Continuation Part Six: Discussion of the Amanda Knox/Raffaele Sollecito case

[Wildhorses]

Here's an abstract from the Journal of Forensic Sciences. Surely, an editor would correct the author if he/she were using the term "sensitivity" improperly.

Do you see anything having to do with false positives? Is it not implied that sensitivity refers to the dilution at which a reaction can be detected by a "blind" observer?

J Forensic Sci. 2010 Sep;55(5):1340-2. doi: 10.1111/j.1556-4029.2010.01427.x.

Sensitivity of the luminol test with blue denim.

Middlestead C, Thornton J.
Forensic Science Graduate Group, University of California, Davis, CA, USA.

Abstract

An article appearing in this journal in 2000 suggested that the sensitivity of the luminol test performed on denim fabric is usually no greater than at a 1:100 dilution of blood. This study shows that the luminol test may be unambiguously interpreted at substantially greater dilutions of blood. In this study, four different types of denim were tested by spraying a swatch of fabric with a typical formulation of the luminol reagent. Testing was conducted of dilutions of blood up to 1:1000, all of which showed distinct chemiluminescence. Diluted blood was applied to denim material in the form of a random number. A successful test was obtained only when a "blind" observer, i.e., an observer who was uninformed of the number, correctly reported the number.

© 2010 American Academy of Forensic Sciences.

 

[LondonJohn]

I think you're mistaken about the definition of sensitivity, LondonJohn. I can't find a formal definition, but I perused several papers and found nothing to back up your view.

Here, for example, is a product-brief-like thing from Blue Star.
http://www.bluestar-forensic.com/pdf/en/Watkins_Brown_luminol_BS.pdf

Under the heading "Luminol and Blue Star Sensitivity Testing," the authors walk you through their sensitivity test procedure. There’s not a word about false positives. It’s clear they consider sensitivity the weakest solution/highest dilution that produces a positive result.

Here's a paper by the Swedish National Laboratory of Forensics.
http://www.imprimus.net/PDF Files/D...ood - Interference and Effect on Analysis.pdf

The authors consider false positives a matter of selectivity -- not sensitivity. Again, there's no formal definition of sensitivity, but it's implied throughout that sensitivity is a function of blood dilution.

A snippet with that implication:
"Under laboratorial conditions CL was detected from luminol treated stains of the used hemoglobin solution (corresponding to blood) diluted up to 5•106 times. A comparably high sensitivity of the luminol test has been reported in other studies [22]."

If your definition is correct, google seems to be telling me a lot of people in the forensic world use the term wrong.

Indeed they do use the term incorrectly, or at the very least they introduce ambiguity by using the terms "sensitive" and "sensitivity" to mean something other than the proper statistical definition.

I have no problem with anyone (whether they are the kit manufacturers or online commentators) using an terms, so long as it's absolutely clear how that term is defined. However the problem (as I see it) was that there was beginning to be a muddy overlap of "sensitivity" used in the sense of "dilution ratio" and "sensitivity" used in its statistical sense of "ratio of false negatives".

The statistical sensitivity of the various blood tests is of significant importance to our case. Sensitivity, in its statistical definition, is the number of true positives one would need to test in order for one of them to generate a false negative result. For example, if you took 10,000 people that you knew had malaria (perhaps because you'd tested them with another failsafe test) and you subjected them to the test under examination, suppose that one of these people was falsely diagnosed by the test as NOT having malaria. This would imply that the sensitivity of this particular test for malaria was 1:10,000. The counterparty of sensitivity is specificity (which is even more pertinent to our case), which measures the ratio of true negatives that test as false positives (e.g. if you took 30,000 people who you already knew did not have malaria, and your test showed that one of these people tested positive for malaria, the specificity of this test for malaria would be 1:30,000).

Here's some info on the statistical definition of sensitivity (and specificity), as it applies to binary classification tests (e.g. is the test positive or negative for the tested substance):

http://www.bmj.com/content/308/6943/1552.full

http://www.cebm.net/index.aspx?o=1042



The other thing to bear in mind is that dilution ratios mean nothing in and of themselves - it's only when one specifies a total amount of substance being tested alongside that dilution ratio that it becomes relevant.

For example, 10ml of a sample diluted 1:10,000 contains the same amount of the matter being tested as 1ml or a sample diluted 1: 1,000. And it's the amount of the matter present that's the important thing.

In the case of the blood testing we're talking about, what matters is the actual number of erythrocytes (red blood cells) in the tested sample. As I said, a 2ml spot that was 1:100,000 dilution would actually contain twice as many erythrocytes as a 0.1ml spot that was 1:10,000 dilution.

 

[LondonJohn]

Here's an abstract from the Journal of Forensic Sciences. Surely, an editor would correct the author if he/she were using the term "sensitivity" improperly.

Do you see anything having to do with false positives? Is it not implied that sensitivity refers to the dilution at which a reaction can be detected by a "blind" observer?

J Forensic Sci. 2010 Sep;55(5):1340-2. doi: 10.1111/j.1556-4029.2010.01427.x.

Sensitivity of the luminol test with blue denim.

Middlestead C, Thornton J.
Forensic Science Graduate Group, University of California, Davis, CA, USA.

Abstract

An article appearing in this journal in 2000 suggested that the sensitivity of the luminol test performed on denim fabric is usually no greater than at a 1:100 dilution of blood. This study shows that the luminol test may be unambiguously interpreted at substantially greater dilutions of blood. In this study, four different types of denim were tested by spraying a swatch of fabric with a typical formulation of the luminol reagent. Testing was conducted of dilutions of blood up to 1:1000, all of which showed distinct chemiluminescence. Diluted blood was applied to denim material in the form of a random number. A successful test was obtained only when a "blind" observer, i.e., an observer who was uninformed of the number, correctly reported the number.

© 2010 American Academy of Forensic Sciences.

See my previous answer.

There are two definitions of "sensitivity". One is very precise, and is a statistical measure applied to binary classification tests. The other is a rather nebulous and imprecise definition that refers to the dilution ratio of samples being tested. My contention is that to use the term "sensitivity" in that second (and poor) definition is incorrect, imprecise, in conflict with the statistical definition, and therefore opens the door wide to confusion and misinterpretation.

I think therefore that scientists should not use the term "sensitivity" to refer to dilution ratios (or similar) - especially where they are discussing binary classification tests, where "sensitivity" already has a clear alternative definition.

And in fact, this paper extract is also lazy science. As I said before, dilution ratios only matter when they are married with absolute volumes being tested. If I have a mug full of water/blood diluted to 1:1,000,000, there will be many more red blood cells present than in a tiny droplet of water/blood diluted to 1:10,000. And ultimately it's the number of red blood cells that's important.

 

[Diocletus]

In July 2009, for example, Manuela Comodi stated in court that the total amount of DNA on the metal bra clasp was 1.4 nanograms.

That's not the amount of DNA on the clasp. That's the amount of DNA after the sample purportedly from the clasp went through a 50-cycle amplification process.

The laboratory turned out the SALs.

The SALs are crap. They are just an summary and are not the contemporaneous record of any testing. Moreover, they are grossly incomplete and are obviously erroneous in many places.

The negative TMB results were just noted in the documentation that the prosecution kindly presented to the defence, despite the defence experts had not attended the meetings they were summoned to with the laboratory staff,

Cite the law that says that the defense should have no access to any resulting lab documentation if they don't send a consultant to some testing session. You can't because such a law would be illegal and doesn't exist. This "you snooze, you loose" argument is a dumb one, and incorrect, but I realize one that ignorant guilters love.

But now you are entangling yourself in your contradictions. If you assert they offered ‘manufactured SALs’, then you should have a shred of evidence of that which you obviously don’t have, and you know that.

Easy. Check the extraction dates. Many samples are assigned extraction dates that are not possible. The SALs are frequently erroneous, and that's why the underlying documentation should have been produced, and always should be produced as a matter of course.

The Stefanoni’s report contains the templates of all tests doen on samples and items.

No it doesn't. Among other things, Stefanoni suppressed from her report any mention of dozens of egrams that are harmful to the prosecution/helpful to the defense.

Now, it happens that not always these templates and pages are correct.

You can say that again. What's really weird is how she is always "incorrect" about the most critical evidence in the case: 36b, footprints, cat's blood, etc.

Moreover in the Italian system the defence not only is not directly provided the documentation (the documentation is only deposited at the chancellery) but in order to have a copy of documentation, they also need to pay to have them (and the payment is not an irrelevant fee: at that time it was something like 70 euros (maybe more) for each support, such as CDs, there was a high fee for paper documents etc.). You may dislike this feature of the system, it’s a legitimate opinion. But it’s an opinion and here we are talking about facts.

What is this nonsense? A blank CD costs like 1 E. Defendants should never be put in the position of considering their finances before seeking disclosure of potentially exculpatory information. Under equality of arms, this preposterous and offensive Italian "custom" seems to have operated illegally in this case.

Are you contending that the prosecution has deposited of record with the court all of the EDFs, egrams, amplification records (including raw data), extraction records, controls, and blood testing information, including copies of all of the original records from which any summaries or reports were generated? I think that they clearly did not, so I don't understand the relevance of your comment about what has been given to the clerk: a document not given to the clerk by the prosecution has been illegally withheld.

I agree instead with Massei when he says there are obvious indication of a cleanup

The only cleanups that are necessary in this case are (i) Stefanoni's lab, (ii) the manifestly illogical Italian justice system, and (iii) the minds of the dirty old men/perverts who run it.

 

[RoseMontague]

My opinion on the Luminol tests is pretty simple. First the pictures. None of these is of sufficient detail or quality to determine who it might belong to with the exception of the one in Amanda's room that was found by Rinaldi to be a match to Amanda. Rinaldi is wrong on that one because the second toe is obviously not a match. He simply ignores the second toe in his report and does not even address the issue. These prints could have been made before the murder, during the murder, or after the murder. What is obvious to me is that they were not cleaned, there is no evidence of streaking that you would see if someone had attempted to clean the surface of the floor.

We don't know when these prints were made, what caused the Luminol reaction, nor do we know who made them. They are essentially worthless as evidence. We are basically arguing over blobs that have no meaning whatsoever.

 

[LondonJohn]

My opinion on the Luminol tests is pretty simple. First the pictures. None of these is of sufficient detail or quality to determine who it might belong to with the exception of the one in Amanda's room that was found by Rinaldi to be a match to Amanda. Rinaldi is wrong on that one because the second toe is obviously not a match. He simply ignores the second toe in his report and does not even address the issue. These prints could have been made before the murder, during the murder, or after the murder. What is obvious to me is that they were not cleaned, there is no evidence of streaking that you would see if someone had attempted to clean the surface of the floor.

We don't know when these prints were made, what caused the Luminol reaction, nor do we know who made them. They are essentially worthless as evidence. We are basically arguing over blobs that have no meaning whatsoever.

Agreed. My own personal view tends towards these amorphous blobs having been deposited by the "crack" forensics team as they tramped in and out of Meredith's room repeatedly on the 2nd-5th November, without bothering to change their shoe covers.

Whatever, they are worthless as probative evidence in this case.

 

[RoseMontague]

Can someone tell me in what document the negative TMB tests were finally reported? It is not in Stefi's DNA presentation ppt, nor are they listed in the test results pdf, those just show the positive Luminol tests. Interesting that these results were not reported in these two documents.

 

[Machiavelli]

No. You don't. Sensitivity is the ratio of false negatives per true positives. If a test has a sensitivity of 1:20,000, this has nothing directly to do with the "dilution" of the substance being tested. It actually means that for every 20,000 samples tested where the substance being tested is truly present (i.e. blood in our case), one will falsely test negative.

We are not makeing academic debates. I use the term sensitivity in the way it is used by literature. You can argue about "the statistical way it should be used" with the authors, not with me.
Forensic papers talk about a concentration ratio.

The almost-amusing thing here is that you don't know why you're wrong.

The most amusing thing, is whenever I prove to be right, your response will be predictable to be a condescending tone.

Unfortunately, you've made the same mistake as many, and that's to conflate Hemastix testing with TMB testing.

Again, I referring to the fact that the device used in forensic videos look like Hemastix strips. I am talking about what Stefanoni actually did. Not about what a theoretical TMB solution is. The "sample" technique was used for the knife, as far as I know; while hemastix was used on the stains.
But anyway this makes no difference, since TMB is not as specific nor as sensitive than luminol, in all the forms in which the test can be performed.

There are two types of TMB testing. There's a "quick and dirty" type, which is Hemastix. Hemastix are card strips with a patch of reagent on the end. They were developed for the analysis of blood in urine, and are still used for this purpose. However, forensic scientists have discovered that they can also be used as a quick presumptive test for blood.

Things that I already know, thanks.

There is a second, full form of TMB testing. This is the one that is recommended to be conducted on samples that show a presumptive positive under Luminol. (...)

As I said, the whole thing is pointless. Hemastix is considered the most specific, while the sample dilution could be more sensitive. Albeit, there are several ways for both using an reading the hemastix strip, methods which would modify the result. But the substance is the same. And it is a property of the substance, that of being reactive to all same substances to which Luminol reacts, and to be less sensitive.

Not the Hemastix test. The full TMB test is a two-step process, which consists of first adding the TMB reagent to the sample, then adding hydrogen peroxide. This two-step approach is critical because it allows for the screening out of virtually all other oxidants: if the sample reacts to the first step, it should be considered inconclusive and thus a negative. If, however, the sample does not react to the first test but reacts to the second, then it's highly likely that haemaglobin constituents are present.

The proces is always chemically identical, both with luminol (which in fact reacts with peroxide) and with hemastix strips (it's the identical oxidation process).
There is nothing lees scientific than your concept of "higlhy likely": it is just not true that a TMB sample reaction would be "more likely" blood than a luminol reaction.

This second test is easy to carry out in the field - although admittedly not quite as easy as rubbing a hemastix strip on the sample.

Which is what was done.

Here, for your information and education, is the difference between the simple (but low specificity) Hemastix test and the full (and high specificity)

You will preferebly use the term "education" with some of your students.

PS: Please, please try to learn the science properly before posting again. It's embarrassing otherwise.

I recall some days ago when you found "strange" that alcoholemic rate was measured by grams per litre (what a strange thing, a mass for a volume.. wasn't that what you said, mr Expert LJ ...?) believe me you have a little to learn too.

 

[Machiavelli]

They got more luminol reactions at Raffaele's than at the crime scene. They get a lot of false positives with luminol.

The reactions at Raffaele's house are blood.

 

[Dan O.]

Here's an abstract from the Journal of Forensic Sciences. Surely, an editor would correct the author if he/she were using the term "sensitivity" improperly.

Do you see anything having to do with false positives? Is it not implied that sensitivity refers to the dilution at which a reaction can be detected by a "blind" observer?

J Forensic Sci. 2010 Sep;55(5):1340-2. doi: 10.1111/j.1556-4029.2010.01427.x.

Sensitivity of the luminol test with blue denim.

Middlestead C, Thornton J.
Forensic Science Graduate Group, University of California, Davis, CA, USA.

Abstract

An article appearing in this journal in 2000 suggested that the sensitivity of the luminol test performed on denim fabric is usually no greater than at a 1:100 dilution of blood. This study shows that the luminol test may be unambiguously interpreted at substantially greater dilutions of blood. In this study, four different types of denim were tested by spraying a swatch of fabric with a typical formulation of the luminol reagent. Testing was conducted of dilutions of blood up to 1:1000, all of which showed distinct chemiluminescence. Diluted blood was applied to denim material in the form of a random number. A successful test was obtained only when a "blind" observer, i.e., an observer who was uninformed of the number, correctly reported the number.

© 2010 American Academy of Forensic Sciences.

Something to recognize is that while Luminol and TMB are both catalytic reactions, the indicator for Luminol is the light emmision which is reaction time dependent and therefore the sensitivity is related to consentration. On the other hand, the indicator for TMB is a color change which is mostely time independent limited only by how long the investigator is willing to wait for the change and concern for natural oxidation that would indicate a false positive.

 

[Machiavelli]

My opinion on the Luminol tests is pretty simple. First the pictures. None of these is of sufficient detail or quality to determine who it might belong to with the exception of the one in Amanda's room that was found by Rinaldi to be a match to Amanda. Rinaldi is wrong on that one because the second toe is obviously not a match. He simply ignores the second toe in his report and does not even address the issue.

In my opinion, you are wrong, because:
1) There is also another footprint which has a suficient quailty, at least for taking measurements and for comparative exclusion. This print was made by another individual, not the same who made the print in Amanda's room
2) The tiny stain which you consider a "small toe" is actually a drop (the size comparable to several others in the same picture), not a toe print. While an actual toe print, such as those existing in Amanda's sample footprint, would be a considerably larger stain (several times larger as those the sample print). That drop is too small to be attributed to a toe print.
3) as a consequence of point 2), the mass of information coming from the luminol print (like the compatibility with Amanda and the fact that it was found in her room) are logically too significant to be 'overturned' by a non-significant detail like a drop near the place of a toe. I piint out, that (as for what I remember of Rinaldi's testimony) the defence did not address the detail neither in Rinaldi's cross-questioning.

 

[Dan O.]

The reactions at Raffaele's house are blood.

The emanations from Machiavelli's mouth are bold.


There was no confirmation that the luminol was reacting to blood and the only DNA matched was of the two people that had been living there.

 

[Machiavelli]

The emanations from Machiavelli's mouth are bold.

There was no confirmation that the luminol was reacting to blood and the only DNA matched was of the two people that had been living there.

There were multiple items on which specific presence of blood was detected (such as a handkerchief, item 86, found in the kitchen, with Sollecito's blood); while the presence of Knox's blood was detected too, thus the luminol prints found on pants and on the door's handle, the drops on the shower and in the bathroom, can be logically linked together as having the same origin. Moreover, several of these stains had human DNA of either Knox or of Both Knox+Sollecito.
Everything indicates they are from blood.

 

[Kaosium]

Tell me instead, first when you think the the prosecution was “caught”.

Thank you for the considered response, Machiavelli. I don't have time at the moment to give it the attention it deserves as I was distracted elsewhere and now don't have the time required as I will be only able to check in sporadically for a while due to other commitments.

 

[RoseMontague]

In my opinion, you are wrong, because:
1) There is also another footprint which has a suficient quailty, at least for taking measurements and for comparative exclusion. This print was made by another individual, not the same who made the print in Amanda's room
2) The tiny stain which you consider a "small toe" is actually a drop (the size comparable to several others in the same picture), not a toe print. While an actual toe print, such as those existing in Amanda's sample footprint, would be a considerably larger stain (several times larger as those the sample print). That drop is too small to be attributed to a toe print.
3) as a consequence of point 2), the mass of information coming from the luminol print (like the compatibility with Amanda and the fact that it was found in her room) are logically too significant to be 'overturned' by a non-significant detail like a drop near the place of a toe. I piint out, that (as for what I remember of Rinaldi's testimony) the defence did not address the detail neither in Rinaldi's cross-questioning.

Looks like a toe to me. It is right where someone with a shorter toe would leave a mark and is more detailed and larger than the marks close to the heel that you are referring to. I don't have Rinaldi's testimony, but would be happy to look at it if you have it, otherwise what you are saying is meaningless. In his report, he ignores it. Torre talks about it as a toe and says to him it is pretty obvious it doesn't match Amanda's.

 

[Grinder]

The thing that makes me mad about this is it is so deceitful. While it is possible that there can be a false negative TMB test it just isn't likely.

And getting away from the literature there is logic, which escapes the prosecution.

While you can see the luminol prints, you can't see them with the naked eye. But why can we see Rudy's prints with a naked eye? If Amanda and Raffaele had spent the evening cleaning, how did they eliminate all their own visible prints and leave Rudy's visible prints? Did they levitate? If the blood was fresh and it would have been, the Luminol prints would have streaked, but they weren't streaked. So that means these prints were made at different time than the murder.

You can't ignore the fact that Stefanoni went through the standard process of performing both tests, so she KNOWS THE PROTOCOL.

Knock your socks off Grinder.

*Lerner, K. Lee, and Brenda Wilmoth Lerner, eds. (2006). The World of Forensic Science. New York: Gale Group.
Newton, Michael. (2008). The Encyclopedia of Crime Scene Investigation. New York: Info Base Publishing,
Siegel, Jay. (2006). Forensic Science: The Basics. Boca Raton, FL: CRC Press,
Websites
www.biologymad.com/cells/microscopy.htm
www.cengage.com/school/forensicscienceadv
www.fbi.gov/hq/cjisd/takingfps.html
www.forensic-medicine.info/fingerprints.html
http://inventors.about.com/od/mstartinventions/a/microscope.htm
www.microscope-microscope.org/basic/microscope-history.htm
www.nfstc.org/pdi/Subject02/pdi_s02_m02_01_a.htm
www.redwop.com/download/dfo.pdf
www.rpi.edu/dept/chem-eng/Biotech-Environ/CHROMO/chromtypes.html

I went to the first four and they don't link to anything about TMB that I could quickly find.

If you have the resources, which I'm not doubting, I would appreciate a link to the page where it is discussed and a quote similar to how Mach did his from the PDF he linked to.

 

[Machiavelli]

Something to recognize is that while Luminol and TMB are both catalytic reactions, the indicator for Luminol is the light emmision which is reaction time dependent and therefore the sensitivity is related to consentration. On the other hand, the indicator for TMB is a color change which is mostely time independent limited only by how long the investigator is willing to wait for the change and concern for natural oxidation that would indicate a false positive.

In fact, as for hemastix's manufacturer indications, they should be read within 60 seconds, after which the result should be considered negative.

The article I quoted point out that Hemastix would react faintly to immersion in the 1/100.000 dilution, but reacction occurs only after 3-4 minutes (hence, this concentration yields to a "negative" result for protocol TMB tests).

 

[Grinder]

Grinder,

Take a look at this document about the use of Luminol. In particular look at the pictures at the end of the article where cleanups took place. Then compare that to the images in the Kercher murder.

Now after looking at them both Grinder. Was there a cleanup?

There were footprints of some sort found by the luminol. It is argued that the prints that are missing couldn't have been cleaned up because there would be smear marks or the glow of bleach (although the length of time they waited seemed to be a problem for bleach IIRC).

That would seem to mean that the prints, being of only one foot, must have been left with only one foot even if they were from another time than the murder night or the cleaning marks should be there as well.

 

[Dan O.]

In fact, as for hemastix's manufacturer indications, they should be read within 60 seconds, after which the result should be considered negative.

The article I quoted point out that Hemastix would react faintly to immersion in the 1/100.000 dilution, but reacction occurs only after 3-4 minutes (hence, this concentration yields to a "negative" result for protocol TMB tests).

Now try to square that with the strong luminol reaction.

 

[RoseMontague]

The reactions at Raffaele's house are blood.

Well they were positive with Luminol, most (Reps 92-103) show no other tests run, TMB or specific. Reps 104-105 were negative both for TMB and the specific test according to the test results pdf. It may be your opinion that they are blood, but it is not confirmed in the test results.