The towel, found hanging next to the bathroom sink, was stained with a small reddish smudge.

In a house splashed with blood, it seemed to fit the scene. A man had been brutally stabbed approximately 27 times early the morning of Dec. 2, 1985. The assailants tracked his blood all over the house in New Milford, Conn. They got it in his sock drawer and all over his cigar box as they rummaged through his belongings.

But when police focused on two teenage suspects, they never found a speck of the victim’s blood on any of their belongings, not even in the stolen Buick police believed the teens used as a getaway car. Police never found any forensic evidence connecting the two teenagers to the murder despite such a messy scene.

What could possibly explain that? Prosecutors offered only one explanation, from just one person: Henry C. Lee, then the Connecticut State Police’s forensic laboratory director and now a high-profile criminologist who has worked on cases all over the world.

The state’s star witness in 1989, Lee testified that there was blood on the bathroom towel. And in turn, that detail gave prosecutors the building blocks to suggest the suspects cleaned themselves off before darting from the scene, helping to explain why they still could have murdered 65-year-old Everett Carr without tracking blood into their car.

But there was just one problem with that theory: The stain on the towel was not blood. It hadn’t even been tested in the laboratory.

Now, 30 years after Ralph Birch and Shawn Henning were convicted of murdering Carr, the Connecticut Supreme Court has vacated their convictions and ordered new trials, faulting prosecutors for failing to correct Lee’s testimony. If jurors had known Lee was wrong, the state Supreme Court reasoned, they could have questioned his entire testimony — and “if that had occurred, the state’s entire case against [Birch and Henning] could very well have collapsed.”

The Friday ruling is a sharp rebuke of the well-known forensic scientist’s work, prompting him to call a news conference Monday to mount a defense and assert he did nothing wrong. Lee spoke to reporters inside the University of New Haven’s Henry C. Lee Institute of Forensic Science, which he founded. Over the years, Lee has testified as an expert for the defense in the trials of O.J. Simpson, Phil Spector and fiction writer Michael Peterson. He has assisted in the forensic investigations of the death of JonBenét Ramsey and the suicide of assistant White House counsel Vincent Foster. But this case, Lee said, marked the first time that he ever felt the need to publicly “protect my reputation."

“In my 57-year career I have investigated over 8,000 cases and never, ever was accused of any wrongdoing or for testifying intentionally wrong,” Lee said at the Monday news conference. (A judge found in 2007 that he hid evidence from prosecutors in the Spector case, which Lee denied.) “This is the first case that I have to defend myself.”

In 1985, police in New Milford started investigating Birch and Henning because they were on the hunt for local burglars, who they suspected may be responsible for the murder, according to the ruling. One thief suggested police take a look at Birch and Henning.

The 18- and 17-year-old didn’t have the best alibi: They were out stealing a car the night of the murder.

At first, they tried to lie about that, which increased suspicion that the teens were covering up the murder, according to the ruling. Before long, they showed police where they hid the stolen Buick Regal — tucked in the woods near a reservoir, coated with dirt and stuffed with food and clothing, pillows and blankets. The state Supreme Court noted that it was “evident that it had not been cleaned,” yet still no blood or stolen items could be found.

Lee started assisting in the investigation the day after the murder, he said Monday. He arrived on scene to perform field tests — including, he claims, on the stained towel — and to analyze the blood on the walls.

The technique is known as bloodstain-pattern analysis, a method Lee used to testify for the defense in the high-profile case of Peterson, who was convicted of bludgeoning his wife and pushing her down the stairs in 2001. Lee’s theory on how the blood splattered the walls at the bottom of the steps was used to support the defense’s theory that Peterson’s wife slipped and fell by accident. Lee also looked at bloodstains in the Simpson case, providing a basis for the defense’s assertion that the blood found on a pair of socks in the former football player’s home might have been planted there.

But bloodstain-pattern analysis has also been called into question over the past decade. In 2009, the National Academy of Sciences questioned whether the technique could be trusted at all, finding “the uncertainties associated with bloodstain-pattern analysis are enormous.” More recently, the technique has been the subject of investigations by ProPublica and the New York Times Magazine, which questioned its accuracy in numerous highlighted cases.

But in the case against Birch and Henning, the bloodstain-pattern analysis from Lee was crucial to the prosecution, the court found.

Lee had a theory: He believed it was possible the suspects didn’t get much blood on themselves because the blood sprayed away from them during the killing, according to the ruling.

Lee’s idea would prove highly important at Birch and Henning’s trials almost four years later.

“To be sure,” Justice Richard Palmer wrote, “the prosecutor’s greatest challenge at trial was to explain how it was possible for two teenagers to have stabbed the victim twenty-seven times in the confines of a narrow hallway, severed his jugular vein, struck him over the head several times, tracked blood all over the house, and yet somehow managed not to leave any trace evidence in their getaway vehicle . . . or elsewhere.”

To “answer this question,” Palmer wrote, the state turned to Lee’s blood-spatter expertise — and to his testimony about the smudge of blood on the towel.

Lee insisted Monday that his preliminary field test on the towel stain indicated it was “maybe” blood, or “consistent with blood,” but that he never said it was “definitely” blood or even that it was human blood. Field tests, Lee said, are not definitive. Evidence must be formally tested in the laboratory, but as he now concedes, that was never done with the towel in this case.

His testimony in 1989, however, showed that he went further than just saying the stain was possibly blood. “This towel had a reddish smear, very light smear,” Lee said then. “Subsequently, that smear was identified to be blood.”

The state Supreme Court does not accuse Lee of intentionally giving false testimony, but says his testimony was incorrect nonetheless, saying it is “inarguable that Lee, as the representative of the state police forensic laboratory, should have known that the bathroom towel had not been tested for blood.”

Birch had served more than 30 years of a 55-year prison sentence at the time of the Supreme Court’s ruling. Henning was sentenced to 50 years but released on probation last year, the Hartford Courant reported.

Henning’s attorneys faulted Lee in a statement to the newspaper, saying Henning was wrongfully convicted thanks to “false and misleading testimony."

“It has taken 30 years to correct this injustice and we are very pleased with the court”s thoughtful decision,” the attorneys, Craig Raabe and James Cousins, said in the statement.

David Shepack, the state attorney in Litchfield, Conn., didn’t respond to a message about whether the state will seek to retry Birch and Henning.