CASE CLOSED … what really happened in the 2001 anthrax attacks?

* the case of the copier that wasn’t used by Dr. Bruce Ivins to photocopy the anthrax letters

Posted by DXer on March 14, 2010


The New York Times says the FBI’s anthrax case has “too many loose ends.” Find out where some of those looses ends might have originated in my novel CASE CLOSED. Sure it’s fiction, but many readers, including a highly respected member of the U.S. Intelligence Community, think my premise is actually “quite plausible.”

* buy CASE CLOSED at amazon *


39 Responses to “* the case of the copier that wasn’t used by Dr. Bruce Ivins to photocopy the anthrax letters”

  1. DXer said

    Dr. Majidi explains in a self-published e-book that:

    “The laboratory identified three “copy imperfections” on the letters to Senators Daschle and Leahy. These markings were compared to …. 1,014 photocopier exemplar sets collected from copy machines located inside or near the vicinity of every known biological laboratory that possessed virulent Ames anthrax in 2001. No matches were found.”

    Translation as applied to an Ivins Theory: Even though the Amerithrax Investigative Summary pointedly discussed Ivins presence in the library with the photocopy machines, a comparison of photocopy exemplars was NOT a match. If the reports were disclosed, GAO could confirm whether the examplars were photocopies made in September or October 2001. In the Amerithrax Investigative Summary, the report should have disclosed that a comparison was made and no match was found and the reports should have been disclosed.

  2. DXer said

    GAO, as an example of the FBI withholding documents and substituting innuendo, compare the withholding of Dr. Bartick’s report excluding the toner of the USAMRIID machines with the withholding of the rabbit formaldehyde protocol and the covance study rabbit notebook. GAO need only look at the innuendo in the report to see that the AUSA’s are using his time in the library on those dates to suggest he is guilty rather than as the exculpatory evidence that it is.

  3. DXer said

    GAO reports: “We will perform independent gap analyses of the scientific and technical methods.”

    I think the first scientist they want to contact is Dr. Bartick, who has deep expertise in the examination of toners. In the Amerithrax Investigative Summary there was innuendo about the time that Dr. Ivins spent in the USAMRIID library where there were photocopy machines. It is basic to the GAO’s mission to fill in the gap on the issue of the examination of the toner that was done. Those documents have not been produced by the FBI in any form.

    • DXer said

      David Willman may not understand that the report withheld by the FBI addresses “toner” — totally separate from the issue of “trash marks” — EXCLUDES the photocopiers at USAMRIID.

      The photocopiers at USAMRIID can be excluded based on research by Dr. Bartick that the FBI has failed to produce.

      The issue of toner is different from the issue of “trash marks.” It involves mass spec.

  4. DXer said

    Were the Goldman Sachs photocopied? There were many dozens of letters and so if not photocopied it, to me, seems a lot of work and evidences a lot of commitment by the writers. Written in red ink and double-spaced, it had your usual generic type message of HUNDREDS WILL DIE.

    (Even if photocopied, it seems a lot of work to address dozens of envelopes).

    While I think block handwriting can be faked or made to look like other block handwriting, the lengthy letter in cursive likely offers a wealth of information. We can look to see how handwriting was used other famous cases.

    This recurring issue points to the critical importance that the NAS reach the blossoming science relating to the examination of photocopy toner. High rates of confidence increasingly have been achieved — even reaching 99% confidence in excluding photocopy machines.

    The AUSA in a footnote implies the USAMRIID photocopy was used but given my quarter-century experience reviewing briefs by attorneys, the fact that Rachel and Ken hid the issue in a footnote relating to time Dr. Ivins was in the library, suggests to me that the science does not point to those machines. She otherwise would have cited the science in support. In either event, the science in fact was used by the FBI and the subject should be addressed by the NAS. Dr. Bartick, judging from his seminal work years ago and his cutting expertise on the forefront of such issues, is highly expert on the subject.

  5. DXer said

    Readers are reminded daily of the importance in this field that the NAS review the FBI’s science relating to photocopied threat letters — whether they include a pathogen or not.

    13 letters with white substance sent to Twin Cities businesses

    The threatening letters were sent to eight Home Depots and five Renewal By Andersen businesses. In 12 of them, the substance was baking soda.

    Star Tribune

    January 8, 2011 – 11:38 AM

    The FBI is investigating 13 threatening letters containing white powder and sent to Twin Cities stores.


    The return addresses are identical on all the letters, with “exactly the same letter that appears to have been photocopied,” the statement said.


    “The use, threat of use, attempted use or conspiracy to use a weapon of mass destruction, such as a hazardous material, is a federal offense,” said the statement from the FBI, which wasn’t releasing further information Friday.

  6. DXer said

    “One note might have been a photocopy of the other, Sheridan said during a news conference at state police headquarters at Sykesville.”

    Incendiary mailings spread from Maryland to D.C.

    Now if the NAS had addressed the FBI’s substantial experience and expertise with photocopy toner examination and confirmed that the FBI’s science was sound, it might now be tested science to apply confidently to this ongoing incident.

    The NAS task was to address the science used – and photocopy toner was used. So why didn’t the NAS address it? Why wasn’t Dr. Bartick invited to present at an open session? His work seems highly important and probative over a wide range of issues.

    As a mechanic, if you bill a million dollars to check an engine and look under the hood, then there is a problem if the “check engine” light is still on.

    There is an electronically available FBI DC Field Office analyst report showing the science that existed even as of 2001. As I recall, it is available from a standards organization and available for a modest charge.

    Although I am not scientifically trained, my reading of the peer-reviewed literature on photocopy toner examination is that the confidence levels have consistently increased due to advances in the scientific techniques. Dr. Bartick appears to be fully up to speed on the most current methods and so he would be an excellent witness for the NAS to hear.

    • DXer said

      Note relatedly that the FBI has never claimed that any piece of the hundreds of pages of paper taken from Dr. Ivins (and he kept old papers) — not a single one — was a match. They incredibly used the fact he wrote letters to the editor as if it was evidence against him without noting that the paper was not a match.

      Determination of Trace Elemental Concentrations in Document Papers for Forensic Comparison Using Inductively Coupled Plasma–Mass Spectrometry

      Article first published online: 29 JUN 2009

      Journal of Forensic Sciences

      Volume 54, Issue 5, pages 1163–1170, September 2009

      Abstract:  With improvements in manufacturing procedures, comparing physical characteristics of paper samples may not offer sufficient discrimination among different vendors. In this work, the potential to differentiate paper samples based on trace element concentrations was investigated. Paper samples from two different vendors were microwave-digested and trace element concentrations (Mg, Al, Mn, Fe, Sr, Y, Ba, Ce, and Nd) were determined using inductively coupled plasma–mass spectrometry. Differences in concentration were assessed statistically using two-way ANOVA and Tukey’s honestly significant differences test. Elemental concentrations were shown to be consistent across a single sheet as well as within a single ream of paper for each vendor. Reams from vendor A were differentiated based on Al and Ba concentration while reams from vendor B were differentiated based on Mg, Mn, and Sr concentrations. Paper was differentiated according to vendor based on significant differences in Ba, Sr, Ce, and Nd concentrations.


      A million dollar review of the science doesn’t buy what it once did.

  7. DXer said

    In a June 27, 2008 302 interview statement, it is noted that:

    “________ was asked for and provided several photocopied documents which were created (photocopied) in the 2000 and 2001 timeframe. (The original photocopies were given to SA ______ for inclusion into a laboratory submission, duplicates are included in the attached 1A.”

    The FBI has failed to produce the lab results showing that the USAMRIID photocopiers can be excluded.

    The examination by Dr. Bartick and his colleagues show that they can be EXCLUDED.

  8. DXer said

    A leading expert in the field of the forensic examination of photocopy toners is the FBI’s own Edward G. Bartick. In a chapter in a 2002 textbook, he explains the science as it existed and was known by the FBI’s Laboratory Division at the time of the Fall 2001 anthrax mailings. See Handbook of Vibrational Spectroscopy.

    Given the NAS’ inexcusable failure to include the FBI’s important science on photocopy toner as a subject at public meetings — accompanied by the all-important required production of documents — the GAO should obtain the documents and hear expert testimony from Dr. Bartick about the Lab’s examination of the USAMRIID photocopy.

    THEY CAN BE EXCLUDED AS A SOURCE OF THE ANTHRAX LETTERS! Now ask yourself: how does that fact established by the science and documentary evidence being willfully withheld for 2 years square with the representations and BS innuendo the AUSA made on this issue in the February 2010 Amerithrax Summary?

    Applications of Vibrational Spectroscopy in Criminal Forensic
    Edward G. Bartick
    Handbook of Vibrational Spectroscopy
    John M. Chalmers and Peter R. Griffiths (Editors)
    John Wiley & Sons Ltd, Chichester, 2002
    Applications of Vibrational Spectroscopy in Criminal
    FBI Academy, Quantico, VA, USA

    This is publication number 01-06 of the Laboratory Division of the Federal Bureau of Investigation….


    With the development of Fourier transform infrared (FT-IR) spectroscopy, the application of IR to forensic analysis became more prevalent because of the increased speed and sensitivity of FT-IR. The development of diffuse reflection (DR) accessories provided ease of sample introduction for several forensic applications. … Document analysis by DR has been reported for copy toners15,16 and inks.17 It was not until the 1990s that the use of FT-IR became more regularly applied in forensic laboratories. The introduction


    2.2 Raman spectroscopy

    Recent technological advancement in Raman spectrometers has provided a reason for exploring this method in forensic applications. While the applications have been slow to find their way into forensic laboratories, the advantages are being recognized and Raman spectrometers are starting
    to find use in forensic analysis.20

    3.2 Copy toners

    Questioned documents involving fraud and threatening letters are often produced on printers, copy machines and facsimile machines. The machine model identification of this common office equipment has been achieved through comparison of the resins of the toners used as ink. These
    “copy toners” have been studied for forensic analysis as a class of polymeric material. An example where copy toner analysis was used to produce an investigative lead was in a case involving a copied address label. A packaged 4 Forensic Applications of Vibrational Spectroscopy
    bomb, mailed to a corporate executive, had an address label that appeared to be an enlarged copy of the company’s return address logo typically used on company envelopes. Investigators suspected that the bomb had been mailed by an employee with access to internal supplies and that the
    person had copied an envelope using equipment within the company. There were over 200 copy machines, involving 62 different copier models, located throughout the facility.

    It was important for the investigators to know the copier model used to narrow the area of the investigation to employees with convenient access to a copier model of the type used to print the label. Sample pages were prepared from each of the 62 models and forwarded to the FBI Laboratory
    for analysis. In the laboratory, the samples were prepared for IR analysis using a heat transfer technique to remove the toner from the documents. The preparation technique involves heating the back of the paper with a soldering iron at a specified temperature and smearing the toner onto aluminum foil attached to a glass microscope slide. Spectra were obtained with an FT-IR microscope by R-A. With this method, the IR beam passes through the sample and is reflected from the aluminum foil to the detector via the microscope optics.22 Figure 2(a) shows the original spectrum of the toner from the bomb package label. This spectrum is sloped due to scattering from the carbon black particles used for the copy image. The baseline flattened spectrum in Figure 2(b) is typical of a styrene/acrylate copolymer. Significant variations in the IR spectra are produced by the polymeric resins which contain
    numerous additives that vary in type and quantity. A visual comparison of the case sample spectrum was made with the 62 spectra of the model types in the building. One Kodak model type matched closely with the case sample spectrum. The spectrum was also searched in an IR database
    of copy toner resins categorized based on over 800 copier and printer models.23 The search software narrowed the toner type to a group containing 24 models of machines. By careful scrutiny of the peaks, it was possible to narrow the spectra to six Kodak models in the database. The Kodak
    copier model from the corporate building was included in the computer search. Therefore, the results of the visual inspection and the computer search of the spectra corroborated. Two-thirds of the binders contained in the spectral database consist of the styrene resins plus additives to provide
    desired properties in particular copy machines. Other types of binders used are phenolic and polyethylene resins.

    The regions boxed off in Figure 2(b) contain small bands from the additives that provide the differentiating spectral features of this toner resin. Because the building contained only eight examples of that particular Kodak model, the results of this analysis permitted the investigators to narrow
    their search to personnel working in limited locations of the building. Thus, a suspect was determined in considerably less time than if the company’s entire personnel required investigation.

  9. DXer said

    About the hard drive of digital copy machines.

  10. DXer said

    A man walked into a very high-tech bar..
    As he sat down on a stool he noticed that the bartender was a robot. The robot clicked to attention and asked, “Sir, what will you have?”

    The man thought a moment then replied, “A martini please.”

    The robot clicked a couple of times and mixed the best martini the man had ever had.

    The robot then asked, “Sir, what is your IQ?”

    The man answered “oh, about 164.”

    The robot then proceeded to discuss the trillion spore concentration of the attack anthrax, microencapsulation, surfactants, and nanoemulsions.

    The man was most impressed.
    He left the bar but thought he would try a
    different tact.. He returned and took a seat.
    Again the robot clicked and asked what he would have? “A Martini please.”

    Again it was superb. The robot again asked “what is your IQ sir?”

    This time the man answered, “Oh about 100”. So the robot started discussing how Dr. Ivins had group therapy meetings the dates he was supposed to be mailing the letters, and various animal studies that he was working hard on (including contemporaneous lab notes being wrongfully withheld), and how the Speed Vac was too small to do the drying.

    The guy had to try it one more time. So he left, returned and took a stool….
    Again a martini, and the question, “What is your IQ?”?

    This time the man drawled out “Uh….. bout 50”.

    The robot clicked then leaned close and very slowly asked,

    d-i-d i-t
    n-o-w t-h-a-t y-o-u
    k-n-o-w t-h-e
    e-x-c-l-u-d-e-d t-h-a-t
    b-u-t K-e-n a-n-d
    R-a-c-h-e-l s-t-i-l-l
    r-e-a-s-o-n-e-d h-e
    u-s-e-d t-h-a-t m-a-c-h-i-n-e?

  11. DXer said

    Bruce Ivins wrote on January 29, 2002, that

    “we have a copy of the mailing label from the USDA. A few days ago I went back to the USAMRIID library and USAMRIID Library archives and looked through everyone of _________ notebooks. Unfortunately, he never mentions writing for the strain or receiving it in any notebook. The first mention of the strain is in September, 1981, when he also says he looks at the Sterne, Vollum (probably Vollum 1B), Colorado, Ames and Texas strains. Hope this helps.

    [Note: Given they kept so many notebooks in the library at 1425, copies of notebook pages were available to serve nicely as photoexemplars for the toner used by the machines and such exemplars were how the FBI was able to exclude the photocopy machine as the one used to copy the anthrax letters– and when AUSA’s Ken and Rachel imply that Dr. Ivins used that photocopy they do so having full access to the FBI report that the DOJ is withholding from the public in violation of FOIA.]

  12. DXer said

    Dr. Bannan, the lead FBI scientist, in 2001 was collections scientist at the Bacteriology Division of American Type Culture Collection, which co-sponsored Ali Al-Timimi’s program. (Ali had unfettered access to the collection and was coordinating with the 911 imam Anwar Aulaqi).

    Were the copiers at ATCC and the Center For Biodefense where Ali shared a suite with the leading anthrax scientist tested?

    The Defense Intelligence Agency, of course, is not dependent on the FBI forensics people to test the xeroxes made at Ft. Detrick library in September 2001 and October 2001.

  13. DXer said

    “First, the paper is examined. There are different kinds of paper varying in weight, strength, size, grade, watermark, treatment, and coating.”

    Forensics the Easy Way, By Harold Henry Trimm (2005)

    The paper was not a match either but Rachel and Ken have not disclosed that fact.

    How many of these forensic examiners contributed to the Amerithrax investigation?

    “Mar 12, 2010 7:02 pm US/Eastern
    DC Prosecutors Miss Target For Review Of FBI Work


    Prosecutors in Washington, D.C., say they need more time to complete a review of cases involving lab work by FBI scientists whose work came under fire in the 1990s.

    A judge asked for the report from the U.S. attorney’s office in December, when he released Donald Eugene Gates after 28 years behind bars.

    Gates’ conviction rested on testimony by FBI agent Michael Malone, one of 13 lab examiners criticized by the Justice Department’s inspector general in 1997.

    D.C. prosecutors say they’ve reviewed 20 cases involving the 13 examiners but found their role wasn’t significant.

    However, they’ve now learned of more than 100 additional cases and need more time.

    They had promised to finish by March 15.”

  14. BugMaster said

    If the FBI (or a quasi-governmental agency, for that matter) didn’t want the actual copy machine location disclosed, then why did they wave a red flag over the issue by this inane statement regarding a copy machine at Fort Detrick.

    Why bring attention to evidence they (claim) they don’t have?

    And then there’s Ed’s post from yesterday!

    “Uh oh. I was hunting for documents to use in a new supplemental page when I happened to notice that one of the FBI’s .pdf files containing the 2,700 pages of supporting data has been modified. Pdf file #847551 previously consisted of 180 pages. Now it consists of 172 pages. And the missing 8 pages are some of the pages I was using for my new supplemental page. Pages 70 – 77 of the copy I downloaded and saved on February 19, 2010, are not in the current version. They contain a description of an FBI interview with Dr. Ivins on June 13, 2008.”

    Lew, do you or anyone else have an archivial of file #847551? Can someone post pages 70-77.

    I am sure that Ed, FBI-worshipping propagandist do-bee that he is won’t let anyone have a peek at his old file.

    Ed’s conclusion:

    “They probably contain something that should have been redacted.”


    • Anonymous scientist said

      DXer has a copy of the original. There is an interview with Ivins especially concentrating on his relationship with 2 female co-workers, including one who left Detrick to take another job. I don’t want to give her name here (which is redacted in the document, but it is fairly easy to figure out her identity)

      • Anonymous scientist said

        Just speculating here but perhaps the ex-female coworker complained to the FBI that the wording around her redacted description made it easy to identify her so that’s why the FBI removed the pages.

        • BugMaster said

          Several pages from interviews with Dr. Ivins have been posted on this website. A portion of one in particular surprised me in what wasn’t redacted.

          Apparently, if a name is that of someone deceased (not Ivins) then the FBI must feel no need to redact it.

        • BugMaster said

          According to Ed’s website, that is not what the deleted sections appear to have been about.

        • DXer said

          I would gladly email the ..551 pdf so you can compare it to the original to what is online now and then post the pages that were omitted. Or Ed can do that.

        • DXer said

          The blogger you mention argues that it is 99% a First Grader wrote the letters and for six-months after Ivins’ death stridently argued that the genetics pointed to Ivins alone and not 350. He cherry-picked a short passage from the first page of what he describes as the June 13, 2008 interview.

          The interview was transcribed on 6/13/2008 but occurred on 6/9. It discusses his relationship with his former colleagues and good friends, Mara Linscott and Patricia Fellows. He had a severe falling out with them due to various stresses, his poor conduct and the growing paranoia and isolation he felt. The note from his wife to Dr. Ivins on the day of his death indicates that the FBI had prompted Mrs. Ivins into thinking that the FBI thought that he was having an affair with Pat and/or Mara.

          It states:

          “IVINS’ personal relation with _______________ is very complex. __________________________________. IVINS does not believe he ever did anything which would result in an EEO complaint, but he ________ did blindfold ____________ to an adult novelty store ______

          When __________ left USAMRIID _______ IVINS not angry, but he was very sad. He knows that she did not leave because of him, but she did leave him. In 2000, IVINS confided everything in ____________ and would do anything for _____. He was “quite taken” with ___ for various reasons. Sometimes he would drive by ____ apartment much like he would for KKG sorority houses. …

          When ________ left USARMIID left ___________________ was not angry , but he was very sad.”

          The FBI agents did not request the records relating to the Egyptian male who thanked both of these women, Mara and Pat, for providing technical assistance, until 2005 — even though he was the lifelong friend of a man of my acquaintance (who consults with the Intelligence Community) who had been recruited by Ayman Zawahiri into the Egyptian Islamic Jihad while he and the DARPA-funded virulent Ames researcher had both been schoolmates at Cairo Medical School. The University of Michigan researcher would visit Cairo from Khartoum as a child where his mother was an accounting professor.

          It has been known since 2001 that Ayman Zawahiri planned to use the cover of charities and universities as a cover for developing anthrax as a weapon. Cheney had insisted that the CIA and FBI share information efficiently because there was no priority greater than figuring out whether the anthrax letters were related to 911. But the record disclosed seems to indicate that the same USG that takes 2 years to make available a stack of Ivins’ emails did not request the pertinent documents from Bruce Ivins until 2005. Given this inefficiency, our prospects of avoiding the next attack simply are not at all good.

          Given it was Ayman Zawahiri whose intention to use anthrax had been publicly announced, and given that it was his group whose signature was mailed letters to people in symbolic position relating to the detention of WTC attackers, it is disappointing — but totally understandable — that Mara and Pat got distracted over the anger they likely felt at Bruce.

          It is also understandable that the busy prosecutors got confused. With Ken Kohl consumed by the Blackwater case, there was never even a remote chance of Amerithrax being sorted out or missteps corrected. These AUSA’s seem to have huge workloads and like so many others are a victim of circumstances.

          Neither Mara or Pat, nor the prosecutors, likely appreciated that Dr. Ivins’ anger stemmed in part from isolation that was due to the DOJ’s apparent cluelessness reflected by their failure to even request from Dr. Ivins the documents relating to the visit of the former associate of Heba Zawahiri until 2005.

          I emailed both Mara L. and Pat F. a couple years ago but understandably neither would respond.

          While I can appreciate the offense they took at Dr. Ivins’ gross and inexcusable invasion of their privacy, just because he was paranoid doesn’t mean he wasn’t being betrayed by the people with whom he had felt close and greatly admired. An unrequited love will always seem really pathetic — but is part of the human experience and we shouldn’t be so quick to judge.

          Why do we indulge’s Ed’s hundreds and hundreds of pornography he maintains for horny 13-year-olds and yet get titillated and disapproving of Dr. Ivins’ pictures of blindfolded women? Or his preoccupation with KKG that he indulged in at his workplace?

          The interview continues:

          “In 2000, IVINS was diagnosed with clinical depression and began taking medication for it.

          After ________ left for _______ IVINS obtained _______ computer password, and he would logon to the computer as ________ to read _____ email. In doing so, IVINS read emails between _________ and _________ in which they made less than favorable comments about him. After reading those emails, he was angered and felt betrayed.

          In November 2005, IVINS wrote an email to himself in which he theorized that ___________ and mailed the anthrax letters. His emotions are not rational, and accusing them of the mailings is a way to get revenge for the derogatory email exchanges about him. …

          IVINS does not know what emotional state he was in when he wrote the email to himself, but he was angry and felt “jilted” when he again accused them in a March 2007 witness conference before he testified in the grand jury. He does not believe ____________________ or any of the people in his laboratory mailed the letters.

          IVINS was shown an email he wrote to ___________ in June 2007 and asked to explain why he tried to scare her. He could offer no explanation other than emotions are irrational and dictate his actions. …

          When pressed on the factual accuracy of the email, such as IVINS’ assertion that individuals could be put to death for not locking a freezer, IVINS claimed innocent people often receive the death penalty. Referring to the portion of the email about people’s personal information being disclosed at judicial proceedings, IVINS referenced a conversation with ______________ about what happens during a trial, but he could not elaborate.

          [Note Attorney Paul Kemp’s recent lament last week that he had said what he said about the death penalty.] As for IVINS’ question in the email about ____________________________________________________________” [Note: the best stuff is always redacted, eh?]

          “Regarding DNA, it is comprised of Thymine, Adenine, Guanine, and Cytosine, which are commonly referred to as T, A, G and C, respectively. Codons are triplicate codes of the As, Cs, Ts and Gs for the 20 different amino acids. Different codons can be the code for the same amino acid. Each amino acid has a single letter designator, but IVINS does not know what they are. IVINS’ laboratory uses amino acids to make medium for the growth of spores, and when referring to the amino acids in his notebooks, IVINS either writes the entire word or uses the first three letters of the amino acid. IVINS is unaware if the amino acids in his lab have the single letter designator on the bottle, and he cannot translate the codons into the amino acids. For example, he does not know which amino acid is represented by the codon “TTT.”

          IVINS is not much of a “Gene Jockey”, but he knows that As and Ts match up, as do Cs and Gs. IVINS did some work on genes in 1986, but he would have to refer to a book to discuss the topic further. __________ is a Gene Jockey who could answer any such questions regarding DNA.

          IVINS has never used codons, As, Cs, Ts or Gs, or DNA to communicate, and he has never read anything on the topic. IVINS has never read or sent any messages using codons, As, Cs, Ts, or Gs, or DNA, and he has never read anything on that topic. He has never read anything or had discussions regarding how to send coded messages with codons, As, Cs, Ts, or Gs, or DNA.”


          “IVINS read about 3/4’s of the book [GODEL, ESCHER, BACH] and found it intellectually fascinating. It is about “Truths that cannot be proven within the system you’re trying to prove them in.” The book also analyzes Bach’s Magnificat, which proclaims, “generation, upon generation, upon generation, shall call me blessed.”

          Because the book is “really cool,” IVINS bought a copy for ___________ a few years ago, but he does not recall giving anyone else a copy of it. IVINS learned of the book in the late 1970’s or early 1980’s from either ______________ or someone with whom he rode a bus to work. He has a “dog eared” copy of the book at home.


          With regard to parental rights, IVINS believes state agencies should intervene in cases of child abuse of neglect. He volunteered that he is not in favor of corporal punishment. IVINS understands there could be justifiable reasons for a state agency to interview a child wtihout the presence of a parent, but he would have to see the specific case before offering an opinion. IVINS does not recall ever expressing an opinion on this topic.

          The AFA (American Family Association) Journal is very “far out” with its positions on many issues. It is possible that IVINS has read an article in the AFA Journal and made a donation to support htem, but he does not specifically recall such an instance.


          All of IVINS KKG related materials were kept in a cabinet over his desk at work. He does not recall ever storing anything at other locations.”

          [Note: He says the reason he often would go into the B3 rather than his office was to avoid having a security guard he didn’t like walk in on him; a colleague told investigators he used his office and the B3 interchangeably and would use the shower and internet at work.]

          When IVINS turned 60 years old, he was concerned he could have a heart attack or unexpectedly die. If that happened, his family or coworkers would find his KKG and fetish materials which would be very embarrassing. Therefore, IVINS decided to throw away all such items. He threw the KKG materials in the dumpster at work and his fetish materials in the regular trash can at home.


          IVINS did not receive the protocol or any instructions on the preparation of the submissions until May 24, 2002, when he received an email from _______”

          [Note: The former colleague of Heba Zawahiri, in addition to thanking Bruce Ivins for supplying virulent Ames, and Mara Linscott and Patricia Fellows for their technical assistance, thanked Kimothy Smith, the FBIs genetics expert in 2002, for providing the B3 space at LSU to do research with 4 characterized strains of virulent anthrax.]
          Yet, based on this record, the FBI only first obtained the documents about the former Zawahiri’s associate from Bruce Ivins in 2005. You would think that their own expert would mention that he had worked with virulent Ames with a former Zawahiri associate.]


          The lengthy Summary appears to be a conclusions in search of evidence and the heaps of innuendo only avoid being demolished by the intentional withholding of highly material evidence, such as the pertinent pages from numerous Lab Notebooks, Ivins emails, the report on the paper, toner and “trash marks” etc.

          Big picture: Lead investigator Lambert once wrote a memo to FBI Director Mueller explaining that compartmentalizing the squads would prevent investigators from “connecting the dots.”

          Agent Lambert was right. Due both to compartmentalization and the difficulties of proof, the Third Squad told Agent Montooth that they would go along with an Ivins Theory. And so while we can appreciate the good faith of the one squad in trying to fashion a viable Ivins Theory, and the inability of the Third Squad to make a different case, the fact is that the record evidence shows that they did not request the record of the former Zawahiri associate until 2005. That, without more, based on the record disclosed indicates that the Department of Justice and Federal Bureau of Investigation seriously dropped the ball in Amerithrax by a lack of diligence. The failure demonstrates a lack of adequate focus on the evidence relating to Ayman Zawahiri and the documentary evidence regarding his specific intent to use anthrax against US targets to retaliate for the rendering and mistreatment of senior EIJ leaders, including his brother Muhammad.

          Dr. Ivins was a deeply flawed man, to be sure, but he was also a victim of very regrettable circumstances.

          Dr. Zawahiri is also a deeply flawed man, to be sure, and was also a victim of very regrettable circumstances.

          The difference is that it is Dr. Zawahiri whose group had announced the specific intent to use anthrax against US targets. It was his group Vanguards of Conquest that had as its signature (not merely modus operandi) the use of lethal letters to NYC and DC newspapers and targets symbolic of the detention of WTC attackers.

    • DXer said

      They contain interview statement about IM’s, I believe, with PF, who was thanked by the former Zawahiri associate for providing technical assistance.

    • Anonymous said

      I think one can get a pretty good idea of why the seven page Ivins interview was removed from 847551.pdf from the following.

      Pages 70-77 of 847551.pdf reproduce an Interview with Dr. Bruce Ivins. The interview begins with the following statements:

      BRUCE EDWARDS IVINS, W/M. DOB: 04/22/46, SSAN 280-44-544-, [REDACTED] was advised of the identities of the interviewing agents and the purpose of the interview. Also present for the interview were Ivins’ attorneys, [REDACTED] and [REDACTED], and AUSAs [REDACTED] and [REDACTED] AUSA [REDACTED] provoded a letter to Ivins and his attorneys detailing the nature of the “off the record” interview…

      Footnote 1, Page 5 of The Amerithrax Investigative Summary states:

      Information derived from sources such as the federal grand jury investigation, sealed court orders, and an “off-the-record” interview of Dr. Bruce E. Ivins, while contributing to the overall investigation, is omitted from this Investigative Summary.


  15. DXer said

    They searched Dr. Ivins’ house in November 2007 and searched for “printers, toner or printer cartridges, photocopy exemplars.” The photocopy exemplars (examples of other things copied), to include things he copied at the USAMRIID library, did NOT match the anthrax letters. Why wasn’t that in the summary?

    • DXer said

      Given that he had attended a 4 hour Red Cross training session on September 22, 2001 (see Red Cross record uploaded 10/1), and we know there was no match with the anthrax letters, isn’t the more logical inference that what he was copying were one of the documents from that week that the FBI seized?

      Isn’t it more likely, for example, he was reading the newspaper, looking for a book, checking the catalog given that we know — from the forensic evidence — he wasn’t copying the anthrax letters?

      • DXer said

        Source: Wall Street Journal, December 11, 2001.
        The Anthrax Probe Ranges Far and Wide As Investigators Scour Tips, Trash for Leads

        Officials are also scrutinizing the photocopied letters. Analyzing the toner and the “feeder marks from the grippers that pull in the paper” can narrow down the type of copier used, says Gideon Epstein. He worked as a document examiner for many years with the Army crime lab and the Bureau of Alcohol, Tobacco and Firearms, and founded the Immigration and Naturalization Service’s forensic document laboratory. He says that document experts will also look for “what we call trash marks — scratches on glass, defects in the drum. These are very individual, so they may tie copies to a particular machine.” FBI officials acknowledge examining the documents for these clues, but won’t reveal what they’ve found.


        For an expert to consult, I commend Bill S.

        Hello, Thomas,

        Yes, DSC can definitely be used to study the physical properties of toners,
        particularly the polymeric binder. I know that the R&D group at Xerox uses
        several DSC’s, as well as DMA and DEA to look at the characteristics of

        I have analyzed toners myself using DSC and you can successfully analyze the
        toner itself as well as the properties of the toners taken from photocopied
        documents. As I recall, different toners (from different manufacturers) will
        give different DSC ‘signatures’, including Tg’s and/or melting peaks
        associated with the polymer(s). In one interesting study, I did when I was
        with Perkin Elmer, I was able to cut portions of photocopied documents to get
        about 10 mg of total sample (paper substrate and toner). I then cut pieces
        of the photocopied document where no toner was present to serve as the DSC
        reference. By heating the sample at 20 C/min, I was able to see the very
        weak DSC signature transitions for two different photocopied documents. This
        test allowed for the identification of the type of toner used to produce the
        particular document.

        In an article on the anthrax-mailer forensics investigation, the NY Times
        picked up on my work and mentioned it in an article. The Times was
        interested in this since the anthrax mail perpetrator had used photocopying.
        I got many calls the next day from Fox News and different news organizations
        asking me about my involvement with the FBI in the anthrax investigation. Of
        course, I wasn’t involved other than trying to come up with new applications
        for DSC! I believe that the applications note that I wrote on the
        characterization of photocopier toners is still posted on the PerkinElmer

        I hope this information is helpful to you.

        Best regards from sunny North Carolina,
        Bill Sichina

  16. DXer said

    “Page 13 explains that none of the 1,014 photocopier exemplar sets collected from copy machines located inside or near the vicinity of every known biological laboratory that possessed virulent Ames anthrax in 2001. No matches were found.”

    Not only was the Ft. Detrick paper not a match as to “trash marks,” however, it used entirely different toner ink and paper. (This is all subject to confirmation from purchase invoices, model number, copies made contemporaneously (to include those actually made by Ivins on that day).

    So why did the US Attorney’s Office try to falsely offer the fact he made copies as evidence of murder and why is the office — now under fire for faulty forensic analysis (see AP article today) — refusing to provide such centrally relevant evidence? Even the same spokesman who refused us any documents is the spokesman on the investigation of past mistakes in forensic analysis by the Office/FBI.

    Certainly, it is a material piece of evidence (the fact it is tucked away in a footnote notwithstanding) (One always first looks at an opposing counsel’s footnotes to see the issues that they cannot handle or the issue where they are especially likely to be hiding the ball.

    • DXer said

      As an example from case law, in one Minnesota case the expert was the state’s “Question Document Examiner” with the Minnesota Bureau of Criminal Apprehension, and also the forensic science supervisor. She had eight years’ experience in the question document section of the laboratory, had been the lead document examiner since 1984, and had been with the BCA since 1970. She had taken two and a half years of training in document examination, and worked several hundred cases in the document section. She attended question document schools at the FBI Academy and the Secret Service. Some of her initial training had specifically addressed photocopies, and she attended three one or two-day workshops on photocopier evidence. She handled 20 to 25 cases involving photocopy evidence.

      I have no doubt that the FBI had top experts available to it. So let’s see the evidence on the toner, the trash marks, and the paper used — the fanciful speculation in the Department of Justice investigative summary is contradicted by the evidence that it improperly is failing to provide while making accusations inconsistent with that evidence.

  17. DXer said

    Will it take 27 years before the District of Columba United States Attorney Office comes clean and acknowledges that the toner, paper and “trash marks” were not a match with the photocopier in the library used by Ivins and others at the Ft. Detrick library? (In fact, by comparing the copies made by others at the time, including Ivins himself, the copier can be excluded based on the “trash marks”, just as it can based on the toner and paper used).

    Suspicions about FBI analysts growing
    Washington Post – Keith L. Alexander – ‎10 hours ago‎

    The US attorney’s office in the District has found more than 100 cases since the mid-1970s that need to be reviewed because of …
    A prosecutor says “I’m sorry” to a wrongfully convicted man. Really!

    True/Slant – Steve Weinberg – ‎Mar 12, 2010‎
    Of all the professional occupations I’ve encountered, prosecutors are the least likely to apologize when they make grievous errors. …
    DC prosecutors miss target for review of FBI work

    Washington Examiner – Sarah Karush – ‎Mar 12, 2010‎
    AP WASHINGTON — Prosecutors in the District of Columbia said Friday they need more time to review more than 100 cases involving lab work by FBI scientists …
    All 4 related articles »

    Some more representative articles about the forensics on the copier:

    Monday, January 21, 2002
    FBI examined photocopiers on campus

    Princetonian Staff Writer

    McLaughlin said the agents made two copies at each copier without any document in the machine — one copy with the lid up and another with the lid down. He explained that these copies could reveal telltale scratches on the copier glass or faults on the machine’s drum.

    The FBI is investigating letters containing anthrax bacteria, which were mailed from the Trenton area in October. The bureau has repeatedly declined to comment on the specifics of its investigation. University Director of Communications Lauren Robinson-Brown ’85 also would not comment on the situation.

    All four of the anthrax-laden letters were prepared on a copying machine. Comparing these letters to samples it has taken from machines, the FBI may be able to determine which machine was used to copy the letters.

    The Times of Trenton reported yesterday that the FBI has been testing other public copiers in the Mercer County area, including machines in local public libraries and the Pequod outlet inside the University Store. Machines at a Piscataway research institute affiliated with Rutgers University have also been tested.


    New York Post – New York, N.Y.
    Date: Jan 16, 2002

    The FBI has been comparing the four anthrax-laced letters mailed from Trenton to photocopies made on copy machines at two New Jersey colleges.

    Microbiology professor Richard Ebright reported that when asked if the photocopier tests at Rutgers were linked to the anthrax probe, one of the agents replied, “We can’t leave any stones unturned, so we’re turning over stones’ – or something to that effect.”


    AUGUST 4, 2008
    Forensics Gave Investigators Little to Work With

    Did the photocopies contain stray marks from the copier — or copiers — used to make them? Yes, but investigators were apparently never able to find the copy machines.


    New York Post – New York, N.Y.
    Author: BRAD HUNTER
    Date: Dec 10, 2001

    FBI agents have now determined that the poisoned letters sent to Sen. Patrick Leahy and Senate Majority Leader Tom Daschle are virtually identical, leading them to look for a possible Kinko’s connection.


    • DXer said

      Why are Rachel and Ken playing “hide the toner?”

      William J. Egan1, 4, Stephen L. Morgan1, Edward G. Bartick2 , Rena A. Merrill2 and Hollis J. Taylor III3

      (1) Department of Chemistry and Biochemistry, The University of South Carolina, Columbia SC 29208, USA
      (2) Counterterrorism and Forensic Science Research Unit, FBI Laboratory, FBI Academy, Quantico, VA 22135, USA
      (3) Questioned Documents Unit, FBI Laboratory, Quantico, VA 22135, USA
      (4) Present address: Vertex Pharmaceuticals, 130 Waverly Street, Cambridge, MA 02139-4242, USA

      Edward G. Bartick

      Received: 14 March 2003 Revised: 19 May 2003 Accepted: 20 May 2003 Published online: 9 July 2003

      Abstract Copy toner samples were analyzed using reflection-absorption infrared microscopy (R-A IR). The grouping of copy toners into distinguishable classes achieved by visual comparison and computer-assisted spectral matching was compared to that achieved by multivariate discriminant analysis. For a data set containing spectra of 430 copy toners, 90% (388/430) of the spectra were initially correctly grouped into the classifications previously established by spectral matching. Three groups of samples that did not classify well contained too few samples to allow reliable classification. Samples from two other pairs of groups were similar and often misclassified. Closer examination of spectra from these groups revealed discriminating features that could be used in separate discriminant analyses to improve classification. For one pair of groups, the classification accuracy improved to 91% (81/89) and 97% (28/29), for the two groups, respectively. The other pair of groups were completely distinguishable from one another. With these additional tests, multivariate discriminant analysis correctly classified 96% of the 430 R-A IR toner spectra into the toner groups found previously by spectral matching.
      Keywords Pattern recognition – Discriminant analysis – Questioned documents – Photocopy toner – Reflection-absorption infrared spectroscopy – Spectral library

      This is publication number 03–03 of the Laboratory Division of the Federal Bureau of Investigation. Names of commercial manufacturers are provided for identification only, and inclusion does not imply endorsement by the Federal Bureau of Investigation.


    • DXer said

      Special Issue Paper

      Forensic discrimination of photocopy and printer toners II. Discriminant analysis applied to infrared reflection-absorption spectroscopy
      William J. Egan1, 4, Stephen L. Morgan1, Edward G. Bartick2 , Rena A. Merrill2 and Hollis J. Taylor III3

      (1) Department of Chemistry and Biochemistry, The University of South Carolina, Columbia SC 29208, USA
      (2) Counterterrorism and Forensic Science Research Unit, FBI Laboratory, FBI Academy, Quantico, VA 22135, USA
      (3) Questioned Documents Unit, FBI Laboratory, Quantico, VA 22135, USA
      (4) Present address: Vertex Pharmaceuticals, 130 Waverly Street, Cambridge, MA 02139-4242, USA

      Edward G. Bartick

      Received: 14 March 2003 Revised: 19 May 2003 Accepted: 20 May 2003 Published online: 9 July 2003

      Abstract Copy toner samples were analyzed using reflection-absorption infrared microscopy (R-A IR). The grouping of copy toners into distinguishable classes achieved by visual comparison and computer-assisted spectral matching was compared to that achieved by multivariate discriminant analysis. For a data set containing spectra of 430 copy toners, 90% (388/430) of the spectra were initially correctly grouped into the classifications previously established by spectral matching. Three groups of samples that did not classify well contained too few samples to allow reliable classification. Samples from two other pairs of groups were similar and often misclassified. Closer examination of spectra from these groups revealed discriminating features that could be used in separate discriminant analyses to improve classification. For one pair of groups, the classification accuracy improved to 91% (81/89) and 97% (28/29), for the two groups, respectively. The other pair of groups were completely distinguishable from one another. With these additional tests, multivariate discriminant analysis correctly classified 96% of the 430 R-A IR toner spectra into the toner groups found previously by spectral matching.
      Keywords Pattern recognition – Discriminant analysis – Questioned documents – Photocopy toner – Reflection-absorption infrared spectroscopy – Spectral library

      This is publication number 03–03 of the Laboratory Division of the Federal Bureau of Investigation. Names of commercial manufacturers are provided for identification only, and inclusion does not imply endorsement by the Federal Bureau of Investigation.


      The use of office and personal photocopying machines has increased dramatically over the last 20 years. As a result, photocopying documents has become simple, fast, and inexpensive. A major disadvantage of photocopy machines is that they are now more accessible for illegal activities such as counterfeiting, fraud, false documents, anonymous letters, confidential materials, and acts of terrorism [1, 2, 3]. Identification of the source of photocopied documents is not an easy task for forensic examiners since chemical and physical characteristics are very similar and numerous manufacturers of photocopy instruments and toner cartridges exist. The ability to match the chemical fingerprints of questioned toner samples to standards could be a valuable tool in questioned document investigations.

      Toner analysis methods that are useful in forensic investigations must be rapidly performed and possess a known degree of accuracy. Totty [4] reviewed analytical techniques that have been used to characterize toners: visual examination, optical microscopy, scanning electron microscopy (SEM), magnetic viewers, infrared spectroscopy (IR), pyrolysis gas chromatography and/or mass spectrometry (Py-GC, Py-GC/MS, Py-MS), and differential scanning calorimetry (DSC). Early work by Kemp and Totty [5] found that 79 toners from various models of photocopier machines could be separated into 10 groups based on their IR spectra. Williams [6] identified numerous resins and the pigment Prussian blue based on characteristic IR absorptions. The possibilities of toner analysis and classification by IR spectroscopy and by diffuse reflection infrared Fourier transform spectroscopy (DRIFTS) have been described by other researchers [7, 8, 9, 10, 11, 12, 13, 14]. Merrill et al. [15] conducted a comparative study of three microscope-based IR techniques and DRIFTS for the analysis of toner samples. Reflection absorption infrared spectroscopy (R-A IR) performed best overall, in terms of low cost, rapid speed of analysis, nondestructiveness, and quality of spectra. Bartick and Merrill [16] also began development of an R-A IR spectral database for copy toners. More recently, toner samples extracted from photocopies with carbon tetrachloride have been analyzed by FT-IR to identify chemical constituents. [17] DRIFTS, SEM, and Py-GC have also been compared for differentiating photocopier toners [18].

      Hundreds of copiers and printers are available from different manufacturers, varying in model, engine, and toner used. Comparison of a large number of spectra is tedious even when computer-assisted spectral matching is employed, and the accuracy of the classification results may not be quantifiable. Multivariate statistical methods offer a potential solution to these problems. As noted in a 1996 review on chemometrics, “[t]here were only a few papers that focused on the application of pattern recognition techniques to forensics, which is surprising in view of the potential impact that multivariate methods can have on this field” [19]. Paper samples have been differentiated on the basis of multivariate statistical analysis of their elemental compositions [20, 21]. Andrasko [13] has also applied two simple measures, the Euclidean distances between spectra and a similarity index, to differentiate black ink and color toner samples by reflection FTIR.

      Merrill and Bartick [22] have previously compared prominent features and used computer-assisted spectral matching to divide 807 R-A IR spectra of toners into 98 subgroups based on the presence, absence, and ratios of peaks in 40 different spectral regions. Group assignments were summarized using a flowchart with nodes representing yes/no decisions with regard to presence, absence, or ratios of spectral peaks at specified wavenumber locations. The groups at the end point of branches represented clusters of similar copy toner spectra for which further discrimination was judged not possible. Linear discriminant analysis (LDA, also known as canonical variates analysis or CVA) is a multivariate statistical method that facilitates the objective evaluation of the classification of objects (in this case, spectra) into groups. We have previously applied LDA to R-A IR spectra from a subgroup of 60 toners having a poly(styrene-co-acrylate) base component [23]. The objective of the present paper is to evaluate whether groupings of R-A IR spectra of photocopy and printer toners for a larger subset of 430 toner samples can be reliably discriminated by statistical pattern recognition.


      Samples of dry photocopier and printer toners on paper were collected by the FBI Laboratory from verified sources that include original manufacturers. The R-A IR data set used in this study (430 spectra classified into 27 groups) is a subset of the complete library of 807 toner samples categorized into 98 groups previously described [22]. The entire set of 807 samples was not available at the time of this study. Toners were transferred from documents to reflective media (heavy duty aluminum foil affixed to standard glass slides with double-sided tape) using a temperature-regulated soldering iron set at 288 °C [15, 22, 24]. The soldering iron was equipped with a screwdriver tip that had been ground off, leaving a flattened round head with a 4.8-mm diameter. Although other materials provide a suitable reflective surface for the reflection-absorption technique, aluminum foil is readily available, inexpensive, and permits the sample to be stored for further studies. The sample preparation is simple, fast, and essentially nondestructive. The document is still legible after transferring the toner sample and only minimal destruction is visible microscopically.

      Samples were analyzed by R-A IR using a Spectra-Tech IR-Plan microscope with a medium-band MCT detector (Shelton, CT). The instrument collected 256 scans at 4-cm–1 resolution over the 650–4,000 cm–1 range for a total of 1,039 data points per spectrum.

      Data analysis
      Data was preprocessed and analyzed using OMNIC v. 3.0 (Nicolet Analytical Instruments, Madison, WI), Microsoft Excel (Microsoft Corporation, Redmond, WA), and programs written in MatLab (The MathWorks, Inc., Natick, MA).

      The baselines of all R-A IR spectra were manually adjusted using the OMNIC software to remove background dispersion effects caused by carbon black in the samples [22]. Baseline adjusted spectra were then arranged in a matrix, X, whose n rows represent different spectra and m columns represent spectral frequencies (wavenumbers): (1)

      The element xij of this matrix is the absorbance intensity at wavenumber j of spectrum i. Each spectrum (row) was then normalized to unit length by dividing each spectral intensity by the square root of the sum of the squares of all spectral intensities in that spectrum. Normalization removes systematic variation associated with size or amount effects in the spectra. After normalization, a typical preprocessing step is to autoscale the absorbances at each wavenumber by subtracting the mean spectral intensity for wavenumber j, and dividing by the standard deviation of spectral intensity for wavenumber j. Autoscaling produces a column autoscaled matrix with elements, (2)
      having a mean of zero and a standard deviation of one for each feature (wavenumber). In Eq. (2), xij represents the intensity for wavenumber j in the ith spectrum, j is the mean intensity for wavenumber j, and sj is the standard deviation of intensity for wavenumber j. This common data transformation removes inadvertent weighting caused by variations in the magnitude of intensity at different spectral frequencies [25]. In autoscaling the present data, the median was used instead of the mean in the numerator of Eq. (2); the median was also employed in place of the mean in calculations of the standard deviation. Error introduced by outliers is caused by a distortion of the location of the center of the data; the use of the median in these calculations may provide a more robust estimate of the true center of the data [26, 27].
      After median autoscaling, the data was further preprocessed using principal component analysis (PCA) [28, 29] via singular value decomposition (SVD) [30] to project the spectra, initially consisting of intensities at 1,039 wavenumbers, into a space of reduced dimensionality. SVD decomposes the data matrix into the product of three matrices: (3)
      where the matrix X refers to the data matrix after any preprocessing. PCA creates linear combinations of the original spectral variables, called eigenvectors or principal components (PCs), which successively account for increasing amounts of variation in the data. The matrix S contains the square roots of the eigenvalues of X (the singular values), ordered largest to smallest from top left to bottom right along the diagonal; the square of these values define the proportional variance explained by each PC. The columns of the matrix V contain the weights of the original variables (the loadings) necessary to form the principal component scores (U × S). If the first few PCs are found to explain a substantial proportion of the variation in the data, the projection of points representing the samples in a two- or three-dimensional plot may be informative concerning their similarity. For each analysis, a number of PCs was retained for discriminant analysis to capture an adequately large fraction of variance in the data.
      After preprocessing and data compression using PCA, linear discriminant analysis was used to construct axes which best separate the groups by maximizing the ratio of their between- to within-group variances [31, 32, 33]. Discriminant analysis is well suited for the analysis of grouped data and has a long history of use in analytical chemistry [20, 34, 35, 36, 37, 38]. The implementation of LDA is based upon three matrices, (4)
      representing the total, within-groups, and between-groups sums of squares and products matrices, respectively. The matrix represents the mean of X, while j designates a group of samples. The canonical variates (CVs) are defined by the eigenvectors of the matrix W–1B, with the proportion of variance accounted for each CV proportional to the eigenvalues. As with PCA, if a sufficiently large proportion of the variability associated with the first few CVs, a projection of the data points (the spectra) in the two- or three-dimensional space of the CVs permits the researcher to visualize clustering and similarity of the data. The clustering of similar samples can be assessed by comparison to the distances between samples (spectra) judged different from one another.
      Group definitions followed the groups experimentally established from the R-A IR spectra by visual comparison and computer-assisted spectral matching [22]. Jackknife cross validation was used to estimate the predictive ability of the LDA model by removing each sample from the data set in turn and recomputing discriminant functions based on the remaining samples [39, 40]. Estimates of the classification error for each sample are obtained without using that sample to calculate the discriminant model. This “leave-one-out” method is useful when a separate test set of data is unavailable, because it provides a way to estimate the classification error rate from the available data. The similarity between the jackknifed sample and the mean vectors for each group, calculated by scaling each vector to unit length and multiplying them together, was used to assign group membership.

      Elliptical confidence regions around the spectra of toner groups were calculated as follows. Each ellipse represents distances which are statistically equidistant from the group mean for a predetermined level of probability. Confidence regions were calculated by transforming each group to a principal component representation. A confidence circle based on Hotelling’s T2 statistic (the multivariate generalization of Student’s t statistic) was calculated and then transformed back into the original variables, forming an ellipse [41]. Note that even if the original two variables are already PCs, the method calculates new PCs for each group separately. Hence, the confidence region may take the form of an ellipse, not a circle, when plotted on the original spectral variables.

      Univariate Fisher ratios [31, 32, 37] were used as an indicator of which spectral features (wavenumbers) were individually most discriminating for separation of any specified groups. Fisher ratios, calculated as the ratio of between- to within-group variability for selected spectral wavelengths, range from zero (a nondiscriminating feature) to an unbounded upper value. Larger Fisher ratios indicate more discriminating features.


      Results and discussion
      Representative spectra of toners from this data set have been previously shown by Merrill and Bartick [22]. R-A IR microspectroscopy enables discrimination of toners by their organic and polymeric components. Although an experienced analyst becomes expert at recognizing distinguishing features in these complex IR spectra, the pattern recognition task is subjective and becomes quite difficult and time-consuming when numerous samples are compared. Pattern recognition techniques such as principal component analysis and multivariate discriminant analysis offer approaches to handling this complexity and to assess the statistical validity of differences observed between different samples.

      PCA was applied as a dimensionality reduction technique to the data set of 430 R-A IR toner samples spanning 27 assigned groups. The wavenumber regions 2,200–2,750 and 3,201-3,998 cm–1 were deleted from the spectral data prior to PCA, because no peaks were located in these regions. The first three PCs comprised 51.13% of the variation, PCs 1–28 comprised 95.18% of the variation, and PCs 1–139 comprised 99.95% of the variation. The first 139 PCs were used as inputs for LDA, reducing the number of variables from the original 1,039 wavenumbers while preserving the majority of the variation in the data. Fig. 1 shows projections into the plane of the first two PCs for the entire R-A IR data set. While the plot is crowded, some clustering by R-A IR groups is revealed. In the preceding paper, R-A IR spectra were categorized into groups using five flowcharts. In Fig. 1, the greatest separation is between spectra from groups appearing in charts 1–3 (left side of the lower plot) compared to spectra from groups appearing in charts 4–5 (right side).

      Fig. 1. The projection of all 430 R-A IR toner spectra into the space of the first two PCs: (upper) samples labeled by their group numbers; (lower) samples labeled by chart designation from the grouping assigned by visual comparison and computer-assisted spectral matching [22]


      LDA of the entire data set produced 26 canonical variate axes. Such a large and complicated data set would be expected to require a large number of CVs to discriminate between groups. Projection of the spectra into the space of the first three CVs (Fig. 2) explains just 63.51% of the dispersion in the data, supporting this reasoning. However, a number of R-A IR groups are well separated by projections on only the first three CVs. Table 1 summarizes the classification accuracy for jackknifed cross validation of classification by LDA using all 26 CVs. The results were quite good: 90.23% (388/430) of the toners were correctly identified into the predetermined groups. Groups 6 and 8 had poor correct classification percentages, but contained only a few samples. It is necessary to increase the number of toners in these groups before better conclusions can be drawn concerning their classification.

      Fig. 2. The projection of all 430 R-A IR toner spectra into the space of the first three canonical variates, labeled by their group numbers. Three CVs explain 63.51% of the total dispersion. R-A IR toner groups 12, 56, 59, 70, 75, 78, and 81 are well separated on these axes


      Two other similar pairs of R-A IR groups were difficult to classify, and, in fact, were most often misclassified as the other group: groups 42 and 49, and groups 64 and 67. These four groups, however, did have enough samples to make further investigation worthwhile. Visual inspection of the mean spectra for groups 42 and 49 shows that the two groups differ only in two small regions: group 42 possesses small peaks at 1,115 and 1,270 cm–1, whereas the peaks present in group 49 do not. Univariate Fisher ratios based on these two groups show that the two peaks are, indeed, the most important spectral features separating these two groups. The groups’ mean spectra and univariate Fisher ratios are shown in Fig. 3. Fig. 4 is an expanded view of Fig. 3. The presence of a peak at 1,268–1,272 cm–1 is a distinguishing characteristic between the groups. The univariate Fisher ratios indicate that the 1,095–1,115 cm–1 is also important for separating the two groups.

      Fig. 3. Mean spectra and univariate Fisher ratios for groups 42 and 49. Note the small peaks centered at 1,115 and 1,270 cm–1 which have high univariate Fisher ratios, indicative of their importance for discriminating between the two groups


      Fig. 4. Expanded view of the mean spectra and univariate Fisher ratios for groups 42 and 49


      To assess the ability of the 1,115 and 1,270 cm–1 peaks to differentiate samples from groups 42 and 49, we selected wavenumber regions with univariate Fisher ratios exceeding 0.15 (1,097–1,115 and 1,261–1,277 cm–1) for further analysis. When PCA was used to compress the data, a bi-plot (Fig. 5) of the projections of the spectra on the first two PCs (99.94% of the total variation) showed that the 95% confidence ellipses overlap. However, using the first 17 normalized PCs (99.99% of the total variation) as inputs for LDA, the jackknifed classification results for groups 42 and 49 improved considerably. The previous correct classification percentages were 88.76% (79/89) and 72.41% (21/29) for group 42 and group 49, respectively, based on the classification model created using all 430 samples. Use of only the two specific spectral regions improved the correct percentages to 91.01% (81/89) and 96.55% (28/29), for groups 42 and 49, respectively. The group 42 copy toners have a poly(styrene:acrylate) base component and were classified in a single group because previous visual analysis could not distinguish them [22]. However, linear discriminant analysis applied to R-A IR spectra was able to discriminate toners from several group 42 toners including AB Dick, Brother, Copystar, Okidata, Newgen, and Texas Instruments [23].

      Fig. 5 . Projection of spectra from groups 42 and 49 into the space of the first two PCs using the 1,097–1,115 and 1,261–1,277 cm–1 spectral ranges. The ellipses represent 95% confidence bounds on samples from each group


      We performed a similar analysis to investigate discrimination between the other problematic groups: 64 and 67. The 2850–2852 cm-1 range has been used to distinguish between samples in these two toner groups [22]. The LDA model for the entire data set misclassified over half the samples in each of these two groups as belonging to the other group. Fig. 6 shows the mean spectra and univariate Fisher ratios for the two groups. An expanded view of the group 64 and 67 samples is shown in Fig. 7. When the 2,841–2,860 cm–1 spectral region was used as the sole input for PCA, the differences between the groups were highlighted and discrimination between the two groups was possible. A bi-plot (Fig. 8) of the spectra projected on the first two PCs (99.94% of the total variation) shows that the 95% confidence ellipses do not overlap; groups 64 and 67 are totally separable when just the 2,841–2,860 cm–1 range is used.

      Fig. 6. Mean spectra and univariate Fisher ratios for groups 64 and 67. Note the small peak centered at 2,850 cm–1 which has a high univariate Fisher ratio, indicative of its importance for discriminating between the two groups


      Fig. 7. Expanded view of the mean spectra and univariate Fisher ratios for groups 64 and 67


      Fig. 8. Projection of spectra from groups 64 and 67 into the space of the first two PCs using the 2,841–2,860 cm–1 spectral range. The ellipses represent 95% confidence bounds on samples from each group


      Despite the difficulties in separating the two very similar pairs of groups, the overall LDA classification results for the R-A IR data are excellent. When all groups were considered, a 90.23% (388/430) correct classification rate was achieved. When more specific analyses highlighting the importance of smaller spectral regions were performed on groups 64, 67, 42, and 49, the overall percentage of correctly classified toners rose to 95.81% (412/430).


      We have demonstrated the successful application of multivariate statistical methods to the differentiation of photocopy and printer toners using reflection-absorption infrared spectroscopy. Discriminating among different types and manufacturer’s brands of toner by visual examination of these relatively complex IR spectra can be time-consuming and subjective. Additionally, the forensic analyst may not be able to fully utilize the fine structure of the pattern due to its complexity. Multivariate pattern recognition methods can take into account the entire spectrum and thus potentially have more information to use for discrimination, but are also sensitive to minor but discriminating spectral features.

      The focus of this work is development of statistical-based strategies for data handling offering improvements in method validation and ease of interpretation. Multivariate techniques provide a greater ability to discriminate between groups in large sample sets compared to visual analysis. Interpretation time can be reduced because the approach has the potential to be automated. The examples discussed illustrate the potential for computer-assisted data interpretation of forensic analytical data to provide decisive forensic identification of questioned samples. For each set of data, a visually interpretable map displaying the quantitative similarity of the IR spectra of forensic samples can be created. In this work, linear discriminant analysis, combined with some further tests based on specific spectral regions, was able to correctly classify 95.81% of the 430 R-A IR toner spectra into the groups previously established by visual comparison and computer-assisted spectral matching. This work has demonstrated the statistical validity of the groups of toner spectra assigned by previous work at the FBI Laboratory [22]. Further work in our laboratories involves comparisons of the discrimination achieved by R-A IR to that achieved using copy toner elemental compositions determined by scanning electron with X-ray dispersive analysis (SEM-EDX) and organic polymer composition analyzed by Py-GC/MS [42].

      Acknowledgements This work was funded by the FBI Academy and by award number 97-LB-VX-0006 from the Office of Justice Programs, National Institute of Justice, Department of Justice. Points of view in this document are those of the authors and do not necessarily represent the official position of the US Department of Justice. Portions of this work were presented at Pittcon ’97 (Atlanta, GA, 18 March 1997), at FACSS ’97 (Providence, RI, 27–31 October, 1997), and at the American Academy of Forensic Sciences Meeting (San Francisco, CA, 12 February 1998).


      1. Cantu AA (1991) Anal Chem 63:847A-854A

      2. Totty RN (1990) Forensic Sci Int 46:121–126

      3. Gilmour CL (1994) Can Soc Forens Sci J 27:245–259

      4. Totty RN (1990) Forensic Sci Rev 2:1–23

      5. Kemp GS, Totty RN (1983) Forensic Sci Int 22:75–83

      6. Williams RL (1983) Forensic Sci Int 22:85–95

      7. Zimmerman J, Mooney D, Kimmett MJ (1986) J Forensic Sci 31:489–493

      8. Lennard CJ, Mazzella WD (1991) J Forensic Sci Soc 31:365–371

      9. Shiver FC, Nelson MS, Nelson LK (1991) J Forensic Sci 36:145–152

      10. Mazzella WD, Lennard CJ, Margot PA (1991) J Forensic Sci 36:449–465

      11. Mazzella WD, Lennard CJ, Margot PA (1991) J Forensic Sci 36:820–837

      12. Andrasko J (1994) J Forensic Sci 39:226–230

      13. Andrasko J (1996) J Forensic Sci 41:812–823

      14. Trzcinska BM, BrozekMucha Z (1997) Mikrochim Acta Suppl 14:235–237

      15. Merrill RA, Bartick EG, Mazzella WD (1996) J Forensic Sci 41:264–271

      16. Bartick EG, Merrill RA (1995) In: Jacob B, Bonte W (eds) Advances in forensic sciences, vol 3: Forensic criminalistics 1. Verlag, Berlin, pp 310–313

      17. Tandon G, Jasuja OP, Sehgal VN (1997) J Forensic Document Examiners 3:119–126

      18. Brandi J, James B, Gutowski SJ (1997) Int J Forensic Document Examiners 3:324–344

      19. Brown SD, Sum ST, Despagne F, Lavine BK (1996) Anal Chem 68:21R–61R

      20. Duewer DL, Kowalski BR (1975) Anal Chem 47:526–530

      21. Simon PJ, Giessen BC, Copeland TR (1977) Anal Chem 49:2285–2288

      22. Merrill RA, Bartick EG, Taylor HJ III (2003) Anal Bioanal Chem Doi 10.1007/s00216-003-2074-z (in press)

      23. Bartick EG, Merrill RA, Egan WJ, Kochanowski BK, Morgan SL (1998) In: de Haseth JA (ed) Fourier transform spectroscopy: 11th international conference, American institute of physics Woodbury, pp 257–259 (13 August 1997); AIP Conference Proceedings 430:257–259

      24. Bartick EG, Tungol MW, Reffner JA (1994) Anal Chim Acta 288:35–42

      25. Sharaf MA, Illman DL, Kowalski BR (1986) Chemometrics. Wiley, New York

      26. Rousseeuw PJ (1991) J Chemom 5:1–20

      27. Egan WJ, Morgan SL (1998) Anal Chem 70:2372–2379

      28. Jolliffe IT (1986) Principal component analysis. Springer, Berlin Heidelberg New York

      29. Wold S, Esbensen K, Geladi P (1987) Chemom Intell Lab Syst 2:37–52

      30. Press WH, Flannery BP, Teukolsky SA, Vetterling WT (1986) Numerical recipes: the art of scientific computing. Cambridge University Press, New York, pp 52–64

      31. Fisher RA (1936) Ann Eugenics 7:179–188

      32. Mardia KV, Kent JT, Bibby JM (1979) Multivariate analysis. Academic Press, San Diego

      33. Campbell NA, Atchley WR (1981) Syst Zool 30:268–280

      34. Windig W, Haverkamp J, Kistemaker PG (1983) Anal Chem 55:81–88

      35. Vallis LV, MacFie HJ, Gutteridge CS (1985) Anal Chem 57:704–709

      36. Devaux MF, Bertrand D, Robert P, Qannari M (1988) Appl Spec 42:1015–1019

      37. Sahota RS, Morgan SL (1992) Anal Chem 64:2383–2392

      38. Ma CY, Bayne CK (1993) Anal Chem 65:772–777

      39. Tukey JW (1958) Ann Math Stat 29:614

      40. Mosteller F, Tukey JW (1986) In: Jones LV (ed) The collected works of John W. Tukey. Wadsworth & Brooks Monterey, vol IV, chap 15

      41. Jackson JE (1980) J Quality Technology 12:200–213

      42. Egan WJ, Galipo RC, Kochanowski BK, Morgan SL, Bartick EG, Ward DC, Mothershead RF II, Miller ML (2003) Anal Bioanal Chem (in press)

    • “The case of the copier that wasn’t used by Dr. Bruce Ivins to photocopy the anthrax letters” that he didn’t send.

      • DXer said

        Good point.

        To date, 807 toner samples, had been analyzed and a searchable spectral library has been created. These samples represent the 72 brands of office equipment. Do you think that there was a match and the DOJ just forgot to mention it?

        FORENSIC DISCRIMINATION OF PHOTOCOPY AND PRINTER TONERS. I. THE DEVELOPMENT OF AN INFRARED SPECTRAL LIBRARY , R.A. Merrill, E.G. Bartick, and J.H.. Taylor III, J. Analyt. and Bioanalyt. Chem. 376, 1272-1278 (2003). Click to download PDF #46. Click to download #46A.

        • DXer said

          For outside expertise, see any of the most recent journal articles on the subject.

          Ablative analysis of black and colored toners using LA-ICP-TOF-MS for the forensic discrimination of photocopy and printer toners
          M. I. Szynkowska 1 *, K. Czerski 1, T. Paryjczak 1, A. Parczewski 2 3

          *Correspondence to M. I. Szynkowska, Institute of General and Ecological Chemistry, Technical University of Lodz, Zeromskiego 116, 90-924 Lodz, Poland.

          Infrared Spectrometry in Discriminant Analysis of Laser Printer and Photocopy Toner on Questioned Documents

          Authors: Elena Gabriela Udritioiu a; Andrei A. Bunaciu b; Hassan Y. Aboul-Enein c; I. Gh. Tnase a
          Affiliations: a Department of Analytical Chemistry, Faculty of Chemistry, University of Bucharest, Sos. Panduri, Romania
          Instrumentation Science & Technology, Volume 37, Issue 2 March 2009 , pages 230 – 240

        • DXer said

          Poison Pen or Toxic Toner?

          by Lisa Wylie published: 2010-03-04

          Back in the days when all documents were prepared on typewriters, a staple ingredient in crime fiction (and even some sensational true-life crime) was the idea that you could identify the typewriter that any given document was written on, thus proving the identity of the murderer, spy, or traitor. The alignment of mechanical parts and wear on the striking heads would vary from machine to machine, and this, coupled with use of a specific ink ribbon, meant that each document typed up on a given typewriter would have a unique signature. In real life, this kind of identification was, and still is, invaluable, for example in forensic analysis of forged documents, or for establishing potential author identities from anonymous letters.

          Laser printers and photocopiers have made this field of forensics rather trickier in recent times. Since these machines use lasers and photoreceptors to create a xerographic copy, the individuality of specific machines is no longer readily identifiable, and fingerprinting the source of counterfeit documents has become much less straightforward, particularly as personal and office use of printers and copiers is now much more commonplace. Mass spectrometric methods do exist for this kind of analysis, but have the drawback of being destructive.

          Help is at hand, however. A research team at the University of Lodz, Jagellonian Univeristy, and the Institute of Forensic Research in Krakow have developed a semi-nondestructive method of identifying a document based on the printer toner used. When a laser beam is directed onto the toner, some of the toner evaporates. This sample is then sucked into the mass spectrometer for analysis. The team tested samples of toner from printed documents against samples from the cartridges used to print them, and found that they could match up the printed document with the printer to about 96% correlation. In cases where they tested the same toner on the same printer model, the correlation rose to over 99%. Coupled with the fact that each toner manufacturer uses a different chemical mixture in its products (and the mixtures are readily identifiable from the spectra), this technique shows tremendous promise for the development of a database of printer and photocopier fingerprints.

          So in that crime novel you’re still writing, don’t give up on the tried and true plot devices; it could just be that analysis of the anonymous ransom note will lead your hero(ine) to the [ spy or traitor ] after all…

        • DXer said

          Classification of Black Powder Toners on the Basis of Integrated Analytical Information Provided by Fourier Transform Infrared Spectrometry and X-Ray Fluorescence Spectrometry
          Beata M. Trzcińska, Ph.D. 1
          1 Institute of Forensic Research, Westerplatte 9, 31-033 Krakow, Poland
          Correspondence to Additional information and reprints requests:
          Beata M. Trzcińska, Ph.D.
          Institute of Forensic Research
          Westerplatte 9
          31-033 Krakow
          forensic sciences • questioned documents • toner • electrophotography • differentiation • FT-IR • XRF
          ABSTRACT: Differentiating between black powder toners used in laser printers and copiers can be challenging for forensic examiners. One hundred and sixty-two samples from 82 different types of cartridges produced by 21 manufacturers that are currently available on the Polish market were studied using Fourier transform infrared spectrometry (FT-IR) and energy-dispersive X-ray fluorescence spectrometry (XRF). The spectra obtained by these two methods were classified into 14 groups and 28 groups using FT-IR and XRF, respectively. These classifications served as a basis for the development of an analytical scheme for differentiating black powder toners. This scheme can be used with any two methods that supply different information about an examined sample. The research will show that if two samples are similar in polymer composition (e.g., FT-IR spectra), additional quantitative elemental composition from XRF analyses may provide more discrimination. It was possible to differentiate 82.5% pairs of examined samples using only FT-IR, 90.8% pairs of examined samples using only XRF, and 95.8% pairs of examined samples using these two methods.

          Obtained spectra and all the available information could be used to create/build a database. The results obtained confirm the multiplicity and compatibility of toners. Additionally, it was stated that not all the samples were original (OEM).

        • DXer said

          “Examination of the type of the toner and how it has been applied to the paper will often allow determination of the maker, and sometimes even the model, of copier used.”
          Henry C. Lee, Howard A. Harris, Physical Evidence in Forensic Science, (2000), p. 120

          What is known about the maker of copier used? What is known of the maker of the copier used in the Ft. Detrick library in 2001?

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: