How bad were Elizabeth Holmes’ emails at 1am?

Today we had the first direct link between the problems in the labs of Theranos and Elizabeth Holmes. The former manager of testing systems, Surekha Gangakhedkar, was preparing blood tests for use on patients. The system was unreliable; She described herself as “stressed and unhappy and worried about the way the launch was going,” Gangakhedkar stopped.

Gangakhedkar met Holmes to explain why she resigned: She didn’t think the Edisons were good enough for patients to use. Holmes told her that Theranos “promised to deliver to the customers and didn’t have much choice” except to go ahead with the launch, Gangakhedkar testified.

Holmes, who is on trial for wire fraud and conspiracy to commit wire fraud, had not been directly linked to laboratory flaws in the testimony of the two previous witnesses. But Holmes was hands-on when it came to the R&D lab, even emailing Gangakhedkar at 1 a.m. to ask how the validation tests were going.

When Gangakhedkar quit, she printed out email correspondence and kept it because “I was worried about the launch,” she said. “I was afraid it wouldn’t go well, and I was also afraid that I would be blamed.” This was a direct violation of the nondisclosure agreement she had signed.

Gangakhedkar’s account confirmed that of whistleblower Erika Cheung. Earlier this week, Cheung testified that secrecy was the norm at Theranos and that she was even discouraged from listing the company on her LinkedIn page. “There was generally guidance not to share” outside the immediate group, Gangakhedkar said.

She found that frustrating. The secrecy meant she couldn’t communicate directly with the hardware team about the issues she was experiencing with the machines. The instructions to keep things secret came from Holmes or her co-defendant Sunny Balwani, who is being tried separately.

We saw quite a lot of emails from Holmes or her in the chain during Gangakhedkar’s direct testimony. Holmes gave specific instructions to Gangakhedkar’s team on work priorities and checked in to see how the tests she wanted validated were doing. One of those emails came at 1:14 a.m. on a Saturday.

It seems late hours were the norm; another email from Balwani, with Holmes in cc’d, suggested that Gangakhedkar’s team wasn’t working hard enough. “Note: The software team was here until 3:07 a.m. — and is here already at 10 a.m.,” he wrote. He then reprimanded Gangakhedkar’s team for not using the machines in the evening.

The pressure and secrecy made Gangakhedkar uncomfortable. The Edison 3.0 and 3.5 devices were not reliable, she said. She went on vacation in August 2013, and when she returned, she found—to her surprise—that those machines would be used to test patient samples.

Another surprise was that they would modify others’ machines to use the smaller monsters Theranos had collected. Holmes pressured Gangakhedkar to approve tests so they could move to the clinical lab.

She stopped taking it in early September. And another member of her team resigned on his own the same day; after accepting the team member’s resignation, Gangakhedkar told her she was also quitting.

Both Cheung, who testified today, and Gangakhedkar said there were significant quality control issues with the Edison devices. Every time a device dropped the QC, it had to be recalibrated, a process that could take more than 12 hours if all went well. “We had people sleeping in the car because it was taking too long,” Cheung said. “Every few days, you ran these monsters over and over.”

Later in the day, Gangakhedkar said one of the delays in validating blood tests was that so few Edison machines were available. There was “a shortage,” she said, because they had to be recalibrated so often.

Among the emails Gangakhedkar printed was one from Theranos VP Daniel Young. Gangakhedkar emailed Holmes to say that data was missing from the server about testing on a version of the Edison devices. In response, Young wrote that the wrong cartridges had been used in those runs and that he recommended repeating them.

That was not true, Gangakhedkar said. And five days after the 1 a.m. email she received from Holmes, she tendered her resignation. “It was difficult because the last eight years I worked at Theranos I felt I was going to make a difference,” Gangakhedkar said. It was a “huge disappointment” that all her hard work would be lost, she said.

But the message she got was, “We’re launching anyway,” she testified. Gangakhedkar, who has immunity from prosecution for her testimony, will return to the stands next week.