Thursday, June 9, 2016

Proposed Uniform Language for Forensic Serology

The second document listed in the announcement of "Uniform Language" (UL) documents for forensic science is the “Proposed Uniform Language for Testimony and Reports For the Forensic Examination of Serology.” It pertains to the detection of sera, not the examination of serology, but you get the idea. The task is one of classification: Is this a blood stain? Is it a semen stain?

The approved statements (with emphasis added) are
• blood or semen may be present on an item of evidence [using "the appropriate presumptive testing procedure(s)"]
• blood or semen was identified on an item of evidence [using "the appropriate confirmatory procedure"].
In terms of accuracy, the difference between "presumptive" and "confirmatory" is that presumptive tests are less specific — more substances will be test positive with a presumptive test. A supporting document explains that confirmatory testing can include either a highly specific test — or just a second, presumptive test, if based on “different chemical principles” and “not subject to the same limitations.” (Or maybe a second and a third screening test are required; it is hard to tell from the document.)

Although the examiner is not allowed to say that a positive result is "infallible," that is only because an examiner might goof in following a protocol that is perfectly discriminating. The confirmatory procedure “identif[ies] blood or semen to the exclusion of all other substances.” (P. 5.)

The documents display little sympathy for probabilistic thinking. Apparently, examiners may not give an error probability for an "identification," for "the analytical processes and procedures ... do not have a calculable error rate due to the unpredictability of human error." This observation motivates the proscription of statements of a "zero error rate," but by this logic, the probability of commercial pilot error, of a mishap in a nuclear power plant, or of medication errors in ICUs could not be calculated.

Aside from this allegedly unquantifiable risk of strictly "human error," identifications are subject to no uncertainty:
• when crystals are formed under the proper testing conditions, there is no doubt of the nature of the stain tested ... no false positives.
• [if] the Takayama test ... is positive, blood is conclusively present.
Despite this certitude in the supporting document, the UL document itself cautions that
An examiner may not state or imply that a level of numerical certainty is calculated to support the identification of blood or semen.
The claims of "no doubt" and "conclusively" positive results can be reconciled with the prohibition on what can be said or reported; after all, that the number corresponding to certainty is not "calculated" is literally true. But if the examiner can report that (1) the testing conditions were proper, (2) a positive result means that "blood is conclusively present," then why cannot the examiner state that (3) "my tests demonstrate that the stain surely contains blood — the probability is 1 — unless I have done something wrong"? An argument might be that expressing conditional certainty as the number 1 is not necessary and might unduly impress readers of reports or legal factfinders.

Closely related postings


  1. Thank you for posting this and the the supporting document. I have several concerns. One, it is difficult for me to see how soybean root (for example) would not test positive by the Takayama test: leghemoglobin has a heme prosthetic group, just as hemoglobin does; the only doubt revolves around the concentration of this protein. Two, this document does not discuss modern confirmatory methods of testing for blood, such as HemaTrace and the RSID test for glycophorin. The reported sensitivities of HemaTrace vary widely, but some of the values compare with the presumptive tests. Three, it is odd that saliva is not mentioned at all.

  2. "or just a second, presumptive test, if based on “different chemical principles” and “not subject to the same limitations.” (Or maybe a second and a third screening test are required; it is hard to tell from the document.)"
    One might try to argue that if one follows a positive luminol result by a positive tetramethylbenzidine (TMB) result, that this would be a confirmatory test for blood. Luminol is a chemiluminescent test, and TMB is a color-change test. And it is common practice to follow up a luminol or fluorescein test with TMB or a Kastle-Meyer (K-M) test. However, the conclusion that a positive result with a TMB or K-M confirms the existence of blood would be a fallacious argument. All four of these tests are really tests for peroxidase activity and give some of the same false positives. A great deal stands or falls on what constitutes a different chemical principle. It is my understanding that some labs have stopped performing true confirmatory tests, instead moving from the color-change tests on to DNA profiling. This is a step backward.