lengthy or technical dictation is an inadequate reason for violating the rule.

In December 1993 Whitehurst submitted a memorandum to the OIG concerning a conversation he had with SSA Don Haldimann on December 15, 1993. According to Whitehurst, Haldimann stated that the Assistant United States Attorneys (AUSAs) in the Trade Center case had grave concerns about the complexity of Whitehurst's dictation and thought the information in the dictation could be damaging to the case. Whitehurst further asserted that Haldimann said that the U.S. Attorney's Office had inquired into means of circumventing my testimony in this matter and is displeased with my expert opinion as it is stated because it offers strength to the defense side in this matter. Whitehurst characterized Haldimann's statements as indicating possible suppressions of evidence by the U.S. Attorney's office . . . [which] can be deemed to be fraudulent and unethical.

At the Rahman trial, Whitehurst testified that after the December 15, 1993, conversation he met with the prosecutors in the World Trade Center case and felt no pressure from the lawyers on the prosecution team. He testified further, however, that in the December 15, 1993, conversation he felt pressure from Haldimann to take out the qualifying statements in his dictation. Whitehurst acknowledged that the conversation with Haldimann occurred at a Christmas party.

In his OIG interview Haldimann stated that the conversation on December 15, 1993, was a personal conversation at a Christmas party and lasted about 10 or 15 minutes. Haldimann stated that in the conversation Haldimann was merely giving his opinion that the dictation was confusing and included superfluous information and that simpler reports would be better. Haldimann insisted in the interview that he was in no way asking or attempting to influence Whitehurst to change the reports ; the reports had already been provided to the defense attorney in discovery, and therefore the point was moot. Haldimann stated in the interview that it was his impression that the AUSAs in the case were distressed about Whitehurst's dictation, and he did tell Whitehurst that the AUSAs did not want to put Whitehurst on the stand. Finally, Haldimann stated in the interview that no one directed him to talk to Whitehurst.

Whitehurst did not change his dictation as a result of the Haldimann conversation, and Whitehurst was agreeable to having Burmeister testify at the Trade Center trials.

Although we are unable to determine the specific words used in the December 15, 1993, Christmas party conversation, we think Whitehurst grossly overstated the matter in his memorandum. Whatever was said in this brief conversation does not constitute or evince suppressions of evidence . . . [which] can be deemed to be fraudulent and unethical. Although both Whitehurst and Haldimann may have raised their voices during this conversation, ultimately it signified nothing.

We are profoundly disturbed by Williams' testimony in the Salameh trial. We conclude that Williams (1) gave inaccurate testimony regarding his role in the manufacture of urea nitrate and regarding whether the urea nitrate was made pursuant to Arabic formulas from bomb-making books; (2) testified beyond his expertise regarding the defendants' capacity to make urea nitrate and in a way that made the testimony appear tailored to the most incriminating result; (3) gave incomplete testimony concerning the VOD of urea nitrate; (4) gave an invalid opinion regarding the VOD of the main charge; (5) gave invalid and misleading opinions on direct examination concerning the explosives that may have been used in the bombing; (6) regarding his identification of the main charge on cross examination, gave an opinion that was based on speculation beyond his scientific expertise and that appears tailored to the most incriminating result; and (7) gave misleading testimony concerning his attempt to modify Whitehurst's dictation. In short, the testimony lacked the objectivity, credibility, and competence demanded of examiners in the FBI Laboratory.

Williams' testimony also suggests the need for certain improvements in Laboratory procedure that we discuss in detail in Part Six of this Report. For example, Williams' testimony about a specific VOD had no precedent in the FBI, and we found it to be scientifically unjustifiable. This error would have been avoided had Williams followed the ASCLD/LAB requirement that new procedures be validated before they are used in casework. Similarly, the need for complete case notes was exemplified by the absence of any notes supporting Williams' claim that he determined the VOD of urea nitrate from conversations with persons outside the Laboratory. Further, Williams' lack of a scientific background may have been the cause of his difficulty with the stoichiometric calculations. Finally, clear guidelines regarding what is within an EU examiner's expertise may have helped Williams avoid other problems identified in this section.

The pre-trial issues present relatively minor matters, but exemplify the need to follow applicable protocols and to have an orderly dispute-resolution procedure within the Laboratory.

#####

The FBI Laboratory: An Investigation into Laboratory Practices and Alleged Misconduct in Explosives-Related and Other Cases (April,1997) Office of the Inspector General, Department of Justice