This is from the randi forum, by user Kaosium:
What the TMB test does, along with common sense forensic methods (a perfectly square 'hit' is highly likely to be where a lamp's metal base sat for years, causing some of the metal to adhere to the floor as opposed to it being blood which just happened to 'pool' in a perfect square) is winnow out the non-blood hits, as TMB is more selective than luminol which lights up everything. It doesn't matter if it's 250 from that one study, or less with Hemaglow or Bluestar so they can sell more of their luminol-like product.
If ILE is using the TMB test it suggests they know this too ( ) as does the FBI which does the same thing. Spray down the scene with luminol, look for the ones whose shape and position suggest they might be blood, and then apply the TMB test which will help verify it. Then they take it to the lab and someone looks at it under the scope and confirms it.
In this case they tried to say they sprayed the luminol, and noted a distinctive CL pattern generally only found with blood and stuff that contains about eight certain metal and vegetable components, (which is a lot more than eight total things if you think about it!) and then when the TMB test was negative they stopped, it not being blood as a negative is a (virtual) absolute. Blood was not 'proven'--period. However, they later didn't reveal the negative TMB test and Comodi tried to pretend it was still possible it was blood, 'or it could be turnip juice, you decide.' This is known as being deceptive in my view, though it's possible being a lawyer who didn't know better that someone fed her that line because it sure makes it sound like it's more possible it was blood, doesn't it?
Now there is a way you can get a positive hit with luminol and a negative one with TMB, and that's if the dilution of the blood is below the 1:1M level, as luminol can detect at extremity blood down to concentrations as low as 1:5M. Thus perhaps it was diluted so much so that TMB couldn't pick it up but luminol could. Maybe it could still be blood, you might be thinking! Nope, because if it is diluted to that extreme you don't get that distinctive pattern, which should be common sense, however it has to do with the reageants in the luminol-based product having only a tiny fraction of things to catalyze with to give off that cool glow.
So if Stefanoni saw that distinctive pattern then the concentration of the blood was undiluted enough that the TMB negative proves it wasn't blood.