On a Saturday evening in late September, about three weeks before the Journal published my first story, an email arrived in the in-box of Gary Yamamoto, the veteran CMS field inspector who had dropped in unannounced at the old Facebook building in 2012 and lectured Sunny Balwani about lab regulations. Under the subject line "CMS Complaint: Theranos Inc.," it began:

I've been nervous to send or even write this letter. Theranos takes confidentiality and secrecy to an extreme level that has always made me scared to say anything… I'm ashamed in myself for not filing this complaint sooner.

The email was from Erika Cheung and it contained a series of allegations, ranging from scientific misconduct to sloppy lab practices. It also said that Theranos's proprietary devices were unreliable, that the company cheated on proficiency testing, and that it had misled the state inspector who surveyed its lab in late 2013. Erika closed the email by saying that she'd resigned from the company because she couldn't live with herself knowing that she could "potentially devastate someones [sic] life by giving them a false and deceiving result."

Dear Gary,

After a while, Sunny Balwani, Daniel Young, Heather King, and Boies Schiller's Meredith Dearborn arrived. They took the two CMS inspectors to a conference room and insisted on giving them a PowerPoint presentation. Although it felt like a diversion tactic, Yamamoto and Bennett politely sat through it. As soon as it was over, they asked for a tour of the lab.

Yamamoto and his superiors at CMS took the complaint so seriously that the agency launched a surprise inspection of Theranos's laboratory less than three days later. On the morning of Tuesday, September 22, Yamamoto and another field inspector from CMS's San Francisco regional office named Sarah Bennett arrived at the Newark facility and explained that they were there to survey the lab. Men in dark suits wearing earpieces denied them entry and told them to wait in a small reception room.

As they headed out of the conference room, they were escorted by more men in dark suits with fingers pressed to their ears. King and Dearborn followed close behind, holding laptops and taking notes. When they got to the lab rooms, they noticed that their doors were equipped with fingerprint scanners and that there was a buzzing sound when you entered. It reminded Yamamoto of the door buzzers in liquor stores.

When they returned, the Journal investigation had been published, ratcheting up pressure on the agency to take action. Yamamoto noticed that security was a bit lighter and that Holmes was there to greet them. Balwani and King were also present again, along with a different set of outside attorneys and some lab consultants. The inspectors split up: Yamamoto roamed through the lab rooms and peppered the lab's personnel with questions, followed everywhere he went by Balwani, while Bennett set up shop in a conference room where King and the other lawyers kept close tabs on her.

Yamamoto and Bennett had initially set aside two days for the inspection, but they found so many problems and Theranos was missing so much basic lab documentation that they concluded they would have to return. Balwani asked for a two-month reprieve. He claimed that the company's new fiscal year was about to start and that it was in the midst of raising new funding. They agreed to come back in mid-November.

This time, they stayed four days. At one point, Bennett asked to conduct a confidential interview with one of the lab associates who worked in Normandy and had direct experience with the Edisons. She was made to wait in a windowless room for a long time until a young woman finally appeared. As soon as she sat down, the woman asked for an attorney. She looked coached and afraid.

Erika Cheung and I had remained in sporadic contact after her parking lot scare in late June, but I didn't know she'd gathered the courage to reach out to a federal regulator. When I first heard about the CMS inspection, I had no idea it had been triggered by her.

Throughout the fall of 2015 and into the winter of 2016, I tried to find out what the inspection had uncovered. After Yamamoto and Bennett completed their second visit in November, there were rumblings from former Theranos employees in contact with current ones that it hadn't gone well, but details were hard to come by. In late January, we were finally able to publish a story reporting that the CMS inspectors had found "serious" deficiencies at the Newark lab, citing sources familiar with the matter. How serious became clear a few days later when the agency released a letter it had sent the company saying they posed "immediate jeopardy to patient health and safety." The letter gave the company ten days to come up with a credible correction plan and warned that failing to come back into compliance quickly could cause the lab to lose its federal certification.

This was major. The overseer of clinical laboratories in the United States had not only confirmed that there were significant problems with Theranos's blood tests, it had deemed the problems grave enough to put patients in immediate danger. Suddenly, Heather King's written retraction demands, which had been arriving like clockwork after each story we published, stopped.

However, Theranos continued to minimize the seriousness of the situation. In a statement, it claimed to have already addressed many of the deficiencies and that the inspection findings didn't reflect the current state of the Newark lab. It also claimed that the problems were confined to the way the lab was run and had no bearing on the soundness of its proprietary technology. It was impossible to disprove these claims without access to the inspection report. CMS usually made such documents public a few weeks after sending them to the offending laboratory, but Theranos was invoking trade secrets to demand that it be kept confidential. Getting my hands on that report became essential.

I called a longtime source of mine in the federal government who had access to it. The most he was willing to do was read some passages over the phone. That was enough for us to report one of the inspection's most serious findings: the lab had continued to run a blood-clotting test for months despite repeated quality-control failures indicating that it was faulty. "Prothrombin time," as the test was known, was a dangerous test to get wrong because doctors relied on it to determine the dosage of blood-thinning medication they prescribed to patients at risk of strokes. Prescribing too much blood thinner could cause patients to bleed out, while prescribing too little could expose them to fatal clots.

Theranos couldn't refute our story, but it argued once more that its proprietary technology was not at issue. The prothrombin time test had been performed on regular venous samples using commercial equipment, it said. When its back was against the wall, the company was willing to admit to using conventional analyzers if doing so could help maintain the illusion that its own devices worked.

As the tug-of-war with Heather King over the inspection report dragged on, news surfaced that Holmes would be hosting a fund-raiser for Hillary Clinton's presidential campaign at Theranos's headquarters in Palo Alto. She had long cultivated a relationship with the Clintons, appearing at several Clinton Foundation events and forging a friendship with their daughter. The fund-raiser was later relocated to the home of a tech entrepreneur in San Francisco, but a photo from the event showed Holmes holding a microphone and speaking to the assembled guests with Chelsea Clinton at her side. With the election eight months away and Clinton considered the front-runner, it was a reminder of how politically connected Holmes was. Enough to make her regulatory problems go away? Anything seemed possible.

To try to force CMS to release the inspection report, I filed a Freedom of Information Act request for any and all documents connected to the Newark lab inspection and requested that it be expedited. But Heather King continued to urge the agency not to make the report public without extensive redactions, claiming that doing so would expose valuable trade secrets. It was the first time the owner of a laboratory under the threat of sanctions had demanded redactions to an inspection report, and CMS seemed unsure how to proceed. With each passing day, I became concerned that the full inspection findings would never be released.

More important, the inspection report showed, citing the lab's own data, that the Edisons produced wildly erratic results. During one month, they had failed quality-control checks nearly a third of the time. One of the blood tests run on the Edisons, a test to measure a hormone that affects testosterone levels, had failed quality control an astounding 87 percent of the time. Another test, to help detect prostate cancer, had failed 22 percent of its quality-control checks. In comparison runs using the same blood samples, the Edisons had produced results that differed from those of conventional machines by as much as 146 percent. And just as Tyler Shultz had contended, the devices couldn't reproduce their own results. An Edison test to measure vitamin B12 had a coefficient of variation that ranged from 34 to 48 percent, far exceeding the 2 or 3 percent common for the test at most labs.

I went back to my source and, this time, cajoled him into leaking the whole inspection report to me. Running 121 pages long, the document was as damning as one could expect. For one thing, it proved that Holmes had lied at the Journal's tech conference the previous fall: the proprietary devices Theranos had used in the lab were indeed called "Edison," and the report showed it had used them for only twelve of the 250 tests on its menu. Every other test had been run on commercial analyzers.

The coup de grâce came a few days later when we obtained a new letter CMS had sent to Theranos. It said the company had failed to correct forty-three of the forty-five deficiencies the inspectors had cited it for and threatened to ban Holmes from the blood-testing business for two years. As with the inspection report, Theranos was desperately trying to keep the letter from becoming public, but a new source had contacted me out of the blue and leaked it to me.

Heather King tried to prevent us from publishing the report, but it was too late. We posted it on the Journal's website and the accompanying story quoted a laboratory expert who said its findings suggested the Edisons' results were no better than guesswork.

When we reported news of the threatened ban, it was no longer possible for Holmes to downplay the gravity of the situation. She had to come out and say something, so she gave an interview to Maria Shriver on NBC's Today show in which she professed to be "devastated." But not enough, it seemed, to apologize to the patients she had put in harm's way. Watching her, I got the distinct impression that her display of contrition was an act. I still didn't sense any real remorse or empathy.

As for the lab itself, it was a mess: the company had allowed unqualified personnel to handle patient samples, it had stored blood at the wrong temperatures, it had let reagents expire, and it had failed to inform patients of flawed test results, among many other lapses.

In May 2016, I returned to the San Francisco Bay Area to try to find out what had happened to Tyler Shultz. It was almost exactly a year to the day since we'd met at the beer garden in Mountain View. Erika had told me Tyler was working on a research project with a nanotechnology professor at Stanford, so I drove my rental car to Palo Alto and searched for him in Stanford's School of Engineering. After asking around, I finally located him in a room in the materials science building.

After all, Theranos's employees, its investors, and its retail partner, Walgreens, had all learned of the inspection's findings and the threatened ban by reading the Wall Street Journal. If Holmes was sincere about making things right, why had she tried so hard to suppress their disclosure?

Tyler wasn't surprised to see me. Erika had given me his real email address and I'd written to him to let him know I was coming back through town. He'd been noncommittal about meeting with me. Now that I was there, though, he relented. We walked over to a nearby cafeteria to grab some lunch and slipped into easy banter.

Tyler seemed in good spirits. He told me he was part of a small group of researchers at Stanford that had teamed up with a Canadian company to compete in the multimillion-dollar Qualcomm Tricorder XPRIZE competition. They were trying to build a portable device capable of diagnosing a dozen diseases from a person's blood, saliva, and vital signs.

When our conversation turned to Theranos, his brows furrowed and he tensed up. He didn't want to discuss the subject in an open place within earshot of other people, he said. He suggested we walk back to the materials science building. We found an empty classroom there and sat down. His relaxed demeanor in the cafeteria had given way to palpable anxiety.

"My lawyers forbade me from talking to you, but I can't keep this bottled up anymore," he said.

For the next forty-five minutes, I listened dismayed as Tyler told me about the ambush at his grandfather's house and the months of legal threats he'd endured. Despite it all, he had never buckled. He had steadfastly refused to sign any document Boies Schiller had put to him. If not for his courage and the more than $400,000 his parents had spent on his attorneys, I might never have been able to get my first article published, I realized. I felt pangs of guilt for having put him through such an ordeal.

I agreed to keep whatever he was about to tell me off the record and only to write about it in the future if he gave me his permission to do so.

As we parted, Tyler thanked me for doggedly pursuing the story. He pointed out that Theranos had consumed the past four years of his life dating back to his summer internship at the company between his junior and senior years in college. I thanked him in turn for helping me get the story out and for withstanding the immense pressure under which he had been placed.

Tyler had heard through his parents that his grandfather continued to believe in the promise of Theranos. In a complete about-face after years of intense secrecy, Holmes was going to unveil the inner workings of her technology at the American Association for Clinical Chemistry's annual meeting on August 1, 2016. George believed that her presentation would silence the doubters. Tyler didn't understand why he couldn't see through her lies. What would it take for him to finally accept the truth?

The most heartbreaking part of it all was Tyler's estrangement from his grandfather. George Shultz had continued to side with Holmes in spite of everything my reporting had revealed. He and Tyler hadn't seen each other for the better part of a year and they communicated only through lawyers. The previous December, the Shultzes had hosted a party at a penthouse apartment they owned in San Francisco to celebrate George's ninety-fifth birthday. Holmes had attended, but not Tyler.

I now suspected Theranos had had both of us under continuous surveillance for a year. And, more than likely, Erika Cheung and Alan Beam too.

Holmes had told Maria Shriver on the Today show that she took responsibility for the Newark lab's failings, but it was Balwani who suffered the consequences. Rather than take the fall herself, she sacrificed her boyfriend. She broke up with him and fired him. In a press release, Theranos dressed up his departure as a voluntary retirement.

Not long afterward, Theranos contacted Tyler's lawyers and told them it knew about our meeting. Since neither of us had told a soul about it, we deduced that Holmes was having one or both of us followed. Fortunately, Tyler didn't seem too worried about it. "Next time maybe I'll take a selfie with you and send it her way to save her the trouble of hiring PIs," he quipped in an email.

A week later, we reported that Theranos had voided tens of thousands of blood-test results, including two years' worth of Edison tests, in an effort to come back into compliance and avoid the CMS ban. In other words, it had effectively admitted to the agency that not a single one of the blood tests run on its proprietary devices could be relied upon. Once again, Holmes had hoped to keep the voided tests secret, but I found out about them from my new source, the one who had leaked to me CMS's letter threatening to ban Holmes from the lab industry. In Chicago, executives at Walgreens were astonished to learn of the scale of the test voidings. The pharmacy chain had been trying to get answers from Theranos about the impact on its customers for months. On June 12, 2016, it terminated the companies' partnership and shut down all the wellness centers located in its stores.

On a muggy summer day at the beginning of August, more than 2,500 people crowded into the grand ballroom of the Pennsylvania Convention Center in Philadelphia. Most were laboratory scientists who had come to hear Holmes speak at the annual meeting of the AACC. "Sympathy for the Devil" by the Rolling Stones was playing on the public announcement system, a choice of music that didn't seem like a coincidence.

In another crippling blow, CMS followed through on its threat to ban Holmes and her company from the lab business in early July. More ominously, Theranos was now the subject of a criminal investigation by the U. S. Attorney's Office in San Francisco and of a parallel civil probe by the Securities and Exchange Commission. In spite of all these setbacks, Holmes felt she still had one card to play to turn public opinion around: wow the world with a display of her technology.

The invitation the association had extended to Holmes was highly controversial among its members. Some had argued forcefully that it should be withdrawn given the events of recent months. But the association's leadership had seen a chance to generate publicity and buzz for the normally staid scientific conference. It was proven right in that regard: several dozen journalists had made the trek to Philadelphia to watch the spectacle.

After a few introductory remarks from AACC president Patricia Jones, Holmes stepped up to the lectern. She wore a white blouse under a dark jacket. Gone was the black turtleneck, which had become an object of ridicule since the previous fall.

What followed wasn't so much a scientific presentation as it was a new product exhibit. Over the next hour, Holmes proceeded to unveil the machine that had been but a malfunctioning prototype when Theranos had gone live with its blood tests nearly three years earlier: the miniLab. Theranos's engineers and chemists had improved the device since that early model, but the company still hadn't conducted a full clinical study to prove that it worked reliably across a wide range of assays using blood pricked from fingertips. While Holmes's presentation included some data, much of it involved venous blood drawn from the arm. The little finger-stick data it contained covered only eleven blood tests and wasn't independently verified or peer-reviewed.

CMS had just banned Holmes from running clinical laboratories, but no matter: she explained that the miniLab connected wirelessly to servers at Theranos headquarters and could be deployed directly in patients' homes, doctors' offices, or hospitals, doing away with the need for a central lab.

In effect, she was pirouetting back to her original vision of portable blood-testing machines operated remotely through Wi-Fi or cellular networks. Of course, after everything that had happened, commercializing such a system without the FDA's approval was out of the question. And putting together the thorough studies the agency would want to see would take years. That was why she had tried to bypass the FDA in the first place.

The odds that Holmes could pull off this latest Houdini act while under criminal investigation were very long, but watching her confidently walk the audience through her sleek slide show helped crystallize for me how she'd gotten this far: she was an amazing saleswoman. She never once stumbled or lost her train of thought. She wielded both engineering and laboratory lingo effortlessly and she showed seemingly heartfelt emotion when she spoke of sparing babies in the NICU from blood transfusions. Like her idol Steve Jobs, she emitted a reality distortion field that forced people to momentarily suspend disbelief.

The spell was broken, however, during the question-and-answer session when Stephen Master, an associate professor of pathology at Weill Cornell Medical Center in New York and one of three panelists invited onstage to ask Holmes questions, pointed out that the miniLab's capabilities fell far short of the original claims she had made. His comment drew a burst of applause from the audience. Reverting to the chastened persona from her Today show interview, Holmes acknowledged that Theranos had a lot of work to do, as she put it, to "engage" with the laboratory community. But she stopped short again of apologizing or admitting fault.

When Dennis Lo, a pathology professor at the Chinese University of Hong Kong, later asked her how the miniLab differed from the technology the company had used in its laboratory on patient samples, she dodged the question. It was a giant issue to sidestep, yet the hundreds of assembled pathologists remained civil and respectful despite her evasiveness. There were no boos or catcalls. The decorum broke down only briefly when Holmes turned to leave the stage at the end of the Q&A. "You hurt people," a voice yelled out from the dispersing crowd.

One of the miniLab tests Holmes had showcased at the conference was for Zika, the mosquito-borne virus that had damaged the brains of thousands of newborns around the world. Theranos had applied to the FDA for emergency-use authorization of the test, billing it as the first finger-stick blood test of its kind. But in another embarrassing setback, FDA inspectors soon discovered that the company had failed to include basic patient safeguards in its study, forcing it to withdraw the application.

In an interview with the Financial Times, Geoffrey Baird, a professor of pathology at the University of Washington, said Holmes's presentation had included "a comically small amount of data" and had "the feel of someone putting together a last-minute term paper late at night." Other lab experts were quick to note that none of the miniLab's various components were novel. All Theranos had done was make them smaller and pack them into one box, they said.

If Holmes had hoped to rehabilitate her image and change the media narrative by unveiling the miniLab, that hope was dashed by the flurry of critical articles published in the wake of the event. A headline in Wired captured the reaction best: "Theranos Had a Chance to Clear Its Name. Instead, It Tried to Pivot."

Most of the other investors opted against litigation, settling instead for a grant of extra shares in exchange for a promise not to sue. One notable exception was Rupert Murdoch. The media mogul sold his stock back to Theranos for one dollar so he could claim a big tax write-off on his other earnings. With a fortune estimated at $12 billion, Murdoch could afford to lose more than $100 million on a bad investment.

The possibility that Holmes might pull a rabbit out of her proverbial hat at the AACC meeting had kept Theranos's restless investors from launching a mutiny. After her appearance was panned and the Zika fiasco made headlines, one of them decided it had had enough: Partner Fund, the San Francisco hedge fund that had invested close to $100 million in the company in early 2014, sued Holmes, Balwani, and the company in Delaware's Court of Chancery, alleging that they had deceived it with "a series of lies, material misstatements, and omissions." Another set of investors led by the retired banker Robert Colman filed a separate lawsuit in federal court in San Francisco. It also alleged securities fraud and sought class-action status.

After initially attempting to appeal the CMS ban, Holmes resigned herself to the inevitable and closed the California lab as well as the company's second lab in Arizona, which had used only commercial analyzers. During an inspection of the Arizona facility days before it was shuttered, CMS found a multitude of problems there too.

David Boies and his law firm, Boies, Schiller & Flexner, stopped doing legal work for Theranos after falling out with Holmes over how to handle the federal investigations. Another big law firm, WilmerHale, took their place. A month after Holmes's AACC appearance, Heather King returned to Boies Schiller as a partner based in its Palo Alto office. Boies left the Theranos board a few months later.

Walgreens, which had sunk a total of $140 million into Theranos, filed its own lawsuit against the company, accusing it of failing to meet the "most basic quality standards and legal requirements" of the companies' contract. "The fundamental premise of the parties" contract -- like any endeavor involving human health -- was to help people, and not to harm them," the drugstore chain wrote in its complaint.

One thing is certain: the chances that people would have died from missed diagnoses or wrong medical treatments would have risen exponentially if the company had expanded its blood-testing services to Walgreens's 8,134 other U. S. stores as it was on the cusp of doing when Pathology Blawg's Adam Clapper reached out to me.

The number of test results Theranos voided or corrected in California and Arizona eventually reached nearly 1 million. The harm done to patients from all those faulty tests is hard to determine. Ten patients have filed lawsuits alleging consumer fraud and medical battery. One of them alleges that Theranos's blood tests failed to detect his heart disease, leading him to suffer a preventable heart attack. The suits have been consolidated into a putative class action in federal court in Arizona. Whether the plaintiffs are able to prove injury in court remains to be seen.

Under a settlement with Arizona's attorney general, Theranos subsequently agreed to pay $4.65 million into a state fund that reimbursed the 76,217 Arizonans who ordered blood tests from the company.