Marsh opened her defense case on a Monday morning with the methodical confidence of someone who had been waiting six weeks to speak.
She'd spent Scott's prosecution case sitting at her table and watching the jury. Not taking notes — watching. Building a map of which jurors were tracking the emotional narrative and which ones were tracking the technical evidence, which arguments had moved them and which had washed over them. Eleanor Marsh had been doing pharmaceutical defense for thirty years. She knew how to use six weeks of opposing case to build her own.
Her first three witnesses were expert biostatisticians. All three attacked Dr. Chen's methodology from different angles — the backup file's chain of custody, the statistical comparison methodology, the interpretation of adverse event threshold changes under evolving FDA guidance. None of the three attacks were strong enough to undermine Chen's testimony individually, but Marsh wasn't trying to win on any one of them. She was creating a texture of technical uncertainty. By the end of day two of her defense case, the jury had heard six different experts say "interpretation" and "ambiguity" and "evolving standards" a combined forty-three times. I was counting.
[ Argument Crusher: Marsh defense strategy — expert bombardment. Goal: create impression of scientific disagreement rather than fraud. Effectiveness: moderate. Two jurors showing recalibration toward uncertainty. Core problem for plaintiff: jurors must believe it was deliberate, not incompetent. Defense is making incompetence look plausible. ]
Day four: the vendor contract.
I watched Marsh walk it through her software vendor witness — a representative from the company that had contracted with Meridian in March 2011. He was professional, technically competent, and entirely sincere in his testimony that his company had performed scheduled maintenance work on CDAT during the relevant window.
"Can you describe the nature of the work performed?" Marsh asked.
"Processing optimization. We reduced the data compilation time, improved server response, updated the reporting interface. Standard maintenance work for a clinical trial management system of this size and complexity."
"Was this work reflected in the contract?"
"Yes. The statement of work is detailed in Section 4 of the vendor agreement." He gestured toward the exhibit. "Processing speed benchmarks, server load specifications, interface update scope. All of it."
"And was this work performed during the window shown in the version history as containing CDAT modifications?"
"Yes. Our work log shows we accessed the system on March 14th, 2011 and the following two days."
Marsh had the satisfaction of someone placing the final piece of a structure she'd been building for weeks.
"Thank you," she said. "No further questions."
The jury had the vendor contract. The vendor had real access. The vendor performed real work on the same day as the modification in Exhibit 47. Two jurors who had been plaintiff-leaning since Dr. Rowe's testimony shifted in their seats.
[ Win Rate Calculator: Vendor contract impact. Plausible alternative explanation introduced. Jury model update: two jurors moved from plaintiff-favorable to uncertain. Current lean: 6-6. Probability: 58% → 51%. ]
The trial had been at sixty-four percent four days ago.
My team spent eight hours on the vendor contract the night it was admitted.
Marcus pulled the full agreement — not just the excerpt Marsh had shown in court, but the complete sixty-seven-page document that had been produced in discovery. Sarah ran the relevant technical terms against the CDAT specification documents that had also been produced.
I read Section 4 of the statement of work at eleven PM with a cup of coffee that had gone cold sometime around nine.
The maintenance work was genuine. Processing speed benchmarks: real. Server load specifications: real. Interface update scope: real.
Classification threshold parameters: not mentioned. Not once, in sixty-seven pages.
The modification that appeared in Exhibit 47 — the change to the adverse event severity classification boundary — was not in the vendor's scope of work. The vendor had performed real work on the same day. Different work. Their access to the system was documented. Their specific changes were documented, in their own work logs, which had been produced in discovery and which listed every parameter they touched.
The classification threshold was not among them.
I wrote the cross-examination sequence in longhand, the way I did when I needed the logic to be linear and complete before I trusted it to examination-room improvisation.
"Mr. Jameson, the maintenance work your company performed — was it reflected in your own work log?"
"Yes."
"Can you tell the jury what CDAT parameters are listed in your work log as modified during the March 14th window?"
He'd have to read the list. The list wouldn't include classification thresholds.
"And looking at Plaintiff's Exhibit 47 — the version history showing the adverse event classification modification — does the parameter changed in that entry appear in your work log?"
Silence would do the rest.
I went to bed at one AM. My back hurt from the conference room chairs. My coffee consumption over the past week had reached levels Donna had specifically flagged as outside the acceptable zone, which I'd acknowledged and not adjusted.
The cross-examination would work.
Gerald Wethering, Meridian's CEO, took the stand in week five of the defense case.
He was sixty-three, silver-haired, with the bearing of a man who had run a pharmaceutical company for eighteen years and understood exactly how to present himself to a jury. He was not going to be aggressive or dismissive. He was going to be measured, thoughtful, and human — the CEO of a company that employed twelve thousand people who were trying to help patients with cardiovascular disease.
He was also going to establish that the executive team had followed proper procedures regarding Dr. Rowe's concerns, that the data review had been conducted through legitimate internal channels, and that the software modification was — in his understanding — part of the vendor maintenance work.
Marsh built his direct examination with the same methodical structure she brought to everything. By the time she sat down, Wethering had established: good intentions, good process, good faith reliance on his scientific team's judgment.
I let Marcus handle the morning's other cross-examinations. I had my notes. I was conserving focus for the afternoon.
Wethering's cross began at 2:15 PM.
"Mr. Wethering, the executive review meeting on March 14th, 2011 — you were present?"
"Yes."
"That meeting's agenda included discussion of the Phase III trial data for Corvalox?"
"Among other items, yes."
"I'd like to direct your attention to Plaintiff's Exhibit 63 — an email dated March 7th, 2011, from Dr. Hendricks to the executive team."
I waited for the document to appear on the jury screens and on Wethering's.
"Is that email in your inbox?"
He read it carefully. Not performing carefulness — he was actually reading it, which told me he'd forgotten the specific email in the intervening six years, which was useful.
"Yes. I received this email."
"Dr. Hendricks is summarizing a concern raised by Dr. Rowe about the adverse event classification data for Corvalox. The email indicates that the concern involves 'statistically significant deviation in severity level 2 cardiac event frequency between Q4 2010 and Q1 2011 review periods.'"
"Yes, I can read the email."
"Thank you. Seven days later — March 14th — the executive review meeting took place."
"That's correct."
"And on March 14th, the CDAT software parameter for adverse event classification was modified, changing the threshold for which cardiac events would be categorized as drug-related."
"I'm not a software engineer, Mr. Roden."
"You don't need to be to read a calendar." I kept my voice even. Not aggressive — conversational. "The email about the statistical deviation arrived on March 7th. The classification threshold changed on March 14th. The FDA application was submitted on March 20th." I stepped back from the documents. "I don't have any further questions about the timeline."
I sat down.
Marsh was on her feet for redirect.
"Mr. Wethering, does the email from Dr. Hendricks indicate that the concern required executive action?"
"No. It indicates the concern had been reviewed and assessed at the scientific level."
"And would it be unusual for a software maintenance window to coincide with an executive review period?"
"Not at all. These things run on parallel tracks."
But the jury had the dates. March 7th: CEO reads a concern about the exact category of data that was subsequently modified. March 14th: modification occurs. March 20th: FDA application filed.
Parallel tracks could coincide by accident.
Or they could coincide by design.
The jury had to decide which.
That evening, the team at Zane & Roden.
The war room had been home for six weeks. Sarah looked like she'd been sleeping on a pull-out couch, which she had been, twice, when late-night preparation ran into early morning filing deadlines. Marcus had lost weight he hadn't had to spare. I'd been running on the specific adrenaline of a trial that mattered, which was its own kind of fuel, and was going to crash spectacularly the day after the verdict.
"Marsh's closing," Sarah said. "What's it going to look like?"
"She's going to argue interpretation versus manipulation." I looked at the timeline on the war room wall. The months of evidence, the witnesses, the cross-examinations, all of it assembled. "She'll say our experts reached one conclusion and her experts reached another. She'll say the vendor contract provides an innocent explanation for the March 14th modification. She'll humanize Meridian — twelve thousand employees, a company genuinely trying to help patients."
"And that's good enough for a defense?"
"It's good enough for reasonable doubt if the jury doesn't make the vendor contract distinction." I looked at the wall. "The closing argument is the last chance to make sure they make it."
[ Win Rate Calculator: Trial assessment post-defense case. Vendor contract partially effective — introduced plausible alternative. CEO cross-examination: planted timeline question without answer. Current probability: 61%. Margin of error: wider than usual due to jury uncertainty variables. ]
Marcus looked at the number.
"Sixty-one is fine," he said. "It was forty-four when she walked in with the banker's box."
"It was," I said.
That was the job. From forty-four to sixty-one, brick by brick.
Now it came down to twenty minutes in front of twelve people.
A LOT More chapters reign FREE upon unwrittenrealm.com.
• Normal | $8 — 15 chapters ahead
• Advanced | $11 — 20 chapters ahead
• Overlord | $19 — 25 chapters ahead, plus request a refinement (plot, character, system) on any of my novels
patreon.com/TheFinex5
▰▰▰▰▰▰▰▰▰▰▰▰▰▰▰▰▰▰▰▰▰▰
― DECREE ―
More chapters reign FREE upon unwrittenrealm.com.
The throne acknowledges.
▰▰▰▰▰▰▰▰▰▰▰▰▰▰▰▰▰▰▰▰▰▰