Fiona,
I only have time for one more lengthy post today. The nine scientists have seen some of the data, and they show an electropherogram in their petition. In comment 572 you wrote, “A retest is certainly desirable. But one clear and unambiguous result is one clear and unambiguous result.” With due respect, I think you are making an overstatement about the data, as I will try to explain.
I quote from the figure legend of the petition: “Figure 1. The low copy DNA electropherogram of the knife described above. Almost all peaks are below 75 RFUs, and most are below 50 RFUs.” This issue is also discussed in the main body of their petititon. It is important because the lower the number of RFUs (relative fluorescence units), the lower the signal-to-noise ratio. When this ratio gets lower and lower, it becomes more difficult to distinguish what is real from what is random (the noise).
This is where the number of steps of PCR becomes important; each step of PCR amplifies the amount of DNA by a factor of two. Stefanoni’s team did not do as many cycles as is typical of LCN analysis, therefore, a low S/N ratio is not surprising. I know that you do not trust this site (
http://www.sciencespheres.com/2009/10/lcn-dna-profiling-part-ii-watch-where.html), but for others who might:
There was so little DNA present that the instrument indicated no DNA until Stefanoni overrode the machine limits. This amplification increase was not achieved by the PCR technique. Once the sample has been chopped up and subjected to electrophoresis, it is too late for that. The increase was performed by other methods, such as lowering the threshold level, or simply changing the display scale until the minute fluorescence peaks were visible.
Matthew provided a link (
http://www.thedailybeast.com/blogs-and-stories/2009-09-14/no-mistrial-for-knox/?cid=bsa:moreauthor2) which is consistent with (but does not demand) the notion that the S/N ratio on the knife was too low:
The key evidence that the prosecution withheld pertains to test results on a knife found in Sollecito’s apartment with Knox’s DNA on the handle. Forensic scientist Patrizia Stefanoni, who testified as a prosecution witness last spring, wrote “too low” in English on initial results, assumed to mean that the samples of Kercher’s DNA on the alleged murder weapon—along with Knox’s DNA—did not meet the normal standards. Writing “too low” in English suggests that she was copying a reading directly from the machine, while she was continuing to test the sample. The implication, according to the defense, is that Stefanoni then had to amplify the tiny sample found on the blade beyond the protocol to find a match to Kercher’s DNA. Why she did that, and who may have ordered her to do so, are troubling questions. The knife evidence has been heavily contested because the sample on the blade was so small it could not be double-tested and because Stefanoni completed the one and only test without defense experts in the laboratory.
The day was not a complete loss for the defense. Their forensic expert, Adriano Taglibracci, testified that the standard protocol in DNA examination requires double testing to prove a match and that Stefanoni made what Taglibracci called intentional errors to mislead the jury. “This sample was so small that there was not enough to double-test it,” he told the court. “There are contradictions between what was recorded in the lab report and the court testimony signed by the same person.”
One final, more general point. It seems to me that both those who accept and those who doubt the DNA evidence should agree that the .fsa files ought to be released, so as to allow independent analysis, as discussed upthread. To date I have not heard this call from those who accept the DNA evidence.
Chris
