Luminol reacts with many things besides fruit juice, such as rust and other metal ions. It also reacts with animal blood. Sara Gino testified (p. 258, Massei Report) that about 50% of the luminol-positive areas were negative by TMB in her experience. It is highly doubtful that this is only because of the supposedly lower limit of detection of luminol over TMB. Using literature references, I
compiled a list of these limits of detection on my blog, and the difference between the two chemicals is small, relative to the large range where both are positive. Of perhaps greater importance is the fact that modern confirmatory tests for blood also have a very low limit of detection, at least equal to that of luminol. Confirmatory tests can distinguish between blood versus chemicals that can produce false positives, and they can also distinguish between primate and non-primate blood.
There are some jurisdictions (including Arkansas and Connecticut) which have decided that a positive luminol result by itself has no probative value. Other jurisdictions (Texas and Kansas) allow the evidence, but the fact that it is only presumptive affects its
weight. Neither of these two positions resembles Comodi's viewpoint, as reported via
Andrea Vogt. "[Comodi] defended the work of police biologist Patrizia Stefanoni and other police forensic investigators and appealed to jurors to use common sense when considering the reliability of defense consultants' testimony.
'At the scene of the crime there is a footprint made in blood on the bathmat and Knox and Sollecito's footprints made in blood on the floor,' Comodi said. 'and these were supposedly made at some different time because they stepped in bleach or rust or fruit juice? It's up to you to decide.'" My common sense tells me to run a confirmatory test; these kinds of situations are why people invested the time to improve them.