The Dumpster, the DNA, and the Paper Towel: What Five Witnesses Put on the Record on Day 4

Day 4 of the Kosowski trial opened with a dog and closed with a photograph. In between, an FBI agent mapped two hours the defendant's own phone went dark, a print examiner rolled Tomasz Kosowski's fingerprints in the courtroom itself, a DNA analyst put a number in front of the court so large he had to explain the scale of it himself, and a blood pattern analyst spent the back half of the day describing a men's room she says shows every sign of a cleanup. Five witnesses. Five Case Story Parts. One long, forensic day that, for the first time in this trial, ended with Kosowski actually fighting back.

Court ran roughly eight and a half hours on Thursday, five witnesses across direct and cross, and it moved with none of the volume of Day 3's eight-witness marathon but considerably more friction, because for the first two-thirds of the day the state's own experts spoke largely uninterrupted, and for the last third, Kosowski turned every one of their conclusions into a fight.

I want to open this the same way I open every one of these, because it does not change from day to day: nothing below is proof of anything. This is a bench trial. Judge Joseph A. Bulone alone will decide whether Kosowski is guilty of anything, and if it comes to that, whether he lives or dies. Kosowski is presumed innocent in this sentence exactly as much as he was in the one before it. But Day 4 is worth separating from the three days that came before it, because the pattern this channel has had to describe over and over, no cross-examination, again, broke here. Not for every witness. He still let the K9 deputy step down untouched. But on the two witnesses who mattered most, the DNA analyst and the blood pattern analyst, Kosowski cross-examined at length, for the first time putting real pressure on the state's own physical evidence instead of only preserving objections for later.

For anyone coming to this coverage for the first time on Day 4, the shape of the case so far is this. Steven Cozzi was opposing counsel in a civil lawsuit Kosowski himself brought and was, by his own summary judgment deadline, about to lose. Cozzi disappeared in March 2023 and has never been found. Day 1 established the relationships and the motive. Day 2 built the surveillance trail connecting a truck to Kosowski's own neighborhood. Day 3 filled that frame in with presumptive forensic testing pulled from three locations at once, a landfill dig, and a paralytic drug traced by lot number to Kosowski's own surgery center. Day 4 is where the forensic case stopped being presumptive and started being specific: confirmed DNA matches, a rolled fingerprint, and a phone record with a very precise hole in it.

The Dog That Sat, and the Phone That Went Dark

The day opened with Deputy William Eric Simons, a 21-year veteran of the Miami-Dade Sheriff's Office and the handler of Connor, a human-remains detection dog trained to separate the scent of real human remains from everything else a search can throw at him. Simons testified Connor has passed that test before, in the field, locating a body submerged in a Davie lake and a buried torso in a Boynton Beach backyard. Days after Kosowski's arrest at the end of March 2023, Largo detectives called the pair out to a wooded lot off the Tamiami Trail near Monroe Station, an overgrown rest area with a few dumpsters scattered in the brush. Working into the wind, Connor went straight for a green dumpster, jumped up, and sat, his trained final alert for the scent of human remains. Nobody found a body inside it. Its own contents, by then, had already been hauled away to the Collier County landfill, and for what Simons remembers as several long, hot days, he and Connor worked through that landfill by hand, past needles and decaying animal carcasses, in a smell he says was bad enough that he threw away the clothes he wore that night. Connor never got another hit there, or on the one other landfill search of his career that also came up empty, in the Cayman Islands. Simons was careful on the stand to describe what a trained final alert actually means and what it does not: it identifies the presence of a scent Connor is trained to detect, not the presence of a body at that exact moment, and a landfill compacting and shifting trash daily for days at a time is, by his own account, close to the hardest environment in Florida to run a search like this one in. Kosowski did not cross-examine him.

Special Agent Kevin Corrigan of the FBI's Cellular Analysis Survey Team followed, a 12-year Bureau veteran certified in cellular analysis in 2024 who has testified as an expert roughly 17 times. He walked the court through call detail records, data sessions and timing-advance records, then plotted Kosowski's own T-Mobile phone against five locations tied to the case: a residence in Tarpon Springs, the law firm where Cozzi worked, a dumpster, a license-plate-reader camera near that dumpster, and a residence in Miami. The records trace the phone leaving the Tarpon Springs residence the morning of March 21, 2023, consistent with heading toward Cozzi's own firm. At 8:08 a.m. the records simply stop. For two hours and sixteen minutes there is no data, no call activity, nothing, until the phone reappears at 10:24 a.m., consistent with being at the law firm itself, around the same time surveillance video the state says shows Kosowski walking in. A second gap follows almost immediately: the records stop again at 11:03 a.m., this time for 59 minutes, resuming just as video shows the same vehicle driving away. Two days later, Corrigan testified, the phone makes the same trip south toward the dumpster on the Tamiami Trail, past a license-plate reader that catches Kosowski's own vehicle. Corrigan's timeline actually begins earlier than the gap itself: he plotted the phone at Kosowski's own home at 7:52 that morning, the same minute surveillance video the state says shows a vehicle tied to Kosowski pulling out of the driveway, before the records go dark sixteen minutes later. Two days after the dumpster trip, the phone heads back toward the same dumpster from Miami and turns around almost immediately, what Corrigan called a very quick turnaround. Along the same timeline, a separate license-plate-reader camera four miles east of the dumpster caught Kosowski's own vehicle at 9:36 that first night, minutes after the phone data placed it in the area. On cross-examination, representing himself, Kosowski asked Corrigan exactly one question: how long did the phone actually stop at that dumpster? Consistent with maybe a minute, the agent answered, and stepped down.

The method behind that timeline is worth a plain explanation, because it is doing real work in this case with no eyewitness behind any of it. A cell phone in regular use is constantly recorded by the carrier's own network: which tower it connects to for a call or a data session, and, with more precise timing-advance data, roughly how far the phone sits from that tower at that moment. Plotted across a full day, those connections trace a path, not a precise street address but a corridor a phone plausibly traveled. What makes a gap in that record meaningful is not the silence itself, a phone can simply have no activity for stretches of an ordinary day, but the specific shape of this one: two hours and sixteen minutes with absolutely nothing, opening the moment the state says Kosowski's own vehicle left his driveway and closing the moment surveillance video the state points to shows him arriving at the exact building where Cozzi worked.

Nicole Hadley, a Pinellas County Sheriff's Office latent print examiner with the agency since 2012 and certified by the International Association for Identification since 2015, closed the block. She testified she has personally conducted thousands of fingerprint comparisons across her career, and she had been assigned to this case under a motion to compel: ordered to report to the courtroom, roll the defendant's fingerprints herself, and compare them to whatever open latent prints remained unresolved in the investigation. She did exactly that, in front of the court, with the state moving to admit the freshly rolled prints as their own exhibit. She then walked the room through the comparison on the monitor, pointer in hand: item number 48, a latent print already in evidence, held up beside the fresh roll she had just taken from Kosowski, ridge detail worked through point by point to a stated conclusion, an identification. Kosowski's cross touched one narrow thing only: who actually lifted that print from the scene in the first place. She was not the one, she told him, and named the colleague who was. No redirect followed.

Each of these three witnesses also has a full version of their own testimony, watchable start to finish, and there is a standalone clip of just the dumpster search. Those release on the channel this weekend; until then, Part 13 above carries all three.

380 Septillion to One

The entire second block of the day belonged to one witness: Chad Summerfield, who spent 17 years as the DNA technical leader at the Pinellas County Forensic Laboratory, writing the policies the lab still runs on, before moving to Florida International University's National Forensic Science Technology Center, where he now trains crime labs nationwide. He opened by walking the court through the underlying science before he ever touched this case's own results: how a reference sample, a toothbrush, a cheek swab, gets compared against evidence collected at a scene, and how a mixture, more than one person's DNA on a single item, gets interpreted and reported using a likelihood ratio rather than a flat yes-or-no match. With no body to draw a reference sample from, investigators had brought the lab a toothbrush represented to be Cozzi's own, submitted while Cozzi was still missing, and Summerfield explained plainly why that works: skin cells shed onto bristles during ordinary use carry the same DNA a cheek swab would, which is why a missing person's own toothbrush, recovered from his own home, functions as a reliable standing-in reference sample in a case with no living or recovered donor to swab directly. Investigators also submitted a cheek swab from Michael Montgomery, Cozzi's husband, so that any male profile recovered from the case could be measured against both men at once.

Against those two references, Summerfield testified that a swab from inside the tailgate of the Toyota Tundra tied to this case returned a DNA profile approximately 380 septillion times more likely to have come from Cozzi than from an unrelated, unknown contributor, a number so large he had to walk the court through the scale of it himself, counting upward past everyday magnitudes most people would never otherwise have reason to name out loud: millions, billions, trillions, quadrillions, quintillions, sextillions, septillions.

Item by item, he went further. A two-person mixture from Kosowski's own garage floor fit both men almost evenly, once the lab had Kosowski's cheek swab to compare it against. The men's room floor drain at the law firm came back with very strong support for Cozzi's DNA, and Kosowski was excluded outright. The men's room's exterior door held a mixture, this time very strong support for both men. Of two ballistic vests tested, one came back negative for blood, and the black one came back positive, scoring at the top of the scale for Cozzi. Asked whether diluting blood with cleaning products would push those numbers lower, Summerfield said you could expect exactly that. Not everything came back that strong: a swab from a dumpster produced no usable profile, the mop tested negative for blood, and a stained paper towel, tested for both men, came back with a far lower number for Kosowski's side of the mixture, low enough that the lab's own report does not call it inconclusive but gives it a specific, standardized qualifier instead: limited support. When the state passed the witness, Kosowski told the court he had a lot of questions. Lunch came first.

Kosowski Presses His Own Number, Then an Analyst Who Never Saw the Room

What followed lunch was close to an hour of Kosowski, representing himself, cross-examining Summerfield directly, and it did not read like the rest of this trial. He went after the analyst's own deposition, where he said the analyst had found none of his DNA in the bathroom at all. He went after the lab's own change in reporting language, results once called inconclusive now reported as limited support, and pressed hard on what a low number on one paper towel actually means. Summerfield held his ground under the pressure: very strong support that both men are present in that mixture, he said, but not proof beyond a shadow of a doubt. Kosowski's closing question made the stakes explicit: would you say my DNA is on this paper towel without a shadow of a doubt? I would not say that, Summerfield answered. I would say it provides very strong support.

It is worth pausing on what limited support and very strong support actually mean as reporting categories, because the difference is doing real work in this case. A modern DNA lab does not report a mixture result as a simple match or no match. It reports a likelihood ratio, a number describing how much more likely the evidence is if a specific person contributed to it than if a random, unrelated person did, and then translates that number onto a standardized verbal scale the whole lab uses consistently, from inconclusive or limited support at the low end up through moderate, strong, very strong, and, in Summerfield's own septillions-scale testimony, essentially unequivocal support at the top. A single low number on one item, Kosowski's own paper towel result, sits at the bottom of that same scale the tailgate swab sits at the top of. Both are real, reportable results. They simply describe very different strengths of statistical support, and Summerfield's job on cross was defending the honesty of reporting a weak result as weak rather than rounding it up or excluding it outright.

The state then called Anna Cox, a former Pinellas County Sheriff's Office forensic specialist who now runs her own practice, Cox Forensic Consulting and Training, and who received the case materials in June 2023, after the scene had already been fully processed, and who built her entire analysis from photographs, body camera footage and lab reports rather than a walk-through of the room itself. She took the court through the men's room at the law firm image by image, working toward the question she said actually drives a bloodstain-pattern analysis of a room like this one: how does blood end up only in the grout lines and not on the tile tops themselves. Small stains low on the stall door and the wall, she said, were caused by an action close to the floor, not by simple dripping, a scene she called dynamic rather than static. Blood diluted by another liquid. A discolored section of wall. Officers who had reported the strong smell of a cleaning solution when they first arrived. Blood sitting in the grout lines rather than on the tile tops, which she said ruled out a simple pool and pointed instead toward wiping. Diluted blood on the wall beneath the paper towel dispenser, and diluted blood on towels sitting in the trash can directly below it. And luminol lighting up stretches of the room no one could see with the naked eye, which she said confirmed the space had been altered after whatever happened in it.

The Paper Towel Outside the Door

Cox finished her direct examination by moving outward from the stall itself. Luminol on the stall door, she testified, suggests the door stood open during part of what happened. A diluted swipe on the outside of the men's room door and a paper towel found on the floor just outside it both came back with Cozzi and Kosowski named as contributors, and she said those were the only two stains found anywhere outside the room. From there she moved to Kosowski's own pickup, identifying transfer stains in the truck bed and on the tailgate, with the tailgate stain testing to Cozzi, to luminol reacting on his garage floor with a sample naming both men, and to a ballistic vest in the trunk of his car carrying a transfer stain that also tested to Cozzi. Her conclusions were careful and specific: at least one bloodletting event involving Cozzi occurred in the stall, close to the floor, and something carrying his blood later touched the truck and the vest. She would not say what caused the injury behind any of it.

Then Kosowski began his cross, and his first target was the paper towel itself: why, he asked, did an early DNA report name only Cozzi on that item, while a later report named both men. Cox sent the question back to the DNA analyst, it was not hers to answer. A technical glitch with a thumb drive stopped the courtroom cold shortly after, and the day took its next break there.

His Longest Cross of the Trial

What came back from that break was the most sustained fight Kosowski has put up in four days of trial. He put two photographs side by side: one taken by the forensic team, timestamped 8:43 p.m., showing the paper towel on the floor outside the men's room door, and one an officer had taken hours earlier, at 12:29 that afternoon, where the towel cannot be seen at all. He pressed Cox on scene security and on exactly when that towel got there. She pointed to written reports from the first responding officers, both of which describe a bloody paper towel outside the door, and did not move from that answer. He called it evidence out of thin air. The judge told him to ask factual questions, no editorials, one of several moments across the day's back half where the court had to rein in the framing of a question rather than its substance, a small but real sign of how much more adversarial this cross ran than anything Kosowski had attempted in the trial's first three days.

He then turned to chemistry. Reading from a journal article on false positives in presumptive blood tests, he pressed Cox on whether bleach and other cleaning agents could produce the identical reaction luminol reads as blood, a claim she disputed on the specifics. He asked whether the stains could have come from a punch, a nosebleed, a gunshot, or a stabbing. She said she could not tell what caused the underlying injury, only that one had occurred. He asked, too, whether luminol itself can measure how much blood was actually present. No, she said, and a later cleanup could easily spread whatever was there to begin with, working against his own suggestion that the amount of blood in the room was necessarily large. His final question summed up her entire testimony in one sentence: a blood spatter event that includes Cozzi's blood, and someone trying to clean it up afterward. That is correct, she answered. On redirect, the state returned to the paper towel and to the literature on presumptive testing, and to what Cox had never been given at all, a medical examiner's report, because no body has ever been found. Court adjourned for the day.

All five of Day 4's witnesses also get their own whole-testimony release, and each one runs longer than the excerpt woven into the day's narrative here. None of them is on the channel yet as this piece goes up: they release Saturday and Sunday, alongside the day's standalone clips, and the five Case Story Parts above carry every one of these witnesses until then. That distinction matters more on a day like this than most: a Case Story Part gives you the moment as it fits into the day's story, cross-examination, objections and all context trimmed to what moves the narrative forward, while a whole testimony gives you the witness's full direct and cross exactly as the court heard it, gavel to gavel for that one person, with nothing summarized out. If a single beat is what brought you here, the standalone clips are the fastest way in once they release. If you want to sit with one witness's full account the way Judge Bulone did, the whole testimony is built for that.

How This Channel Handles a Name Before It's Settled

Every witness named in this piece, Deputy William Eric Simons, Kevin Corrigan, Nicole Hadley, Chad Summerfield and Anna Cox, spelled their own name for the court reporter, clearly, on the record, under oath, and every one of them cleared this case's own verification before this piece went out. That is the standard every name on this page meets, checked against the case's internal name file the same way it was checked on Day 1, Day 2 and Day 3. It is also why a handful of other people mentioned only briefly during Day 4, officers heard once, spelled more than one way, or named only in passing, do not appear here by name. Where this channel cannot confirm a spelling against the person's own sworn word or an independent, dated source, it holds the name back and describes the person by role instead. That rule does not bend because a name would make a sentence read more smoothly.

Where Day 4 Actually Leaves This

By the close of the day, the state had put a coherent forensic case on the record across five witnesses: a cadaver dog's confirmed alert at a dumpster with nothing found inside it, a two-hour hole in the defendant's own phone records that ends at the same law firm surveillance video shows him entering, a fingerprint identification made in the courtroom itself, a DNA match measured in the hundreds of septillions tying blood in the defendant's own truck to the missing man, and a blood pattern analyst's read of a men's room that she says was bled in and then cleaned. Against most of it, Kosowski still did not mount a fight in the moment. Against the two witnesses carrying the state's most technical evidence, he did.

The shift from Day 3 to Day 4 is itself worth naming directly. Day 3 built its entire forensic case out of presumptive testing, a chemical color-change reaction that tells an investigator where to look harder, never a confirmed identification, and more than one Day 3 witness said as much on the stand, unprompted. Day 4 is where that presumptive groundwork actually got converted into confirmed DNA results with specific statistical weight behind them, at the top of the scale for the truck tailgate and the men's room drain, in the middle for the mixture on the garage floor and the exterior door, and at the bottom for the one paper towel Kosowski spent an hour fighting over. That progression, screening test to statistical confirmation, is the ordinary shape of how a forensic case like this one is built over time, and Day 4 is where the state, for the first time, started closing that gap.

Real openings remain, and naming them plainly matters more on a day like this than on a quieter one. A cadaver dog's alert confirms the presence of a scent, not a body, and no body was recovered at the dumpster or in the landfill dig that followed it. Every DNA and blood pattern conclusion in this case still rests on statistical likelihood ratios and a chemical reaction to human tissue, not a body to anchor a cause or manner of death. The paper towel dispute Kosowski raised, whether it was present at the earliest documented photograph of the scene, is a real chain-of-evidence question that the state answered with written officer reports rather than a photograph placing the towel there from the start, and it remains open on this record. Kosowski's own literature on presumptive-test false positives was disputed by the state's witness but not resolved. The FBI's own cell-site timeline, for all its precision, still describes where a phone was, not who was holding it, a distinction Kosowski did not raise on cross but one that remains available to any future defense case. None of that adds up to an answer. It is simply what is still unsettled going into the next day of testimony.

The suppression fight the defense has been quietly preserving since the trial's opening days, raised again and again through objection after objection but never once argued on its merits in front of Judge Bulone, carried into Day 4 unchanged. The standing objection to residence-search evidence was invoked again on this record, and it was preserved again rather than pressed. Whether that fight ever gets argued on its own terms, in front of the judge who will decide this case, rather than banked purely for a future appeal, remains the single largest procedural question still hanging over this trial, separate from anything the forensic evidence itself proves or fails to prove.

What Day 4 actually did was put the state's forensic case at its most technical and its most numerically precise, and it did something else worth noting on its own: it is the first day this trial where Kosowski, representing himself against the death penalty, fought two of the state's own experts at length rather than letting their testimony stand unchallenged. Across Days 1 through 3, the record this channel had to describe again and again was a defendant who did not cross-examine, witness after witness stepping down with no cross at all, while Kosowski's own objections stayed largely procedural, preserved for a later argument rather than pressed in the moment. Day 4 is the first day that pattern actually broke, and it broke specifically where the state's case is hardest to argue with in plain language, statistics and chemistry, which is itself worth noticing. A defendant with no formal legal training chose to fight the two witnesses whose testimony is built on numbers and reactions rather than eyewitness memory, going after the language of the reports themselves, inconclusive versus limited support, a shadow of a doubt versus very strong support, rather than trying to contest whether the witnesses were telling the truth about what they saw.

It is also worth being honest about what that cross-examination did and did not accomplish. Kosowski got Summerfield to restate, more than once, that very strong support is not certainty, a true and unremarkable statement about how forensic DNA statistics work that the analyst never disputed in the first place. He got Cox to concede she cannot identify a cause of injury from bloodstain pattern analysis alone, which she had already said on direct. Neither concession moved the state's central claims: that Cozzi's blood is in that bathroom, in that truck, and on that vest, and that someone worked to clean part of it up afterward. The paper towel timing question is the one place Day 4 left something genuinely unresolved that was not already conceded, a real gap between a photograph that does not show the item and a written report that says it was there, and it is worth watching whether the defense returns to it once its own case, if it puts one on, actually begins.

Court resumes Friday at 10:15, the judge's own words at adjournment, "1015 tomorrow is what you see," confirmed a moment later with a plain "Right, 1015," closing out a day that, more than any before it, showed the defendant meeting the state's own case head on.

For the whole day, gavel to gavel, with every recess cut out and every witness in order, watch the full edition below.